Health Policy and Planning

Wireko,, Ishmael;Béland,, Daniel;Kpessa-Whyte,, Michael

doi: 10.1093/heapol/czaa080pmid: 32989440

Abstract Contributing to the ongoing debate about policy feedback in comparative public policy research, this article examines the evolution of healthcare financing policy in Ghana. More specifically, this article investigates the shift in healthcare financing from full cost recovery, known as ‘cash-and-carry’, to a nation-wide public health insurance policy called the National Health Insurance Scheme (NHIS). It argues that unintended, self-undermining feedback effects from the existing health policy constrained the menu of options available to reformers, while simultaneously opening a window of opportunity for transformative policy change. The study advances the current public policy scholarship by showing how the interaction between policy feedbacks and other factors—particularly ideas and electoral pressures—can bring about path-departing policy change. Given the dearth of scholarship on self-undermining policy feedback effects in the Global South, this contribution’s originality lies in its application of the novel theory to the sub-Saharan African context. Health care, health insurance, policy feedback, policy change, National Health Insurance Scheme, Ghana Key Messages First, we argue that, in Ghana, the unintended and negative consequences of the existing health arrangements, which we define as ‘self-undermining policy feedbacks’, shaped the advent of a nation-wide public health insurance policy, the National Health Insurance Scheme (NHIS). Second, and more generally, we show how policy feedbacks can interact with other factors, particularly ideas and electoral pressures to bring about path-departing policy change. Introduction The study of public policy has received far less attention in African countries than other jurisdictions. However, drawing insights from public policy research can shed light on the complex dynamics of policy stability and change on the continent. For instance, the growing interest in, and debate about, the mediating role of existing policy legacies in either reinforcing or disrupting institutional policy pathways have been instructive (e.g. Pierson, 1993; Campbell, 2003; Mettler and Soss, 2004; Béland, 2010; Weaver, 2010; Mettler and SoRelle, 2014; Jacobs and Weaver, 2015; Béland and Schlager, 2019). Although countries in sub-Saharan African have been instituting major policy reforms that are, for the most part, materially and ideationally inspired by transnational policy actors (Walt et al., 2008; Kpessa and Béland, 2012), little research has situated these regional developments within the broader framework of policy feedback research. This article draws on insights from the policy feedback literature in order to explain the shift in Ghanaian healthcare financing from full cost recovery, known as ‘cash-and-carry’, to a nation-wide public health insurance policy called the National Health Insurance Scheme (NHIS). In so doing, this article shows that the cash-and-carry policy unintentionally undercut itself by creating self-undermining feedbacks: costly and unaffordable healthcare services, reduced healthcare access, institutional corruption, negligible user fee exemptions for the poor and marginalized, and overt abuse of these vulnerable populations in healthcare facilities. These self-undermining feedbacks not only opened a window of opportunity for policy change but also generated an alternative feedback effect from below. Indeed, communities began organizing their own health-financing arrangements—referred to as Community-Based Health Insurance Schemes (CBHISs)—as a means of circumventing the increasingly ineffective cash-and-carry policy. Over time, the fame achieved by these CBHISs convinced political actors of different ideological dispositions of the need to roll out similar health-financing schemes across the country. It was within this context of interacting feedback effects that the NHIS was established in 2003. This study advances the self-undermining policy feedback literature by incorporating a case from the Global South. In analysing evidence from sub-Saharan Africa, this case study empirically demonstrates how feedback effects between self-undermining policies and other factors give rise to policy change. To understand this phenomenon in all of its complexity, one must first be introduced to the formal theory behind self-undermining feedbacks. Therefore, the next section begins with an overview of the methods and the analytical framework that guide this article’s analysis. The overview is then followed by the Results section, which explores the contextual and historical variables pertaining to Ghanaian health care. Afterwards, the Discussion section uses insights from the policy feedback literature in order to explain the shift from cash-and-carry to national health insurance. To conclude, the article summarizes and reflects on the key dimensions of self-undermining feedback effects that create opportunities for policy reforms. Methods and analytical framework The article employs a qualitative case study method, focusing on the transition to NHIS in Ghana. It examines the self-undermining factors that shaped the policy change by drawing on insights from the policy feedback literature. Given the historical nature of the phenomenon under investigation, the article combines data from the existing literature with semi-structured, in-depth interviews for triangulation. We conducted the interviews in Ghana between February and July 2013, using an interview guide. The interview guide was designed based on what the theoretical literature has identified as the key self-undermining factors that shape health policy change. Some of these factors include problem pressure, interest mobilization, the generation of new policy alternatives and the role of political actors in seizing opportunity windows to pursue policy change. An interview consent form for each participant was designed, and both the interview guide and the interview consent form were later approved by a university-based research ethics committee. In all, 22 major actors involved at various stages of the NHIS development process were selected for the interviews, using purposive and snowball sampling methods (see Table 1). Each interview lasted between 30 and 60 min and was taped before being transcribed and uploaded into NVivo 11 software for coding and analysis. Two participants chose a pseudonym to protect their confidentiality, while the remaining participants opted to have their real names mentioned in the research report. Table 1 Summary of interviewees Actor . Stage . Number . Ministry of Health Agenda setting and formulation 5 Ghana Health Service Formulation 3 Ministry of Finance Formulation 1 Parliament Legislation 4 National Health Insurance Authority Implementation 2 Civil Society Organizations Various stages 3 International Actors Various stages 4 Total 22 Actor . Stage . Number . Ministry of Health Agenda setting and formulation 5 Ghana Health Service Formulation 3 Ministry of Finance Formulation 1 Parliament Legislation 4 National Health Insurance Authority Implementation 2 Civil Society Organizations Various stages 3 International Actors Various stages 4 Total 22 Open in new tab Table 1 Summary of interviewees Actor . Stage . Number . Ministry of Health Agenda setting and formulation 5 Ghana Health Service Formulation 3 Ministry of Finance Formulation 1 Parliament Legislation 4 National Health Insurance Authority Implementation 2 Civil Society Organizations Various stages 3 International Actors Various stages 4 Total 22 Actor . Stage . Number . Ministry of Health Agenda setting and formulation 5 Ghana Health Service Formulation 3 Ministry of Finance Formulation 1 Parliament Legislation 4 National Health Insurance Authority Implementation 2 Civil Society Organizations Various stages 3 International Actors Various stages 4 Total 22 Open in new tab A significant body of the data used in this article’s analysis is drawn from the existing literature on cross-temporal health politics and policies in Ghana. Through the lens of feedback and policy legacy perspectives, our analytical framework retrospectively examined policies over time in relation to the already available scientific data. This method allowed us to situate causal claims in their appropriate historical context. Not only were prior analyses of the existing material on health-financing politics in Ghana instrumental in providing us with the needed perspectives to frame the questions that foregrounded our interviews, but also prior analyses provided a framework for validating claims and assertions that were articulated by the interviewed actors. Secondary literature sources included published articles in scientific journals, new media sources, government documents and literature obtained from various political parties. Policy feedback perspectives that specifically explore the enabling and constraining roles of existing public policies on future action informed the analytical framework. The analytical framework, in turn, was used in order to process the data collected from the interviews and secondary sources. To grasp the analytical framework in more detail, one must first understand what exactly the policy feedback perspective is about. Simply put, the policy feedback perspective seeks to explain how existing policies shape the ideas and behaviour of political actors and mass publics over time (e.g. Skocpol, 1992; Pierson, 1993; Campbell, 2003; Mettler and Soss, 2004; Béland, 2010; Weaver, 2010; Mettler and SoRelle, 2014; Jacobs and Weaver, 2015; Béland and Schlager, 2019). For example, scholars have investigated how existing public policies generate strong constituencies of citizens and interest groups that are likely to mobilize in order to preserve said policies (Pierson, 1993; Campbell, 2003; for a critical overview, see Béland, 2010). This example identifies self-reinforcing feedback effects that are conducive to policy stability and even path dependence. In his work on welfare state reform during the Reagan and Thatcher eras in the USA and the UK, Pierson (1994) perfectly illustrates the potential link between self-reinforcing feedback effects and path dependency. In particular, he notes how the development of entrenched constituencies composed of supportive interest groups and voters exacerbated the fear of adverse electoral consequences, thereby leading Conservative governments in both nations to back off from initial plans to pursue fundamental welfare state restructuring during the 1980s. This discussion about self-reinforcing feedback effect and its relationship with path dependence points to a stability bias in the early research on policy stability and change, which tended to overemphasize stability over change, especially transformative change (Weaver, 2010; Jacobs and Weaver, 2015). Over the past 15 years, however, a growing number of scholars have explored endogenous and gradual forms of transformative institutional change, moving away from the stability bias of previous scholarship (Campbell, 2004; Hacker, 2004; Thelen, 2004). This literature has led a number of scholars to revisit the concept of policy feedback and directly criticize the stability bias of scholars who tend to emphasize self-reinforcing mechanisms at the expense of what they describe as self-undermining mechanisms. Scholars engaging in this new approach describe these self-undermining mechanisms as being widespread and crucial for explaining transformative policy change (Weaver, 2010; Jacobs and Weaver, 2015; Béland et al., 2019). For example, the recent history of accelerated population ageing across advanced industrial countries has shown that excessive costs of expansive pension programmes generate self-undermining feedback effects. These feedback effects may generate the political momentum necessary for path-departing change, as seen in Sweden’s 1998 pension reform that significantly transformed pension policies (Weaver, 2010). What is particularly helpful about works such as Weaver (2010), Jacobs and Weaver (2015) and Oberlander and Weaver (2015) is that they draw attention to self-undermining feedback effects as a potential source of transformative policy change. Indeed, these works identify the factors that may facilitate self-undermining feedback effects (Jacobs and Weaver, 2015); e.g. these works show that, over time, certain demographic and economic conditions make specific public policies either costly for ordinary citizens or less fiscally sustainable for the state. Moreover, they explain that another factor includes the ‘mechanisms through which policies can undermine their own support base in the mass public by creating conditions under which voters can be persuaded of the need for or acceptability of major policy change’ (Jacobs and Weaver, 2015, p. 447). Finally, they identify that problems generated by self-undermining policies are likely to spur policy alternatives, which key political actors can use to replace the existing policies that have become prohibitively costly and less popular over time (Jacobs and Weaver, 2015, p. 449). This means that, as Jacobs and Weaver (2015) point out, if self-undermining policy feedback is to lead to transformative policy change, it should trigger the involvement of key political actors who can bring about such change by exploiting what Kingdon (1984) calls ‘policy windows’. It is worth noting that self-undermining feedback as a source of transformative policy change is not necessarily a deterministic process. Rather, it is a political reality shaped, in part, by the agency and strategy of political actors. These factors are illustrated in the following discussion of Ghanaian healthcare politics. Results Cash-and-Carry health-financing policy After independence in 1957, Ghana introduced a single-payer model of health care that was financed through general tax revenue called the National Health Service (NHS) system. At the time, the system’s goal was to make healthcare access a universal right. Thus, health care was provided free of charge at the point of use to everybody, including foreigners residing in the country (Wireko and Béland, 2013). In the face of the economic crisis of the 1970s and 1980s, however, the country’s economic resources plummeted, making it impossible to sustain the NHS system. Therefore, beginning in the 1970s, gradual reforms introducing individual cost bearing responsibility were initiated. By the early 1980s, these reforms had grown into general healthcare financing reforms, aiming to generate the revenue needed to support efficient healthcare delivery. At the centre of these reforms was the introduction of healthcare user fees payable at the point of service use, popularly referred to as the cash-and-carry healthcare financing policy. According to Asenso-Okyere, Ghana’s cash-and-carry health policy was ‘the most comprehensive cost recovery legislation in West Africa’ (1995, p. 87). Indeed, it covered a wide array of healthcare services: outpatient consultations; drugs and pharmaceuticals; laboratory, X-ray and other examinations; medical, dental and surgical services; and even hospital catering and accommodations (Waddington and Enyimayew, 1989, 1990; Coleman, 1997). However, the government did attempt to protect the average Ghanaian from potentially catastrophic expenditures associated with the healthcare user fees. To this end, the government ensured that the imposed fees were graduated in favour of rural dwellers vis-à-vis urban dwellers. The government also ensured that user fee exemptions were introduced to cater to the poor and marginalized, who would be unable to afford the standard user fees (Adibo, 2013). Initially, health service providers were authorized to deposit the revenue accruing from user fee charges in the national treasury. However, by 1992, health facilities had been advised to retain the entirety of their user fees revenue to ensure they had readily accessible income for service delivery improvements (Adjei, 2013). According to some interviewed actors, introducing user fees was necessary at the time because government subventions to the health sector had significantly declined (Adibo, 2013; Akor, 2013). This decline had such a detrimental impact on healthcare delivery that ‘providers could not do their work because of shortages of basic tools’ (Agyepong, 2013). Policymakers believed that the additional revenue collected through the cash-and-carry policy would help to restore service provision to the standard it had been before the economic crisis (Adibo, 2013). Thus, at the time of its introduction, the cash-and-carry policy attracted significant support from various stakeholders—especially the frontline service providers who were already ‘pushing that if government could not pay for those services, they should be allowed to charge user fees’ to ensure quality healthcare delivery (Agyepong, 2013). As anticipated, the implementation of the cash-and-carry policy brought an initial improvement in the fiscal position of the healthcare system. The improvement was reflected in service providers having an enhanced capacity to generate additional revenue for supporting the system. For instance, as noted by an officer in the Ministry of Health during the interviews, ‘the cash and carry system was able to help us hold the drug revolving fund, and we had so much money in the bank that we were not dependent on government to give us money anymore to order drugs from abroad’ (Adjei, 2013). Awittey (2013) made a similar observation, noting that ‘it was a plus for the facilities in terms of having money and making their own decisions as to what to buy and what not to buy’ because ‘they didn’t have to wait for government to give them a subvention before payments were made, or even for budgets to go through before drugs and other supplies were purchased’. However, in a country with high levels of poverty, it soon became obvious that the Ghanaian cash-and-carry policy made health care unaffordable for a large majority of the population. The following section shows that the citizenry’s reaction to unaffordable health care manifested in ways that undermined the newly adopted cash-and-carry policy (Akor, 2013). Self-undermining feedback of the cash-and-carry policy Besides improving the fiscal position of the health system, the cash-and-carry policy ultimately generated unforeseen, self-undermining feedback effects for major political players, including its original proponents. First, the policy reduced the citizenry’s access to health care. As some of the officers in the Ministry of Health at the time noted during the interviews: Between 1985 and 1989, when we were looking at the trend, we noticed that there has [sic] been a massive reduction in attendance of out-patients. In fact, the evidence showed that attendance had dropped, from every person attending hospital once in a year on average to about a person attending a hospital once in three years instead of once in a year (Adjei, 2013). A similar account was given by another officer who specifically noted that the indicator, per capita utilization of health services, which is simply the number of time [sic] the population is likely to use health services was low, it was at 0.3 or 30% so the population was actually using services which was way below the globally accepted norm of at least one. That was very, very bad. So, we knew that there was something wrong (Seddoh, 2013). Healthcare access was reduced because the cash-and-carry policy made the healthcare system costly and, indeed, ‘unaffordable to the public, especially the poor [emphasis added]’ (Awittey, 2013). While some user fees exemptions were introduced to protect the poor and vulnerable from potential catastrophic expenditure, the exemption system could not be implemented effectively. That is to say, the fee exemptions ‘never worked out’ (Seddoh, 2013). Data from the interviews point to the fact that the implementation of said exemptions was undermined by several challenges. Firstly, the implementation guidelines neither specified exactly what should be exempted for each exemption category nor identified who would pay for those exemptions if they were given. As Adams (2013) reported during the interviews, ‘if children under five are to be exempted, for what? Everything?’ Adams went on to say, ‘we don’t have any business giving free services; somebody should have to buy the service for the poor’. Moreover, per the guidelines, any exemption that is granted under the scheme had to be sanctioned by staff in the top echelon of the health system instead of those who directly provide the service (Seddoh, 2013). That requirement bureaucratized the exemption process, deterring most providers and potential beneficiaries alike from taking advantage of the exemption scheme (Agyepong, 2013; Awittey, 2013). Secondly, there was also the problem of identification. As Adams (2013) lamented during the interview, ‘we are not a social welfare institution; we are a service-oriented institution’, adding that ‘if you ask us to provide the exemption, we were hopeless at that, we couldn’t do it, even though we were trying to implement it’. Because professionals who could properly identify the potential exempts were not incorporated in the implementation process, providers were generally helpless at implementing the exemption scheme. The situation continued to deteriorate after 1992, once the service providers were authorized to retain all their user fees revenue as a means of building revolving funds. According to Adjei (2013), that decision made providers ‘so concerned about recovering their capital that even if you were on the exemption list, they did not exempt you’. To make matters worse, ‘fees of varying shades of legality’ were also introduced across the healthcare system, as facilities were granted permission to determine their own fees (Agyepong, 1999, p. 60). For instance, ‘>70% of all institutions at all levels charged for gauze, gloves, needles, and syringes’, although fees for such services were supposed to be waived for patients (Coleman, 1997, p. 19). Moreover, indigents were not supposed to pay for services, but ‘fee collection efforts [were] insistent’ on them (Kraus, 1991, p. 143). Nyonator and Kutzin (1999) classified those fees into three (3) main types: (1) the cash-and-carry programme for drugs, (2) locally authorized charges instituted by the various health facilities on their own and (3) illegal or unauthorized charges made by staff of the facilities for various services. These unauthorized charges included commissions for selling drugs and other consumables, as well as commissions for the disposal of placentas, among other things. After the introduction of the multiple fee types, most service providers charged de facto fees that went far beyond those stipulated by the de jure user fees policies, which were already unaffordable. By consequence, this fee system became overly exorbitant for most patients. By 1996, adult outpatient consultation in many of the district health facilities in the Brong Ahafo region costed about ¢350 (Ghana’s currency) instead of the ¢50 recognized by the policy; X-ray examinations had risen from ¢200 to ¢2000; caesarean deliveries had risen from ¢1000 to ¢55000; and hernia treatment had increased from ¢500 to ¢28300. For major services such as surgeries, the average charge had risen to about 5000% above the de jure price (Coleman, 1997). Nyonator and Kutzin (1999) observed a similar situation in many of the country’s deprived areas. The situation was particularly dire in the Volta region, where the costs of outpatient consultation had shot up, on average, by over 700% irrespective of the level and place of care. By the time, the NHIS was being implemented, the situation had reached a ‘catastrophic’ level: It was estimated that 11.0% of households in Ghana spent over 5% of their total household expenditure on health care OOP [user fees]. Also 10.7% of households spent more than 10% of their non-food consumption expenditure on OOP health care payments. Furthermore, about 2.6% of households are observed to have spent in excess of 20% of their total household income on health care OOP (Akazili et al., 2017, p. 1). The adverse effects of the cash-and-carry policy had also spread from the poor, the middle class and the higher-income brackets to now include politicians who needed to support the poorer members of their communities, who were struggling to survive the catastrophic health expenditures (Andoh, 2013). Despite the high user fees that were charged by the various health facilities, service providers remained generally unable to meet their target of generating 15–20% of their recurrent expenditure through user fees (Akanzige, 2013; Amoh, 2013). Several reasons accounted for this failure, including the reduction in hospital attendance and malpractice by some staff members and customers (Akorsah, 2013). For instance, many staff members pocketed fees that were collected from patients’ consultation cards before they were allowed to see their doctors (Seddoh, 2013). Moreover, there were frequent reports of health administrators under-invoicing and under-issuing receipts. As Nyonator and Kutzin (1999, p. 334) noted in the case of the Volta region: If you pay ¢14 000, you are given receipt covering ¢8000… In government hospitals, only 39% of those who paid registration fees and 41% of those who paid for laboratory services were given receipts compared with 93% of those who paid for drugs. The same trend was observed in health centres, i.e. 28, 47 and 92% respectively. Patients who needed to bypass the usually long queues in the facilities were also asked to make side-payments, yet those monies would not be accounted for by staff (Kuntulo, 2013; Nortey, 2013).1 Unfortunately, instead of having a one-stop-shop for service payment, most of the facilities had multiple payment points, increasing the frequency and the overall instances of bribery within the system (Nyonator and Kutzin, 1999). Furthermore, there were irregular cases of patients absconding without paying their bills. Such reports pertained mostly to emergencies, where healthcare providers were required to request payment after service delivery. After receiving treatment, many patients would run away because they could not afford such high bills. Many of the health facilities adopted mitigation strategies, including detaining patients after recovery and discharge until payments were made. As Akor (2013) noted during the interviews, ‘pregnant women would go and deliver at Kole Bu hospital and because they couldn’t pay they were detained and the media was on government at that time showing patients that have been detained at the ward because they couldn’t afford to pay their bills’. In general, by the time that many of those patients had gathered enough money to pay their healthcare bills, the amount would have increased even further providing them with an additional impetus to bolt away without paying their bills (Singleton, 2006; Wireko, 2016). As Asenso-Okyere (1995) noted, in such circumstances, most facilities could barely raise even 10% of their recurrent expenditure through user fees—a situation that became worrying not just for the facilities but also for the Ministry of Health and the government, which had believed that the cash-and-carry policy would address the financial woes of the health sector. On a general level, the root of these challenges lies in the weak capacity of the state bureaucracy to effectively implement its devised policies. On a more fine-grained level, however, specific institutional factors directly challenged the policies. As the section below illustrates, cash-and-carry’s adverse impacts not only generated a heavy dossier of self-undermining feedback effects against the policy but also opened a window of opportunity for key policy actors to pursue a shift from cash-and-carry towards a nation-wide health insurance programme that would be financed through premium payments, payroll deductions and tax revenues. Simply put, the section below shows that the specific choice of health insurance as a policy alternative and the political processes that foregrounded said reform were shaped by the self-undermining feedbacks associated with the cash-and-carry policy. Policy feedback creating a policy window for change The political movement to find a replacement for the problematic cash-and-carry policy began under the Provisional National Defence Council (PNDC) military government in the late 1980s. Data collected from our interviews revealed that the PNDC government recruited a health insurance consultant and approved a feasibility report that recommended the rollout of an NHIS in 1988. The aim of this rollout was to ‘limit the burden of paying for hospital fees by the ordinary Ghanaian’, however, the report went no further than this stage of the policy development process (Adibo, 2013). When, in 1992, politics was liberalized and democratic rule was re-established in Ghana, the introduction of health insurance to counter the adverse effects of the cash-and-carry policy became a huge political issue (Afriyie, 2013). For example, in their manifesto for the 1992 elections, the National Democratic Congress (NDC)—a party formed out of the PNDC military regime—indicated that it would continue the ‘preliminary work’ that had ‘already been done on the design of a National Health Insurance Scheme’ and that the NDC would ‘pursue this as a matter of urgency, with the view to putting such a scheme in place at the earliest opportunity’ (NDC, 1992). The main opposition party at the time, the New Patriotic Party (NPP), made a similar promise to ensure efficient management of the healthcare system and the introduction of health insurance as a means of making health care affordable (NPP, 1992). As the elections were approaching, the NDC formed an alliance with two other social democratic parties (National Convention Party and Every Ghanaian Living Everywhere). That coalition was called the Progressive Alliance and was led by the NDC candidate and incumbent head of state Flight Lieutenant (rtd) Jerry John Rawlings. Under the leadership of Rawlings, the Progressive Alliance won the 3 November 1992 presidential election. However, the opposition parties boycotted the subsequent parliamentary election on 22 December 1992, claiming that the earlier presidential election’s results reflected a stolen verdict. The boycott culminated in the NDC winning virtually all the 200 parliamentary seats (Gyimah-Boadi, 1994). With a monopoly on legislative initiatives, the NDC-led Progressive Alliance continued the PNDC administration’s NHIS project, as promised, by commissioning additional feasibility studies, research projects, and consultancy services. Each of these initiatives aimed to materialize the proposed healthcare insurance policy. For example, based on one feasibility study’s recommendation, the government decided in 1995 to establish an independent company that would implement their proposed health insurance policy, focusing on contributors to the Social Security and National Insurance Trust (SSNIT) and cocoa farmers. According to the then-Director of Medical Services who later became the deputy health minister, the Ministry of Health took some preliminary steps that were meant to implement the NDC’s proposed health insurance policy. Indeed, the Director of Medical Services began negotiations with, the chairman of Cocoa Board [i.e. Ghana Cocoa Board] at the time and his assistant, proposing that Cocoa Board pay the health insurance premium on behalf of the farmers… The idea was that Cocoa Board would pay in foreign exchange from the proceeds of the cocoa it purchased from the farmers and sold on the international market… If Cocoa Board was able to pay the premium in lump sum for all cocoa farmers, then part of the money would straight away be used in buying the equipment and other resources needed to run the [health insurance] system (Adibo, 2013). However, it was later realized, in the words of the former Director of Medical Services, that health insurance policy was not going to work [effectively] in Ghana because it was only about 5% of the population that had registered with SSNIT and that was mainly the formal sector. So, if we were going to rely on those in the formal sector to design a health insurance, then it meant that we were going to exclude about 95% of the population. We realized that such health insurance scheme was going to be discriminatory (Adibo, 2013). All this to say, the health insurance policy was finally discontinued. Around the same time, however, various religious organizations and community associations, which were predominantly based in the country’s poor and rural areas, began exploring alternatives to the cash-and-carry policy that would enable their members to ‘pay their medical bills’ (Ofori, 2013). During its exploits in 1992, the St. Theresa’s Catholic Hospital in the town of Nkoranza in the Brong Ahafo Region introduced a pro-poor health insurance arrangement, known as Nkoranza CBHIS. The Nkoranza CBHIS allowed community members to pool resources together—similar to what was being done in many localities to cater for emergency situations such as funerals—so that community members would not be financially overburdened when members were admitted in the hospital (Ofori, 2013). Having begun in Nkoranza in 1992, this programme eventually spread to other regions like Damango, Nkwanta and Drobo (Akor, 2013). While the emergence of these CBHIS initiatives was a form of local resistance against national government policy, it ultimately shaped the path chosen for reforming the cash-and-carry policy. According to the former Director of Medical Services, the emergence of those CBHISs informed the Ministry of Health that ‘social support systems like susu, where a group of people collected monies on a monthly basis … to support things like weddings, outdoorings, funerals and other social events, could also be turned into sickness funds or … a unique health insurance system’ (Adibo, 2013). In light of this information, the Ministry of Health ‘instructed every region [in Ghana] to collect data on the type of social support systems they had in their areas’ (Adibo, 2013). Additionally, ‘a feasibility study to experiment [with] how the insurance scheme could be implemented was initiated in Dodowa, then Dangme West District … and they were to be called Mutual Health Organizations’ (Adibo, 2013). However, the guiding principle of the scheme, as noted by the former Director of Medical Services, was that ‘we should go slow and learn as we go along so that we don’t repeat mistakes’ (Adibo, 2013). While this experimentation enabled the ministry to achieve its intended goal, the cautious approach, among other factors, critically delayed the Dodowa health insurance project. Indeed, the scheme was still at the exploratory stage by the time of the next election in 1996. Prior to the 1996 general elections, the NPP elected a new flag-bearer who also ‘made it one of his campaign promises that if ever the Ghanaian people voted for him, he would make sure that health insurance was implemented’ (Afriyie, 2013). Thus, in their manifesto for the 1996 elections, the NPP criticized the incumbent NDC administration for its inability to replace the cash-and-carry financing system, arguing that the policy of ‘cost recovery at the point of service has proven notoriously callous and inhumane’ (NPP, 1996, p. 37). To overcome this institutional failure, the NPP promised that the system of financing healthcare services would be thoroughly overhauled and that the cash-and-carry system would be ‘reviewed with a view for evolving a more equitable system including health insurance and other repayment schemes’ (NPP, 1996, p. 37). In response to these policy objectives, the NDC indicated that their ‘feasibility studies on the proposed health insurance scheme have been completed’ and that ‘the scheme will be tested on a pilot basis’ (NDC, 1996). They said that, ‘if successful, it will be introduced on a national scale as part of the NDC’s health policy’ (NDC, 1996). Simply put, these parties were competing for the authority to realize the future of Ghanaian healthcare policy. Following the NDC’s victory in the 1996 general elections and given increased pressure on political leaders to replace the policy with a different financing arrangement, President Jerry John Rawlings announced in the 1997 State of the Nation address that ‘a pilot Insurance Scheme will be implemented in the Eastern region to test the work done so far’ and that ‘its performance will be studied as well as the existing rural health Insurance Schemes … so that problems can be identified and eliminated before the implementation can be done on National scale’ (Rawlings, 1997). This time, with the active involvement of the Ghana Health Service, the government launched pilot health insurance schemes in four municipalities in the Eastern Region (Mensah, 2011). These schemes were directly modelled on the CBHISs (Mensah, 2011), with only a minor modification so as to ensure that the pilot schemes were rolled out and managed by the Ministry of Health rather than the local communities (Adibo, 2013; Agyepong, 2013). Indeed, interview data demonstrate that the Ministry of Health did groundwork for the implementation of the pilot schemes. For example, ‘a whole directorate was set up within the Ministry of Health to implement the scheme in the three pilot districts and to subsequently spread the scheme across the entire country’ (Akor, 2013). Additionally, ‘there were also some preliminary studies … and a regional secretariat was set up in Eastern region, Koforidua’ (Akor, 2013). In turn, the health insurance directorate launched various education and sensitization programmes on the health insurance schemes. The directorate also conducted technical studies in order to assess the pattern of healthcare utilization within the pilot districts and to determine the appropriate premium and payment methods for enrollees (Adjei, 2013). Despite all the aforementioned initiatives, the pilot health insurance schemes in the Eastern region ‘never took off … funds were [rather] misdirected and it came to nothing’ (Afriyie, 2013). Andoh (2013) confirmed the above observation, noting that the pilot schemes ‘did not move from design stage to implementation stage and nobody received any insurance coverage under the scheme’. Once the Ministry of Health learned that the pilot health insurance schemes would have a long implementation period, it returned to SSNIT. This time, however, the Ministry of Health had the financial aid for incentivizing SSNIT to start a health insurance scheme that would not only target SSNIT contributors and cocoa farmers but also target people in the informal sector (Agyepong, 2013; Alhassan, 2013). According to one of the officers who worked on the SSNIT health insurance project, although their health insurance policy did not reach the implementation stage, ‘preparations were advanced to the extent that letters [to launch the policy] had gone to the presidency and invitation cards had also gone out…. So, there was a model that was ready to be rolled out’ before the next election in December 2000 (Alhassan, 2013). In the build-up to the 2000 general elections, health financing became a major—if not the most important—issue. Indeed, almost all the competing political parties, including first-timers such as the United Ghana Movement (UGM) and the National Reform Party (NRP), campaigned on replacing the cash-and-carry policy with health insurance (Afriyie, 2013; Benya, 2013). In that election, the NDC once again argued that it had piloted health-financing schemes modelled off of the country’s many community initiatives, and promised that, when re-elected, health insurance will be a major strategy for mobilizing additional resources and for ensuring financial access in time of need. The pilot work already carried out will form the basis for a mix of insurance schemes, both public and private, national and local, to cater for salaried employees, the self-employed as well as both urban and rural communities (NDC, 2000). Not withstanding its liberal leaning, the NPP made a similar promise to pursue care for the sick on an equitable basis so that the needs of the sick are addressed according to their circumstances rather than their ability to pay…. Under NPP Government, workers and their Employers, Local Communities Religious Bodies, etc. would be encouraged to establish their own health insurance schemes. The public health facilities will be improved to give quality service at affordable cost even to those not covered by any insurance scheme. A special institution will be created for supervising health insurance. That institution shall be managed independently of Government (NPP, 2000). In general, each party’s proposed health insurance model during the 2000 elections was no different from the existing CBHISs (Adjei, 2013). In the end, the NPP won the December 2000 general elections and followed through on its electoral commitment by immediately assembling a team to design the NHIS based on the experiences of the CBHISs. In this case, the existing CBHISs were directly incorporated into the NHIS. To ensure a quick spread of the NHIS countrywide, various District Assemblies were authorized to establish their own schemes (Afriyie, 2013). Payroll deductions from formal sector workers and an NHIS levy (2.5% on selected goods and services) were also introduced to cultivate a stronger NHIS funding source and provide free insurance coverage for the marginalized (Amoh, 2013). Some studies have shown that the NDC lost the 2000 elections partly because of its failure to implement the proposed NHIS as a replacement to the cash-and-carry policy (Rajkotia, 2009; Blanchet et al., 2012). In a sense, the NDC had lost the trust and credibility necessary for establishing issue ownership over health care. In the context of the NDC’s shortcoming, other parties were perceived as more legitimate contenders for delivering effective healthcare policy. This perception was affirmed during the interviews by Andoh (2013), who emphasized that ‘the opposition managed to convince the population that if they were voted into power they were going to leave them with the Health Insurance Scheme, and they won the election so it was like a social contract with the people’. In sum, the history of partisan contestation over healthcare issue ownership provides further evidence of self-undermining feedback effects shaping electoral politics. Discussion In contrast to the conventional wisdom in the policy feedback literature which suggests that existing policies reinforce themselves and constrain reforms through various self-reinforcing mechanisms (Jacobs and Weaver, 2015), a critical analysis of health policy trajectories in Ghana shows that feedback from cash-and-carry (1) provided windows of opportunity to move health-financing reform to the policy agenda, (2) structured the menus of options available to policymakers and (3) shaped the electoral politics of the reforms. Feedbacks as policy window for change Feedbacks from the cash-and-carry healthcare policy can be classified into two categories: self-reinforcing effects and self-undermining effects. On the one hand, the self-undermining feedback effects included the costly and unaffordable healthcare services, reduced access to health care by the citizenry, corrupt institutions, weak design of exemptions granted to the poor, high prices leading to self-medication and clearly inadequate treatment of patients by public healthcare facilities. On the other hand, the self-reinforcing feedback effects included the initial improvement in the fiscal position of the health system and the related improvement in the ability of service providers to buy needed tools, equipment and other resources for quality service delivery. The aforementioned self-reinforcing feedbacks would normally interact with various other self-reinforcing mechanisms to put the cash-and-carry policy on a path-dependent trajectory. One important self-reinforcing mechanism includes the development of supportive coalitions and institutions. Other self-reinforcing mechanisms include the role of ‘adaptive expectations created as societal actors adjust their behaviour to current policies’, learning, sunk-costs and uncertainty effects that are expected to make policy change increasingly costly and unattractive, over time (Weaver, 2010, p. 138). Indeed, some of these self-reinforcing mechanisms were at work in the Ghanaian context. For example, strong constituencies of frontline health service providers and administrators rallied around the cash-and-carry policy; these venal individuals stood to profit from the dysfunctional policy by bribing and raising user fees at will. It is also probable that the (P)NDC government accounted for a sizeable share of the pro-policy constituency since cash-and-carry alleviated the government’s burden of bearing the entire cost of health care. However, these self-reinforcing mechanisms were clearly unable to prevent the NHIS’ introduction in 2003. As indicated, the policy change from cash-and-carry to NHIS was largely the result of the cash-and-carry policy’s unintended self-undermining feedbacks on major political players. Chiefly among these major political players were the poor and marginalized Ghanaians, who formed the majority of voters (Nathan, 2019). Voters in the middle- and upper-income classes were also major players given that they had to support the needy in their families and communities pay their healthcare bills. When coupled with staff corruption, the reduced access to health care led to a wide-ranging failure of the healthcare system—including facilities, the Ministry of Health and the government—to generate the expected revenue for service delivery. The aforementioned self-undermining feedbacks made the cash-and-carry policy increasingly unpopular and unacceptable, creating a window of opportunity for policy change. Proof of these feedbacks lies in the emergence and increased proliferation of CBHISs in Ghana. Simply put, as Seddoh (2013) noted during the interviews, ‘it became easy for politicians to say I am going to do something about cash-and-carry’ (Seddoh, 2013). Feedback structured the menus of options available to policymakers In addition to creating a window of opportunity for policy change, the self-undermining feedbacks from the cash-and-carry policy structured the menu of options available to the reformers. According to the theory of self-reinforcing mechanism, the structuring role of policy feedback on policy change occurs when policy entrepreneurs like ‘politicians, bureaucrats, and policy experts … undertake a search for new alternatives to address … problems [within the status quo]’ (Jacobs and Weaver, 2015, p. 449). With respect to the NHIS’ introduction in Ghana, this feedback phenomenon was reflected in the ways the (P)NDC and NPP governments, the Ministry of Health and the various actors (SSNIT, consultants and committees) shaped the policy process. In addition to these formal actors, however, were the religious groups and local communities who—out of pressure to circumvent the unaffordable healthcare system by taking their healthcare needs into their own hands—spearheaded the CBHIS programmes. Most importantly, these CBHIS programmes eventually became the nursing bed for the NHIS design. Beyond reinforcing the argument in the self-undermining feedback literature that an expansion in the menu of options increases the chances of major policy change (Jacobs and Weaver, 2015), the Ghanaian case shows how formal and non-formal policy entrepreneurs can work together to ensure policy change. Policy feedback and the electoral politics of the reform As a military regime, the PNDC was able to resist any form of opposition to the cash-and-carry policy. After all, there were no veto players and the political regime was not based on electoral accountability. Following the country’s transition to constitutional rule, however, health financing became a major electoral issue; from 1992—when Ghana transitioned from military rule to democracy—up to the 2000 elections, the adverse effects of the cash-and-carry healthcare policy became a salient campaign issue. For instance, the opposing NPP persistently engaged the NDC government on the cash-and-carry policy’s negative effects, framing them as obnoxious, callous and inhumane (NPP, 1996). To counter these accusations, the NDC highlighted the government’s efforts to arrest the situation, stating that they were expanding the scope of the user fees exemption scheme and exploring health insurance as an alternative to the cash-and-carry policy. These responses were part of the NDC’s strategy to, at least, minimize blame for those adverse effects and ultimately persuade the electorate to renew the party’s mandate at the polls. Although the NDC won both the 1992 and 1996 elections, it lost the 2000 elections to the NPP opposition. As already suggested, the failure of the NDC to replace the problematic cash-and-carry healthcare policy was a key factor in the party’s 2000 electoral defeat. Arguably, from the perspective of voters, the NDC had ample time to undo its own cash-and-carry policy that had deprived many Ghanaians of quality health care; from their perspective, by 2000, the party appeared too slow given the severity of policy’s negative effects. Having noticed that the NDC was suffering a credibility crisis on the issue of health financing, the NPP promised the electorate that, if elected, it would repeal and replace the cash-and-carry policy with an alternative policy that would meet the population’s needs and expectations. Thus, this analysis provides even further evidence to substantiate the claim that self-undermining feedback from status quo policies can shape policy change. Besides these self-undermining factors, several other factors had coalesced to place the creation of a new health insurance programme firmly on the policy agenda by the time that the NPP administration assumed office in 2001 (Seddoh et al., 2011). For instance, Singleton (2006) notes that although the introduction of health insurance was not generally of high priority for the NPP administration, its overwhelming popularity and the looming ‘2004 general elections may have created the impetus needed to encourage law-makers to make such sweeping legislative changes so as to demonstrate to constituents their efficacy’ (p. 30). Kusi-Ampofo et al. (2015, p. 207) confirm that ‘the NPP was mindful of its promise, fully aware of the dire consequence if it failed to abolish cash-and-carry before the [then pending] 2004 general election’. This assessment is further reinforced by opinion polls conducted in five districts in the Eastern and greater Accra regions, which revealed that over 90% of the public supported changes to health-financing arrangements similar to the CBHIS (Asenso-Okyere et al., 1997; Edoh and Brenya, 2002). Furthermore, the global discourse on development had shifted from earlier versions of neoliberalism that emphasized commodification of social services based on the Washington consensus to a slightly different narrative in the Post-Washington Consensus that prioritized socially sensitive policies under the broader umbrella of ‘poverty reduction’. As a leading actor in the reform process, Akor (2013) notes that the introduction of NHIS in Ghana was not in sync with the NPP’s market-friendly disposition, but the design of Ghana’s NHIS, nonetheless, found inspiration in global policy discourse shifts and drew lessons from the CBHIS. All these factors pushed the NHIS to the policy agenda by the 2004 general elections. Conclusion This study has demonstrated how the self-undermining feedback effects of the cash-and-carry policy created a window of opportunity for policy change to occur. These self-undermining factors included the costly and comprehensive introduction of healthcare user fees, which adversely affected access to health care across the country. Over time, the situation empowered forces within the status quo to explore alternative measures, of which the CBHIS was the most popular. Even though it was at odds with the ideological disposition of the NPP, the government drew lessons from the CBHIS and instituted the NHIS in 2003. By demonstrating how the NHIS rose to the top of the government’s policy agenda despite conflicting with the NPP’s liberal ideology, this study also contributes to the ongoing debate about policy feedback. Indeed, it shows how self-undermining policy feedback interacts with other factors—including ideas and electoral pressure—to spur path-departing policy change. These interactive mechanisms may serve as a useful basis for future scholars exploring the impact of policy feedback on policy change. Moreover, this article’s findings provide crucial lessons for public policymaking in general. In particular, the findings suggest the need to incorporate policy feedback into the policymaking process. The failure to adequately consider policy feedback places policies at high risk of self-undermining feedback effects that ultimately truncate policies over time. Building some rectification mechanisms into policies for monitoring and detecting these self-undermining feedbacks would go a long way to safeguard their long-term sustainability. Conflict of interest statement. None declared. Ethical approval. Ethical clearance was obtained at the University of Saskatchewan. Acknowledgements The authors thank Declan Gemmill and the reviewers for their useful comments and suggestions. References Adams I. 2013 . Ministry of Health, Ghana. Personal Interview. April 11. Adibo E. 2013 . Former Deputy Minister of Health and Director General of the Ghana Health Service, Ghana. Personal Interview. July 9. Adjei S. 2013 . Ghana Health Service and a key meber of the design team for the National Health Insurance Scheme, Ghana. Personal Interview. April 3. Afriyie K. 2013 . Minister of Health that led the development of the National Health Insurance Scheme, Ghana. Personal Interview. June 23. Agyepong I. 2013 . Formerly of the Ghana Health Service and Secretary for the design team for the National Health Insurance Scheme, Ghana. Personal Interview. June 13. Agyepong IA. 1999 . Reforming service delivery at district level: the perspective of a Ghanaian District Medical Officer . Health Policy and Planning 14 : 59 – 69 . Google Scholar Crossref Search ADS PubMed WorldCat Akanzige P. 2013 . Ghana Health Service and the Ministerial Task Force on the design of the National Health Insurance scheme. Personal Interview. April 16. Akazili J , McIntyre D , Kanmiki EW et al. 2017 . Assessing the catastrophic effects of out-of-pocket healthcare payments prior to the uptake of a Nationwide Health Insurance Scheme in Ghana . Global Health Action 10 : 1289735 . Google Scholar Crossref Search ADS PubMed WorldCat Akor S. 2013 . Chairman of the Design Team for the National Health Insurance Scheme and first Executive Secretary of the National Health Insurance Council, Ghana in 2004. Personal Interview. March 28. Akorsah B. 2013 . Former Director General of the Ghana Health Service and president of the Ghana Medical Association. Personal Interview. June 10. Alhassan Y. 2013 . Social Security and National Insurance Trust (SSNIT), Ghana. Personal Interview. April 12. Amoh KN. 2013 . Consultant for the National Health Insurance Scheme design process, Ghana. Personal Interview. June 30. Andoh AF. 2013 . National Health Insurance Authority. He was a key member of the ministerial task force that provided technical expertise in drafting the National Health Insurance Scheme Act, 2003 and regulations and a special assistant to the Minister of Health during the development of the National health Insurance Scheme. Personal Interview. July 16. Asenso-Okyere WK. 1995 . Financing health care in Ghana . World Health Forum 16 : 86 – 91 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Asenso-Okyere WK , Osei-Akoto I , Anum A , Appiah EN. 1997 . Willingness to pay for health insurance in a developing economy. A pilot study of the informal sector of Ghana using contingent valuation . Health Policy 42 : 223 – 37 . p Google Scholar Crossref Search ADS PubMed WorldCat Awittey SD. 2013 . Executive Secretary, Pharmaceutical Society of Ghana. Personal Interview. May 20. Béland D. 2010 . Reconsidering policy feedback: how policies affect politics . Administration & Society 42 : 568 – 90 . Google Scholar Crossref Search ADS WorldCat Béland D , Rocco P , Waddan A. 2019 . Policy feedback and the politics of the Affordable Care Act . Policy Studies Journal 47 : 395 – 422 . p Google Scholar Crossref Search ADS WorldCat Béland D , Schlager E. 2019 . Varieties of policy feedback research: looking backward, moving forward . Policy Studies Journal 47 : 184 – 205 . p Google Scholar Crossref Search ADS WorldCat Benya E. 2013 . World Bank, Accra Office. Personal Interview. June 6. Blanchet NJ , Fink G , Osei-Akoto I. 2012 . The effect of Ghana’s National Health Insurance Scheme on health care utilisation . Ghana Medical Journal 46 : 76 – 84 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Campbell AL. 2003 . How Policies Make Citizens: Senior Political Activism and the American Welfare State . Princeton, New Jersey: Princeton University Press . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Campbell JL. 2004 . Institutional Change and Globalization . Princeton, New Jersey: Princeton University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Coleman NA. 1997 . The Uneven Implementation of User Fee Policy in Ghana. Research Paper, Harvard School of Public Health, Boston. http://www.hsph.harvard.edu/takemi/Reseaarch% 20papers.htm, accessed June 10, 2020. Edoh D , Brenya A. 2002 . A community-based feasibility study of National Health Insurance Scheme in Ghana . African Journal of Health Sciences 9 : 41 – 50 . p Google Scholar PubMed OpenURL Placeholder Text WorldCat Gyimah-Boadi E. 1994 . Ghana’s uncertain political opening . Journal of Democracy 5 : 75 – 86 . Google Scholar Crossref Search ADS WorldCat Hacker JS. 2004 . Privatizing risk without privatizing the welfare state: the hidden politics of social policy retrenchment in the United States . American Political Science Review 98 : 243 – 60 . p Google Scholar Crossref Search ADS WorldCat Jacobs AM , Weaver RK. 2015 . When policies undo themselves: self‐undermining feedback as a source of policy change . Governance 28 : 441 – 57 . p Google Scholar Crossref Search ADS WorldCat Kingdon JW. 1984 . Agendas, Alternatives, and Public Policies . Boston, MA : Little, Brown and Company . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kpessa M , Béland D. 2012 . Transnational actors and the politics of pension reform in Sub-Saharan Africa . Review of International Political Economy 19 : 267 – 91 . p Google Scholar Crossref Search ADS WorldCat Kraus J. 1991 . The political economy of stabilization and structural adjustment in Ghana. In: Rothschild D (ed). Ghana: The Political Economy of Recovery . Boulder, CO : Lynne Reinner , 119 – 55 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kuntulo DA. 2013 . Trades Union Congress of Ghana. Personal Interview. March 27. Kusi-Ampofo O , Church J , Conteh C , Heinmiller BT. 2015 . Resistance and change: a multiple streams approach to understanding health policy making in Ghana . Journal of Health Politics, Policy and Law 40 : 195 – 219 . Google Scholar Crossref Search ADS PubMed WorldCat Mensah J. 2011 . The impact of National Health Insurance on health delivery in Brong Ahafo. Master’s Thesis. Kwame Nkrumah University of Science and Technology. Mettler S , SoRelle M. 2014 . Policy feedback theory . Theories of the Policy Process 3 : 151 – 81 . Google Scholar OpenURL Placeholder Text WorldCat Mettler S , Soss J. 2004 . The consequences of public policy for democratic citizenship: bridging policy studies and mass politics . Perspectives on Politics 2 : 55 – 73 . Google Scholar Crossref Search ADS WorldCat Nathan NL. 2019 . Does participation reinforce patronage? Policy preferences, turnout and class in urban Ghana . British Journal of Political Science 49 : 229 – 55 . Google Scholar Crossref Search ADS WorldCat National Democratic Congress (NDC). 1992 . Manifesto 1992: Continuity and Stability . Accra : NDC . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC National Democratic Congress (NDC). 2000 . Manifesto 2000: Spreading the Benefits of Development . Accra : NDC . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC New Patriotic Party (NPP). 1992 . Manifesto 1992: Development in Freedom . Accra : NPP . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC New Patriotic Party (NPP). 1996 . Manifesto: Development in Freedom and Agenda for Change . Accra : NPP . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC New Patriotic Party (NPP). 2000 . Manifesto 2000: Development in Freedom, Agenda for Change . Accra : NPP . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Nortey G. 2013 . Secretary, Ghana Registered Nurses Association. Personal Interview. April 18. Nyonator F , Kutzin J. 1999 . Health for some? The effects of user fees in the Volta Region of Ghana . Health Policy and Planning 14 : 329 – 41 . p Google Scholar Crossref Search ADS PubMed WorldCat Oberlander J , Weaver RK. 2015 . Unraveling from within? The affordable care act and self-undermining policy feedbacks . The Forum 13 : 37 – 62 . Google Scholar Crossref Search ADS WorldCat Ofori M. 2013 . Christian Health Association of Ghana. Personal Interview. June 21. Pierson P. 1993 . When effect becomes cause: policy feedback and political change . World Politics 45 : 595 – 628 . Google Scholar Crossref Search ADS WorldCat Pierson P. 1994 . Dismantling the Welfare State? Reagan, Thatcher, and the Politics of Retrenchment . New York, NY : Cambridge University Press . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Rajkotia Y. 2009 . National Health Insurance in Ghana: politics, adverse selection, and the use of child health services. PhD dissertation. The Johns Hopkins University. Rawlings JJ. 1997 . State of the Nation, Republic of Ghana . Accra : Government of Ghana . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Seddoh T. 2013 . Ghana College of Surgeons and Physician. Member, the Ministerial Task Force on the design of the National Health Insurance scheme. Personal Interview. July 2. Seddoh A , Adjei S , Nazzar A. 2011 . Ghana’s National Health Insurance Scheme: Views on Progress, Observations, and Commentary . Accra, Ghana : Centre for Health and Social Services , 20 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Singleton JL. 2006 . Negotiating Change: An Analysis of the Origins of Ghana’s National Health Insurance Act. Paper 4. Honors Projects. Macalester College Skocpol T. 1992 . Protecting Soldiers and Mothers: The Political Origins of Social Policy in the United States . Cambridge, MA : Belknap Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Thelen K. 2004 . How Institutions Evolve: The Political Economy of Skills in Germany, Britain, the United States, and Japan . Cambridge : Cambridge University Press . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Waddington CJ , Enyimayew KA. 1989 . A price to pay: the impact of user charges in Ashanti‐Akim district, Ghana . International Journal of Health Planning and Management 4 : 17 – 47 . p Google Scholar Crossref Search ADS WorldCat Waddington C , Enyimayew KA. 1990 . A price to pay, part 2: the impact of user charges in the Volta region of Ghana . International Journal of Health Planning and Management 5 : 287 – 312 . p Google Scholar Crossref Search ADS WorldCat Walt G , Shiffman J , Schneider H et al. 2008 . Doing’ health policy analysis: methodological and conceptual reflections and challenges . Health Policy and Planning 23 : 308 – 17 . Google Scholar Crossref Search ADS PubMed WorldCat Weaver K. 2010 . Paths and forks or chutes and ladders? Negative feedbacks and policy regime change . Journal of Public Policy 30 : 137 – 62 . Google Scholar Crossref Search ADS WorldCat Wireko I. 2016 . Explaining radical change in Ghanaian health care policy. PhD dissertation. University of Saskatchewan. Wireko I , Béland D. 2013 . The challenge of healthcare accessibility in Sub-Saharan Africa: the role of ideas and culture . European Journal of International Management 7 : 171 – 86 . Google Scholar OpenURL Placeholder Text WorldCat Footnotes 1 There were very few health facilities in Ghana. The long queues were, therefore, the result of inadequate health infrastructure, rather than an increase in utilization. © The Author(s) 2020. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. All rights reserved. For permissions, please e-mail: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Lakha,, Fatim;Suriyawongpaisul,, Paibul;Sangrajrang,, Suleeporn;Leerapan,, Borwornsom;Coker,, Richard

doi: 10.1093/heapol/czaa063pmid: 33212481

Abstract Thailand has successfully implemented Universal Health Coverage (UHC) and embedded the 2030 Agenda for Sustainable Development into its Thailand 4.0 policy. Breast cancer is a growing challenge in Thailand, as it is globally. It serves as a perfect medium through which to interrogate UHC and demonstrate areas of the health system which require further strengthening if UHC is to be sustainable in the longer term. We conducted a situation analysis and used a Systemic Rapid Assessment (SYSRA) framework to examine the challenges posed to UHC through the lens of breast cancer. We identified a number of challenges facing UHC including (1) continued political commitment; (2) the need for coordinated scale-up of strategic investments involving increased financing and fine-tuning of the allocation of resources according to health needs; (3) reducing inequities between health insurance schemes; (4) investing in innovation of technologies, and more critically, in technology transfer and capacity building; (5) increasing capacity, quality and confidence in the whole primary healthcare team but especially family medicine doctors. This would subsequently increase both efficiency and effectiveness of the patient pathway, as well as allow patients wherever possible to be treated close to their homes, work and family; (6) developing and connecting information systems to facilitate understanding of what is working, where needs are and track trends to monitor improvements in patient care. Our findings add to an existing body of evidence which suggest, in light of changing disease burden and increasing costs of care, a need for broader health system reforms to create a more enabling platform for integrated healthcare as opposed to addressing individual challenging elements one vertical system at a time. As low- and middle-income countries look to realize the 2015 Sustainable Development Goals and sustainable UHC this analysis may provide input for policy discussion at national, regional and community levels and have applicability beyond breast cancer services alone and beyond Thailand. Health system strengthening, universal healthcare, breast cancer, equity, challenges KEY MESSAGES Breast cancer is a growing problem and provides an opportune lens to examine the policy and health system challenges epidemiological and demographic transition are placing on sustainability and advancement of Universal Health Coverage (UHC). Implementation is just the first step. For sustainability and advancement of UHC, there is a need, both at implementation and thereafter for integrated healthcare. Macro policy efforts are necessary to ensure resources are directed to address gaps, pressures and shortages in workforce, information systems and investment in building research and development capacity in health as populations age and medical technologies and treatments expand across regions. The challenges identified, in exploring breast cancer services in a country which is a leader in implementation of UHC, are equally applicable and relevant to other countries and settings as well as other health services which either are implementing or have implemented UHC. Programme and policy efforts to ensure financial sustainability of UHC must continually review and address any imbalance in equitable access and ensure available health insurance schemes are aligned so that resources are focussed on those who are more disadvantaged. Introduction Thailand is one of the few middle-income countries (MICs) to have successfully implemented Universal Health Coverage (UHC). Key factors that have contributed to its success include political commitment, committed health personnel and active involvement of Civil Society Organizations. As a result, it is a health leader amongst MICs and in South East Asia (SEA), with 99% of the Thai population receiving free healthcare at the point of delivery since 2002. Challenges remain, however, in the form of increased workload and rising costs to meet the demands of epidemiological and demographic transition in a nation with a high proportion of people living in relative poverty (Thaiprayoon and Wibulpolprasert, 2017; Sumriddetchkajorn et al., 2019). This in turn leads to a potential for a large treatment gap. Non-communicable diseases (NCDs) are an emerging challenge for Thailand and addressing NCDs and their risk factors is not merely a priority for economic development but the most powerful way of improving longevity and healthy life expectancy for the foreseeable future—especially in both low- and middle-income countries (LMICs) (Magnusson and Patterson, 2014). NCD deaths accounted for 71% of the total 501 000 deaths in Thailand in 2014 and are predicted to continue to rise rapidly (WHO, 2014). The estimated economic impact in 2013 from major NCDs to Thai society was 280 billion Thai baht ($8.8 billion). Cancer alone has become a critical challenge for Thailand with rising incidence rates and cancer deaths accounting for approximately the same number and percentage of deaths as communicable diseases, maternal, perinatal and nutritional conditions together in 2014 (17–18% of total deaths in Thailand) (Sankaranarayan et al., 2014; WHO, 2014; Bray et al., 2018; Globocan, 2018a; WHO, 2018). The 2012–14 ASEAN costs in oncology study found cancer to be a significant driver of poverty and inequality in South East Asia. It demonstrated that even though Thailand has achieved UHC, families affected by cancer can suffer catastrophic health expenditures due to gaps in the current health financing scheme (ACTION Study Group et al., 2017). Studies to date have identified that the risk of dying from cancer and facing catastrophic costs have been associated with more advanced disease stage and with gender. Women have been found to be particularly vulnerable to financial catastrophe associated with cancer (ACTION Study Group, 2015; ACTION Study Group et al., 2017). Both globally and in Thailand, breast cancer, specifically, is the most common cancer in women (incidence globally 46.3/100 000 people per year; Thailand 35.7/100 000 per year) as well as the leading cause of cancer-related mortality in women (global ASR 13/100 000; Thailand age-standardised rate (ASR) 14/100 000) (DeSantis et al., 2015; Bray et al., 2018). In Thailand, incidence rates have been steadily increasing with rates rising from 20.5 cases to 35.7 per 100 000 person-years in the last 20 years (Virani et al., 2014; Tassanasunthornwong et al., 2015; Lalitwongsa et al., 2016; Sangkittipaiboon et al., 2016; Globocan, 2018b; Virani et al., 2018a). This trend is expected to continue and at a steeper gradient throughout the country over the next decade secondary to demographic and epidemiological transition (Porter, 2008; Youlden et al., 2012; 2014; Virani et al., 2017). Mortality rates have continued to remain high and despite diagnostics, medicines and treatments, survival rates have not improved at the same rate as in HICs (5 years ASR 69% in 2018 in Thailand) (Miller et al., 2017; Allemani et al., 2018; Gliniewicz et al., 2018). Prognostic and predictive factors associated with better survival are lower stage, lower grade, being aged between 40 and 69 years, absence of comorbidities and a favourable genetic profile (including a non-expressive HER2 receptor) (Youlden et al., 2012). This puts further impetus on identifying those with breast cancer earlier and initiating treatment without delay. As such the potentially large ‘treatment gap’ in breast cancer services and care offers a perfect medium through which to examine UHC and the policy and health system challenges which require to be surmounted, as well as develop our understanding of what elements must be put in place at national, regional and community levels to support the longer-term sustainability of UHC. We conducted a situation analysis and used this to examine the challenges posed to UHC through the lens of breast cancer. Methods We used a systematic approach to gather information about a complex health system. For this, we drew upon the Systemic Rapid Assessment (SYSRA) framework which we adapted for our purpose of assessing breast cancer. The SYSRA framework is a conceptual and analytic framework, initially developed by Atun et al. (2004) to evaluate health systems and communicable disease programmes. The SYSRA analytical framework provides a conceptual, analytical framework and tool to evaluate health interventions that take into account disease programmes, the general health system and the wider socioeconomic and political context. Our adapted SYSRA framework comprised two core elements: a horizontal ‘health and pharmaceutical system’ element and a vertical ‘breast cancer programme element’ (Figure 1). The ‘horizontal assessment’ focussed on structures and functionality of two overall interlinked systems—health and pharmaceutical whilst the ‘vertical’ element assessed the specific breast cancer programme components embedded within the health and pharmaceutical systems. For both horizontal and vertical assessments’ the perspectives reviewed included five key component areas: external environment; health systems structure and organization; finances and resource allocation; healthcare delivery and information systems. Figure 1 Open in new tabDownload slide Adapted SYSRA for situation analysis. Adapted from Atun et al. (2004). Figure 1 Open in new tabDownload slide Adapted SYSRA for situation analysis. Adapted from Atun et al. (2004). This study was conducted in September 2017 to March 2019. For each of the rapid systemic assessment perspectives quantitative and qualitative data were obtained based on a review of the available literature, reports and interviews with key informants. The first stage comprised review of secondary data and documentation to determine what information was available and what the information gaps were. For the second stage, 10 semi-structured interviews were conducted with key informants using a predefined semi-structured questionnaire, focussing especially on gaps and queries identified in the initial literature/documentation review. Questions were focussed on the key component areas. During the second stage alongside the interviews, additional qualitative and quantitative information were gathered using an iterative process of information gathering. The third stage comprised synthesis of information gathered followed by further discussions with both a subgroup of key informants spoken to in stage two and new key informants to corroborate our findings. Key informants were selected on the basis of their expertise in various health system, pharmaceutical system and breast cancer programme components. They were chosen from different institutions thus providing the opportunity for triangulation of the information and views provided. These included senior individuals from within: The Ministry of Public Health—the department of medical sciences, the department of communicable diseases and the department of workforce planning; the three Health Insurance Schemes; the National Cancer Institute (NCI); Oncologists; Pharmacists; Breast Surgeons; Primary Care doctors and academics. No remuneration was offered to any informants. Results In examining UHC using the lens of breast cancer, we have used five component areas to synthesize key points, lessons learned and challenges yet to be conquered on the path to sustainable UHC in Thailand. External environment (political, legal, economic, socioeconomic and technological) Since the early 2000s, the Thai Government’s policy has been to achieve health equity and UHC via two major reform strands: supply-side investment (infrastructure and extensive geographical coverage over the past decades) and minimizing financial barriers to healthcare through expansion of financial risk protection to the whole population (Phongpaichit and Baker, 1992; McManus, 2012; Vasavid et al., 2003; Tangcharoensathien et al., 2004; Towse et al., 2004; Somkotra and Lagrada, 2008; Jongudomsuk et al., 2015; WHO, 2017). Challenges remain. These include an aging population, increasing rates of NCDs including cancer, antimicrobial resistance and high levels of air pollution (Porter, 2008; Youlden et al., 2012; OECD, 2013). Thailand recently embedded the Sustainable Development Goals (SDGs) into its ‘Thailand 4.0’ policy as part of its 20-year plan. The 2030 Agenda for Sustainable Development was the first time that the global development agenda recognized that NCDs, including cancer, constitute a major health and development challenge. This led to an individual target within SDG3 to reduce premature mortality from NCDs, including cancer and another to achievement of sustainable UHC. Progressive steps taken to date include undertaking a leadership role within the Cancer Control Leadership Forum (Singh et al., 2017); engagement with multilateral external partners such as the World Health Organisation (WHO), the International Agency for Research on Cancer (IARC), Breast Health Global Initiative and ASEAN to ensure good data systems and improve data collection and analysis (Singh et al., 2017); being an active member of the Union of International Cancer Control; the NCI’s drive to implement the National Cancer Control Plan (NCCP) and having a multi-stranded cancer strategy. Health system financing, structure and organization The health system is mainly financed by general taxation, followed by direct out of pocket (OOP) payment, Social Health Insurance (SHI) and private insurance premiums. Healthcare coverage is provided by three public health insurance schemes—civil servants and their dependents by the Civil Servant Medical Benefit Scheme (CSMBS) (8%), private sector employees by the Social Health Insurance (SHI) scheme (16%) and the rest of the population by the Universal Coverage Scheme (UCS) (75%) (Table 1) (Jongudomsuk et al., 2015). Analysis by the National Statistical Office found that 50% of UCS beneficiaries are in the poorest two quintiles whilst approximately half of all beneficiaries of CSMBS and SHI are in the most affluent quintiles (Patcharanarumol et al., 2018a). A small percentage of the population (2% in 2012) hold additional private health insurance. Table 1 Characteristics of the three public health insurance schemes (Tangcharoensathien et al., 2018) . CSMBS . SHI . UCS . Population covered (2017) 4.4 million (8%) 10.6 million (16%) 48 million (75%) Beneficiaries Civil servants and their dependents Formal sector employees but not dependents except maternity leave Those not enrolled in the CSMBS and SHI Source of finance (2016) (Thai baht) Tax based, non-contributory Tripartite contribution by employer, employee and government: . 37.7 billion Tax based, non-contributory 71.02 billion 109.3 billion Purchaser Comptroller General’s Department, Ministry of Finance Social Security Office, Ministry of Labour NHSO Provider choice Free choice of public providers. Private in emergency Annual choice of contracted public and private hospitals (>100 beds and other required facilities are eligibility criteria as contractor hospitals Limited choice: annual registration with public and private contracting unit for primary care, mostly district hospital and its network of health centres in their domicile district Benefit package Comprehensive curative and rehabilitation, no prevention and health promotion services Comprehensive curative and rehabilitation, no prevention and health promotion services Comprehensive curative and rehabilitation services Prevention and promotion was managed by NHSO for all Thai citizens Payment OP: fee-for-service, reimbursement to providers IP: DRG since 2009 No global budget applied for IP Cannot apply hard budget OP: capitation inclusive for OP and IP where IP has additional pay for high DRG weights Hard budget was applied OP: age-adjusted capitation IP: DRG + global budget Hard budget was applied . CSMBS . SHI . UCS . Population covered (2017) 4.4 million (8%) 10.6 million (16%) 48 million (75%) Beneficiaries Civil servants and their dependents Formal sector employees but not dependents except maternity leave Those not enrolled in the CSMBS and SHI Source of finance (2016) (Thai baht) Tax based, non-contributory Tripartite contribution by employer, employee and government: . 37.7 billion Tax based, non-contributory 71.02 billion 109.3 billion Purchaser Comptroller General’s Department, Ministry of Finance Social Security Office, Ministry of Labour NHSO Provider choice Free choice of public providers. Private in emergency Annual choice of contracted public and private hospitals (>100 beds and other required facilities are eligibility criteria as contractor hospitals Limited choice: annual registration with public and private contracting unit for primary care, mostly district hospital and its network of health centres in their domicile district Benefit package Comprehensive curative and rehabilitation, no prevention and health promotion services Comprehensive curative and rehabilitation, no prevention and health promotion services Comprehensive curative and rehabilitation services Prevention and promotion was managed by NHSO for all Thai citizens Payment OP: fee-for-service, reimbursement to providers IP: DRG since 2009 No global budget applied for IP Cannot apply hard budget OP: capitation inclusive for OP and IP where IP has additional pay for high DRG weights Hard budget was applied OP: age-adjusted capitation IP: DRG + global budget Hard budget was applied DRG, diagnostic-related group; IP, in-patient; OP, out-patient. Open in new tab Table 1 Characteristics of the three public health insurance schemes (Tangcharoensathien et al., 2018) . CSMBS . SHI . UCS . Population covered (2017) 4.4 million (8%) 10.6 million (16%) 48 million (75%) Beneficiaries Civil servants and their dependents Formal sector employees but not dependents except maternity leave Those not enrolled in the CSMBS and SHI Source of finance (2016) (Thai baht) Tax based, non-contributory Tripartite contribution by employer, employee and government: . 37.7 billion Tax based, non-contributory 71.02 billion 109.3 billion Purchaser Comptroller General’s Department, Ministry of Finance Social Security Office, Ministry of Labour NHSO Provider choice Free choice of public providers. Private in emergency Annual choice of contracted public and private hospitals (>100 beds and other required facilities are eligibility criteria as contractor hospitals Limited choice: annual registration with public and private contracting unit for primary care, mostly district hospital and its network of health centres in their domicile district Benefit package Comprehensive curative and rehabilitation, no prevention and health promotion services Comprehensive curative and rehabilitation, no prevention and health promotion services Comprehensive curative and rehabilitation services Prevention and promotion was managed by NHSO for all Thai citizens Payment OP: fee-for-service, reimbursement to providers IP: DRG since 2009 No global budget applied for IP Cannot apply hard budget OP: capitation inclusive for OP and IP where IP has additional pay for high DRG weights Hard budget was applied OP: age-adjusted capitation IP: DRG + global budget Hard budget was applied . CSMBS . SHI . UCS . Population covered (2017) 4.4 million (8%) 10.6 million (16%) 48 million (75%) Beneficiaries Civil servants and their dependents Formal sector employees but not dependents except maternity leave Those not enrolled in the CSMBS and SHI Source of finance (2016) (Thai baht) Tax based, non-contributory Tripartite contribution by employer, employee and government: . 37.7 billion Tax based, non-contributory 71.02 billion 109.3 billion Purchaser Comptroller General’s Department, Ministry of Finance Social Security Office, Ministry of Labour NHSO Provider choice Free choice of public providers. Private in emergency Annual choice of contracted public and private hospitals (>100 beds and other required facilities are eligibility criteria as contractor hospitals Limited choice: annual registration with public and private contracting unit for primary care, mostly district hospital and its network of health centres in their domicile district Benefit package Comprehensive curative and rehabilitation, no prevention and health promotion services Comprehensive curative and rehabilitation, no prevention and health promotion services Comprehensive curative and rehabilitation services Prevention and promotion was managed by NHSO for all Thai citizens Payment OP: fee-for-service, reimbursement to providers IP: DRG since 2009 No global budget applied for IP Cannot apply hard budget OP: capitation inclusive for OP and IP where IP has additional pay for high DRG weights Hard budget was applied OP: age-adjusted capitation IP: DRG + global budget Hard budget was applied DRG, diagnostic-related group; IP, in-patient; OP, out-patient. Open in new tab UHC has reduced unmet need to a similar level to that of OECD countries household; reduced OOP payments from 34% of total health expenditure in 2000 (before UCS) to 12% of total health expenditure in 2014 and increased the likelihood of having annual check-ups, especially for women (Ghislandi et al., 2015; The World Bank, 2018). Additionally, the National Health Security Office (NHSO)’s range of strategic purchasing actions have led to substantial cost savings whilst ensuring quality and the opportunity to provide more services for patients in the UCS (Tangcharoensathien et al., 2015; 2018). Almost half of Thailand’s total health expenditure is on medicines. Imported medicines make up one-third of the medicines used in the country; however, the expense incurred is two-thirds of all pharmaceutical cost due to the high cost of imported patented essential medicines (Akaleephan, 2013). With 75–80% of healthcare costs being financed through the government due to UHC the Thai Food and Drug Administration Agency (FDA) developed a National List of Essential Medicines (NLEM) to contain costs and ensure equity. Public hospitals are mandated that >70% of prescribed medicines must be from the NLEM. All new medicines made available through the NLEM undergo economic evaluation and must demonstrate both cost and clinical effectiveness, feasibility of implementation and budget impact (Mohara et al., 2012). Government reimbursement is biased towards locally made, low-cost generics as opposed to high-cost imported medicines. Nevertheless, there remains an imbalance in equitable access, secondary to different entitlements for the three health insurance schemes, and there is widespread concern about the sustainability of a predominantly tax-financed policy keeping pace with increased requirements. Increasing healthcare costs for civil servants have led to expenditure per capita being four times that of the UCS (Patcharanarumol et al., 2018a; Tangcharoensathien et al., 2018). The main causes are felt to be increased expectations from patients in the CSMBS for new and expensive interventions (OECD, 2013) and indeed increased access to a wider range of medicines and technologies whilst UCS and SHI beneficiaries are restricted to those medicines on the NLEM and thus usually locally manufactured medicines. CSMBS recipients also have free choice of public provider and immediate access to tertiary centres and teaching hospitals which those on the UCS and SHI scheme do not. The disparity in access for breast cancer patients to technologies for screening is evident. The screening programme in Thailand is opportunistic. Women are encouraged to perform routine breast self-examination and to attend their doctor for clinical breast examination (CBE) and mammography/USS if they detect an abnormality (Mukem et al., 2014). There has been a drive over the past 10 years to increase awareness of breast cancer and educate regarding BSE. Establishment of once in a lifetime nationwide organized breast cancer screening by mammography has not been found to be cost-effective and the three insurance schemes cover CBE, diagnostics and treatment but not mammography for screening purposes (Anothaisinthawee et al., 2013; Yoo et al., 2013). Screening mammograms, therefore, whilst encouraged are not reimbursed and require to be paid OOP. The tariff of 2000–3000 Baht (US$67–100 at an exchange rate of 30baht/US$) in the public sector and 8000–12 000 Baht (US$267–400) in the private sector can be unaffordable for many people. Mukem et al. (2016) analysed the 2007 Health and Welfare Study and the 2009 Reproductive Health Study and found that being a CSMBS or SHI beneficiary was a significant positive predictor of having had a mammogram compared with being a UCS beneficiary after adjusting for socioeconomic and demographic factors. Equally disparities in access to treatment exist for many antineoplastic therapies. Whilst many innovative technologies are available the eligibility criteria varies by health insurance scheme leading to those on UHC having to pay OOP for therapies which are fully reimbursed for those on the CSMBS. Increasing access to clinically beneficial cancer medicines is a challenge in every country due to their high cost. Since the inception of UHC, a number of policies and programmes have been initiated to facilitate access to medicines (Sruamsiri et al., 2015). Even where the government has negotiated ‘equitable pricing’ for imported medicines, and market access is high, most Thai citizens are unable to afford imported medicines which require to be paid OOP as strict criteria need to be met to obtain them at zero cost and these are dependent on the patient’s health insurance scheme (Ngorsuraches and Chaiyakan, 2015; Thailand Board of Investment, 2016; Saerekul et al., 2018). Similarly, whilst there are ∼3000 computed tomography, magnetic resonance imaging and mammography machines in Thailand. A quarter are in the Bangkok Metropolitan Area and the remainder spread throughout the 76 provinces. Approximately 35% of machines are in the private sector and not available to 75% of the population who are covered by the UCS (Tangcharoensathien et al., 2016; Virani et al., 2017). Requirements to travel can be a deterrent to both presentation for diagnosis and to uptake of treatment for many reasons including not only due to loss of livelihood and associated costs but because of caregiving responsibilities and in some cultures also hierarchical structures within the family (ACTION Study Group, 2015). Information systems Integrated health information systems (HIS) which interact with multiple information systems are essential to inform decisions, and monitor progress towards sustainable UHC and the SDGs. Implementation of UHC prompted insurance schemes, hospitals and health facilities to invest in HIS to speed claim and reimbursement processes leading to improved quality and quantity. Software developers also took the opportunity to promote hospital information system packages leading to data being available at district and provincial level in many areas. The central data repository for the country’s cancer registry, within the NCI, has responsibility for provision of cancer-related information, cancer-related intelligence and policy-oriented recommendations. There are sentinel sites throughout Thailand which report back to a team of data analysts responsible for converting hospital data to population health data and tracing back cases where necessary. All health facilities are encouraged to submit cancer reports and information to the NCI, however, it is not mandatory to do so. Consequently, there is a lack of strong commitment on the part of the private sector and often other public health facilities also do not always comply leading to incomplete datasets and moderate data flows (Aljunid et al., 2012). In 2017, 28 of 77 provinces provided data and population coverage was 49% (NCI data). Individual hospitals, the Thai Cancer Information Network, NHSO and regional population-based cancer registries, whilst rich sources of data, are not integrated. Healthcare delivery Thailand is self-reliant in health-care workforce production and infrastructure development. There has been heavy investment, and strong political commitment, to ensure extensive geographical coverage of primary healthcare and hospital services, and expansion of the workforce (Figure 2). Figure 2 Open in new tabDownload slide Healthcare delivery system numbers, 2015 (Sintavanarong, 2014; Patcharanarumol et al., 2018b). Figure 2 Open in new tabDownload slide Healthcare delivery system numbers, 2015 (Sintavanarong, 2014; Patcharanarumol et al., 2018b). Numbers of healthcare professionals (HCPs) have increased significantly, especially nursing professionals, however, opportunities for private practice, financial benefits and social prestige have led to a pull for HCPs overall to urban areas, for doctors, nurses, dentists and pharmacists to private healthcare, and for doctors to become specialists in secondary and tertiary care as opposed to in primary healthcare (PHC) (Jongudomsuk et al., 2015). These losses have led to shortages in PHC workforce especially rurally and overburdening of secondary and tertiary services with minor illnesses or routine healthcare which could be provided at primary care level. Multiple government interventions have been put in place including educational strategies, mandatory rural service, financial incentives, social recognition and significant task-shifting, particularly in remote, rural communities (Treerutkuarkul, 2008). Whilst successful in increasing numbers of doctors, midwives and nurses overall (24.7/100 000 in 2013) (Scheil-Adlung, 2013) these interventions have been less successful in reducing inequitable access. Access remains unequally distributed with residents of Bangkok having far greater access to medical services, as measured by physician numbers and medical appliances per 100 000 people, than residents in other regions. By the same measure, poor people in urban areas have less access to medical services than better-off urban residents resulting in the poor being more vulnerable to receiving inadequate medical care ( McManus, 2012, 2001–2010; Towse et al., 2004). In connection to breast cancer, patients tend to present to either their local primary care clinic or directly to the hospital nurse. Conversations with experts suggested that breast cancer services are stretched. Insufficient numbers of qualified family physicians mean patients often bypass the PHC system and self-refer directly to tertiary or regional centres to reduce time to diagnosis which then further negatively impacts the efficiency and effectiveness of current resources. There are limited numbers of state-run specialist breast clinics or one-stop services in Thailand, though there are hopes to develop these over the next 2 years in tertiary sites. In most hospitals, patients are first seen by a hospital nurse followed by a general surgeon who requests the appropriate tests and investigations. Care is provided by a multidisciplinary team though service pressures mean patients may not see a breast surgeon until directly pre-surgery and do not see a medical oncologist until their final pathology is available. This indicates there is a need for both breast surgeons and medical oncologists especially with rising breast cancer incidence. Tertiary services are overstretched as many primary care units and district hospitals are not adequately trained to manage post-surgical patients. For patients requiring palliative care, there is an active national network aiming to foster collaboration among key players at national scale for the development of palliative care and specifically to strengthen community-based palliative care where there is exponential growth and demand (Jongudomsuk et al., 2015). Recognizing that the PHC workforce is critical in contributing to health services and outcomes and to addressing the challenge of NCDs, including cancer, the most recent PHC reform (2016) aims to increase the capacity and quality of the PHC workforce, especially family medicine doctors (Office of the Council of State, 2017). It is planned to achieve this through development of primary care clusters with strategic overview being provided by the department of workforce development (PIHWD) within the MOPH and collaboration with the Thai Medical Council. Pharmaceutical landscape Thailand’s domestic biopharmaceutical production capacity is concentrated on generics and traditional medicines as opposed to the development of new molecules. All cancer medicines to date are imported. The landscape of current cancer therapies has transformed over the last decade from cytotoxic drugs to various newer classes of pharmacological drug agents (e.g. monoclonal antibodies and kinase inhibitors) as well as innovative, targeted therapies such as Trastuzumab which have better safety, tolerability and efficacy profiles compared with conventional cancer treatments yet also have high development costs making them inaccessible to many patients despite various policies and programmes to facilitate access (Akaleephan, 2013; Ghislandi et al., 2015; Sruamsiri et al., 2015). Discussion In this analysis, breast cancer was used as lens to explore the health systems challenges facing a country which has already successfully implemented UHC, is transitioning both epidemiologically and demographically and is committed to ensuring sustainability of UHC and to achieving the 2015 SDGs. Our assessment of the challenges to UHC through the lens of breast cancer identified a number of key findings. Despite several achievements including implementation of UHC and political commitment to addressing NCDs, the current breast cancer programme is affected by a number of challenges, as described above, due to a number of inter-related health system factors (Jongudomsuk et al., 2015; Tangcharoensathien et al., 2018). Firstly, whilst cancer has been the leading cause of death since 2000, policymakers have been slow to recognize and act on this transition from infectious diseases to NCDs. To date, there is no National cancer act and ‘cancer’ is not officially recognized as being under the umbrella of NCDs. This has led to a disjointed service where responsibility for different arms of the national cancer strategy lies within different arms of the MOPH and we strongly advocate that there is a critical need to address this divergence. Secondly, disparities in access to technologies and affordable medicines are possibly a sizable driver and responsible for a substantial proportion of the poorer prognosis, secondary to later presentation with more aggressive cancer and non-adherence, secondary to rising expenditure on medicines, travel and loss of work (Supakul et al., 2006; Limwattananon et al., 2007). Whilst there have been many efforts to address the imbalance in equitable access between the three health insurance schemes there has been little progress primarily due to resistance from CSMBS beneficiaries, who feared a reduction in their entitlements, to merge the three health insurance schemes (Towse et al., 2004; Sumriddetchkajorn et al., 2019). We suspect this is one of the root causes of stagnation of successful equitable cancer care and UHC advancement in Thailand and we propose that for UHC to remain financially sustainable there is a need for more strenuous cost control of CSMBS, and especially a need to curtail the current fee-for-service payment system and bring it in line with the close-ended system used by UCS, so that overall fiscal health cost is manageable and more resources become available for reducing disparities in access to health services. Addressing this will be a true challenge due to resistance from both health service providers where the passive purchasing model can be used to advantage to create income and CSMBS beneficiaries who often choose to work in the public sector because of access to the CSMBS for themselves and their families (OECD, 2013; Tangcharoensathien et al., 2018). Thirdly, we found many existing health information sources and subsystems and a lot of data being collected yet insufficient resource and data connectedness to be able to utilize the data to its full potential for policy and programme development, strategic planning and quality improvement of services. A central complete data observatory as seen in many HICs would potentially enable improved connectedness, the ability to link datasets between health and other areas as well as at different levels plus more efficient use of human resource. We found few available breast cancer data. Incidence rates in Thailand are thought to currently be 35.7/100 000 but with no national screening programme in situ it is difficult to assess what proportion of all incident cancers are being diagnosed and what proportion are being missed. Whilst cancer registries were first initiated in 1988, it is not mandatory for health facilities to submit data to the NCI and hence not all are complete and some are not fully validated. There is a need for an effective country wide validated cancer registry, and sufficient resource allocated to analyse it, which can be used to reliably inform trends, inform resource planning, guide hypotheses and policy decisions (Virani et al., 2014; 2017; 2018a,b). Fourth, we discovered that whilst UHC, alongside other healthcare reforms to increase workforce and infrastructure, has been transformative there remains inequitable access to both. Significant ‘pull’ factors have led to shortages in the healthcare workforce, overburdening of existing state-run secondary and tertiary hospitals and an under-skilled workforce at primary and community hospital level. There is a critical need for systematically improving resource allocation to those schemes with the greatest need, increased autonomy for hospitals to manage their own financial and human resource, upskilling of those working at community hospital and primary care level in cancer care and implementation of the NCCP. Fifth and finally our analysis identified Thailand’s pharmaceutical R&D capacity as lagging. Globally the biopharmaceutical industry has experienced significant growth and with people living longer and more people being likely to experience cancer, the demand for new cancer technologies and especially personalized medicines has grown (Tunpaiboon, 2017). In Thailand whilst the country has transitioned epidemiologically, the necessary R&D to combat the burden of disease has not, leading to Thailand being a net importer of biopharmaceuticals at substantial cost to the country. This economic burden of cancer treatments to health systems, individuals and their households is expected to grow as the availability of medical technologies and treatments expands across regions. Thailand has a well-developed health technology assessment programme, with cost-effectiveness and budget impact analysis formally embedded as part of coverage decisions. However, policymakers could further contribute by creating a regulatory environment, likely based on financial incentives, that both supports investment in research infrastructure and capacity building and enables scaling up of new approaches that have evolved globally over the past decade, such as advance market commitments, product development partnerships, the Medicines Patent Pool and tiered pricing, for developing and delivering medicines and technologies (Tantivess et al., 2017; Witty, 2017). Overall our findings align closely with the World Health Organization’s key health system strengthening ‘building blocks’ of governance, financing, human resource development, service provision and information systems. Research and development is included as an additional cross-cutting component and is in keeping with Thailand’s 4.0 policy which aims to create a value-based economy that is driven by innovation, technology and creativity. Even within the duration of the situational analysis, many strategies have been initiated primarily to address findings three and four: there is a large 10-year programme of work to upskill and increase the primary healthcare workforce, especially family medicine practitioners; palliative care has significantly developed and become a specialty in its own right; the NHSO which is responsible for financing UCS has been and continues to explore ways to identify cost-cutting measures both by working closely with the Health Intervention and Technology Assessment Program (HITAP) to ensure a complete and up to date NLEM as well as to facilitate access to therapies and diagnostics to those on the UCS. However, our findings suggest that it will take more than addressing each challenge individually one vertical system at a time. Even if we consider strengthening of cancer health systems alone our findings suggest that in light of the changing burden of disease and the increasing cost of technologies there is a need for broader health system reforms to create a more enabling platform for integrated healthcare as opposed to addressing individual elements within one vertical system one at a time. The elements, we identified, which require to be put in place to achieve a financially sustainable, more equitable, health-care system which is responsive, both, to challenges arising from demographic and epidemiological transition and to addressing continuing gaps, pressures and shortages in workforce, infrastructure, technologies, financing mechanisms, information systems and research are not dissimilar to those of other situation analyses looking at health system strengthening of vertical NCDs (Mugisha et al., 2017; Petersen et al., 2019; Semrau et al., 2019). Mugisha et al. (2017), when undertaking a situation analysis of mental health systems in LMICs, similarly found a need for updating of legislation and policies in engendering commitment to allocation of resources equitably both in delivery and in monitoring and evaluation. Petersen et al. (2019) then went on to further assess the challenges and propose strategies to tackle issues identified by Mugisha et al. (2017). They identified broadly similar challenges to ourselves. Thus, our findings are in keeping with, and add to, an existing body of evidence which suggests that our findings may be generalizable to other NCDs and to other low-income, middle-income and upper middle-income countries—whether trying to ensure sustainable UHC or in the case of countries which are in the process of planning for, or implementing, UHC—learning from Thailand’s experience. Our findings indicate that Thailand’s greatest challenges to achieving sustainable UHC are the inequity between the three different health insurance schemes and the political challenge of questioning this inequity given the imbalance between those served by the different schemes. This analysis provides input for policy discussion at national, regional and community levels on how to leverage equitable, integrated healthcare. As Thailand looks to realize the 2030 Agenda for Sustainable Development and sustainable UHC it will require not only knowledge generation but also strong political will and active involvement of civic service organizations, similar to that which was required to initiate and implement UHC ‘The Triangle that moves the mountain’. Acknowledgements The authors would like to acknowledge the contributions of all those individuals within the MOPH, the NCI, NHSO, Mahidol, Chulalongkorn and Songkla Universities, FDA and various district hospitals who gave their time and expertise in helping to inform this piece of work. Any errors are our authors’ own. F.L. and R.C. are funded through a Global Challenges Research Fund grant via UK Research Institute BB/P02789X/1. Conflict of interest statement. None declared. Ethical approval. No ethical approval was required for this study. References ACTION Study Group. 2015 . Financial catastrophe, treatment discontinuation and death associated with surgically operable cancer in SEA: results from the ACTION Study . Surgery 157 : 971 – 82 . Crossref Search ADS PubMed WorldCat ACTION Study Group Bhoo-Pathy N , Yip C et al. 2017 . Policy and priorities for national cancer control planning in low-and middle-income countries: lessons from the Association of South East Asian Nations (ASEAN) costs in oncology prospective cohort study . European Journal of Cancer 74 : 26 – 37 . Google Scholar Crossref Search ADS PubMed WorldCat Akaleephan C. 2013 . Drug Patenting and Self-Reliance . Thailand : International Health Policy Programme . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Aljunid S , Srithamrongsawat S , Chen W et al. 2012 . Health-care data collecting, sharing, and using in Thailand, China mainland, South Korea, Taiwan, Japan, and Malaysia . Value in Health 15 : S132 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat Allemani C , Matsuda T , Di Carlo V et al. 2018 . Global surveillance of trends in cancer survival 2000-14 (CONCORD-3): analysis of individual records for 37513 025 patients diagnosed with one of 18 cancers from 325 population-based registries in 71 countries . The Lancet 391 : 1023 – 75 . Google Scholar Crossref Search ADS WorldCat Anothaisinthawee T , Tantai N , Teerawattananon Y. 2013 . Cost-utility of once-in-a-lifetime breast cancer screening with mammography in Thai women . Journal of Health Systems Research 7 : 413 – 20 . Google Scholar OpenURL Placeholder Text WorldCat Atun R , Lennox-Chhugani N , Drobniewski F , Samyshkin Y , Coker R. 2004 . A framework and toolkit for capturing the communicable disease programmes within health systems: tuberculosis control as an illustrative example . The European Journal of Public Health 14 : 267 – 73 . Google Scholar Crossref Search ADS PubMed WorldCat Bray F , Ferlay J , Soerjomataram I et al. 2018 . Global Cancer Statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries . CA: A Cancer Journal for Clinicians 68 : 394 – 424 . Google Scholar Crossref Search ADS PubMed WorldCat DeSantis C , Bray F , Ferlay J et al. 2015 . International variation in female breast cancer incidence and mortality rates. Cancer . Cancer Epidemiology Biomarkers & Prevention 24 : 1495 – 506 . Google Scholar Crossref Search ADS WorldCat Ghislandi S , Manachotphong W , Perego V. 2015 . The impact of universal health care consumption and risky behaviours: evidence from Thailand . Health Economics, Policy and Law 10 : 251 – 66 . Google Scholar Crossref Search ADS WorldCat Gliniewicz A , Zielinska A , Kwiatkowska K , Dudek-Godeau D , Bielska-Lasota M. 2018 . Survival in women diagnosed with breast and cervical cancer in Poland—compared to European countries, based on CONCORD-3 programme . Przeglad Epidemiologiczny 72 : 499 – 508 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Globocan. 2018 a. Breast Cancer Fact Sheet 2018 . France : International Agency for Research on Cancer, World Health Organisation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Globocan. 2018 b. Thailand Factsheet 2018 . France : International Agency for Research on Cancer, World Health Organisation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Jongudomsuk P , Srithamrongsawat S , Patcharanarumol W et al. 2015 . The Kingdom of Thailand: Health System Review . Geneva : WHO on behalf of : Asia Pacific Observatory on Health Systems and Policies. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Lalitwongsa S , Pongnikorn D , Daoprasert K , Sriplung H , Bilheem S. 2016 . Breast cancer in Lampang, a province in Northern Thailand: analysis of 1993-2012 incidence data and future trends . Asian Pacific Journal of Cancer Prevention 16 : 8327 – 33 . Google Scholar Crossref Search ADS WorldCat Limwattananon S , Prakongsai P , Tangcharoensathien V. 2007 . Catastrophic and poverty impacts of health payments: results from National Household Surveys in Thailand . Bulletin of the World Health Organization 85 : 600 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat Magnusson R , Patterson D. 2014 . The role of law and governance reform in the global response to non-communicable diseases . Globalization and Health 10 : 44 . Google Scholar Crossref Search ADS PubMed WorldCat McManus J. 2012 . Thailand’s Universal Coverage Scheme: Achievements and Challenges. An Independent Assessment of the First 10 Years 2001-2010 . Nonthaburi : Health Insurance System Research Office . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Miller J , Smith J , Ryerson A , Tucker T , Allemani C. 2017 . Disparities in breast cancer survival in the United States (2001-9): findings from the CONCORD-2 study . Cancer 123 : 5100 – 18 . Google Scholar Crossref Search ADS PubMed WorldCat Mohara A , Youngkong S , Velasco RP et al. 2012 . Using health technology assessment for informing coverage decisions in Thailand . Journal of Comparative Effectiveness Research 1 : 137 – 46 . Google Scholar Crossref Search ADS PubMed WorldCat Mugisha J , Abdulmalik J , Hanlon C et al. 2017 . Health systems context(s) for integrating mental health into primary health care in six Emerald countries: a situation analysis . International Journal of Mental Health Systems 11 : 7 . Google Scholar Crossref Search ADS PubMed WorldCat Mukem S , Meng Q , Sriplung H , Tangcharoensathien V. 2016 . Low coverage and disparities of breast and cervical cancer screening in Thai women: analysis of National Representative Household Surveys . Asian Pacific Journal of Cancer Prevention 16 : 8541 – 51 . Google Scholar Crossref Search ADS WorldCat Mukem S , Sriplung H , McNeil E , Tangcharoensathien V. 2014 . Breast cancer screening among women in Thailand: analyses of population-based household surveys . Journal of Medical Association of Thailand 97 : 1106 – 18 . Google Scholar OpenURL Placeholder Text WorldCat Ngorsuraches S , Chaiyakan K. 2015 . Equitable prices of single-source drugs in Thailand . Applied Health Economics and Health Policy 13 : 389 – 97 . Google Scholar Crossref Search ADS PubMed WorldCat OECD. 2013 . Structural Policy Challenges for Southeast Asian Countries . Southeast Asian Economic Outlook 2013: With Perspectives on China and India. Paris: OECD Publishing. 10.1787/saeo-2013-8-en. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Office of the Council of State. 2017 . “Constitution of the Kingdom of Thailand” B.E. 2560 (unofficial translation). Bangkok: The council of State. www.constitutionalcourt.or.th accessed 15 June 2020. Patcharanarumol W , Panichkriangkrai W , Sommanuttaweechai A et al. 2018 a. Strategic purchasing and health system efficiency: a comparison of two financing schemes in Thailand . PLoS One 13 : e0195179 . Google Scholar Crossref Search ADS PubMed WorldCat Patcharanarumol W , Pongutta S , Witthayapipopsakul W , Viriyathorn S , Tangcharoensathien V. 2018 b. Thailand: Number of healthcare facilities, 2015. In: Legido-Quigley H and Asgari-Jirhandeh N (eds). Resilient and People-Centred Health Systems: Progress, Challenges and Future Direction in Asia [Internet]. New Delhi: World Health Organisation, Regional Office for South-East Asia, 360. http://www.searo.who.int/entity/asia_pacific_observatory/publications/country_comparative_studies/thailand.pdf?ua=1. Petersen I , Vean-Rensburg A , Kigozi F et al. 2019 . Scaling up integrated primary mental health in six low- and middle-income countries: obstacles, synergies and implications for systems reforms . British Journal of Psychiatry Open 5 : e69 . Google Scholar Crossref Search ADS PubMed WorldCat Phongpaichit P , Baker C. 1992 . The politics of economic policy reform in Thailand. In: Timberman D (ed). The Politics of Economic Reform in South East Asia: The Experiences of Thailand, Indonesia, and the Philippines . Metro Manila, Philippines : Asian Institute of Management, 33–52 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Porter P. 2008 . Westernising Women’s risks? Breast cancer in lower income countries . New England Journal of Medicine 358 : 213 – 6 . Google Scholar Crossref Search ADS WorldCat Saerekul P , Limsakun T , Anantachoti P , Sakulbumrongsil R. 2018 . Access to medicines for breast, colorectal and lung cancer in Thailand . The Thai Journal of Pharmaceutical Sciences 42 : 221 – 9 . Google Scholar OpenURL Placeholder Text WorldCat Sangkittipaiboon S , Leklob A , Sriplung H , Bilheem S. 2016 . Breast cancer in Lopburi, a province in Central Thailand: analysis of 2001-2010 incidence and future trends . Asian Pacific Journal of Cancer Prevention 16 : 8359 – 64 . Google Scholar Crossref Search ADS WorldCat Sankaranarayan R , Ramadas K , Qiao Y. 2014 . Managing the changing burden of cancer in Asia . BMC Medicine 12 : Google Scholar OpenURL Placeholder Text WorldCat Scheil-Adlung X. 2013 . Health workforce benchmarks for universal coverage and sustainable development . Bulletin of the World Health Organization 91 : 888 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat Semrau M , Alem A , Ayuso-Mateos J et al. 2019 . Strengthening mental health systems in low- and middle-income countries: recommendations from the Emerald programme . BJPsych Open 5 : e73 . Google Scholar Crossref Search ADS PubMed WorldCat Singh T , Pearlman P , Kostelecky B. 2017 . Supporting evidence-based national cancer control planning: the Asia-Pacific Phase II Leadership Forum . Journal of Cancer Policy 12 : 75 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat Sintavanarong P. 2014 . Health Care System in Thailand: Past, Present and Where is the Future? Nonthaburi : MOPH Thailand . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Somkotra T , Lagrada L. 2008 . Payments for health care and its effects on catastrophe and impoverishment: experience from the transition to Universal Coverage in Thailand . Social Science & Medicine 67 : 2027 – 35 . Google Scholar Crossref Search ADS WorldCat Sruamsiri R , Ross-Degnan D , Lu C , Chaiyakunapruk N , Wagner A. 2015 . Policies and programs to facilitate access to targeted therapies in Thailand . PLos One 10 : e0119945 . Google Scholar Crossref Search ADS PubMed WorldCat Sumriddetchkajorn K , Shimazaki K , Ono T et al. 2019 . Universal health coverage and primary care . Bulletin of the World Health Organization 97 : 415 – 22 . Google Scholar Crossref Search ADS PubMed WorldCat Supakul S , Sooksriwong C , Chiersilpa A , Hartzema A. 2006 . Cost of breast cancer treatment with adjuvant therapy in Thai women . Mahidol University Journal of Pharmaceutical Sciences 33 : 9 – 14 . Google Scholar OpenURL Placeholder Text WorldCat Tangcharoensathien V , Limwattananon S , Patcharanarumol W et al. 2015 . Achieving universal health coverage goals in Thailand: the vital role of strategic purchasing . Health Policy and Planning 30 : 1152 – 61 . Google Scholar Crossref Search ADS PubMed WorldCat Tangcharoensathien V , Patcharanarumol W , Panichkriangkrai W. 2016 . H ealth Systems Review: Achievements and Challenges. Policy Note: Thailand Health Systems in Transition . Asia Pacific Observatory on Health Systems and Policies . https://apps.who.int/iris/bitstream/handle/10665/246421/WPR_2016_DHS_010_eng.pdf;jsessionid=393E85D30A664FB43CC609AF70306979?sequence=1. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tangcharoensathien V , Wibulpholprasert S , Nitayaramphong S. 2004 . Knowledge-Based Changes to Health Systems: The Thai Experience in Policy Development . Geneva : World Health Organisation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tangcharoensathien V , Witthayapipopsakul W , Panichkriangkrai W , Patcharanarumol W , Mills A. 2018 . Health systems development in Thailand: a solid platform for successful implementation of universal health coverage . The Lancet 391 : 1205 – 23 . Google Scholar Crossref Search ADS WorldCat Tantivess S , Chalkidou K , Tritasavit N , Teerawattananon Y. 2017 . Health technology assessment capacity development in low- and middle-income countries: experiences from the international units of HITAP and NICE . F1000Research 6 : 2119 . Google Scholar Crossref Search ADS PubMed WorldCat Tassanasunthornwong S , Chansaard W , Sriplung H , Bilheem S. 2015 . Breast cancer in Surat Thani, a province in Southern Thailand: analysis of 2004-2012 incidence and future trends . Asian Pacific Journal of Cancer Prevention 16 : 6735 – 40 . Google Scholar Crossref Search ADS PubMed WorldCat Thailand Board of Investment. 2016 . Thailand’s Bioeconomy Industry . Thailand : Thailand Board of Investment . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Thaiprayoon S , Wibulpolprasert S. 2017 . Political and Policy Lessons from Thailand’s UHC Experience. New Delhi, London: Observer Research Foundation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC The World Bank. 2018 . Out-of-Pocket Expenditure (% of Current Health Expenditure). World Bank. https://data.worldbank.org/indicator/SH.XPD.OOPC.CH.ZS, accessed 16 June 2020. Towse A , Mills A , Tangcharoensathien V. 2004 . Learning from Thailand’s health reforms . BMJ 328 : 103 – 5 . Google Scholar Crossref Search ADS PubMed WorldCat Treerutkuarkul A. 2008 . Thailand’s unsung heroes . Bulletin of the World Health Organization 86 : 5 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat Tunpaiboon N. 2017 . Pharmaceuticals . Thailand : Krungsri . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Vasavid C , Tisayatikom K , Patacharanarumol W , Hanvoravongchai P , Tangcharoensathien V. 2003 . Preliminary Report EQUITAP Project Bangkok . Thailand : IHPP . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Virani S , Bilheem S , Chansaard W et al. 2017 . National and subnational population-based incidence of cancer in Thailand: assessing cancers with the highest burdens . Cancers 9 : 108 . Google Scholar Crossref Search ADS WorldCat Virani S , Chindaprasirt J , Wirasorn K et al. 2018 a. Breast cancer incidence trends and projections in North Eastern Thailand . Journal of Epidemiology 28 : 323 – 30 . Google Scholar Crossref Search ADS PubMed WorldCat Virani S , Sriplung H , Rozek L , Meza R. 2014 . Escalating burden of breast cancer in Southern Thailand: analysis of 1990-2010 incidence and prediction of future trends . Cancer Epidemiology 38 : 235 – 43 . Google Scholar Crossref Search ADS PubMed WorldCat Virani S , Wetzel E , Laohawiriyakamol S et al. 2018 b. Ethnic disparity in breast cancer survival in Southern Thailand . Cancer Epidemiology 54 : 82 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat WHO. 2014 . Non-Communicable Diseases Country Profile . Geneva : World Health Organisation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC WHO. 2017 . WHO Country Cooperation Strategy 2017-2021 . Thailand : World Health Organisation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC WHO. 2018 . Noncommunicable Diseases Country Profile 2018. Geneva: World Health Organisation. Contract No.: License: CC BY-NC-SA 3.0 IGO. Witty A. 2017 . Scaling up for universal health coverage . The Lancet 389 : 343 – 5 . Google Scholar Crossref Search ADS WorldCat Yoo B , Kwon J , Cho E et al. 2013 . Is mammography for breast cancer screening cost-effective in both Western and Asian countries? Results of a systematic review . Asian Pacific Journal of Cancer Prevention 14 : 4141 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat Youlden D , Cramb S , Dunn N et al. 2012 . The descriptive epidemiology of female breast cancer: an international comparison of screening, incidence, survival and mortality . Cancer Epidemiology 36 : 237 – 48 . Google Scholar Crossref Search ADS PubMed WorldCat Youlden D , Cramb S , Yip C , Baade P. 2014 . Incidence and mortality of female breast cancer in the Asia-Pacific region . Cancer Biology & Medicine 11 : 101 – 15 . Google Scholar PubMed OpenURL Placeholder Text WorldCat © The Author(s) 2020. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. All rights reserved. For permissions, please e-mail: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Root, Elisabeth, Dowling;Zarychta,, Alan;Tapia, Bertha, Bermudez;Grillos,, Tara;Andersson,, Krister;Menken,, Jane

doi: 10.1093/heapol/czaa084pmid: 33026429

Abstract Health systems strengthening is at the forefront of the global health agenda. Many health systems in low-resource settings face profound challenges, and robust causal evidence on the effects of health systems reforms is lacking. Decentralization has been one of the most prominent reforms, and after more than 50 years of implementation and hundreds of studies, we still know little about whether these policies improve, harm or are inconsequential for the performance of health systems in less-developed countries. A persistent problem in existing studies is the inability to isolate the effect of decentralization on health outcomes, struggling with heterogeneous meanings of decentralization and missing counterfactuals. We address these shortcomings with a quasi-experimental, longitudinal research design that takes advantage of a unique staggered reform process in Honduras. Using three waves of household survey data over 10 years for a matched sample of 65 municipalities in Honduras, we estimated difference-in-difference models comparing changes in outcomes over time between local health systems that were decentralized using one of three types of organizations [municipal governments, associations of mayors or non-governmental organization (NGOs)] and those that remained centrally administered. We find evidence of overall improvements between 2005 and 2016 in several service delivery-related outcomes, and additional improvements in decentralized municipalities governed by NGOs. NGO-led municipalities saw a 15% decrease in home delivery relative to centralized municipalities in 2016, a 12.5% increase in MCH facility delivery and a 7% increase in the use of a skilled birth attendant. There were no detectable positive treatment effects for vaccination, and a slight decline in the weight-for-length z-scores in NGO municipalities, but we find no systematic evidence of decentralization negatively impacting any maternal and child health outcomes. These findings highlight the importance of considering implementation context, namely organization type, when assessing the effects of decentralization reform. Decentralization, health systems strengthening, health system governance, maternal and child health, health services, Honduras Key Messages When trying to understand the impact of decentralization reforms of health service delivery and health outcomes, explicit attention to the organizational form of the local managing organization is critical for understanding the overall effects for the reform. Policymakers designing decentralization reforms, and researchers studying those reforms, should focus greater attention on the role of governance structures and the contextual conditions that may facilitate or undermine the effective decentralization of local health systems. Despite reasonable concerns about decentralization and performance-based contracting, we found no evidence of systematically poorer health services or outcomes for women and children in decentralized municipalities. Introduction One of the greatest challenges to improving health in low-resource settings is effectively governing the health system. Without effective governance in the health sector—namely decision-making processes that create and enforce socially binding agreements about shared resources and responsibilities (Andersson et al., 2009)—countries cannot scale-up health programmes that are required to meet national and global goals, nor provide continuous and quality care to all individuals. A good health system delivers high-quality services to the entire population in a timely and effective manner. The exact configuration of services varies from country to country, but in all cases requires a robust governance system that can attract and allocate financial resources; institutes transparency and accountability in decision-making routines; has a well-trained and adequately paid workforce; acquires reliable data to inform decisions and policies; and maintains facilities and logistics to deliver quality medicines and technologies. Many health systems in low-resource settings, however, face profound governance challenges that keep them from addressing basic health disparities and preventing human suffering. Strengthening health systems to ensure the delivery of quality healthcare is at the forefront of the global health agenda (Al-Janabi et al., 2018) and global health interventions increasingly target the governance of local health systems for improvement. In fact, the most recent report from the Lancet Commission on High Quality Health Systems in the SDG Era specifically highlighted the need for focused research on contemporary health sector governance strategies that involve partnerships between governments, non-governmental organizations (NGOs) and civil society groups, as well as the contextual factors that promote or hinder efforts at effective health systems reform (Kruk et al., 2018). This message has been echoed by others (Frenk, 2010; Leslie et al., 2018) who have reinforced the critical role that health system quality plays in achieving major global health priorities. Hybrid and mixed services delivery models are common features of health sector reform in many less-developed countries (LDCs): decentralization, collaboration, contracting-out and public–private partnerships have all been used in efforts to increase access to and improve the quality of public health services for marginalized communities (Bossert, 1998; Loevinsohn and Harding, 2005; Bertelli and Smith, 2010; El Bcheraoui et al., 2017; Post et al., 2017; Rowe et al., 2018). While the broad literature on the effects of decentralization is decidedly mixed, a recent quality-adjusted review and a quasi-experiment study of Spain show that health systems decentralization can credibly reduce infant mortality rates (Jiménez-Rubio and García-Gómez, 2017), and in some cases improves the match between patients’ preferences and services offered by the health system (Channa and Faguet, 2016). Contracting for health service delivery has a growing body of literature supporting improved effectiveness (Loevinsohn and Harding, 2005), and some early results on performance-based strategies also point towards success (Basinga et al., 2011; El Bcheraoui et al., 2017), though concerns over perverse incentives and muddled accountability remain relevant (Brodkin, 2008, 2011; Lieberman, 2011; Barasa et al., 2017). Robust causal evidence on how these reforms affect health outcomes in developing countries, however, is missing, partly because of missing longitudinal data, and partly because of the prevalent problem of ‘missing counterfactuals’ in policy reform studies (Schulze and Sjahrir, 2014; Wright et al., 2016). The causal mechanisms underlying effectiveness, as well as moderating conditions shaping how well or how poorly decentralization works under different contextual conditions, continue to be important areas needing further investigation. In short, the state of evidence about decentralized governance of health systems lags far behind enthusiasm for these reforms among policymakers and donors. Decentralization was first introduced as a strategy for health sector reform after the World Health Organization (WHO) emphasized the importance of community participation and local resources in its 1978 Alma-Ata Declaration. We are left with a rather bizarre paradox today: after >50 years of policy experimentation and hundreds of studies, we still know very little about whether decentralization improves, harms or is inconsequential for the performance of health systems in LDCs. While the methodological challenges are great, research design that permits valid causal analysis is essential to providing the evidential base for compelling evaluation of decentralization (Frenk, 2010; Leslie et al., 2018). In this study, we employ a longitudinal quasi-experimental design to address some of the weaknesses in prior studies. Our analysis goes beyond the basic question of whether decentralization affects health services and outcomes to examine the extent to which the organizational structure for governing local health systems influences the effectiveness of decentralization. Data and methods Health sector decentralization in Honduras The Honduran Ministry of Health (MOH) piloted decentralized health service delivery in 2005 and began fully implementing the reform in 2007. This governance reform is a blend of de-concentration and delegation; the MOH passes specific authorities and responsibilities for health system management and health services provision to a local intermediary organization using performance-based agreements while retaining its oversight role as regulator and financier through the relevant Regional Health Authority (Rondinelli et al., 1989; Ministry of Health (MOH) and Government of Honduras, 2009, 2010). Three types of organizations have been utilized as intermediaries under this version of decentralization: single municipal governments, associations of municipalities and NGOs (see Supplementary Appendix for details). Municipal governments will typically manage the health centres in their one municipality under this reform, while associations and NGOs often manage multiple municipalities. As of 2015, over 400 health centres in 94 municipalities, or 28% of the country's primary care health units, were managed under the decentralized health services delivery model. In addition to its direct importance for the healthcare received by rural populations across the country, several characteristics of decentralization in Honduras make it an excellent case through which to generate knowledge about institutional conditions shaping local health system performance. First, focusing our analysis on a single country minimizes confounding risks of historical and cultural factors. Second, staggered implementation allows for a quasi-experimental research design including a credible comparison group. Third, important subnational variation on key institutional attributes enables us to begin to distinguish among possible mechanisms linking governance reform to improved community health. Finally, this reform combines three design features that are emblematic of major debates over the governance of primary care health services: decentralization, contracting-out via intermediary organizations and performance-based incentives. This case thus provides an unusual and exceptionally valuable empirical setting with which to assess persistent and ongoing debates about the conditions under which policy reforms like decentralization can succeed in strengthening local health systems within developing countries. Study design We used difference-in-difference methods combined with propensity score matching to identify effects of decentralization on maternal and child health and health services. This form of quasi-experimental methods has been used extensively to evaluate population-level health interventions (Wagstaff and Yu, 2007; Bonell et al., 2011; Craig et al., 2012; Dimick and Ryan, 2014). Our Encuestas de Salud y Gobernanza (ENSAGO 2016) study took advantage of the partial and staggered rollout of decentralization that began in 2007. It set up a de facto quasi-experimental design in which we could compare decentralized municipalities to centrally administered municipalities. Honduras has a total of 298 municipalities of which 94 municipalities had been decentralized at the time of research design planning in 2015. To define the treatment group, we selected all municipalities with at least 4 years under decentralization by 2015 (42 of 79 fully decentralized, non-pilot municipalities, see Supplementary Appendix for additional details). The 37 municipalities we excluded were decentralized between 2011 and 2015. Our rationale was that a 4-year buffer was necessary to allow sufficient time for restructuring to take effect and institutional changes to impact early childhood health. We matched the 42 decentralized municipalities to controls (drawn from the 204 municipalities that remained under central government control) using standard propensity score methods. We modelled treatment (decentralization) for municipalities as a function of politics, health outcomes, supply of health services, demographics and socioeconomics [see Zarychta et al. (2019a) and its Supplementary Appendix, Zarychta et al. (2019b), for details of study design, propensity score matching and municipality selection]. A total of 23 controls were selected. There was a higher ratio of intervention cases to controls because we wanted to examine potential effects of different organization types utilized under the common decentralization policy (governance strategies). In summary, the final sample included 65 municipalities, 42 that were decentralized during the early or mid-part of the 10-year study period and 23 that remained centrally administered during the entire study period. Post-matching data were reasonably balanced. Overall distance for the treatment group was 0.62 and for the control group was 0.54 [see Zarychta et al. (2019b) for details]. All included municipalities were rural, as these were the kinds of areas targeted by the Honduran MOH. Figure 1 shows their geographic distribution. Figure 1 Open in new tabDownload slide ENSAGO study municipalities by governance type, 2011. Figure 1 Open in new tabDownload slide ENSAGO study municipalities by governance type, 2011. Survey data Data for this study are derived from three household surveys. First, we used the 2005/06 and 2011/12 Honduras Demographic and Health (DHS) Surveys. The DHS programme is funded by USAID and was co-ordinated by Macro International. Data collection was carried out by the Instituto Nacional de Estadística (INE). These data sets are in the public domain and are available from the DHS website along with a detailed sampling methodology (Secretaría de Salud Honduras et al., 2013a,b). Second, we conducted our own household survey (the 2016 ENSAGO survey) modelled after the Honduras DHS. The 2005/06 survey was administered prior to the decentralization reform and 2011/12 was the cut-off for a decentralized municipality to be included in our sample. We implemented a shortened version of the DHS survey following the same sampling methodology so we could compare data over the10-year period covered by the three surveys. This created a Before-After Control-Intervention (BACI) study with comparable data at the municipality level. The ENSAGO household survey sampled ∼9000 households in 65 municipalities, or about 140 households per municipality, roughly double the DHS sample per municipality in each year. We use a standard two-stage sampling design based on municipalities and census enumeration districts (segmentos censales). Within each municipality, we randomly generated a list of seven segmentos censales to sample. INE provided maps of these segments and 20 households within each segment were selected at random. We interviewed the head of household and collected anthropometrics from all children <5 years old. In a slight deviation from the DHS surveys, which interviewed all women in the household aged 18–45, for practical reasons, we randomly chose one woman in this age group to interview. We then constructed a longitudinal data set using 2005/06 and 2011/12 Honduras DHS subsets from our 65 municipalities and the 2016 ENSAGO Household Survey. We developed survey weights reflecting the probabilities that the municipality was selected, the household was selected and the woman chosen for interview (DuGoff et al. 2014). Outcome measures This analysis focuses on five domains representing maternal and child health indicators: child growth outcomes, vaccination, birth and delivery care, maternal health services and women’s health services. We focused on maternal and child health for three reasons. First, the Honduran MOH prioritized maternal and child health as outcomes that health sector reform was designed to improve, particularly in impoverished rural areas (see Supplementary Appendix for details). Second, by 2016, the reform had been in place long enough to plausibly see changes in indicators related to maternal care, birth outcomes and early childhood health. Finally, since vaccination programmes remained under central MOH control even after reform, we rationalized that they were a good check for our findings since we would not expect to see differential effects in this area due to the reform. The 19 specific outcomes used to assess each domain are outlined in Table 1. Child health outcomes are measured using standardized WHO z-scores for child weight-for-age, length-for-age, weight-for-length and body mass index (BMI). Vaccination variables measured whether a child received the full schedule of each vaccine. Other service-related variables capture whether a mother or child received a particular service when needed. Detailed information about measurement of outcomes can be found in the DHS documentation (Secretaría de Salud Honduras et al., 2013b). Table 1 Domains and outcomes use to examine maternal and child health Domain . Outcome measure . Child health z-score for weight-for-age z-score for length-for-age z-score for weight-for-length z-score for BMI Vaccination Completed BCG series Completed DTP series Completed polio series Completed measles series Full vaccination (BCG + DPT + polio + measles) Birth/delivery care Home delivery Maternal Child Health Center (Institutional) Delivery Birth with skilled attendant (Doctor or Trained Nurse) Birth by caesarean section Maternal/child care 4 or more antenatal care visits Maternal ion supplementation Maternal tetanus vaccine Child monitored as part of a weight control programme Women’s health Pap smear done Mammogram done Domain . Outcome measure . Child health z-score for weight-for-age z-score for length-for-age z-score for weight-for-length z-score for BMI Vaccination Completed BCG series Completed DTP series Completed polio series Completed measles series Full vaccination (BCG + DPT + polio + measles) Birth/delivery care Home delivery Maternal Child Health Center (Institutional) Delivery Birth with skilled attendant (Doctor or Trained Nurse) Birth by caesarean section Maternal/child care 4 or more antenatal care visits Maternal ion supplementation Maternal tetanus vaccine Child monitored as part of a weight control programme Women’s health Pap smear done Mammogram done Open in new tab Table 1 Domains and outcomes use to examine maternal and child health Domain . Outcome measure . Child health z-score for weight-for-age z-score for length-for-age z-score for weight-for-length z-score for BMI Vaccination Completed BCG series Completed DTP series Completed polio series Completed measles series Full vaccination (BCG + DPT + polio + measles) Birth/delivery care Home delivery Maternal Child Health Center (Institutional) Delivery Birth with skilled attendant (Doctor or Trained Nurse) Birth by caesarean section Maternal/child care 4 or more antenatal care visits Maternal ion supplementation Maternal tetanus vaccine Child monitored as part of a weight control programme Women’s health Pap smear done Mammogram done Domain . Outcome measure . Child health z-score for weight-for-age z-score for length-for-age z-score for weight-for-length z-score for BMI Vaccination Completed BCG series Completed DTP series Completed polio series Completed measles series Full vaccination (BCG + DPT + polio + measles) Birth/delivery care Home delivery Maternal Child Health Center (Institutional) Delivery Birth with skilled attendant (Doctor or Trained Nurse) Birth by caesarean section Maternal/child care 4 or more antenatal care visits Maternal ion supplementation Maternal tetanus vaccine Child monitored as part of a weight control programme Women’s health Pap smear done Mammogram done Open in new tab Statistical methods We estimate a multivariate regression specification of a difference-in-difference (DiD) model to compare changes in outcomes over time between intervention (decentralized) and control (centrally administered) municipalities. Our primary analysis compared three variants of decentralization corresponding to organizational structures—single municipal government (MUNI), association of municipalities (ASSN) and NGO—against control municipalities which remained centrally administered (CENT). We compare differences between centrally administered municipalities and each organizational type under decentralization in 2005 vs 2011 and 2005 vs 2016. The year effects correspond to the three different surveys—DHS2005, DHS2011 and ENSAGO2016. We focus on the difference-in-difference estimates for the change between 2005 and 2016, using 2011 as a check of similarity in trend before most of the intervention had been implemented. The regression takes the form: yi=β0+β1Xit+β2Year 2011+ β3Year2016+β4NGO+β5ASSN + β6MUNI+β7 Year2011*NGO+β8 Year2011*ASSN+β9 Year2011*MUNI+β10Year2016*NGO+β11 Year2016*ASSN+β12 Year2016*MUNI where yi is an outcome for individual i, Xit is a set of household or maternal characteristics for individual i at time t, β2 and β3 are fixed effects for year and β4 ⁠, β5 and β6 are fixed effects for treatment (decentralization) by governance type (NGO, ASSN and MUNI, respectively). Because we examined the effect of several different treatment effects—decentralization by governance type—over multiple years, we introduce several additional fixed effects (⁠ β7 through β12 ⁠) for the interaction between year * treatment. These terms all compare change from 2005 for decentralized organization types in 2011 and 2016 to change for centralized municipalities. A significant coefficient indicates that the trend for the governance type differs from the trend for centralized municipalities. Standard errors are clustered by municipality and DHS/ENSAGO survey weights applied. While propensity score matching ensured that municipality characteristics were similar at baseline, we also included control variables to correct for potential imbalances in household or maternal characteristics between centralized and decentralized areas. These variables were derived from household modules of DHS and ENSAGO surveys and included: household size, maternal age, number of children, maternal education and literacy, and household wealth. The latter was developed using standard DHS methodology from a listing of household assets (ICF, 2019). We applied survey weights and calculated robust standard errors clustered by municipality to correct for correlation of error terms across households within municipalities. To interpret results, we report model coefficients and 95% confidence intervals for the year, treatment and year*treatment effects by outcome. We also report the difference-in-difference estimate (for per cents or means) between each treatment group and the control group between 2005 and 2011 and 2005 and 2016. All statistical analyses were done with STATA (version 15). The presence of multiple outcomes increases the probability of a chance finding of significance. Therefore, we first adjusted statistical tests for multiple comparisons whenever possible. Second, we organized conceptually similar outcomes into five pre-specified domains (Table 1). We then developed a ‘standardized’ domain measure by creating an index from equally weighted outcomes within that domain and testing for effects on its z-score. We excluded the women’s health domain because having only two outcomes precluded developing a standardized score. We conducted a wide range of secondary analyses to test the robustness and sensitivity of our results. Across those analyses, we found that results remained qualitatively the same as those presented here (see Supplementary Appendix). To examine the parallel trends assumption, we verified that trends in each outcome were similar between decentralized and centrally administered municipalities prior to the complete implementation of the decentralization reform, using data from the 2005 and 2011 DHS. Second, we examined the sensitivity of the treatment effect to propensity score matching by comparing results for the full sample of 65 municipalities with those for a reduced, stronger match sample of 50 municipalities. Finally, we checked that findings were unaffected when we altered the control variables or used different variable classifications (e.g. continuous maternal age vs categorized). We also estimated models in which governance structures were classified only into centralized and decentralized. In this case, we found that grouping all decentralized organization types together masked the differential effect governance organizations, especially NGOs, had in leading the management of local health systems under the decentralization reform. Results The DHS2005 included 2009 women and DHS2011 included 1328 women. ENSAGO2016 enumerated 4142 women. At baseline, child z-scores and some vaccination and women’s health indices differed across the treatment groups (Table 2). There were also differences in most control variables, suggesting demographic and socioeconomic heterogeneity across municipalities. Table 2 Characteristics of the study sample by survey wave and treatment status . 2005 (n = 2009) . 2011 (n = 1068) . 2016 (n = 4142) . Outcome/variable . CENT (n = 548) . NGO (n = 541) . MUNI (n = 323) . ASSN (n = 597) . P-valuea . CENT (n = 451) . NGO (n = 241) . MUNI (n = 260) . ASSN (n = 376) . P-valuea . CENT (n = 1486) . NGO (n = 682) . MUNI (n = 653) . ASSN (n = 1321) . P-valuea . Child health  Weight-for-age −0.845 (1.06) −0.903 (1.05) −0.735 (1.14) −0.905 (1.05) 0.003 −0.700 (1.00) −0.821 (1.05) −0.762 (1.17) −0.789 (1.07) 0.002 −0.509 (1.36) −0.935 (1.52) −0.577 (1.27) −0.604 (1.43) 0.009  Length-for-age −1.668 (1.07) −1.797 (1.21) −1.622 (1.27) −1.583 (1.16) 0.011 −1.554 (1.07) −1.797 (1.21) −1.622 (1.27) −1.583 (1.16) 0.901 −0.954 (1.47) −1.087 (1.63) −1.137 (1.62) −0.822 (1.56) 0.048  Weight-for-length 0.158 (1.06) 0.277 (0.94) 0.171 (1.00) 0.124 (1.00) 0.415 0.226 (0.95) 0.251 (0.92) 0.220 (1.01) 0.155 (0.97) 0.001 −0.026 (1.35) −0.379 (1.53) 0.089 (1.41) −0.185 (1.52) 0.139  BMI 0.335 (1.08) 0.495 (0.96) 0.306 (1.01) 0.302 (1.04) 0.187 0.411 (0.96) 0.481 (0.90) 0.436 (1.00) 0.342 (0.96 0.701 −0.043 (1.40) −0.268 (1.47) 0.207 (1.52) −0.101 (1.56) 0.097 Vaccination  Measles vaccination 60.7 67.0 58.4 61.8 0.197 88.3 89.0 85.7 85.5 0.730 93.8 91.4 83.5 88.7 0.126  Polio Vaccination 78.2 83.1 71.1 77.8 0.028 97.1 98.8 97.2 97.7 0.525 87.8 90.5 83.3 92.7 <0.001  BCG Vaccination 80.4 81.9 78.0 81.7 0.863 87.2 94.2 89.7 87.5 0.322 94.8 93.3 86.5 90.0 0.086  DPT Vaccination 81.7 86.7 78.1 84.0 0.007 97.2 98.1 97.0 97.4 0.947 94.3 92.3 83.9 89.2 0.088  Complete Vaccination 58.1 64.9 57.1 60.0 0.160 84.1 86.2 80.2 82.0 0.665 92.7 90.5 78.7 87.7 0.057 Delivery care  Home delivery 60.0 50.6 51.4 57.9 0.622 39.4 27.6 37.8 35.8 0.574 39.6 8.3 20.6 13.7 <0.001 MCH facility delivery 5.1 2.8 8.5 3.8 0.687 5.0 14.1 14.2 11.9 0.081 7.5 20.3 14.3 21.0 <0.001  Skilled birth attendant 38.2 47.6 50.0 41.3 0.556 61.2 71.7 61.5 62.2 0.604 69.3 92.3 80.6 85.5 <0.001  Caesarean birth 7.0 8.8 9.7 6.5 0.313 11.4 13.1 7.7 13.9 0.196 8.0 22.1 13.4 12.0 0.008 Maternal/child care  Child weight control programme 16.0 22.7 10.1 21.3 0.141 29.6 51.4 12.1 30.7 0.032 39.5 58.4 38.3 46.0 0.049  4 antenatal care visits 82.1 86.6 85.9 86.7 0.412 89.6 94.3 86.5 94.7 0.025 84.4 93.9 92.6 92.7 0.015  Iron supplementation 22.2 16.8 14.3 19.6 0.391 37.0 27.8 26.8 33.0 0.074 94.3 88.1 84.1 91.1 <0.001  Maternal tetanus 84.8 94.9 87.4 89.7 0.055 93.7 94.8 94.1 91.7 0.677 83.4 89.1 83.7 83.9 0.171 Women's health  Pap smear done 56.9 36.3 48.0 42.7 <0.001 63.6 57.2 50.9 55.9 0.011 56.2 50.2 56.9 51.1 0.386  Mammogram done 2.7 2.3 3.1 1.6 0.701 4.8 3.5 3.5 3.7 0.363 3.3 1.7 3.8 1.7 0.025 Control variables  Maternal age at birth 27.0 (7.10) 25.6 (6.74 26.7 (7.36) 26.6 (7.04) 0.004 25.5 (7.1) 26.1 (6.76) 26.1 (7.14) 26.1 (6.73) 0.002 25.6 (6.88) 27.3 (0.66) 24.5 (6.72) 24.9 (6.35) 0.004  Maternal age 28.9 (7.11) 27.7 (6.87) 28.7 (7.45) 28.6 (7.09) 0.004 28.1 (7.41) 28.6 (7.04) 28.0 (7.13) 28.7 (6.95) 0.582 27.7 (7.04) 27.3 (7.04) 27.1 (6.91) 27.4 (6.47) 0.661  Household size 7.0 (2.8) 6.9 (2.7) 7.0 (2.7) 6.8 (2.9) 0.893 6.1 (2.4) 6.1 (2.2) 6.1(2.3) 6.2 (2.7) 0.967 6.2 (2.2) 6.0 (2.3) 6.3 (2.5) 6.3 (2.8) 0.893  Number of children 4.2 (2.8) 3.8 (2.6) 4.1 (3.0) 4.1 (2.8) 0.364 3.1 (2.0) 3.4 (2.1) 3.6 (2.6) 3.4 (2.4) 0.069 3.0 (1.9) 2.7 (1.8) 3.0 (2.0) 2.6 (1.6) 0.002  Maternal literacy 78.7 77.8 80.2 76.9 0.934 84.9 81.2 85.0 81.5 0.765 84.6 91.7 89.6 88.4 0.124  Wealth index 2.12 (1.41) 2.20 (1.49) 2.40 (1.46) 2.05 (1.46) 0.007 2.72 (1.46) 2.49 (1.49) 2.88 (1.65) 2.44 (1.53) 0.823 2.87 (1.31) 3.06 (1.45) 2.88 (1.31) 3.35 (1.38) 0.355 . 2005 (n = 2009) . 2011 (n = 1068) . 2016 (n = 4142) . Outcome/variable . CENT (n = 548) . NGO (n = 541) . MUNI (n = 323) . ASSN (n = 597) . P-valuea . CENT (n = 451) . NGO (n = 241) . MUNI (n = 260) . ASSN (n = 376) . P-valuea . CENT (n = 1486) . NGO (n = 682) . MUNI (n = 653) . ASSN (n = 1321) . P-valuea . Child health  Weight-for-age −0.845 (1.06) −0.903 (1.05) −0.735 (1.14) −0.905 (1.05) 0.003 −0.700 (1.00) −0.821 (1.05) −0.762 (1.17) −0.789 (1.07) 0.002 −0.509 (1.36) −0.935 (1.52) −0.577 (1.27) −0.604 (1.43) 0.009  Length-for-age −1.668 (1.07) −1.797 (1.21) −1.622 (1.27) −1.583 (1.16) 0.011 −1.554 (1.07) −1.797 (1.21) −1.622 (1.27) −1.583 (1.16) 0.901 −0.954 (1.47) −1.087 (1.63) −1.137 (1.62) −0.822 (1.56) 0.048  Weight-for-length 0.158 (1.06) 0.277 (0.94) 0.171 (1.00) 0.124 (1.00) 0.415 0.226 (0.95) 0.251 (0.92) 0.220 (1.01) 0.155 (0.97) 0.001 −0.026 (1.35) −0.379 (1.53) 0.089 (1.41) −0.185 (1.52) 0.139  BMI 0.335 (1.08) 0.495 (0.96) 0.306 (1.01) 0.302 (1.04) 0.187 0.411 (0.96) 0.481 (0.90) 0.436 (1.00) 0.342 (0.96 0.701 −0.043 (1.40) −0.268 (1.47) 0.207 (1.52) −0.101 (1.56) 0.097 Vaccination  Measles vaccination 60.7 67.0 58.4 61.8 0.197 88.3 89.0 85.7 85.5 0.730 93.8 91.4 83.5 88.7 0.126  Polio Vaccination 78.2 83.1 71.1 77.8 0.028 97.1 98.8 97.2 97.7 0.525 87.8 90.5 83.3 92.7 <0.001  BCG Vaccination 80.4 81.9 78.0 81.7 0.863 87.2 94.2 89.7 87.5 0.322 94.8 93.3 86.5 90.0 0.086  DPT Vaccination 81.7 86.7 78.1 84.0 0.007 97.2 98.1 97.0 97.4 0.947 94.3 92.3 83.9 89.2 0.088  Complete Vaccination 58.1 64.9 57.1 60.0 0.160 84.1 86.2 80.2 82.0 0.665 92.7 90.5 78.7 87.7 0.057 Delivery care  Home delivery 60.0 50.6 51.4 57.9 0.622 39.4 27.6 37.8 35.8 0.574 39.6 8.3 20.6 13.7 <0.001 MCH facility delivery 5.1 2.8 8.5 3.8 0.687 5.0 14.1 14.2 11.9 0.081 7.5 20.3 14.3 21.0 <0.001  Skilled birth attendant 38.2 47.6 50.0 41.3 0.556 61.2 71.7 61.5 62.2 0.604 69.3 92.3 80.6 85.5 <0.001  Caesarean birth 7.0 8.8 9.7 6.5 0.313 11.4 13.1 7.7 13.9 0.196 8.0 22.1 13.4 12.0 0.008 Maternal/child care  Child weight control programme 16.0 22.7 10.1 21.3 0.141 29.6 51.4 12.1 30.7 0.032 39.5 58.4 38.3 46.0 0.049  4 antenatal care visits 82.1 86.6 85.9 86.7 0.412 89.6 94.3 86.5 94.7 0.025 84.4 93.9 92.6 92.7 0.015  Iron supplementation 22.2 16.8 14.3 19.6 0.391 37.0 27.8 26.8 33.0 0.074 94.3 88.1 84.1 91.1 <0.001  Maternal tetanus 84.8 94.9 87.4 89.7 0.055 93.7 94.8 94.1 91.7 0.677 83.4 89.1 83.7 83.9 0.171 Women's health  Pap smear done 56.9 36.3 48.0 42.7 <0.001 63.6 57.2 50.9 55.9 0.011 56.2 50.2 56.9 51.1 0.386  Mammogram done 2.7 2.3 3.1 1.6 0.701 4.8 3.5 3.5 3.7 0.363 3.3 1.7 3.8 1.7 0.025 Control variables  Maternal age at birth 27.0 (7.10) 25.6 (6.74 26.7 (7.36) 26.6 (7.04) 0.004 25.5 (7.1) 26.1 (6.76) 26.1 (7.14) 26.1 (6.73) 0.002 25.6 (6.88) 27.3 (0.66) 24.5 (6.72) 24.9 (6.35) 0.004  Maternal age 28.9 (7.11) 27.7 (6.87) 28.7 (7.45) 28.6 (7.09) 0.004 28.1 (7.41) 28.6 (7.04) 28.0 (7.13) 28.7 (6.95) 0.582 27.7 (7.04) 27.3 (7.04) 27.1 (6.91) 27.4 (6.47) 0.661  Household size 7.0 (2.8) 6.9 (2.7) 7.0 (2.7) 6.8 (2.9) 0.893 6.1 (2.4) 6.1 (2.2) 6.1(2.3) 6.2 (2.7) 0.967 6.2 (2.2) 6.0 (2.3) 6.3 (2.5) 6.3 (2.8) 0.893  Number of children 4.2 (2.8) 3.8 (2.6) 4.1 (3.0) 4.1 (2.8) 0.364 3.1 (2.0) 3.4 (2.1) 3.6 (2.6) 3.4 (2.4) 0.069 3.0 (1.9) 2.7 (1.8) 3.0 (2.0) 2.6 (1.6) 0.002  Maternal literacy 78.7 77.8 80.2 76.9 0.934 84.9 81.2 85.0 81.5 0.765 84.6 91.7 89.6 88.4 0.124  Wealth index 2.12 (1.41) 2.20 (1.49) 2.40 (1.46) 2.05 (1.46) 0.007 2.72 (1.46) 2.49 (1.49) 2.88 (1.65) 2.44 (1.53) 0.823 2.87 (1.31) 3.06 (1.45) 2.88 (1.31) 3.35 (1.38) 0.355 Note: All statistics have been generated by applying survey weights. Numbers reported are either the mean (SD) or a per cent. a Based on a year-specific regression of the outcome variable on governance types: yi= β0+β1NGO + β2ASSN + β3MUNI and a joint test that β1−β3 equal zero. Open in new tab Table 2 Characteristics of the study sample by survey wave and treatment status . 2005 (n = 2009) . 2011 (n = 1068) . 2016 (n = 4142) . Outcome/variable . CENT (n = 548) . NGO (n = 541) . MUNI (n = 323) . ASSN (n = 597) . P-valuea . CENT (n = 451) . NGO (n = 241) . MUNI (n = 260) . ASSN (n = 376) . P-valuea . CENT (n = 1486) . NGO (n = 682) . MUNI (n = 653) . ASSN (n = 1321) . P-valuea . Child health  Weight-for-age −0.845 (1.06) −0.903 (1.05) −0.735 (1.14) −0.905 (1.05) 0.003 −0.700 (1.00) −0.821 (1.05) −0.762 (1.17) −0.789 (1.07) 0.002 −0.509 (1.36) −0.935 (1.52) −0.577 (1.27) −0.604 (1.43) 0.009  Length-for-age −1.668 (1.07) −1.797 (1.21) −1.622 (1.27) −1.583 (1.16) 0.011 −1.554 (1.07) −1.797 (1.21) −1.622 (1.27) −1.583 (1.16) 0.901 −0.954 (1.47) −1.087 (1.63) −1.137 (1.62) −0.822 (1.56) 0.048  Weight-for-length 0.158 (1.06) 0.277 (0.94) 0.171 (1.00) 0.124 (1.00) 0.415 0.226 (0.95) 0.251 (0.92) 0.220 (1.01) 0.155 (0.97) 0.001 −0.026 (1.35) −0.379 (1.53) 0.089 (1.41) −0.185 (1.52) 0.139  BMI 0.335 (1.08) 0.495 (0.96) 0.306 (1.01) 0.302 (1.04) 0.187 0.411 (0.96) 0.481 (0.90) 0.436 (1.00) 0.342 (0.96 0.701 −0.043 (1.40) −0.268 (1.47) 0.207 (1.52) −0.101 (1.56) 0.097 Vaccination  Measles vaccination 60.7 67.0 58.4 61.8 0.197 88.3 89.0 85.7 85.5 0.730 93.8 91.4 83.5 88.7 0.126  Polio Vaccination 78.2 83.1 71.1 77.8 0.028 97.1 98.8 97.2 97.7 0.525 87.8 90.5 83.3 92.7 <0.001  BCG Vaccination 80.4 81.9 78.0 81.7 0.863 87.2 94.2 89.7 87.5 0.322 94.8 93.3 86.5 90.0 0.086  DPT Vaccination 81.7 86.7 78.1 84.0 0.007 97.2 98.1 97.0 97.4 0.947 94.3 92.3 83.9 89.2 0.088  Complete Vaccination 58.1 64.9 57.1 60.0 0.160 84.1 86.2 80.2 82.0 0.665 92.7 90.5 78.7 87.7 0.057 Delivery care  Home delivery 60.0 50.6 51.4 57.9 0.622 39.4 27.6 37.8 35.8 0.574 39.6 8.3 20.6 13.7 <0.001 MCH facility delivery 5.1 2.8 8.5 3.8 0.687 5.0 14.1 14.2 11.9 0.081 7.5 20.3 14.3 21.0 <0.001  Skilled birth attendant 38.2 47.6 50.0 41.3 0.556 61.2 71.7 61.5 62.2 0.604 69.3 92.3 80.6 85.5 <0.001  Caesarean birth 7.0 8.8 9.7 6.5 0.313 11.4 13.1 7.7 13.9 0.196 8.0 22.1 13.4 12.0 0.008 Maternal/child care  Child weight control programme 16.0 22.7 10.1 21.3 0.141 29.6 51.4 12.1 30.7 0.032 39.5 58.4 38.3 46.0 0.049  4 antenatal care visits 82.1 86.6 85.9 86.7 0.412 89.6 94.3 86.5 94.7 0.025 84.4 93.9 92.6 92.7 0.015  Iron supplementation 22.2 16.8 14.3 19.6 0.391 37.0 27.8 26.8 33.0 0.074 94.3 88.1 84.1 91.1 <0.001  Maternal tetanus 84.8 94.9 87.4 89.7 0.055 93.7 94.8 94.1 91.7 0.677 83.4 89.1 83.7 83.9 0.171 Women's health  Pap smear done 56.9 36.3 48.0 42.7 <0.001 63.6 57.2 50.9 55.9 0.011 56.2 50.2 56.9 51.1 0.386  Mammogram done 2.7 2.3 3.1 1.6 0.701 4.8 3.5 3.5 3.7 0.363 3.3 1.7 3.8 1.7 0.025 Control variables  Maternal age at birth 27.0 (7.10) 25.6 (6.74 26.7 (7.36) 26.6 (7.04) 0.004 25.5 (7.1) 26.1 (6.76) 26.1 (7.14) 26.1 (6.73) 0.002 25.6 (6.88) 27.3 (0.66) 24.5 (6.72) 24.9 (6.35) 0.004  Maternal age 28.9 (7.11) 27.7 (6.87) 28.7 (7.45) 28.6 (7.09) 0.004 28.1 (7.41) 28.6 (7.04) 28.0 (7.13) 28.7 (6.95) 0.582 27.7 (7.04) 27.3 (7.04) 27.1 (6.91) 27.4 (6.47) 0.661  Household size 7.0 (2.8) 6.9 (2.7) 7.0 (2.7) 6.8 (2.9) 0.893 6.1 (2.4) 6.1 (2.2) 6.1(2.3) 6.2 (2.7) 0.967 6.2 (2.2) 6.0 (2.3) 6.3 (2.5) 6.3 (2.8) 0.893  Number of children 4.2 (2.8) 3.8 (2.6) 4.1 (3.0) 4.1 (2.8) 0.364 3.1 (2.0) 3.4 (2.1) 3.6 (2.6) 3.4 (2.4) 0.069 3.0 (1.9) 2.7 (1.8) 3.0 (2.0) 2.6 (1.6) 0.002  Maternal literacy 78.7 77.8 80.2 76.9 0.934 84.9 81.2 85.0 81.5 0.765 84.6 91.7 89.6 88.4 0.124  Wealth index 2.12 (1.41) 2.20 (1.49) 2.40 (1.46) 2.05 (1.46) 0.007 2.72 (1.46) 2.49 (1.49) 2.88 (1.65) 2.44 (1.53) 0.823 2.87 (1.31) 3.06 (1.45) 2.88 (1.31) 3.35 (1.38) 0.355 . 2005 (n = 2009) . 2011 (n = 1068) . 2016 (n = 4142) . Outcome/variable . CENT (n = 548) . NGO (n = 541) . MUNI (n = 323) . ASSN (n = 597) . P-valuea . CENT (n = 451) . NGO (n = 241) . MUNI (n = 260) . ASSN (n = 376) . P-valuea . CENT (n = 1486) . NGO (n = 682) . MUNI (n = 653) . ASSN (n = 1321) . P-valuea . Child health  Weight-for-age −0.845 (1.06) −0.903 (1.05) −0.735 (1.14) −0.905 (1.05) 0.003 −0.700 (1.00) −0.821 (1.05) −0.762 (1.17) −0.789 (1.07) 0.002 −0.509 (1.36) −0.935 (1.52) −0.577 (1.27) −0.604 (1.43) 0.009  Length-for-age −1.668 (1.07) −1.797 (1.21) −1.622 (1.27) −1.583 (1.16) 0.011 −1.554 (1.07) −1.797 (1.21) −1.622 (1.27) −1.583 (1.16) 0.901 −0.954 (1.47) −1.087 (1.63) −1.137 (1.62) −0.822 (1.56) 0.048  Weight-for-length 0.158 (1.06) 0.277 (0.94) 0.171 (1.00) 0.124 (1.00) 0.415 0.226 (0.95) 0.251 (0.92) 0.220 (1.01) 0.155 (0.97) 0.001 −0.026 (1.35) −0.379 (1.53) 0.089 (1.41) −0.185 (1.52) 0.139  BMI 0.335 (1.08) 0.495 (0.96) 0.306 (1.01) 0.302 (1.04) 0.187 0.411 (0.96) 0.481 (0.90) 0.436 (1.00) 0.342 (0.96 0.701 −0.043 (1.40) −0.268 (1.47) 0.207 (1.52) −0.101 (1.56) 0.097 Vaccination  Measles vaccination 60.7 67.0 58.4 61.8 0.197 88.3 89.0 85.7 85.5 0.730 93.8 91.4 83.5 88.7 0.126  Polio Vaccination 78.2 83.1 71.1 77.8 0.028 97.1 98.8 97.2 97.7 0.525 87.8 90.5 83.3 92.7 <0.001  BCG Vaccination 80.4 81.9 78.0 81.7 0.863 87.2 94.2 89.7 87.5 0.322 94.8 93.3 86.5 90.0 0.086  DPT Vaccination 81.7 86.7 78.1 84.0 0.007 97.2 98.1 97.0 97.4 0.947 94.3 92.3 83.9 89.2 0.088  Complete Vaccination 58.1 64.9 57.1 60.0 0.160 84.1 86.2 80.2 82.0 0.665 92.7 90.5 78.7 87.7 0.057 Delivery care  Home delivery 60.0 50.6 51.4 57.9 0.622 39.4 27.6 37.8 35.8 0.574 39.6 8.3 20.6 13.7 <0.001 MCH facility delivery 5.1 2.8 8.5 3.8 0.687 5.0 14.1 14.2 11.9 0.081 7.5 20.3 14.3 21.0 <0.001  Skilled birth attendant 38.2 47.6 50.0 41.3 0.556 61.2 71.7 61.5 62.2 0.604 69.3 92.3 80.6 85.5 <0.001  Caesarean birth 7.0 8.8 9.7 6.5 0.313 11.4 13.1 7.7 13.9 0.196 8.0 22.1 13.4 12.0 0.008 Maternal/child care  Child weight control programme 16.0 22.7 10.1 21.3 0.141 29.6 51.4 12.1 30.7 0.032 39.5 58.4 38.3 46.0 0.049  4 antenatal care visits 82.1 86.6 85.9 86.7 0.412 89.6 94.3 86.5 94.7 0.025 84.4 93.9 92.6 92.7 0.015  Iron supplementation 22.2 16.8 14.3 19.6 0.391 37.0 27.8 26.8 33.0 0.074 94.3 88.1 84.1 91.1 <0.001  Maternal tetanus 84.8 94.9 87.4 89.7 0.055 93.7 94.8 94.1 91.7 0.677 83.4 89.1 83.7 83.9 0.171 Women's health  Pap smear done 56.9 36.3 48.0 42.7 <0.001 63.6 57.2 50.9 55.9 0.011 56.2 50.2 56.9 51.1 0.386  Mammogram done 2.7 2.3 3.1 1.6 0.701 4.8 3.5 3.5 3.7 0.363 3.3 1.7 3.8 1.7 0.025 Control variables  Maternal age at birth 27.0 (7.10) 25.6 (6.74 26.7 (7.36) 26.6 (7.04) 0.004 25.5 (7.1) 26.1 (6.76) 26.1 (7.14) 26.1 (6.73) 0.002 25.6 (6.88) 27.3 (0.66) 24.5 (6.72) 24.9 (6.35) 0.004  Maternal age 28.9 (7.11) 27.7 (6.87) 28.7 (7.45) 28.6 (7.09) 0.004 28.1 (7.41) 28.6 (7.04) 28.0 (7.13) 28.7 (6.95) 0.582 27.7 (7.04) 27.3 (7.04) 27.1 (6.91) 27.4 (6.47) 0.661  Household size 7.0 (2.8) 6.9 (2.7) 7.0 (2.7) 6.8 (2.9) 0.893 6.1 (2.4) 6.1 (2.2) 6.1(2.3) 6.2 (2.7) 0.967 6.2 (2.2) 6.0 (2.3) 6.3 (2.5) 6.3 (2.8) 0.893  Number of children 4.2 (2.8) 3.8 (2.6) 4.1 (3.0) 4.1 (2.8) 0.364 3.1 (2.0) 3.4 (2.1) 3.6 (2.6) 3.4 (2.4) 0.069 3.0 (1.9) 2.7 (1.8) 3.0 (2.0) 2.6 (1.6) 0.002  Maternal literacy 78.7 77.8 80.2 76.9 0.934 84.9 81.2 85.0 81.5 0.765 84.6 91.7 89.6 88.4 0.124  Wealth index 2.12 (1.41) 2.20 (1.49) 2.40 (1.46) 2.05 (1.46) 0.007 2.72 (1.46) 2.49 (1.49) 2.88 (1.65) 2.44 (1.53) 0.823 2.87 (1.31) 3.06 (1.45) 2.88 (1.31) 3.35 (1.38) 0.355 Note: All statistics have been generated by applying survey weights. Numbers reported are either the mean (SD) or a per cent. a Based on a year-specific regression of the outcome variable on governance types: yi= β0+β1NGO + β2ASSN + β3MUNI and a joint test that β1−β3 equal zero. Open in new tab Nationally, we find evidence of sizeable improvement between 2005 and 2016. Of the 19 measures in Table 3, significant improvement was recorded in just over half, as indicated by the bolded coefficients in the year column. Concerning child health, both weight-for-age and length-for-age increased, but weight-for-age did not increase proportionately. As a result, the z-scores for both weight-for-length and BMI decreased somewhat. Taken together, this still indicates less stunting and somewhat better nutrition. All vaccination measures increased significantly over time. Concerning delivery care, home delivery declined and the presence of skilled birth attendant increased. In the maternal/child care domain, enrolment in a weight monitoring programme and iron supplementation increased. Pap smear prevalence also increased in 2011. These changes resulted in improvement in three of the four domain scores (Table 4). Table 3 Estimated effect of three decentralization types on child health and maternal and child healthcare services Outcomea . n . Year . Year . NGO . MUNI . ASSN . Treatment . Year*Treatment . . Treatment . Year*Treatment . . Treatment . Year*Treatment . . β (95% CI) . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . Child health  Weight-for-age 5043 2005 −0.057 (−0.30, 0.18) 0.079 (−0.17, 0.33) −0.056 (−0.23, 0.12) 2011 −0.00 (−0.21, 0.20) −0.010 (−0.30, 0.28) −0.010 −0.083 (−0.48, 0.12) −0.182 0.021 (−0.24, 0.29) 0.021 2016 0.142 (−0.11, 0.39) −0.166 (−0.51, 0.17) −0.165 −0.196 (−0.57, 0.18) −0.195 −0.184 (−0.52, 0.15) −0.184  Length-for-age 5014 2005 −0.245 (−0.67, 0.19) 0.158 (−0.20, 0.52) −0.053 (−0.30, 0.20) 2011 −0.102 (−0.28, 0.07) 0.080 (−0.19, 0.35) 0.079 −0.275 (−0.66, 0.10) −0.275 0.103 (−0.18, 0.38) 0.102 2016 0.449 (0.05, 0.83) 0.187 (−0.35, 0.73) 0.186 −0.258 (−0.78, 0.27) −0.258 −0.079 (−0.51, 0.35) −0.079  Weight-for-length 4990 2005 0.118 (−0.07, 0.30) 0.002 (−0.21, 0.22) −0.030 (−0.23, 0.17) 2011 0.043 (−0.21, 0.30) −0.082 (−0.42, 0.25) −.0821 −0.033 (−0.35, 0.29) −0.033 −0.029 (−0.34, 0.28) −0.029 2016 −0.197 (−0.37, −0.01) −0.344 (−0.65, −0.02) −0.343 −0.046 (−0.43, 0.37) −0.046 −0.180 (−0.53, 0.17) −0.179  BMI 4991 2005 0.162 (−0.05, 0.37) −0.032 (−0.25, 0.19) −0.026 (−0.24, 0.19) 2011 0.075 (−0.20, 0.35) −0.090 (−0.45, 0.27) −0.090 0.038 (−0.30, 0.38) 0.038 −0.039 (−0.37, 0.29) −0.039 2016 −0.274 (−0.46, −0.08) −0.367 (−0.70, −0.02) −0.367 0.028 (−0.39, 0.44) 0.027 −0.149 (−0.50, 0.20) −0.148 Vaccination  Measles vaccination 5911 2005 0.315 (0.06, 0.56) −0.053 (−0.52, 0.42) 0.056 (−0.22, 0.33) 2011 1.750 (1.20, 2.29) −0.297 (−1.28, 0.69) −0.070 −0.209 (−0.92, 0.51) −0.016 −0.342 (−0.99, 0.30) −0.045 2016 2.368 (1.63, 3.09) −0.541 (−1.65, 0.57) −0.087 −0.880 (−1.79, 0.03) −0.072 −0.581 (−1.41, 0.24) −0.053  Polio vaccination 6257 2005 0.327 (0.04, 0.61) −0.366 (−0.79, 0.05) −0.042 (−0.29, 0.21) 2011 2.011 (1.18, 2.83) 0.830 (−0.91, 2.57) −0.025 0.693 (−0.48, 1.86) 0.080 0.559 (−0.59, 1.71) 0.022 2016 1.438 (0.89, 1.98) −0.461 (−1.82, 0.25) −0.061 −0.511 (−1.31, 0.29) −0.010 0.076 (−0.57, 0.72) 0.009  BCG vaccination 6224 2005 0.167 (−0.11, 0.44) −0.066 (−0.74, 0.60) 0.132 (−0.23, 0.49) 2011 0.623 (0.22, 1.02) 0.666 (−0.60, 1.94) 0.041 0.373 (−0.33, 1.07) 0.041 −0.130 (−0.75, 0.49) −0.020 2016 1.502 (0.91, 2.09) −0.379 (−1.53, 0.77) −0.038 −0.804 (−1.75, 0.14) −0.058 −0.725 (−1.50, 0.05) −0.062  DPT vaccination 5999 2005 0.370 (0.06, 0.67) −0.221 (−0.50, 0.06) 0.135 (−0.17, 0.44) 2011 2.089 (1.56, 2.61) −0.030 (−1.41, 1.35) −0.041 0.124 (−0.97, 1.22) 0.032 −0.079 (−0.94, 0.78) −0.017 2016 1.236 (0.52, 1.94) −0.598 (−1.8, 0.61) −0.064 −0.742 (−1.62, 0.14) −0.050 −0.741 (−1.63, 0.15) −0.065  Complete vaccination 5624 2005 0.329 (0.71, 0.58) 0.0002 (−0.48, 0.48) 0.088 (−0.20, 0.37) 2011 1.477 (1.11, 1.84) −0.213 (−1.11, 0.69) −0.062 −0.296 (−0.92, 0.33) −0.042 −0.272 (−0.79, 0.25) −0.046 2016 2.307 (1.59, 3.02) 0.522 (−1.68, 0.63) −0.091 −1.067 (−1.97, −0.16) −0.114 −0.571 (−1.40, 0.26) −0.062 Delivery care  Home delivery 6413 2005 −0.384 (−0.94, 0.17) −0.292 (−1.05, 0.46) −0.085 (−0.54, 0.37) 2011 −0.422 (−0.67, −0.17) −0.460 (−1.16, 0.24) −0.076 0.106 (−0.55, 0.76) 0.021 −0.344 (−0.85, 0.16) −0.063 2016 −0.656 (−0.96, −0.34) −1.174 (−1.72, −0.62) −0.156 −0.384 (−0.95, 0.19) −0.060 −0.959 (−1.658, −0.33) −0.154  MCH facility delivery 6413 2005 −0.702 (−2.43, 1.02) 0.507 (−0.96, 1.97) −0.368 (−2.31, 1.57) 2011 −0.137 (−1.21, 0.93) 1.837 (0.11, 3.56) 0.117 0.678 (−1.04, 2.39) 0.061 1.295 (−0.91, 3.50) 0.083 2016 0.277 (−0.55, 1.10) 1.666 (−0.05, 3.38) 0.125 0.104 (−1.37, 1.50) 0.019 1.504 (−0.57, 3.58) 0.139  Skilled birth attendant 5513 2005 0.384 (−0.22, 0.99) 0.446 (−0.30, 1.19) 0.137 (−0.31, 0.59) 2011 0.569 (0.33, 0.80) 0.361 (−0.34, 1.06) 0.058 −0.353 (−0.95, 0.24) −0.070 0.131 (−0.40, 0.66) 0.023 2016 1.134 (0.83, 1.43) 0.636 (0.06, 1.20) 0.069 −0.173 (−0.74, 0.40) −0.043 0.368 (−0.18, 0.92) 0.053  Caesarean birth 6096 2005 0.229 (−0.36, 0.81) 0.288 (−0.28, 0.86) −0.104 (−0.74, 0.53) 2011 0.262 (−0.32, 0.85) −0.009 (−0.96, 0.94) 0.002 −0.695 (−1.38, −0.00) −0.059 0.331 (−0.39, 1.05) 0.031 2016 −0.423 (−1.19, 0.35) 0.903 (0.04, 1.76) 0.081 0.556 (−0.34, 1.46) 0.042 0.489 (−0.38, 1.35) 0.033 Maternal/child care  Child in weight control programme 6252 2005 0.407 (−0.44, 1.26) −0.506 (−1.48, 0.47) 0.331 (−0.51, 1.17) 2011 0.762 (−0.14, 1.66) 0.546 (−0.63, 1.72) 0.154 −0.459 (−1.54, 0.62) −0.095 −0.270 (−1.34, 0.80) −0.036 2016 1.112 (0.58, 1.64) 0.306 (−0.52, 1.14) 0.106 0.42 (−0.55, 1.39) 0.038 0.053 (−0.62, 0.73) 0.040  4+ antenatal visits 4452 2005 0.314 (−0.00, 0.63) 0.240 (−0.29, 0.77) 0.353 (−0.10, 0.80) 2011 0.330 (−0.16, 0.82) 0.365 (−0.89, 1.62) 0.015 −0.483 (−1.18, 0.21) −0.050 0.470 (−0.34, 1.28) 0.019 2016 −0.231 (−0.72, 0.26) 0.702 (−0.14, 1.55) 0.065 0.847 (0.17, 1.51) 0.077 0.232 (−0.40, 0.86) 0.029  Iron supplementation 5374 2005 −0.320 (−1.00, 0.36) −0.504 (−1.25, 0.24) −0.117 (−0.72, 0.48) 2011 0.682 (1.79, 1.85) −0.081 (−0.71, 0.55) −0.035 0.0464 (−1.71, 0.80) −0.021 −0.048 (−0.68, 0.58) −0.017 2016 4.213 (3.35, 5.07) −0.766 (−1.85, 0.31) −0.031 −0.911 (−1.93, 0.11) −0.047 −0.693 (−1.75, 0.37) −0.034  Maternal tetanus 4554 2005 1.275 (0.41, 2.13) 0.210 (−0.73, 1.15) 0.048 (−0.17, 1.13) 2011 0.958 (−0.42, 1.49) −1.028 (−2.74, 0.69) −0.088 −0.180 (−1.36, 1.00) −0.022 −0.756 (−1.58, 0.07) −0.069 2016 −0.726 (−1.85, 0.40) −0.235 (−1.27, 0.80) 0.047 0.354 (−0.86, 1.56) 0.068 −0.020 (−1.19, 1.15) 0.026 Women's health  Pap smear done 12715 2005 −1.112 (−1.69, −0.52) −0.595 (−1.14, −0.50) −0.887 (−1.70, −0.06) 2011 0.515 (0.22, 0.80) 0.710 (0.25, 1.16) 0.110 −0.200 (−0.71, 0.31) −0.025 0.425 (−0.16, 1.01) 0.067 2016 0.091 (−0.41, 0.59) 1.139 (0.28, 1.99) 0.171 0.942 (0.09, 1.79) 0.138 0.876 (0.02, 1.72) 0.131  Mammogram done 12717 2005 −0.023 (−0.80, 0.76) −0.005 (−1.00, 0.99) −0.632 (−1.62, 0.36) 2011 0.398 (−0.25, 1.04) −0.283 (−1.08, 0.54) −0.010 −0.458 (−1.43, 0.51) −0.015 0.189 (−0.72, 1.10) −0.000 2016 −0.042 (−0.69, 0.60) −0.462 (−1.54, 0.61) −0.010 0.334 (−0.72, 1.39) 0.011 −0.236 (−1.42, 0.95) −0.002 Outcomea . n . Year . Year . NGO . MUNI . ASSN . Treatment . Year*Treatment . . Treatment . Year*Treatment . . Treatment . Year*Treatment . . β (95% CI) . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . Child health  Weight-for-age 5043 2005 −0.057 (−0.30, 0.18) 0.079 (−0.17, 0.33) −0.056 (−0.23, 0.12) 2011 −0.00 (−0.21, 0.20) −0.010 (−0.30, 0.28) −0.010 −0.083 (−0.48, 0.12) −0.182 0.021 (−0.24, 0.29) 0.021 2016 0.142 (−0.11, 0.39) −0.166 (−0.51, 0.17) −0.165 −0.196 (−0.57, 0.18) −0.195 −0.184 (−0.52, 0.15) −0.184  Length-for-age 5014 2005 −0.245 (−0.67, 0.19) 0.158 (−0.20, 0.52) −0.053 (−0.30, 0.20) 2011 −0.102 (−0.28, 0.07) 0.080 (−0.19, 0.35) 0.079 −0.275 (−0.66, 0.10) −0.275 0.103 (−0.18, 0.38) 0.102 2016 0.449 (0.05, 0.83) 0.187 (−0.35, 0.73) 0.186 −0.258 (−0.78, 0.27) −0.258 −0.079 (−0.51, 0.35) −0.079  Weight-for-length 4990 2005 0.118 (−0.07, 0.30) 0.002 (−0.21, 0.22) −0.030 (−0.23, 0.17) 2011 0.043 (−0.21, 0.30) −0.082 (−0.42, 0.25) −.0821 −0.033 (−0.35, 0.29) −0.033 −0.029 (−0.34, 0.28) −0.029 2016 −0.197 (−0.37, −0.01) −0.344 (−0.65, −0.02) −0.343 −0.046 (−0.43, 0.37) −0.046 −0.180 (−0.53, 0.17) −0.179  BMI 4991 2005 0.162 (−0.05, 0.37) −0.032 (−0.25, 0.19) −0.026 (−0.24, 0.19) 2011 0.075 (−0.20, 0.35) −0.090 (−0.45, 0.27) −0.090 0.038 (−0.30, 0.38) 0.038 −0.039 (−0.37, 0.29) −0.039 2016 −0.274 (−0.46, −0.08) −0.367 (−0.70, −0.02) −0.367 0.028 (−0.39, 0.44) 0.027 −0.149 (−0.50, 0.20) −0.148 Vaccination  Measles vaccination 5911 2005 0.315 (0.06, 0.56) −0.053 (−0.52, 0.42) 0.056 (−0.22, 0.33) 2011 1.750 (1.20, 2.29) −0.297 (−1.28, 0.69) −0.070 −0.209 (−0.92, 0.51) −0.016 −0.342 (−0.99, 0.30) −0.045 2016 2.368 (1.63, 3.09) −0.541 (−1.65, 0.57) −0.087 −0.880 (−1.79, 0.03) −0.072 −0.581 (−1.41, 0.24) −0.053  Polio vaccination 6257 2005 0.327 (0.04, 0.61) −0.366 (−0.79, 0.05) −0.042 (−0.29, 0.21) 2011 2.011 (1.18, 2.83) 0.830 (−0.91, 2.57) −0.025 0.693 (−0.48, 1.86) 0.080 0.559 (−0.59, 1.71) 0.022 2016 1.438 (0.89, 1.98) −0.461 (−1.82, 0.25) −0.061 −0.511 (−1.31, 0.29) −0.010 0.076 (−0.57, 0.72) 0.009  BCG vaccination 6224 2005 0.167 (−0.11, 0.44) −0.066 (−0.74, 0.60) 0.132 (−0.23, 0.49) 2011 0.623 (0.22, 1.02) 0.666 (−0.60, 1.94) 0.041 0.373 (−0.33, 1.07) 0.041 −0.130 (−0.75, 0.49) −0.020 2016 1.502 (0.91, 2.09) −0.379 (−1.53, 0.77) −0.038 −0.804 (−1.75, 0.14) −0.058 −0.725 (−1.50, 0.05) −0.062  DPT vaccination 5999 2005 0.370 (0.06, 0.67) −0.221 (−0.50, 0.06) 0.135 (−0.17, 0.44) 2011 2.089 (1.56, 2.61) −0.030 (−1.41, 1.35) −0.041 0.124 (−0.97, 1.22) 0.032 −0.079 (−0.94, 0.78) −0.017 2016 1.236 (0.52, 1.94) −0.598 (−1.8, 0.61) −0.064 −0.742 (−1.62, 0.14) −0.050 −0.741 (−1.63, 0.15) −0.065  Complete vaccination 5624 2005 0.329 (0.71, 0.58) 0.0002 (−0.48, 0.48) 0.088 (−0.20, 0.37) 2011 1.477 (1.11, 1.84) −0.213 (−1.11, 0.69) −0.062 −0.296 (−0.92, 0.33) −0.042 −0.272 (−0.79, 0.25) −0.046 2016 2.307 (1.59, 3.02) 0.522 (−1.68, 0.63) −0.091 −1.067 (−1.97, −0.16) −0.114 −0.571 (−1.40, 0.26) −0.062 Delivery care  Home delivery 6413 2005 −0.384 (−0.94, 0.17) −0.292 (−1.05, 0.46) −0.085 (−0.54, 0.37) 2011 −0.422 (−0.67, −0.17) −0.460 (−1.16, 0.24) −0.076 0.106 (−0.55, 0.76) 0.021 −0.344 (−0.85, 0.16) −0.063 2016 −0.656 (−0.96, −0.34) −1.174 (−1.72, −0.62) −0.156 −0.384 (−0.95, 0.19) −0.060 −0.959 (−1.658, −0.33) −0.154  MCH facility delivery 6413 2005 −0.702 (−2.43, 1.02) 0.507 (−0.96, 1.97) −0.368 (−2.31, 1.57) 2011 −0.137 (−1.21, 0.93) 1.837 (0.11, 3.56) 0.117 0.678 (−1.04, 2.39) 0.061 1.295 (−0.91, 3.50) 0.083 2016 0.277 (−0.55, 1.10) 1.666 (−0.05, 3.38) 0.125 0.104 (−1.37, 1.50) 0.019 1.504 (−0.57, 3.58) 0.139  Skilled birth attendant 5513 2005 0.384 (−0.22, 0.99) 0.446 (−0.30, 1.19) 0.137 (−0.31, 0.59) 2011 0.569 (0.33, 0.80) 0.361 (−0.34, 1.06) 0.058 −0.353 (−0.95, 0.24) −0.070 0.131 (−0.40, 0.66) 0.023 2016 1.134 (0.83, 1.43) 0.636 (0.06, 1.20) 0.069 −0.173 (−0.74, 0.40) −0.043 0.368 (−0.18, 0.92) 0.053  Caesarean birth 6096 2005 0.229 (−0.36, 0.81) 0.288 (−0.28, 0.86) −0.104 (−0.74, 0.53) 2011 0.262 (−0.32, 0.85) −0.009 (−0.96, 0.94) 0.002 −0.695 (−1.38, −0.00) −0.059 0.331 (−0.39, 1.05) 0.031 2016 −0.423 (−1.19, 0.35) 0.903 (0.04, 1.76) 0.081 0.556 (−0.34, 1.46) 0.042 0.489 (−0.38, 1.35) 0.033 Maternal/child care  Child in weight control programme 6252 2005 0.407 (−0.44, 1.26) −0.506 (−1.48, 0.47) 0.331 (−0.51, 1.17) 2011 0.762 (−0.14, 1.66) 0.546 (−0.63, 1.72) 0.154 −0.459 (−1.54, 0.62) −0.095 −0.270 (−1.34, 0.80) −0.036 2016 1.112 (0.58, 1.64) 0.306 (−0.52, 1.14) 0.106 0.42 (−0.55, 1.39) 0.038 0.053 (−0.62, 0.73) 0.040  4+ antenatal visits 4452 2005 0.314 (−0.00, 0.63) 0.240 (−0.29, 0.77) 0.353 (−0.10, 0.80) 2011 0.330 (−0.16, 0.82) 0.365 (−0.89, 1.62) 0.015 −0.483 (−1.18, 0.21) −0.050 0.470 (−0.34, 1.28) 0.019 2016 −0.231 (−0.72, 0.26) 0.702 (−0.14, 1.55) 0.065 0.847 (0.17, 1.51) 0.077 0.232 (−0.40, 0.86) 0.029  Iron supplementation 5374 2005 −0.320 (−1.00, 0.36) −0.504 (−1.25, 0.24) −0.117 (−0.72, 0.48) 2011 0.682 (1.79, 1.85) −0.081 (−0.71, 0.55) −0.035 0.0464 (−1.71, 0.80) −0.021 −0.048 (−0.68, 0.58) −0.017 2016 4.213 (3.35, 5.07) −0.766 (−1.85, 0.31) −0.031 −0.911 (−1.93, 0.11) −0.047 −0.693 (−1.75, 0.37) −0.034  Maternal tetanus 4554 2005 1.275 (0.41, 2.13) 0.210 (−0.73, 1.15) 0.048 (−0.17, 1.13) 2011 0.958 (−0.42, 1.49) −1.028 (−2.74, 0.69) −0.088 −0.180 (−1.36, 1.00) −0.022 −0.756 (−1.58, 0.07) −0.069 2016 −0.726 (−1.85, 0.40) −0.235 (−1.27, 0.80) 0.047 0.354 (−0.86, 1.56) 0.068 −0.020 (−1.19, 1.15) 0.026 Women's health  Pap smear done 12715 2005 −1.112 (−1.69, −0.52) −0.595 (−1.14, −0.50) −0.887 (−1.70, −0.06) 2011 0.515 (0.22, 0.80) 0.710 (0.25, 1.16) 0.110 −0.200 (−0.71, 0.31) −0.025 0.425 (−0.16, 1.01) 0.067 2016 0.091 (−0.41, 0.59) 1.139 (0.28, 1.99) 0.171 0.942 (0.09, 1.79) 0.138 0.876 (0.02, 1.72) 0.131  Mammogram done 12717 2005 −0.023 (−0.80, 0.76) −0.005 (−1.00, 0.99) −0.632 (−1.62, 0.36) 2011 0.398 (−0.25, 1.04) −0.283 (−1.08, 0.54) −0.010 −0.458 (−1.43, 0.51) −0.015 0.189 (−0.72, 1.10) −0.000 2016 −0.042 (−0.69, 0.60) −0.462 (−1.54, 0.61) −0.010 0.334 (−0.72, 1.39) 0.011 −0.236 (−1.42, 0.95) −0.002 Significant improvement was recorded in just over half, as indicated by the bolded coefficients in the year column. a All results are from models with maternal age, maternal age squared, maternal literacy, number of household members, number of children in the household and a household wealth index. b The difference-in-difference (DiD) was calculated using the margins command and report the difference in means or proportions between treatment and control (NGO or ASSN or MUNI vs CENT) at two different time points (2005 vs 2011 and 2005 vs 2016). Models for binary variables (e.g. vaccination, delivery care, maternal/child care outcomes) display the difference in probabilities as a per cent rather than a proportion. Open in new tab Table 3 Estimated effect of three decentralization types on child health and maternal and child healthcare services Outcomea . n . Year . Year . NGO . MUNI . ASSN . Treatment . Year*Treatment . . Treatment . Year*Treatment . . Treatment . Year*Treatment . . β (95% CI) . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . Child health  Weight-for-age 5043 2005 −0.057 (−0.30, 0.18) 0.079 (−0.17, 0.33) −0.056 (−0.23, 0.12) 2011 −0.00 (−0.21, 0.20) −0.010 (−0.30, 0.28) −0.010 −0.083 (−0.48, 0.12) −0.182 0.021 (−0.24, 0.29) 0.021 2016 0.142 (−0.11, 0.39) −0.166 (−0.51, 0.17) −0.165 −0.196 (−0.57, 0.18) −0.195 −0.184 (−0.52, 0.15) −0.184  Length-for-age 5014 2005 −0.245 (−0.67, 0.19) 0.158 (−0.20, 0.52) −0.053 (−0.30, 0.20) 2011 −0.102 (−0.28, 0.07) 0.080 (−0.19, 0.35) 0.079 −0.275 (−0.66, 0.10) −0.275 0.103 (−0.18, 0.38) 0.102 2016 0.449 (0.05, 0.83) 0.187 (−0.35, 0.73) 0.186 −0.258 (−0.78, 0.27) −0.258 −0.079 (−0.51, 0.35) −0.079  Weight-for-length 4990 2005 0.118 (−0.07, 0.30) 0.002 (−0.21, 0.22) −0.030 (−0.23, 0.17) 2011 0.043 (−0.21, 0.30) −0.082 (−0.42, 0.25) −.0821 −0.033 (−0.35, 0.29) −0.033 −0.029 (−0.34, 0.28) −0.029 2016 −0.197 (−0.37, −0.01) −0.344 (−0.65, −0.02) −0.343 −0.046 (−0.43, 0.37) −0.046 −0.180 (−0.53, 0.17) −0.179  BMI 4991 2005 0.162 (−0.05, 0.37) −0.032 (−0.25, 0.19) −0.026 (−0.24, 0.19) 2011 0.075 (−0.20, 0.35) −0.090 (−0.45, 0.27) −0.090 0.038 (−0.30, 0.38) 0.038 −0.039 (−0.37, 0.29) −0.039 2016 −0.274 (−0.46, −0.08) −0.367 (−0.70, −0.02) −0.367 0.028 (−0.39, 0.44) 0.027 −0.149 (−0.50, 0.20) −0.148 Vaccination  Measles vaccination 5911 2005 0.315 (0.06, 0.56) −0.053 (−0.52, 0.42) 0.056 (−0.22, 0.33) 2011 1.750 (1.20, 2.29) −0.297 (−1.28, 0.69) −0.070 −0.209 (−0.92, 0.51) −0.016 −0.342 (−0.99, 0.30) −0.045 2016 2.368 (1.63, 3.09) −0.541 (−1.65, 0.57) −0.087 −0.880 (−1.79, 0.03) −0.072 −0.581 (−1.41, 0.24) −0.053  Polio vaccination 6257 2005 0.327 (0.04, 0.61) −0.366 (−0.79, 0.05) −0.042 (−0.29, 0.21) 2011 2.011 (1.18, 2.83) 0.830 (−0.91, 2.57) −0.025 0.693 (−0.48, 1.86) 0.080 0.559 (−0.59, 1.71) 0.022 2016 1.438 (0.89, 1.98) −0.461 (−1.82, 0.25) −0.061 −0.511 (−1.31, 0.29) −0.010 0.076 (−0.57, 0.72) 0.009  BCG vaccination 6224 2005 0.167 (−0.11, 0.44) −0.066 (−0.74, 0.60) 0.132 (−0.23, 0.49) 2011 0.623 (0.22, 1.02) 0.666 (−0.60, 1.94) 0.041 0.373 (−0.33, 1.07) 0.041 −0.130 (−0.75, 0.49) −0.020 2016 1.502 (0.91, 2.09) −0.379 (−1.53, 0.77) −0.038 −0.804 (−1.75, 0.14) −0.058 −0.725 (−1.50, 0.05) −0.062  DPT vaccination 5999 2005 0.370 (0.06, 0.67) −0.221 (−0.50, 0.06) 0.135 (−0.17, 0.44) 2011 2.089 (1.56, 2.61) −0.030 (−1.41, 1.35) −0.041 0.124 (−0.97, 1.22) 0.032 −0.079 (−0.94, 0.78) −0.017 2016 1.236 (0.52, 1.94) −0.598 (−1.8, 0.61) −0.064 −0.742 (−1.62, 0.14) −0.050 −0.741 (−1.63, 0.15) −0.065  Complete vaccination 5624 2005 0.329 (0.71, 0.58) 0.0002 (−0.48, 0.48) 0.088 (−0.20, 0.37) 2011 1.477 (1.11, 1.84) −0.213 (−1.11, 0.69) −0.062 −0.296 (−0.92, 0.33) −0.042 −0.272 (−0.79, 0.25) −0.046 2016 2.307 (1.59, 3.02) 0.522 (−1.68, 0.63) −0.091 −1.067 (−1.97, −0.16) −0.114 −0.571 (−1.40, 0.26) −0.062 Delivery care  Home delivery 6413 2005 −0.384 (−0.94, 0.17) −0.292 (−1.05, 0.46) −0.085 (−0.54, 0.37) 2011 −0.422 (−0.67, −0.17) −0.460 (−1.16, 0.24) −0.076 0.106 (−0.55, 0.76) 0.021 −0.344 (−0.85, 0.16) −0.063 2016 −0.656 (−0.96, −0.34) −1.174 (−1.72, −0.62) −0.156 −0.384 (−0.95, 0.19) −0.060 −0.959 (−1.658, −0.33) −0.154  MCH facility delivery 6413 2005 −0.702 (−2.43, 1.02) 0.507 (−0.96, 1.97) −0.368 (−2.31, 1.57) 2011 −0.137 (−1.21, 0.93) 1.837 (0.11, 3.56) 0.117 0.678 (−1.04, 2.39) 0.061 1.295 (−0.91, 3.50) 0.083 2016 0.277 (−0.55, 1.10) 1.666 (−0.05, 3.38) 0.125 0.104 (−1.37, 1.50) 0.019 1.504 (−0.57, 3.58) 0.139  Skilled birth attendant 5513 2005 0.384 (−0.22, 0.99) 0.446 (−0.30, 1.19) 0.137 (−0.31, 0.59) 2011 0.569 (0.33, 0.80) 0.361 (−0.34, 1.06) 0.058 −0.353 (−0.95, 0.24) −0.070 0.131 (−0.40, 0.66) 0.023 2016 1.134 (0.83, 1.43) 0.636 (0.06, 1.20) 0.069 −0.173 (−0.74, 0.40) −0.043 0.368 (−0.18, 0.92) 0.053  Caesarean birth 6096 2005 0.229 (−0.36, 0.81) 0.288 (−0.28, 0.86) −0.104 (−0.74, 0.53) 2011 0.262 (−0.32, 0.85) −0.009 (−0.96, 0.94) 0.002 −0.695 (−1.38, −0.00) −0.059 0.331 (−0.39, 1.05) 0.031 2016 −0.423 (−1.19, 0.35) 0.903 (0.04, 1.76) 0.081 0.556 (−0.34, 1.46) 0.042 0.489 (−0.38, 1.35) 0.033 Maternal/child care  Child in weight control programme 6252 2005 0.407 (−0.44, 1.26) −0.506 (−1.48, 0.47) 0.331 (−0.51, 1.17) 2011 0.762 (−0.14, 1.66) 0.546 (−0.63, 1.72) 0.154 −0.459 (−1.54, 0.62) −0.095 −0.270 (−1.34, 0.80) −0.036 2016 1.112 (0.58, 1.64) 0.306 (−0.52, 1.14) 0.106 0.42 (−0.55, 1.39) 0.038 0.053 (−0.62, 0.73) 0.040  4+ antenatal visits 4452 2005 0.314 (−0.00, 0.63) 0.240 (−0.29, 0.77) 0.353 (−0.10, 0.80) 2011 0.330 (−0.16, 0.82) 0.365 (−0.89, 1.62) 0.015 −0.483 (−1.18, 0.21) −0.050 0.470 (−0.34, 1.28) 0.019 2016 −0.231 (−0.72, 0.26) 0.702 (−0.14, 1.55) 0.065 0.847 (0.17, 1.51) 0.077 0.232 (−0.40, 0.86) 0.029  Iron supplementation 5374 2005 −0.320 (−1.00, 0.36) −0.504 (−1.25, 0.24) −0.117 (−0.72, 0.48) 2011 0.682 (1.79, 1.85) −0.081 (−0.71, 0.55) −0.035 0.0464 (−1.71, 0.80) −0.021 −0.048 (−0.68, 0.58) −0.017 2016 4.213 (3.35, 5.07) −0.766 (−1.85, 0.31) −0.031 −0.911 (−1.93, 0.11) −0.047 −0.693 (−1.75, 0.37) −0.034  Maternal tetanus 4554 2005 1.275 (0.41, 2.13) 0.210 (−0.73, 1.15) 0.048 (−0.17, 1.13) 2011 0.958 (−0.42, 1.49) −1.028 (−2.74, 0.69) −0.088 −0.180 (−1.36, 1.00) −0.022 −0.756 (−1.58, 0.07) −0.069 2016 −0.726 (−1.85, 0.40) −0.235 (−1.27, 0.80) 0.047 0.354 (−0.86, 1.56) 0.068 −0.020 (−1.19, 1.15) 0.026 Women's health  Pap smear done 12715 2005 −1.112 (−1.69, −0.52) −0.595 (−1.14, −0.50) −0.887 (−1.70, −0.06) 2011 0.515 (0.22, 0.80) 0.710 (0.25, 1.16) 0.110 −0.200 (−0.71, 0.31) −0.025 0.425 (−0.16, 1.01) 0.067 2016 0.091 (−0.41, 0.59) 1.139 (0.28, 1.99) 0.171 0.942 (0.09, 1.79) 0.138 0.876 (0.02, 1.72) 0.131  Mammogram done 12717 2005 −0.023 (−0.80, 0.76) −0.005 (−1.00, 0.99) −0.632 (−1.62, 0.36) 2011 0.398 (−0.25, 1.04) −0.283 (−1.08, 0.54) −0.010 −0.458 (−1.43, 0.51) −0.015 0.189 (−0.72, 1.10) −0.000 2016 −0.042 (−0.69, 0.60) −0.462 (−1.54, 0.61) −0.010 0.334 (−0.72, 1.39) 0.011 −0.236 (−1.42, 0.95) −0.002 Outcomea . n . Year . Year . NGO . MUNI . ASSN . Treatment . Year*Treatment . . Treatment . Year*Treatment . . Treatment . Year*Treatment . . β (95% CI) . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . Child health  Weight-for-age 5043 2005 −0.057 (−0.30, 0.18) 0.079 (−0.17, 0.33) −0.056 (−0.23, 0.12) 2011 −0.00 (−0.21, 0.20) −0.010 (−0.30, 0.28) −0.010 −0.083 (−0.48, 0.12) −0.182 0.021 (−0.24, 0.29) 0.021 2016 0.142 (−0.11, 0.39) −0.166 (−0.51, 0.17) −0.165 −0.196 (−0.57, 0.18) −0.195 −0.184 (−0.52, 0.15) −0.184  Length-for-age 5014 2005 −0.245 (−0.67, 0.19) 0.158 (−0.20, 0.52) −0.053 (−0.30, 0.20) 2011 −0.102 (−0.28, 0.07) 0.080 (−0.19, 0.35) 0.079 −0.275 (−0.66, 0.10) −0.275 0.103 (−0.18, 0.38) 0.102 2016 0.449 (0.05, 0.83) 0.187 (−0.35, 0.73) 0.186 −0.258 (−0.78, 0.27) −0.258 −0.079 (−0.51, 0.35) −0.079  Weight-for-length 4990 2005 0.118 (−0.07, 0.30) 0.002 (−0.21, 0.22) −0.030 (−0.23, 0.17) 2011 0.043 (−0.21, 0.30) −0.082 (−0.42, 0.25) −.0821 −0.033 (−0.35, 0.29) −0.033 −0.029 (−0.34, 0.28) −0.029 2016 −0.197 (−0.37, −0.01) −0.344 (−0.65, −0.02) −0.343 −0.046 (−0.43, 0.37) −0.046 −0.180 (−0.53, 0.17) −0.179  BMI 4991 2005 0.162 (−0.05, 0.37) −0.032 (−0.25, 0.19) −0.026 (−0.24, 0.19) 2011 0.075 (−0.20, 0.35) −0.090 (−0.45, 0.27) −0.090 0.038 (−0.30, 0.38) 0.038 −0.039 (−0.37, 0.29) −0.039 2016 −0.274 (−0.46, −0.08) −0.367 (−0.70, −0.02) −0.367 0.028 (−0.39, 0.44) 0.027 −0.149 (−0.50, 0.20) −0.148 Vaccination  Measles vaccination 5911 2005 0.315 (0.06, 0.56) −0.053 (−0.52, 0.42) 0.056 (−0.22, 0.33) 2011 1.750 (1.20, 2.29) −0.297 (−1.28, 0.69) −0.070 −0.209 (−0.92, 0.51) −0.016 −0.342 (−0.99, 0.30) −0.045 2016 2.368 (1.63, 3.09) −0.541 (−1.65, 0.57) −0.087 −0.880 (−1.79, 0.03) −0.072 −0.581 (−1.41, 0.24) −0.053  Polio vaccination 6257 2005 0.327 (0.04, 0.61) −0.366 (−0.79, 0.05) −0.042 (−0.29, 0.21) 2011 2.011 (1.18, 2.83) 0.830 (−0.91, 2.57) −0.025 0.693 (−0.48, 1.86) 0.080 0.559 (−0.59, 1.71) 0.022 2016 1.438 (0.89, 1.98) −0.461 (−1.82, 0.25) −0.061 −0.511 (−1.31, 0.29) −0.010 0.076 (−0.57, 0.72) 0.009  BCG vaccination 6224 2005 0.167 (−0.11, 0.44) −0.066 (−0.74, 0.60) 0.132 (−0.23, 0.49) 2011 0.623 (0.22, 1.02) 0.666 (−0.60, 1.94) 0.041 0.373 (−0.33, 1.07) 0.041 −0.130 (−0.75, 0.49) −0.020 2016 1.502 (0.91, 2.09) −0.379 (−1.53, 0.77) −0.038 −0.804 (−1.75, 0.14) −0.058 −0.725 (−1.50, 0.05) −0.062  DPT vaccination 5999 2005 0.370 (0.06, 0.67) −0.221 (−0.50, 0.06) 0.135 (−0.17, 0.44) 2011 2.089 (1.56, 2.61) −0.030 (−1.41, 1.35) −0.041 0.124 (−0.97, 1.22) 0.032 −0.079 (−0.94, 0.78) −0.017 2016 1.236 (0.52, 1.94) −0.598 (−1.8, 0.61) −0.064 −0.742 (−1.62, 0.14) −0.050 −0.741 (−1.63, 0.15) −0.065  Complete vaccination 5624 2005 0.329 (0.71, 0.58) 0.0002 (−0.48, 0.48) 0.088 (−0.20, 0.37) 2011 1.477 (1.11, 1.84) −0.213 (−1.11, 0.69) −0.062 −0.296 (−0.92, 0.33) −0.042 −0.272 (−0.79, 0.25) −0.046 2016 2.307 (1.59, 3.02) 0.522 (−1.68, 0.63) −0.091 −1.067 (−1.97, −0.16) −0.114 −0.571 (−1.40, 0.26) −0.062 Delivery care  Home delivery 6413 2005 −0.384 (−0.94, 0.17) −0.292 (−1.05, 0.46) −0.085 (−0.54, 0.37) 2011 −0.422 (−0.67, −0.17) −0.460 (−1.16, 0.24) −0.076 0.106 (−0.55, 0.76) 0.021 −0.344 (−0.85, 0.16) −0.063 2016 −0.656 (−0.96, −0.34) −1.174 (−1.72, −0.62) −0.156 −0.384 (−0.95, 0.19) −0.060 −0.959 (−1.658, −0.33) −0.154  MCH facility delivery 6413 2005 −0.702 (−2.43, 1.02) 0.507 (−0.96, 1.97) −0.368 (−2.31, 1.57) 2011 −0.137 (−1.21, 0.93) 1.837 (0.11, 3.56) 0.117 0.678 (−1.04, 2.39) 0.061 1.295 (−0.91, 3.50) 0.083 2016 0.277 (−0.55, 1.10) 1.666 (−0.05, 3.38) 0.125 0.104 (−1.37, 1.50) 0.019 1.504 (−0.57, 3.58) 0.139  Skilled birth attendant 5513 2005 0.384 (−0.22, 0.99) 0.446 (−0.30, 1.19) 0.137 (−0.31, 0.59) 2011 0.569 (0.33, 0.80) 0.361 (−0.34, 1.06) 0.058 −0.353 (−0.95, 0.24) −0.070 0.131 (−0.40, 0.66) 0.023 2016 1.134 (0.83, 1.43) 0.636 (0.06, 1.20) 0.069 −0.173 (−0.74, 0.40) −0.043 0.368 (−0.18, 0.92) 0.053  Caesarean birth 6096 2005 0.229 (−0.36, 0.81) 0.288 (−0.28, 0.86) −0.104 (−0.74, 0.53) 2011 0.262 (−0.32, 0.85) −0.009 (−0.96, 0.94) 0.002 −0.695 (−1.38, −0.00) −0.059 0.331 (−0.39, 1.05) 0.031 2016 −0.423 (−1.19, 0.35) 0.903 (0.04, 1.76) 0.081 0.556 (−0.34, 1.46) 0.042 0.489 (−0.38, 1.35) 0.033 Maternal/child care  Child in weight control programme 6252 2005 0.407 (−0.44, 1.26) −0.506 (−1.48, 0.47) 0.331 (−0.51, 1.17) 2011 0.762 (−0.14, 1.66) 0.546 (−0.63, 1.72) 0.154 −0.459 (−1.54, 0.62) −0.095 −0.270 (−1.34, 0.80) −0.036 2016 1.112 (0.58, 1.64) 0.306 (−0.52, 1.14) 0.106 0.42 (−0.55, 1.39) 0.038 0.053 (−0.62, 0.73) 0.040  4+ antenatal visits 4452 2005 0.314 (−0.00, 0.63) 0.240 (−0.29, 0.77) 0.353 (−0.10, 0.80) 2011 0.330 (−0.16, 0.82) 0.365 (−0.89, 1.62) 0.015 −0.483 (−1.18, 0.21) −0.050 0.470 (−0.34, 1.28) 0.019 2016 −0.231 (−0.72, 0.26) 0.702 (−0.14, 1.55) 0.065 0.847 (0.17, 1.51) 0.077 0.232 (−0.40, 0.86) 0.029  Iron supplementation 5374 2005 −0.320 (−1.00, 0.36) −0.504 (−1.25, 0.24) −0.117 (−0.72, 0.48) 2011 0.682 (1.79, 1.85) −0.081 (−0.71, 0.55) −0.035 0.0464 (−1.71, 0.80) −0.021 −0.048 (−0.68, 0.58) −0.017 2016 4.213 (3.35, 5.07) −0.766 (−1.85, 0.31) −0.031 −0.911 (−1.93, 0.11) −0.047 −0.693 (−1.75, 0.37) −0.034  Maternal tetanus 4554 2005 1.275 (0.41, 2.13) 0.210 (−0.73, 1.15) 0.048 (−0.17, 1.13) 2011 0.958 (−0.42, 1.49) −1.028 (−2.74, 0.69) −0.088 −0.180 (−1.36, 1.00) −0.022 −0.756 (−1.58, 0.07) −0.069 2016 −0.726 (−1.85, 0.40) −0.235 (−1.27, 0.80) 0.047 0.354 (−0.86, 1.56) 0.068 −0.020 (−1.19, 1.15) 0.026 Women's health  Pap smear done 12715 2005 −1.112 (−1.69, −0.52) −0.595 (−1.14, −0.50) −0.887 (−1.70, −0.06) 2011 0.515 (0.22, 0.80) 0.710 (0.25, 1.16) 0.110 −0.200 (−0.71, 0.31) −0.025 0.425 (−0.16, 1.01) 0.067 2016 0.091 (−0.41, 0.59) 1.139 (0.28, 1.99) 0.171 0.942 (0.09, 1.79) 0.138 0.876 (0.02, 1.72) 0.131  Mammogram done 12717 2005 −0.023 (−0.80, 0.76) −0.005 (−1.00, 0.99) −0.632 (−1.62, 0.36) 2011 0.398 (−0.25, 1.04) −0.283 (−1.08, 0.54) −0.010 −0.458 (−1.43, 0.51) −0.015 0.189 (−0.72, 1.10) −0.000 2016 −0.042 (−0.69, 0.60) −0.462 (−1.54, 0.61) −0.010 0.334 (−0.72, 1.39) 0.011 −0.236 (−1.42, 0.95) −0.002 Significant improvement was recorded in just over half, as indicated by the bolded coefficients in the year column. a All results are from models with maternal age, maternal age squared, maternal literacy, number of household members, number of children in the household and a household wealth index. b The difference-in-difference (DiD) was calculated using the margins command and report the difference in means or proportions between treatment and control (NGO or ASSN or MUNI vs CENT) at two different time points (2005 vs 2011 and 2005 vs 2016). Models for binary variables (e.g. vaccination, delivery care, maternal/child care outcomes) display the difference in probabilities as a per cent rather than a proportion. Open in new tab Table 4 Estimated effect of 3 decentralization types on standardized treatment effect (z−scores) by domain Domaina . n . Year . Year . NGO . MUNI . ASSN . Treatment . Year*Treatment . . Treatment . Year*Treatment . . Treatment . Year*Treatment . . β (95% CI) . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . Child health z−scores 5662 2005 −0.002 (−0.16, 0.15) 0.060 (−0.15, 0.28) −0.042 (−0.19, 0.10) 2011 0.003 (−0.18, 0.19) −0.025 (−0.28, 0.23) −0.025 −0.107 (−0.36, 0.14) −0.107 0.016 (−0.21, 0.25) 0.016 2016 0.030 (−0.15, 0.21) −0.107 (−0.36, 0.14) −0.181 −0.181 (−0.43, 0.07) −0.147 −0.157 (-0.43, 0.52) −0.156 Vaccination 4965 2005 0.162 (0.43, 0.28) −0.088 (−0.37, 0.19) 0.037 (−0.10, 0.17) 2011 0.620 (0.51, 0.72) −0.099 (−0.33, 0.13) −0.099 0.057 (−0.21, 0.32) 0.057 −0.071 (−0.24, 0.10) −0.071 2016 0.692 (0.53, 0.85) −0.188 (−0.44, 0.06) −0.187 −0.225 (−0.54, 0.09) −0.224 −0.133 (−0.35, 0.08) −0.133 Delivery care 5197 2005 0.128 (−0.12, 0.37) 0.178 (−0.14, 0.49) 0.0234 (−0.21, 0.25) 2011 0.230 (0.10, 0.35) 0.225 (−0.11, 0.56) 0.224 −0.061 (−0.35, 0.23) −0.061 0.132 (−0.12, 0.38) 0.131 2016 0.480 (0.37, 0.58) 0.197 (−0.13, 0.52) 0.197 −0.032 (−0.30, 0.23) −0.032 0.191 (0.07, 0.45) 0.190 Maternal and child care 4335 2005 0.138 (−0.17, 0.44) −0.093 (−0.39, 0.21) 0.134 (−0.18, 0.45) 2011 0.462 (0.09, 0.82) −0.028 (−0.44, 0.38) −0.028 −0.231 (−0.64, 0.17) −0.230 −0.218 (−0.65, 0.22) −0.218 2016 0.880 (0.68, 1.07) 0.170 (−0.01, 0.35) 0.169 0.165 (−0.07, 0.40) 0.165 0.086 (−0.12, 0.30) 0.086 Domaina . n . Year . Year . NGO . MUNI . ASSN . Treatment . Year*Treatment . . Treatment . Year*Treatment . . Treatment . Year*Treatment . . β (95% CI) . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . Child health z−scores 5662 2005 −0.002 (−0.16, 0.15) 0.060 (−0.15, 0.28) −0.042 (−0.19, 0.10) 2011 0.003 (−0.18, 0.19) −0.025 (−0.28, 0.23) −0.025 −0.107 (−0.36, 0.14) −0.107 0.016 (−0.21, 0.25) 0.016 2016 0.030 (−0.15, 0.21) −0.107 (−0.36, 0.14) −0.181 −0.181 (−0.43, 0.07) −0.147 −0.157 (-0.43, 0.52) −0.156 Vaccination 4965 2005 0.162 (0.43, 0.28) −0.088 (−0.37, 0.19) 0.037 (−0.10, 0.17) 2011 0.620 (0.51, 0.72) −0.099 (−0.33, 0.13) −0.099 0.057 (−0.21, 0.32) 0.057 −0.071 (−0.24, 0.10) −0.071 2016 0.692 (0.53, 0.85) −0.188 (−0.44, 0.06) −0.187 −0.225 (−0.54, 0.09) −0.224 −0.133 (−0.35, 0.08) −0.133 Delivery care 5197 2005 0.128 (−0.12, 0.37) 0.178 (−0.14, 0.49) 0.0234 (−0.21, 0.25) 2011 0.230 (0.10, 0.35) 0.225 (−0.11, 0.56) 0.224 −0.061 (−0.35, 0.23) −0.061 0.132 (−0.12, 0.38) 0.131 2016 0.480 (0.37, 0.58) 0.197 (−0.13, 0.52) 0.197 −0.032 (−0.30, 0.23) −0.032 0.191 (0.07, 0.45) 0.190 Maternal and child care 4335 2005 0.138 (−0.17, 0.44) −0.093 (−0.39, 0.21) 0.134 (−0.18, 0.45) 2011 0.462 (0.09, 0.82) −0.028 (−0.44, 0.38) −0.028 −0.231 (−0.64, 0.17) −0.230 −0.218 (−0.65, 0.22) −0.218 2016 0.880 (0.68, 1.07) 0.170 (−0.01, 0.35) 0.169 0.165 (−0.07, 0.40) 0.165 0.086 (−0.12, 0.30) 0.086 Significant improvement was recorded in just over half, as indicated by the bolded coefficients in the year column. a All results are from models with maternal age, maternal age squared, maternal literacy, number of household members, number of children in the household and a household wealth index. b The difference-in-difference (DiD) was calculated using the margins command and reports the difference in means between treatment and control (NGO or MAN or MUNI vs CENT) at two different time points (2005 vs 2011 and 2005 vs 2016). Open in new tab Table 4 Estimated effect of 3 decentralization types on standardized treatment effect (z−scores) by domain Domaina . n . Year . Year . NGO . MUNI . ASSN . Treatment . Year*Treatment . . Treatment . Year*Treatment . . Treatment . Year*Treatment . . β (95% CI) . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . Child health z−scores 5662 2005 −0.002 (−0.16, 0.15) 0.060 (−0.15, 0.28) −0.042 (−0.19, 0.10) 2011 0.003 (−0.18, 0.19) −0.025 (−0.28, 0.23) −0.025 −0.107 (−0.36, 0.14) −0.107 0.016 (−0.21, 0.25) 0.016 2016 0.030 (−0.15, 0.21) −0.107 (−0.36, 0.14) −0.181 −0.181 (−0.43, 0.07) −0.147 −0.157 (-0.43, 0.52) −0.156 Vaccination 4965 2005 0.162 (0.43, 0.28) −0.088 (−0.37, 0.19) 0.037 (−0.10, 0.17) 2011 0.620 (0.51, 0.72) −0.099 (−0.33, 0.13) −0.099 0.057 (−0.21, 0.32) 0.057 −0.071 (−0.24, 0.10) −0.071 2016 0.692 (0.53, 0.85) −0.188 (−0.44, 0.06) −0.187 −0.225 (−0.54, 0.09) −0.224 −0.133 (−0.35, 0.08) −0.133 Delivery care 5197 2005 0.128 (−0.12, 0.37) 0.178 (−0.14, 0.49) 0.0234 (−0.21, 0.25) 2011 0.230 (0.10, 0.35) 0.225 (−0.11, 0.56) 0.224 −0.061 (−0.35, 0.23) −0.061 0.132 (−0.12, 0.38) 0.131 2016 0.480 (0.37, 0.58) 0.197 (−0.13, 0.52) 0.197 −0.032 (−0.30, 0.23) −0.032 0.191 (0.07, 0.45) 0.190 Maternal and child care 4335 2005 0.138 (−0.17, 0.44) −0.093 (−0.39, 0.21) 0.134 (−0.18, 0.45) 2011 0.462 (0.09, 0.82) −0.028 (−0.44, 0.38) −0.028 −0.231 (−0.64, 0.17) −0.230 −0.218 (−0.65, 0.22) −0.218 2016 0.880 (0.68, 1.07) 0.170 (−0.01, 0.35) 0.169 0.165 (−0.07, 0.40) 0.165 0.086 (−0.12, 0.30) 0.086 Domaina . n . Year . Year . NGO . MUNI . ASSN . Treatment . Year*Treatment . . Treatment . Year*Treatment . . Treatment . Year*Treatment . . β (95% CI) . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . β (95% CI) . β (95% CI) . DiDb . Child health z−scores 5662 2005 −0.002 (−0.16, 0.15) 0.060 (−0.15, 0.28) −0.042 (−0.19, 0.10) 2011 0.003 (−0.18, 0.19) −0.025 (−0.28, 0.23) −0.025 −0.107 (−0.36, 0.14) −0.107 0.016 (−0.21, 0.25) 0.016 2016 0.030 (−0.15, 0.21) −0.107 (−0.36, 0.14) −0.181 −0.181 (−0.43, 0.07) −0.147 −0.157 (-0.43, 0.52) −0.156 Vaccination 4965 2005 0.162 (0.43, 0.28) −0.088 (−0.37, 0.19) 0.037 (−0.10, 0.17) 2011 0.620 (0.51, 0.72) −0.099 (−0.33, 0.13) −0.099 0.057 (−0.21, 0.32) 0.057 −0.071 (−0.24, 0.10) −0.071 2016 0.692 (0.53, 0.85) −0.188 (−0.44, 0.06) −0.187 −0.225 (−0.54, 0.09) −0.224 −0.133 (−0.35, 0.08) −0.133 Delivery care 5197 2005 0.128 (−0.12, 0.37) 0.178 (−0.14, 0.49) 0.0234 (−0.21, 0.25) 2011 0.230 (0.10, 0.35) 0.225 (−0.11, 0.56) 0.224 −0.061 (−0.35, 0.23) −0.061 0.132 (−0.12, 0.38) 0.131 2016 0.480 (0.37, 0.58) 0.197 (−0.13, 0.52) 0.197 −0.032 (−0.30, 0.23) −0.032 0.191 (0.07, 0.45) 0.190 Maternal and child care 4335 2005 0.138 (−0.17, 0.44) −0.093 (−0.39, 0.21) 0.134 (−0.18, 0.45) 2011 0.462 (0.09, 0.82) −0.028 (−0.44, 0.38) −0.028 −0.231 (−0.64, 0.17) −0.230 −0.218 (−0.65, 0.22) −0.218 2016 0.880 (0.68, 1.07) 0.170 (−0.01, 0.35) 0.169 0.165 (−0.07, 0.40) 0.165 0.086 (−0.12, 0.30) 0.086 Significant improvement was recorded in just over half, as indicated by the bolded coefficients in the year column. a All results are from models with maternal age, maternal age squared, maternal literacy, number of household members, number of children in the household and a household wealth index. b The difference-in-difference (DiD) was calculated using the margins command and reports the difference in means between treatment and control (NGO or MAN or MUNI vs CENT) at two different time points (2005 vs 2011 and 2005 vs 2016). Open in new tab We find evidence of differential change by governance type (Table 3 and Figures 2–4). Between 2005 and 2016, NGO municipalities showed additional improvement in all four delivery care measures. Both NGO and ASSN municipalities saw greater decrease in the probability of home delivery relative to CENT municipalities in 2016 (β = −1.174; 95% CI = −1.73, −0.62 and β = −0.959; 95% CI = −1.65, −0.33, respectively), which corresponds to over 15 percentage points greater decline (DiD in Table 3). NGO municipalities also experienced a greater increase between 2005 and 2016 in the probability of MCH facility delivery. Here we find a 12.5 percentage point greater increase for NGO relative to CENT municipalities. For the probability of a skilled birth attendant, NGO municipalities saw a slightly smaller but significant increase (7.0 percentage point) relative to CENT municipalities between 2005 and 2016 (β  =  0.64; 95% CI = 0.06, 1.20). A consistent finding is a positive change across all three decentralized governance structures between 2005 and 2016 in the probability of a woman receiving a pap smear. The provision of this service increased more rapidly among women in decentralized municipalities compared with centrally administered ones (difference in percentage ranged from 13.1 to 17.1). Figure 2 Open in new tabDownload slide Simple difference in model estimated means between decentralized and centralized municipalities for child health outcomes, by governance type (NGO, MUNI and ASSN) and year. Note: The value shown for NGO2016 is (NGO2016_predicted mean—CENT2016_predicted mean). Figure 2 Open in new tabDownload slide Simple difference in model estimated means between decentralized and centralized municipalities for child health outcomes, by governance type (NGO, MUNI and ASSN) and year. Note: The value shown for NGO2016 is (NGO2016_predicted mean—CENT2016_predicted mean). Figure 3 Open in new tabDownload slide Simple difference in model estimated probability (shown as a %) between decentralized and centralized municipalities for select maternal and child health services, by governance type (NGO, MUNI and ASSN) and year. Notes: The value shown for NGO2016 is (NGO2016_predicted probability—CENT2016_predicted probability) * 100. *Indicates a difference-in-difference effect of P < 0.05. Figure 3 Open in new tabDownload slide Simple difference in model estimated probability (shown as a %) between decentralized and centralized municipalities for select maternal and child health services, by governance type (NGO, MUNI and ASSN) and year. Notes: The value shown for NGO2016 is (NGO2016_predicted probability—CENT2016_predicted probability) * 100. *Indicates a difference-in-difference effect of P < 0.05. Figure 4 Open in new tabDownload slide Simple difference in model estimated means between decentralized and centralized municipalities for standardized domain scores, by governance type (NGO, MUNI and ASSN) and year. Notes: The value shown for NGO2016 is (NGO2016_predicted probability—CENT2016_predicted probability) * 100. *Indicates a difference-in-difference effect of P < 0.05; †indicates a difference-in-difference effect of P < 0.06. Figure 4 Open in new tabDownload slide Simple difference in model estimated means between decentralized and centralized municipalities for standardized domain scores, by governance type (NGO, MUNI and ASSN) and year. Notes: The value shown for NGO2016 is (NGO2016_predicted probability—CENT2016_predicted probability) * 100. *Indicates a difference-in-difference effect of P < 0.05; †indicates a difference-in-difference effect of P < 0.06. There were a few negative effects detected under decentralization. NGO municipalities showed a negative effect for weight-for-length and therefore for BMI, though largely because of gains made in length without corresponding gains in weight. MUNI municipalities had lower rates of complete vaccination (Table 3 and Figures 2 and 3), though not any of the individual vaccines measured, and so it is possible this could be a spurious result. There were no detectable treatment effects under any variant of decentralization for child health indicators weight and length or maternal and child care (weight control programme, maternal iron supplementation or tetanus vaccination; Table 3 and Figures 2 and 3). The vaccination findings serve as an important check in our analysis. Since very strong central control over vaccination programmes was maintained within the MOH through its Expanded Immunization Programme (EPI) even after decentralization (Molina-Aguilera et al., 2012), we did not expect to see differential improvement in vaccination across treatment and control municipalities. We did find significant main effects over time, indicating improvements in vaccination across all municipalities in our study. The increases in completed vaccination schedules of over 20 percentage points across organization types and in enrolment of children in weight control (growth monitoring) programmes of over 25 percentage points speak to Honduras’ achievements towards international maternal and child health goals. Results of the standardized treatment effect by domain (Table 4 and Figure 4) indicate positive change between 2005 and 2016 at the national level in all domains but child health (due to weight increase not keeping up with increase in length). The year effects indicate overall improvement in vaccination, delivery care and maternal and child health services across Honduras. NGO and MUNI municipalities saw a 0.17 SD increase in the maternal/child care domain relative to CENT municipalities in 2016 (NGO β = 0.17; 95% CI = −0.01, 0.35, MUNI β = 0.17; 95% CI = −0.07, 0.40), which is significant at the P = 0.10 level. We note that, had we considered only centralized vs decentralized municipalities, our results would be both less nuanced and lead to different overall conclusions about the effectiveness of the reform (Supplementary AppendixTable S1). The national improvement in nearly all indicators over the 10-year study period remains. However, the only differences detected between decentralized vs centrally administered municipalities, are a decline in home delivery, increase in per cent of women having 4+ antenatal visits and increase in pap smears. The success of NGO municipalities in improving delivery care would be missed entirely and a decline in vaccinations of all types would be attributed to decentralized municipalities as a single group. Discussion Existing research has produced inconsistent evidence concerning impacts of decentralization on health system performance, with some studies showing positive (Channa and Faguet, 2016) and others showing null or negative effects (Muñoz et al., 2017). We make three major contributions to this ongoing and critically important debate about how best to strengthen health systems in LDCs. First, our findings point to the importance of considering governance structures, and particularly organizational context, when assessing health effects of decentralization. While we are not the first to note the weaknesses of a dichotomous conceptualization of decentralization (Rondinelli et al., 1989), our study does highlight decentralization as a governance reform. This focuses us on the ways in which governments may choose to deliver services through hybrid arrangements involving NGOs and associations as part of an overarching decentralization strategy. As such, the resulting organizational structures are critical to understanding the effectiveness of the reform. Our findings also indicate that decentralization had a positive effect on the provision of maternal health services, but these results were not similarly strong across all governance structures. Local health systems administered by NGOs showed greater improvement in health services over centrally administered municipalities, while those managed by associations of municipalities or single municipal governments generally did not. Overall, these results are consistent with economic arguments for NGOs being able to overcome market and government failures, as well as political arguments that they address some principal-agent problems in social services delivery when properly regulated (Weisbrod, 1986; Lipsky and Smith, 1989; Brass et al., 2018). To the extent that mission-driven NGOs are better able to motivate their staff (Zarychta et al., 2020) or potentially garner supplemental resources from donors, these can be ancillary benefits of bringing them into a decentralized health service delivery system, and warrant further assessment as potential mechanisms underpinning the positive effects of NGOs identified here. Additionally, our supplementary analyses of all study outcomes using a binary indicator of decentralization (Supplementary AppendixTable S1) shows that where there were positive effects of the binary indicator of decentralization, those effects were almost exclusively driven by NGO-led decentralization, with the exception of pap smears which increased across all three organization types. This implies that the type of decentralized administration is an important factor in the success of health system decentralization. It also suggests that is easy to overlook the true impact of governance reforms if differences in organizational context within the reformed jurisdictions are ignored. Second, our results indicate that health services and outcomes in Honduras have improved virtually across the board during the last decade. In nearly all areas we examined, and irrespective of decentralization, women and children received more health services and were healthier in 2016 relative to 2006. In addition to the substantive importance of this descriptive finding, our analysis also reinforces the persistent criticisms of weak research design and poor-quality data in the health sector reform literature (Channa and Faguet, 2016); without being able to account for these trends, it is impossible to identify the effects of a governance reform like decentralization. Third, our results show that decentralization did not harm maternal and child health services and outcomes. Across all 19 indicators analysed in the five domains, in only a very few cases did we find that decentralized municipalities, dichotomous or disaggregated by governance structures, performed slightly but not significantly worse than their centrally administered counterparts. This is important to note because prior studies have found decentralization to be harmful in some cases (Muñoz et al., 2017). Additionally, many people criticize contracting with NGOs to deliver social services, particularly when including performance incentives, and there are general concerns in public administration about hollowing-out the state through hybrid arrangements of this type (Milward and Provan, 2000; Brass et al., 2018). Future research should analyse additional indicators, namely considering whether performance-based management might harm lower- or non-priority health services and outcomes, as well as assessing consequences of decentralization for satisfaction, attitudes and behaviours of patients and staff. Lastly, for several health services and outcomes analysed here, all variants of decentralization were indistinguishable from central administration. Additional research is warranted to better understand if this is a limitation of the post-reform length of time being considered here, or if barriers or conditions preventing effective decentralization can be identified and addressed. There are several limitations to this study that should be discussed. First, while Honduras is an importantly and analytically useful case because of the way the roll out of the reform allowed for a quasi-experimental study design, it is still only one country. Additional work is needed to assess how our findings apply to other settings. Similarly, the reform in Honduras is targeted to rural areas and our results may not be generalizable to urban areas which often have larger and more concentrated populations, as well as more complex health systems with a diverse mix of health providers. Other limitations result from use of survey data to assess use of health services. Inaccurate respondent recall of the use of health services may lead to greater uncertainty and variation in the survey responses, but there is no reason to believe inaccurate responses would be correlated with decentralization. Future research will triangulate our survey-based findings with administrative data obtained from the Honduran MOH. Finally, while we identify differences in health outcomes and the utilization of services among organization types under the same decentralization policy, we do not explore the mechanisms by which some organization types outperform others. This topic will be covered in detail in future work drawing on a wide range of data sources, including additional original surveys, behavioural games and qualitative case studies (Zarychta et al., 2019b). Conclusion We used quasi-experimental evidence from 10 years of experience with decentralized health services delivery in Honduras to help shed light on strategies that can work to strengthen local health systems in LDCs. The partial and staggered implementation of health sector decentralization in Honduras over time provided the rare opportunity to develop a credible research design for analysing the reform’s effects, which is a major strength of this study relative to much of the decentralization literature (Zarychta et al., (2019a). Drawing on existing household survey data from DHS and original data from our own ENSAGO survey, our analysis showed that Honduras made notable improvements in a wide range of health services and outcomes for women and children between 2006 and 2016, and we found that decentralization improved birth and delivery care, though primarily in systems managed by NGOs. Moreover, despite reasonable concerns about decentralization and performance-based contracting, we found no evidence of systematically poorer health services or outcomes for women and children in decentralized municipalities. Our results contribute to a more nuanced understanding of how local governance structures influence effectiveness of decentralization reforms and suggest that the large body of prior research on this subject has perhaps been asking the wrong question. Rather than examining whether decentralization works to improve health services and outcomes, researchers should consider the extent to which certain health sector governance structures and processes perform better than others when decentralization does occur. What are the key organizational and structural characteristics of successful health systems? How do these contribute to increases in health services, and ultimately, improved population health outcomes? Future work will develop more refined measures characterizing decentralized governance structures for health services delivery in Honduras to examine what drives the apparent performance improvements in some health systems. Supplementary data Supplementary data are available at Health Policy and Planning online. Acknowledgements This research was funded through a grant from the United States National Science Foundation (https://www.nsf.gov/; Award #1328688) and with assistance from the University of Colorado Population Center (CUPC) funded by the National Institutes of Health, National Center for Child Health and Human Development (Award #2P2CHD066613). Conflict of interest statement. None declared. Ethical approval. This study was reviewed and approved by the Institutional Review Board at the University of Colorado, Boulder, USA. References Al-Janabi A , Al-Wahdani B , Ammar W et al. 2018 . Bellagio Declaration on high-quality health systems: from a quality moment to a quality movement . The Lancet Global Health 6 : e1144 – 5 . Google Scholar Crossref Search ADS PubMed WorldCat Andersson K , Gordillo G , Van Laerhoven F. 2009 . Local Governments and Rural Development: Comparing Lessons from Brazil, Chile, Mexico, and Peru . Tucson, AZ: University of Arizona Press. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Barasa EW , Manyara AM , Molyneux S , Tsofa B. 2017 . Recentralization within decentralization: county hospital autonomy under devolution in Kenya . PLoS One 12 : e0182440 . Google Scholar Crossref Search ADS PubMed WorldCat Basinga P , Gertler PJ , Binagwaho A et al. 2011 . Effect on maternal and child health services in Rwanda of payment to primary health-care providers for performance: an impact evaluation . The Lancet 377 : 1421 – 8 . Google Scholar Crossref Search ADS WorldCat Bertelli AM , Smith CR. 2010 . Relational contracting and network management . Journal of Public Administration Research and Theory 20 : i21 – 40 . Google Scholar Crossref Search ADS WorldCat Bonell CP , Hargreaves J , Cousens S et al. 2011 . Alternatives to randomisation in the evaluation of public health interventions: design challenges and solutions . Journal of Epidemiology & Community Health 65 : 582 – 7 . Google Scholar Crossref Search ADS WorldCat Bossert T. 1998 . Analyzing the decentralization of health systems in developing countries: decision space, innovation and performance . Social Science & Medicine (1982) 47 : 1513 – 27 . Google Scholar Crossref Search ADS PubMed WorldCat Brass JN , Longhofer W , Robinson RS , Schnable A. 2018 . NGOs and international development: a review of thirty-five years of scholarship . World Development 112 : 136 – 49 . Google Scholar Crossref Search ADS WorldCat Brodkin EZ. 2008 . Accountability in street-level organizations . International Journal of Public Administration 31 : 317 – 36 . Google Scholar Crossref Search ADS WorldCat Brodkin EZ. 2011 . Policy work: street-level organizations under new managerialism . Journal of Public Administration Research and Theory 21 : i253 – 77 . Google Scholar Crossref Search ADS WorldCat Channa A , Faguet J-P. 2016 . Decentralization of health and education in developing countries: a quality-adjusted review of the empirical literature . The World Bank Research Observer 31 : 199 – 241 . Google Scholar Crossref Search ADS WorldCat Craig P , Cooper C , Gunnell D et al. 2012 . Using natural experiments to evaluate population health interventions: new Medical Research Council guidance . Journal of Epidemiology and Community Health 66 : 1182 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat Dimick JB , Ryan AM. 2014 . Methods for evaluating changes in health care policy: the difference-in-differences approach . JAMA 312 : 2401 – 2 . Google Scholar Crossref Search ADS PubMed WorldCat DuGoff EH , Schuler M , Stuart EA. 2014 . Generalizing observational study results: applying propensity score methods to complex surveys . Health Services Research 49 : 284 – 303 . Google Scholar Crossref Search ADS PubMed WorldCat El Bcheraoui C , Palmisano EB , Dansereau E et al. 2017 . Healthy competition drives success in results-based aid: lessons from the Salud Mesoamérica Initiative . PLoS One 12 : e0187107 . Google Scholar Crossref Search ADS PubMed WorldCat Frenk J. 2010 . The global health system: strengthening national health systems as the next step for global progress . PLoS Medicine 7 : e1000089 . Google Scholar Crossref Search ADS PubMed WorldCat ICF. Wealth Index Construction. The DHS Program website. 2019. https://www.dhsprogram.com/topics/wealth-index/Wealth-Index-Construction.cfm, accessed 25 March 2019 . Jiménez-Rubio D , García-Gómez P. 2017 . Decentralization of health care systems and health outcomes: evidence from a natural experiment . Social Science & Medicine 188 : 69 – 81 . Google Scholar Crossref Search ADS WorldCat Kruk ME , Gage AD , Arsenault C et al. 2018 . High-quality health systems in the sustainable development goals era: time for a revolution . The Lancet Global Health 6 : e1196 – 252 . Google Scholar Crossref Search ADS PubMed WorldCat Leslie HH , Hirschhorn LR , Marchant T et al. 2018 . Health systems thinking: a new generation of research to improve healthcare quality . PLoS Medicine 15 : e1002682 . Google Scholar Crossref Search ADS PubMed WorldCat Lieberman ES. 2011 . The perils of polycentric governance of infectious disease in South Africa . Social Science & Medicine 73 : 676 – 84 . Google Scholar Crossref Search ADS WorldCat Lipsky M , Smith SR. 1989 . Nonprofit organizations, government, and the welfare state . Political Science Quarterly 104 : 625 – 48 . Google Scholar Crossref Search ADS WorldCat Loevinsohn B , Harding A. 2005 . Buying results? Contracting for health service delivery in developing countries . The Lancet 366 : 676 – 81 . Google Scholar Crossref Search ADS WorldCat Milward HB , Provan KG. 2000 . Governing the hollow state . Journal of Public Administration Research and Theory 10 : 359 – 80 . Google Scholar Crossref Search ADS WorldCat Ministry of Health (MOH), Government of Honduras. 2009 . Marco Conceptual Político y Estratégico de La Reforma Del Sector de Salud. Tegucigalpa, Honduras. Ministry of Health (MOH), Government of Honduras. 2010 . Plan Nacional de Salud 2010-2014. Tegucigalpa, Honduras: Ministry of Health. Molina-Aguilera IB , Mendoza-Rodriguez LO , Palma-Rios MA , Danovaro-Holliday MC. 2012 . Integrating health promotion and disease prevention interventions with vaccination in Honduras . The Journal of Infectious Diseases 205 : S77 – 81 . Google Scholar Crossref Search ADS PubMed WorldCat Muñoz DC , Amador PM , Llamas LM , Hernandez DM , Sancho J. 2017 . Decentralization of health systems in low and middle income countries: a systematic review . International Journal of Public Health 62 : 219 – 29 . Google Scholar Crossref Search ADS PubMed WorldCat Post AE , Bronsoler V , Salman L. 2017 . Hybrid regimes for local public goods provision: a framework for analysis . Perspectives on Politics 15 : 952 – 66 . Google Scholar Crossref Search ADS WorldCat Rondinelli DA , McCullough JS , Johnson RW. 1989 . Analysing decentralization policies in developing countries: a political-economy framework . Development and Change 20 : 57 – 87 . Google Scholar Crossref Search ADS WorldCat Rowe AK , Rowe SY , Peters DH et al. 2018 . Effectiveness of strategies to improve health-care provider practices in low-income and middle-income countries: a systematic review . The Lancet Global Health 6 : e1163 – 75 . Google Scholar Crossref Search ADS PubMed WorldCat Schulze GG , Sjahrir BS. 2014 . Decentralization, governance and public service delivery . Regional Dynamics in a Decentralized Indonesia 501 : 186 . Google Scholar OpenURL Placeholder Text WorldCat Secretaría de Salud Honduras, Instituto Nacional de Estadística Honduras, ICF International. 2013 a. Honduras: Standard DHS, 2005-2006 & 2011-12 [Datasets]. Rockville, MD: ICF [Distributor]. Secretaría de Salud Honduras, Instituto Nacional de Estadística Honduras, ICF International. 2013 b. Honduras Encuesta Nacional de Salud y Demografía 2011-2012. Tegucigalpa, Honduras: SS, INE and ICF International. Wagstaff A , Yu S. 2007 . Do health sector reforms have their intended impacts? The World Bank’s Health VIII project in Gansu province . Journal of Health Economics 26 : 505 – 35 . Google Scholar Crossref Search ADS PubMed WorldCat Weisbrod BA. 1986 . Toward a theory of the voluntary nonprofit sector in a three-sector economy. In: The Economics of Nonprofit Institutions . New York, NY: Oxford University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Wright GD , Andersson KP , Gibson CC , Evans TP. 2016 . Decentralization can help reduce deforestation when user groups engage with local government . Proceedings of the National Academy of Sciences 113 : 14958 – 63 . Google Scholar Crossref Search ADS WorldCat Zarychta A , Andersson KP , Root ED , Menken J , Grillos T. 2019 a. Assessing the impacts of governance reforms on health services delivery: a quasi-experimental, multi-method, and participatory approach . Health Services and Outcomes Research Methodology 19 : 241 – 58 . Google Scholar Crossref Search ADS WorldCat Zarychta A , Andersson KP , Root ED , Menken J , Grillos T. 2019 b. Supplemental Appendix for “Assessing the impacts of governance reforms on health services delivery: a quasi-experimental, multi-method, and participatory approach” . Health Services and Outcomes Research Methodology 19 : 241 – 58 . Google Scholar Crossref Search ADS WorldCat Zarychta A , Grillos T , Andersson KP. 2020 . Public sector governance reform and the motivation of street-level bureaucrats in developing countries . Public Administration Review 80 : 75 – 91 . Google Scholar Crossref Search ADS WorldCat © The Author(s) 2020. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. All rights reserved. For permissions, please e-mail: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Mugo,, Cyrus;Njuguna,, Irene;Nduati,, Margaret;Omondi,, Vincent;Otieno,, Verlinda;Nyapara,, Florence;Mabele,, Elizabeth;Moraa,, Hellen;Sherr,, Kenneth;Inwani,, Irene;Maleche-Obimbo,, Elizabeth;Wamalwa,, Dalton;John-Stewart,, Grace;Slyker,, Jennifer;Wagner, Anjuli, D

doi: 10.1093/heapol/czaa089pmid: 32944754

Abstract Stakeholder engagement between researchers, policymakers and practitioners is critical for the successful translation of research into policy and practice. The Counseling and Testing for Children at Home (CATCH) study evaluated a paediatric index case testing model, targeting the children of HIV-infected adults in care in Kenya. Researchers collaborated with stakeholders in the planning, execution and evaluation, and dissemination phases of CATCH. They included a community advisory board, the national HIV programme, County health departments, institutional ethics review bodies, a paediatric bioethics group, facility heads and frontline healthcare workers . Stakeholder analysis considered the power and interest of each stakeholder in the study. All stakeholders had some power to influence the success of the project in the different phases. However, support from institutions with higher hierarchical power increased acceptance of the study by stakeholders lower in the hierarchy. During the planning, execution and evaluation, and dissemination phases, the study benefitted from deliberate stakeholder engagement. Through engagement, changes were made in the approach to recruitment to ensure high external validity, placing recruitment optimally within existing clinic flow patterns. Choices in staffing home visits were made to include the appropriate cadre of staff. Adaptations were made to the consenting process that balanced the child’s evolving autonomy and risks of HIV disclosure. Dissemination involved delivering site-specific results in each HIV clinic, local and international conferences and sharing of study tools, resulting in the study approach being scaled up nationally. The deliberate engagement of stakeholders early in intervention development optimized study validity and accelerated adoption of the CATCH approach in nationwide HIV testing campaigns by the Ministry of Health and inclusion of paediatric index-case testing in national HIV testing guidelines. Involving policymakers and frontline healthcare workers throughout the study cycle builds capacity in the implementing team for quick adoption and scale-up of the evidence-based practice. Stakeholder engagement, community-based participatory research, stakeholder analysis, paediatric HIV testing, index-case testing KEY MESSAGES Identifying key stakeholders in a research project and involving them in the design, implementation and dissemination of findings enables faster translation of the research into policy and practice. Stakeholder analysis is helpful in identifying the relative power and interest each potential stakeholder in a project has, and their relative contribution to determine effort required in engaging the stakeholder. Introduction Stakeholder engagement is critical in designing successful research projects and scaling the solutions they produce (Laura Esmail et al., 2015). The involvement of patients, communities and policymakers as active collaborators beyond their roles as participants and consumers of research ensure that there are meaningful and culturally appropriate research procedures and material (Israel et al., 2008; Amirav et al., 2017; Goodman and Sanders Thompson, 2017). The Good Participatory Practice guidelines for stakeholder engagement developed by UNAIDS (UNAIDS and AIDS Vaccine Advocacy Coalition, 2011) established a framework for involving stakeholders in clinical trials and other HIV research. They recommend that stakeholder engagement be a continuous process throughout the life of the study from formulating research questions, protocol development, execution and dissemination of results. Despite the importance of stakeholder engagement, researchers often struggle to systematically engage diverse stakeholders throughout the lifecycle of a research project (Hoffman et al., 2010). Two systematic reviews noted that slightly less than half of studies involve stakeholders in the early stages of the study, including protocol development, (Day et al., 2018) and only 10% involve them in the dissemination stage (Concannon et al., 2014; Day et al., 2018). Consequently, these research projects may be tested in conditions that do not take into account ‘real-life’ implementation challenges (Glasgow et al., 2003). Even in settings where stakeholder engagement does happen, there are challenges in having well defined roles for each stakeholder that allow for meaningful contributions and avoid the risk of tokenism, where researchers are seen to be only interested in ‘checking a box’ (Harrison et al., 2019). On the other hand, policymakers and practitioners who are not involved in research often do not have ready access to scientific literature about effective interventions, and researchers are often uninvolved in implementation through programmes (Haines et al., 2004). As a result, there is often a long gap, averaging 17 years (Balas and Boren, 2000), in the time between high-quality trial results and widespread adoption of an intervention (Glasgow et al., 2003; Haines et al., 2004). While many studies have documented stakeholder engagement, fewer expound on the methodology used to assess the importance and results of that engagement (Martinez et al., 2019). One method of evaluating stakeholder engagement is using the ‘power and interest’ model. In this model, the importance of any stakeholder is determined by the power they have to impact the success of a research project and the interest they have in the research (Chinyio and Akintoye, 2008; Horton and Pilkington, 2014; Colvin et al., 2020). This model can be applied prospectively to guide the researchers on the stakeholders whose influence could impact feasibility of the research project. Retrospective analysis of the power and interest different stakeholders had in the study can explain the reach and acceptance of the study by the stakeholders and the target community. Researchers can also apply a role-based approach to determine the important stakeholders to engage (Horton and Pilkington, 2014). In this approach, the researchers rate the importance of different roles in the success of the study, and the stakeholders who perform the most important roles, regardless of their power, are engaged more purposefully. Paediatric HIV has high morbidity and mortality, particularly when diagnosis is delayed (Wamalwa et al., 2012). At the time when this study was conducted, prevention of mother-to-child transmission of HIV (PMTCT) programmes had scaled up widely in Kenya, resulting in a reduction in new infant infections [National AIDS and STI Control Programme (NASCOP), 2014]. However, ∼60% of HIV-infected children remained undiagnosed [National AIDS and STI Control Programme (NASCOP), 2014] and many were first identified in hospital when seriously ill when they could not fully benefit from antiretroviral therapy (Njuguna et al., 2016). Among adults in HIV care, only 45% ever tested their children [National AIDS and STI Control Programme (NASCOP), 2014]. This article shares lessons learned from stakeholder engagement in research design through national scale-up of a paediatric HIV index case testing model to address the aforementioned testing gaps in Kenya, the Counseling and Testing for Children at Home (CATCH) study. Methods CATCH study The CATCH study was designed in 2012 and implemented in six health facilities in Nairobi County and one facility in Kisumu County in Kenya between the years 2013 and 2016. Caregivers in HIV care who had children aged 0–12 years of unknown HIV status were offered an opportunity to test their children either in clinic or at home (Wagner et al., 2016). In 2012, when the study was designed, index case testing as an approach to reaching children at high risk of being HIV infected was not widely used and was not explicitly included in the national HIV testing guidelines (Ministry of Public Health and Sanitation, 2010). CATCH was a collaborative study between University of Nairobi (UoN) and University of Washington (UW) (Wagner et al., 2016). The UoN/UW collaborative group had a longstanding collaboration in paediatric HIV prevention, testing and treatment research projects and bi-directional training programmes involving the universities, the largest referral and teaching hospital in Kenya [Kenyatta National Hospital (KNH)] and the Kenya Ministry of Health (MOH). The CATCH study adopted a research approach informed by community-based participatory research (Israel et al., 2008), which included recognition of healthcare workers (HCW) expertise in HIV programming, building on existing strengths within community HIV programmes, involvement of community and government stakeholders in design, conduct and evaluation of the project and capacity building/knowledge dissemination through implementation of study tools into the national programme (Satcher, 2005). Stakeholders in the CATCH study The CATCH study engaged diverse stakeholders throughout the study lifecycle, including the UoN/Kenyatta National Hospital Ethics and Research Committee (ERC), UW Institutional Review Board (IRB) and Seattle Children’s Hospital’s Treuman Katz Center for Pediatric Bioethics. A community advisory board (CAB) was convened, comprised of HCW with experience working with persons living with HIV, community health workers (CHW), peer counsellors, HIV-infected caregivers and a child protection officer. Following constitutional changes in Kenya in 2010, county health departments managed the bulk of public health facilities and programmes including HIV prevention and treatment programmes. The National AIDS & STI Control Programme (NASCOP), housed in MOH, which previously managed the HIV programme countrywide, was tasked with policy development and monitoring (Barker et al., 2014). This required that the study engage both county and national levels of government, along with heads of facilities and HIV care clinics and HIV clinic staff. Data sources Communications to and from the ethics bodies and the paediatrics bioethics centre were done through letters. Engagement with the CAB, County, NASCOP, facility heads and frontline HCW was documented through meeting minutes, presentations, field notes and reflections from the study coordinators, study nurses and the principle investigators. The CAB held a series of meetings prior to the study rollout and met every 6 months for the duration of the study. Meetings were held before the rollout of the study between the research team and NASCOP, the County HIV coordination teams, facility heads and frontline HCW. Thereafter, the study coordinator and study nurses had informal meetings with sub-County HIV programme managers, facility heads and frontline HCW at least monthly and formal meetings with the same stakeholders after completing study activities at each facility. Formal dissemination meetings with all stakeholders were convened after completion of the pilot study (year 1) and completion of the study in year 4. Discussions in all meetings were guided by presentations made by the research team. In addition, results were disseminated in local and international conferences through posters and oral presentations. Analysis of stakeholder engagement To describe the involvement of stakeholders in the CATCH study, we utilized a modified ‘power-interest’ stakeholder analysis model. While the original model classifies stakeholders as having high or low absolute power—where the stakeholder with the most power and influence is regarded as more important in the success of the project (Chinyio and Akintoye, 2008; Horton and Pilkington, 2014)—the context in the CATCH study was different. Given the decentralized but hierarchical structure of health service delivery within Kenya, we determined that all stakeholders had the power to enable or disable the project at any level, but there were institutions with higher hierarchical power and their support resulted in increased acceptance of the study by other stakeholders lower in the hierarchy. Moreover, we also applied the role-based model (Horton and Pilkington, 2014) to accurately describe the importance of the stakeholder engagement in the study. For example, engagement of the frontline HCW was instrumental in the successful execution of the study due to their role in adapting data collection tools and integrating study activities into clinic flow. However, their engagement was dependent on the study buy-in by the facility heads and sub-County HIV programme managers. Furthermore, the County HIV coordination team first required the research team to engage and receive approval from NASCOP and ethics teams, who only played a peripheral role in the actual execution of the study. The Kenya and US-based ethics bodies are mandated to determine ethical questions especially on consenting and study procedures. However, the contribution made by the bioethics centre and the CAB, despite not having hierarchical power, was pivotal, as described in the Results section. In each phase of the study, the authors retrospectively determined which stakeholders were engaged and their roles using the aforementioned data sources. They then made a judgement via consensus on the interest each stakeholder demonstrated and the impact they had in each of the three phases of the study (planning, execution and evaluation, and dissemination). Finally, they considered the hierarchical position of each stakeholder. Results Roles of various stakeholders Diverse stakeholders had evolving roles throughout the planning, execution and evaluation and dissemination phases of the study. The UW and UoN ethics reviewers and experts were heavily engaged during the planning phase of the study due to their regulatory responsibilities. The study design and study population targeted by CATCH made it critical to define how the study would protect children and their caregivers from potential psychological harms associated with paediatric HIV testing. After the ethics review committees reviewed and approved the protocol, their role shifted to oversight and compliance, with periodic monitoring and annual reports. The Treuman Katz Center for Pediatric Bioethics played a crucial role in designing the CATCH study during the planning phase. The study team sought input from the bioethicists on specific issues related to study participant consenting and assenting procedures. In the planning phase, MOH and CAB were engaged closely through a series of meetings due to their crucial role in ensuring acceptable and appropriate entry into facilities and client homes. In the execution and evaluation stage, The sub-County HIV programme managers and heads of facilities had substantive roles in the study, informing the integration of the study within existing HIV programme structures and in supervision, ensuring that study staff adhered to existing HIV testing protocols. The frontline HCW in the HIV clinics were engaged in the execution and evaluation phase of the study and recruiting participants and, through regular meetings, helped optimize study procedures. MOH and the CAB were kept informed through regular brief communications but did not have a direct supervisory role. In the dissemination phase, MOH played a more significant role since they made decisions on how the study results influenced national HIV testing policy and made the dissemination forums available. The study team kept close contact with MOH through frequent meetings and developed packages that helped them interpret the results correctly. The frontline HCW, facility heads and county health departments were kept informed by presenting facility-specific or county-specific results to them, while the CAB received results from all facilities (Figure 1). Figure 1 Open in new tabDownload slide Summary of contribution from stakeholders engaged in the CATCH study Figure 1 Open in new tabDownload slide Summary of contribution from stakeholders engaged in the CATCH study Aspects of the study most heavily impacted by stakeholder engagement Consent and assent During the planning stage, the study struggled with how to conduct appropriate consent and assent to balance the diverse needs and rights of children and adolescents aged 0–18 years. Kenya’s HIV testing guidelines required parental consent for child testing, but no guidance on child assent was provided. The ERC and CAB noted that children in the Kenyan context were not routinely asked to assent for medical procedures and that requiring assent for children might make caregivers reluctant to enrol in the study, limiting generalizability. They also noted that an assent that explicitly mentioned HIV testing could inadvertently disclose either a child’s or a caregiver’s HIV status to their children, making enrolment unacceptable to them. On the other hand, UW IRB and Trueman Katz Bioethics Center suggested that older children and adolescents had a right to receive their HIV test results. In view of the competing priorities, the bioethicists suggested that we restrict the study to include only children 0–12 years for testing and, further, offer an optional assent for children aged 7–12 years that did not mention HIV. The assent was offered to the child only if the caregiver agreed. They further suggested conducting a separate formative study to understand appropriate adolescent engagement approaches, which resulted in the development of the Developing Adolescent Strategies for HIV testing study (Wagner et al., 2017; Wilson et al., 2017). Approach to home visits During the planning phase, the study team considered which cadre of staff should conduct the home-based HIV testing visits, paying particular attention to the risk of stigmatizing families and potential for accidental disclosure, as well as the potential of programmatic scale-up. The CAB advised that a CHW should accompany study nurses during home visits to minimize stigmatization and to match with the role and scope of this cadre in programmatic care. The study engaged the CHW in charge of the community unit in which the household we visited for home-based testing was located. Since the standard scope of work for CHW did not already include HIV testing, transport reimbursement was important in successfully engaging this cadre and was provided by the study. The MOH representatives were engaged to support appropriate study design, staffing and testing procedures to ensure that the intervention was operationalized in a way that could be integrated into existing health systems. For example, children under 18 months who were tested at home required confirmatory polymerase chain reaction (PCR) testing; collecting this sample and returning results to the family at home or by phone would have competed with existing PMTCT systems and threatened appropriate linkage to care. MOH advised that these children should have their samples collected at home and immediately referred to their local PMTCT clinic to receive their PCR results. Integration with existing health systems The study aimed to leverage, and integrate into, existing systems for HIV test kit procurement and reporting. The study team engaged with sub-County HIV programme teams who supplied HIV test kits, allocated CHWs for home visits and kept the study team informed of other HIV testing activities in and around the study sites to improve scientific inference on intervention effectiveness. In turn, the study shared recruitment and testing data for reporting purposes and helped clinics diagnose historic gaps in service delivery. The study team also regularly engaged frontline HCW during execution and evaluation, providing regular updates and incorporating HCW feedback in operationalization. For example, HCW were instrumental in characterizing patient flow, identifying the optimal point of recruitment to maximize contact with all clients and refining study eligibility logs to meet the study objectives, but not be overly burdensome. In return, HCW were either reimbursed as individuals ($2.50/day), as a group, or nominated a peer educator that was hired by the study for participant recruitment and referral; while this was deemed necessary in a research context, this would likely not be feasible at programmatic scale. Dissemination CATCH study results were disseminated widely. In addition to conventional international conferences, the study team presented site-specific results in each of the participating clinics, at county headquarters and at various national MOH and local scientific meetings. During presentation of the site-specific results, the HCW provided context and possible explanations for the results presented especially on the proportion of caregivers reached and test completion. The main study results were published after many of these dissemination meetings. Results interpretation in this manuscript incorporated local feedback from the dissemination meetings (Wagner et al., 2016). Impact of stakeholder engagement on study validity Deliberate engagement of diverse stakeholders was instrumental in ensuring scientifically valid and pragmatic study findings and in enabling integration of the CATCH intervention in the national HIV programme. The study also benefitted from unique partnerships in staffing. Pairing CHWs with study nurses during home visits enhanced the completion of testing and linkage of HIV-infected children to HIV care. Strategic positioning of recruiting staff in the triage and pharmacy areas of HIV care clinics ensured that all patients in a day were approached about testing their children. Stakeholder engagement allowed the study to be culturally and locally informed. The optional child assent approach enabled the study to maximize external validity by not excluding a population who felt uncomfortable with child assent, and to balance respect for both parental and child rights regarding participation. Awareness of overlapping HIV testing mandates between research and programmes allowed quick adaptation to aid in achieving synergistic goals and helped to correctly interpret study results in the absence of a parallel control arm. Figure 2 summarizes the contribution each stakeholder made across the three phases of the study (planning, execution and evaluation and dissemination), and the effect their contribution had on the conduct of the study or interpretation of study results. Figure 2 Open in new tabDownload slide Illustration of stakeholder power and interest in the three phases of the CATCH study Figure 2 Open in new tabDownload slide Illustration of stakeholder power and interest in the three phases of the CATCH study Synthesis of stakeholder engagement using the modified power–interest framework In the planning phase, the ethics bodies had high hierarchical power due to their position as regulators. The study was not considered high risk and so their interest beyond performance of their initial regulatory roles was low. The national Ministry of Health also had high hierarchical power due to their policymaking functions and were highly interested in the actual study since it would inform their policy on HIV testing. The CAB had lower hierarchical power but had an oversight role detailed in the study protocol. Due to the composition of the CAB, they had high interest in the conduct and outcome of the study. The paediatrics bioethics centre had low hierarchical power and little interest beyond advising the study on the consenting and target population questions. In the execution and evaluation phase, the County Department of Health and facility heads had high hierarchical power since they managed the study sites and had a supervisory role on testing and referral procedures. They had high interest in the study outcome since the study was offering HIV testing services—one of their deliverables. Although the frontline HCW had lower hierarchical power than the facility heads and County, they had some power that they used to determine when and where to approach caregivers and refer them to study staff. HIV testing was an important deliverable of their usual role in the clinic, which made them have high interest on the study outcome. The ethics bodies, MOH and the CAB maintained their passive oversight roles and had no direct supervisory roles. In the dissemination phase, MOH showed high interest in the study outcome due to the implications it had on HIV testing policy. Their position as the policymaking body conferred high hierarchical power to determine when and how the dissemination was done. The County, facilities, CAB and attendees of dissemination meetings showed high interest in the study outcome due to the potential to reach untested children and gaps in testing the study showed. The County and facility heads, however, had limited hierarchical power that enabled them to dictate where and when County- and site-specific dissemination was done (Figure 2). Discussion Preliminary CATCH study results were released in 2015 allowing MOH to document CATCH as a best practice in paediatric HIV testing and linkage to care. The study team shared scripts, data collection tools and training material with the programme. Within 5 months of tool sharing, the Kenyan national HIV programme adopted the CATCH approach in a nationally scaled Rapid Results Initiative, testing >500 000 children for HIV. Testing children in clinics following index case identification was also incorporated into national HIV testing guidelines 2015 (Ministry of Health, 2015). The same toolkit was shared with the World Health Organization Inter-Agency Task Team on paediatric HIV. Since completion of CATCH in 2016, in Kenya, the approach was utilized on a wider scale to close the gap in HIV testing for contacts of persons living with HIV. For example, for the period between October 2016 and March 2018, clinics supported by the US Center for Disease Control and Prevention completed >200 000 HIV tests for children <15 years of adults in care and identified over 1400 children living with HIV (Lasry et al., 2019). Several African countries including Malawi and Zimbabwe (Ahmed et al., 2017; Chikwari et al., 2019; Lasry et al., 2019) have since adopted a similar model to reach more children at high risk of being HIV infected with HIV testing. There is limited literature about stakeholder engagement in research studies in low- and middle-income countries (Day et al., 2018). Previous studies have suggested that engaging stakeholders in these settings may be more challenging due to communication barriers, fewer resources and a lack of trust of research by communities (Newman et al., 2015); this is despite the appreciation that stakeholder engagement is even more important when conducting research in these contexts (Newman and Rubincam, 2014). Previous literature on stakeholder engagement demonstrated that few studies rigorously engage government stakeholders (11%) and healthcare stakeholders (21–46%), and that study participants (80%) and community leaders and community advisory groups (71%) are more frequently engaged (Concannon et al., 2014; Day et al., 2018). Stakeholder engagement in the CATCH study was done throughout the study cycle, including dissemination, differing from other research studies that mostly engage stakeholders during the initial stages only (Concannon et al., 2014; Day et al., 2018). Furthermore, CATCH engaged several key stakeholders including patient and community representatives (CAB), research stakeholders (ethics bodies), healthcare stakeholders (HCW) and government stakeholders (MOH/County), which presented a learning opportunity of how to engage diverse stakeholders with varying power, interest and roles. The success of the stakeholder engagement for the CATCH study with government stakeholders could also be attributed to institutional culture. Kenya’s HIV programme is regarded as an early adopter of evidence-based innovations in HIV prevention and treatment. It was among the first African countries to scale up the comprehensive prevention of mother-to-child transmission programme that included Option B (3-drug regimen of antiretroviral drugs during pregnancy) and Option B-plus (3-drug regimen of antiretroviral drugs for life) (Government of Kenya, 2012; du Plessis et al., 2014). It is also among the first countries to scale up HIV self-testing (World Health Organization, 2015), testing of newborns at birth and pre-exposure prophylaxis (Ministry of Health Kenya, 2018). In the formulation of these guidelines, Kenya’s HIV programme has routinely engaged researchers in academic institutions and other implementing partners. The data sources used in this analysis, including meeting minutes, field notes and reflections of the research team, have been utilized in previous stakeholder analyses (Martinez et al., 2019). The consideration of stakeholder’s power in analysing the stakeholder’s contribution, although widely used in stakeholder literature (Roome and Wijen, 2006; Thaler and Levin-Keitel, 2016; Colvin et al., 2020), is rarer in health-related literature, and even rarer from low- and middle-income countries. The CATCH study benefitted demonstrably from comprehensive and diverse stakeholder engagement by having a culturally sound intervention, buy-in from frontline HCW and adoption of the intervention in the national HIV care guidelines. The stakeholders had direct benefit (HCW, facility heads, County and MOH) due to the study helping them accomplish HIV testing, one of their core deliverable services and indirect benefit for patients who learned their children’s HIV status. This demonstration of the benefits of stakeholder engagement has previously been noted as a gap in stakeholder literature (Laura Esmail et al., 2015). This demonstration of sustained stakeholder engagement and its benefits to all parties should encourage researchers to invest time and other resources and to design studies that substantively include stakeholder engagement in their study procedures. Conclusions Engaging stakeholders early and frequently during intervention design and execution and evaluation ensures scientific validity, programmatic relevance and policy impact. This process should include parties at all levels of MOH, heath facilities and the community targeted for research. Involvement of frontline practitioners in the study improved morale and ownership of the intervention. It is also important to tailor compensation for research procedures to practitioners to what is most acceptable with the staff and management. Facility management can be engaged directly by offering supervisory roles in the study. This ensures experience is gained on how to run the intervention if adopted into programmatic practice. Active dissemination, including of preliminary results, methodology used in the research and data collection tools and material, is necessary to achieve scale. Using study designs that are more pragmatic in nature allow researchers to test 'real-life' scenarios, enhancing external validity and ensuring more rapid adoption of effective interventions.. Acknowledgements The CATCH study was funded by the University of Washington Royalty Research Fund (UW RRF; A83526) and the National Institutes of Health (NIH; R21 HD079637). C.M., V.O., V.O., M.N., F.N., E.M. and H.M. were supported by UW RRF A83526; A.D.W. was supported by the NIH F31 MH099988; I.N. and D.W. were supported by the NIH R01 HD023412; K.S. was supported by K02TW009207; G.J.-S. was supported by NIH R01 HD023412 and NIH K24 HD054314; and J.S. was supported by NIH K01 AI087369. We also thank UW Global Center for Integrated Health of Women, Adolescents and Children (Global WACh), ‘University of Washington/Fred Hutch Center for AIDS Research’ and International AIDS Research and Training Program (IARTP) for supporting the authors during preparation of this article. We appreciate the CATCH study caregivers and children without whom this research would have not been possible. We also thank participating site staff, County Directors of Health in Nairobi and Kisumu and their HIV management teams, Kenya National AIDS and STI Control Programme and the CAB for their support and constant engagement in the study. The contents of this paper are solely the responsibility of the authors and do not necessarily represent the official views of the NIH or UW. The funding sources were not involved in the preparation of this article. . Author contributions C.M., M.N., F.N., V.O., V.O., E.M., I.N., H.M., K.S., I.I., E.M.-O., D.W., G.J.-S., J.S. and A.D.W. were involved in data collection and dissemination activities and writing and review of this article. A.D.W., J.S., D.W., K.S. and G.J.-S. were involved in study design. Conflict of interest. None of the authors report any conflict of interest. Ethical approval. Kenyatta National Hospital Ethics and Research Committee and University of Washington Institutional Review Board provided ethical approval for the ‘Counseling and Testing of Children at Home (CATCH)’ study. References Ahmed S , Sabelli RA , Simon K et al. 2017 . Index case finding facilitates identification and linkage to care of children and young persons living with HIV/AIDS in Malawi . Tropical Medicine & International Health 22 : 1021 – 9 . Google Scholar Crossref Search ADS WorldCat Amirav I , Vandall-Walker V , Rasiah J , Saunders L. 2017 . Patient and researcher engagement in health research: a parent’s perspective . Pediatrics 140 : e20164127 . Google Scholar Crossref Search ADS PubMed WorldCat Balas EA , Boren SA. 2000 . Managing clinical knowledge for health care improvement . Yearbook of Medical Informatics 9 : 65 – 70 . Google Scholar Crossref Search ADS WorldCat Barker C , Mulaki A , Mwai D , Dutta A ; Health Policy Project and Futures Group. 2014 . Devolution of Healthcare in Kenya: Assessing County Health System Readiness in Kenya: A review of Selected Health Inputs. https://www.healthpolicyproject.com/pubs/479_KenyaPETSCountyReadinessFINAL.pdf, accessed 16 September 2018. Chikwari CD , Bandason T , Simms V et al. 2019 . Finding the Missing Children with HIV: Index-Linked Testing in Clinics & Communities . Seattle, WA : CROI . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Chinyio EA , Akintoye A. 2008 . Practical approaches for engaging stakeholders: findings from the UK . Construction Management and Economics 26 : 591 – 9 . Google Scholar Crossref Search ADS WorldCat Colvin R , Witt GB , Lacey J. 2020 . Power, perspective, and privilege: the challenge of translating stakeholder theory from business management to environmental and natural resource management . Journal of Environmental Management 271 : 110974 . Google Scholar Crossref Search ADS PubMed WorldCat Concannon TW , Fuster M , Saunders T et al. 2014 . A systematic review of stakeholder engagement in comparative effectiveness and patient-centered outcomes research . Journal of General Internal Medicine 29 : 1692 – 701 . Google Scholar Crossref Search ADS PubMed WorldCat Day S , Blumberg M , Vu T et al. 2018 . Stakeholder engagement to inform HIV clinical trials: a systematic review of the evidence . Journal of the International Aids Society 21 : e25174 . Google Scholar Crossref Search ADS PubMed WorldCat Du Plessis E , Shaw SY , Gichuhi M et al. 2014 . Prevention of mother-to-child transmission of HIV in Kenya: challenges to implementation . BMC Health Services Research 14 : S10 . Google Scholar Crossref Search ADS PubMed WorldCat Esmail L , Moore E , Rein A. 2015 . Evaluating patient and stakeholder engagement in research: moving from theory to practice . Journal of Comparative Effectiveness Research 4 : 133 – 45 . Google Scholar Crossref Search ADS PubMed WorldCat Glasgow RE , Lichtenstein E , Marcus AC. 2003 . Why don't we see more translation of health promotion research to practice? Rethinking the efficacy-to-effectiveness transition . American Journal of Public Health 93 : 1261 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat Goodman MS , Sanders Thompson VL. 2017 . The science of stakeholder engagement in research: classification, implementation, and evaluation . Translational Behavioral Medicine 7 : 486 – 91 . Google Scholar Crossref Search ADS PubMed WorldCat Government of Kenya 2012 . Guidelines for Prevention of Mother to Child Transmission of HIV/AIDS in Kenya, 4th edn. http://guidelines.health.go.ke : 8000/media/Guidelines_for_PMTCT_of_HIVAIDS_in_Kenya-1.pdf, accessed 21 July 2019. Haines A , Kuruvilla S , Borchert M. 2004 . Bridging the implementation gap between knowledge and action for health . Bulletin of the World Health Organization 82 : 724 – 31 . discussion 732. Google Scholar PubMed OpenURL Placeholder Text WorldCat Harrison JD , Auerbach AD , Anderson W et al. 2019 . Patient stakeholder engagement in research: a narrative review to describe foundational principles and best practice activities . Health Expectations 22 : 307 – 16 . Google Scholar Crossref Search ADS PubMed WorldCat Hoffman A , Montgomery R , Aubry W , Tunis SR. 2010 . How best to engage patients, doctors, and other stakeholders in designing comparative effectiveness studies . Health Affairs 29 : 1834 – 41 . Google Scholar Crossref Search ADS PubMed WorldCat Horton L , Pilkington A. 2014 . Rolling back from the power/interest matrix: a new approach for role based stakeholder engagement in projects . PM World Journal 111 : 121 – 5 . Google Scholar OpenURL Placeholder Text WorldCat Israel BA , Schulz AJ , Parker EA , Becker AB. 2008 . Critical issues in developing and following community-based participatory research principles. Community-Based Participatory Research for Health. San Francisco, CA: Jossey-Bass, 47 – 66 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Lasry A , Medley A , Behel S et al. 2019 . Scaling up testing for human immunodeficiency virus infection among contacts of index patients—20 countries, 2016–2018 . Morbidity and Mortality Weekly Report (MMWR) 68 : 474 – 7 . Google Scholar Crossref Search ADS WorldCat Martinez J , Wong C , Piersol CV et al. 2019 . Stakeholder engagement in research: a scoping review of current evaluation methods . Journal of Comparative Effectiveness Research 8 : 1327 – 41 . Google Scholar Crossref Search ADS PubMed WorldCat Ministry of Health. 2015 . The Kenya HIV Testing Guidelines. Nairobi: Ministry of Health. https://aidsfree.usaid.gov/sites/default/files/hts_policy_kenya_2015.pdf, accessed 21 July 2019. Ministry of Health Kenya. 2018 . Guideline On Use of Antiretroviral Drugs for Treating and Preventing HIV in Kenya [Online]. Nairobi. Available: http://cquin.icap.columbia.edu/wp-content/uploads/2017/04/ICAP_CQUIN_Kenya-ARV-Guidelines-2018-Final_20thAug2018.pdf, accessed October 17 2019]. Ministry of Public Health and Sanitation. 2010 . National Guidelines for HIV Testing and Counseling in Kenya. Nairobi: Ministry of Health. http://guidelines.health.go.ke : 8000/media/National_Guidelines_for_HTC_in_Kenya_2010_dWuc0Rr.pdf, accessed 15 September 2018. National AIDS and STI Control Programme (NASCOP). 2014 . Kenya AIDS Indicator Survey 2012: Final Report. Nairobi: Government of Kenya. https://nacc.or.ke/wp-content/uploads/2015/10/KAIS-2012.pdf, accessed 15 September 2018. Newman PA , Rubincam C. 2014 . Advancing community stakeholder engagement in biomedical HIV prevention trials: principles, practices and evidence . Expert Review of Vaccines 13 : 1553 – 62 . Google Scholar Crossref Search ADS PubMed WorldCat Newman PA , Rubincam C , Slack C et al. 2015 . Towards a science of community stakeholder engagement in biomedical HIV prevention trials: an embedded four-country case study . PLoS One 10 : e0135937 . Google Scholar Crossref Search ADS PubMed WorldCat Njuguna IN , Wagner AD , Cranmer LM et al. 2016 . Hospitalized children reveal health systems gaps in the mother–child HIV care cascade in Kenya . AIDS Patient Care and STDs 30 : 119 – 24 . Google Scholar Crossref Search ADS PubMed WorldCat Roome N , Wijen F. 2006 . Stakeholder power and organizational learning in corporate environmental management . Organization Studies 27 : 235 – 63 . Google Scholar Crossref Search ADS WorldCat Satcher D. 2005 . Methods in Community-Based Participatory Research for Health. John Wiley & Sons, Chapter 1 , 3 – 26 . Thaler T , Levin-Keitel M. 2016 . Multi-level stakeholder engagement in flood risk management—a question of roles and power . Environmental Science & Policy 55 : 292 – 301 . Google Scholar Crossref Search ADS WorldCat UNAIDS and AIDS Vaccine Advocacy Coalition. 2011 . Good Participatory Practice: Guidelines for Biomedical HIV Prevention Trials. https://www.avac.org/good-participatory-practice, accessed 30 June 2020. Wagner AD , Mugo C , Bluemer-Miroite S et al. 2017 . Continuous quality improvement intervention for adolescent and young adult HIV testing services in Kenya improves HIV knowledge . Aids 31(Suppl 3 ): S243 –S 252 . Google Scholar Crossref Search ADS PubMed WorldCat Wagner AD , Mugo C , Njuguna IN et al. 2016 . Implementation and operational research: active referral of children of HIV-positive adults reveals high prevalence of undiagnosed HIV . Journal of Acquired Immune Deficiency Syndromes 73 : e83 – e89 . Google Scholar Crossref Search ADS PubMed WorldCat Wamalwa D , Benki-Nugent S , Langat A et al. 2012 . Survival benefit of early infant antiretroviral therapy is compromised when diagnosis is delayed . The Pediatric Infectious Disease Journal 31 : 729 – 31 . Google Scholar Crossref Search ADS PubMed WorldCat Wilson KS , Beima-Sofie KM , Moraa H et al. 2017 . “At our age, we would like to do things the way we want:” a qualitative study of adolescent HIV testing services in Kenya . AIDS 31 : S213 – S220 . Google Scholar Crossref Search ADS PubMed WorldCat World Health Organization. 2015 . Consolidated Guidelines on HIV Testing Services. https://apps.who.int/iris/bitstream/handle/10665/179870/9789241508926_eng.pdf, accessed 21 July 2019. © The Author(s) 2020. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. All rights reserved. For permissions, please e-mail: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Kazerooni,, Yasaman;Mishra,, Brijesh;Gibran,, Nicole;Adu,, Emmanual;Clarke,, Damian;Pham,, Tam;Stewart, Barclay, T

doi: 10.1093/heapol/czaa031pmid: 33450768

Abstract The incidence of chemical assault is increasing globally. In response, a number of countries are proposing legislation. However, current legislative initiatives are uncoordinated and do not address the spectrum of activities to prevent and care for victims of chemical assaults. To fill this gap, we aimed to review legislation, policies and regulations relevant to chemical assault to classify and catalogue current strategies into a legislative framework. Terms related to chemical assault were used to systematically search the academic, lay and legal literatures. Chemical assault was defined as the use of acid or another caustic or corrosive substance or vitriol by one person against another with the intent to injure or disfigure. Reports that described the use of chemical weapons in warfare were excluded. A second search of national legislations of countries with reports of chemical assaults was performed to identify enacted laws and gaps in legislative approaches to chemical assault control. Data regarding relevant laws, policies and regulations were extracted. The findings were used to classify and catalogue current strategies into a legislative framework using content analysis. The search returned 3769 records. Chemical assaults were reported from 104 countries. Using legislation from those countries, a framework consisting of five legislative priorities was developed: (a) apply a public health approach; (b) adopt legal definitions specific to chemical assault; (c) control chemical supply, sales and procurement; (d) facilitate justice; and (e) support survivors. Although many countries have implemented one or more of these legislative priorities, no country has a comprehensive address to chemical assault control. Comparative policy analysis and assessments of the impacts of legislative efforts may further inform strategies to prevent, prosecute and mitigate the impact of chemical assaults. In the meantime, advocacy groups and governments might consider implementation and enforcement of one or more components of this legislative framework to control the growing epidemic of chemical assault. Chemical assault, acid attack, burn, health policy, legislation, global surgery Key Messages Despite a number of national and subnational initiatives to control chemical assaults, no country has implemented comprehensive legislation to address each of the causes and effects of chemical assault. Using existing legislative efforts and notable gaps, we created a compressive legislative framework that might inform strategies to prevent, prosecute and mitigate the impacts of chemical assaults: apply a public health approach; adopt legal definitions specific to chemical assault; control chemical supply, sales and procurement; facilitate justice; and support survivors. Introduction Chemical assault—the use of acid or another caustic or corrosive substance or vitriol by one person against another with the intent to injure or disfigure—has been perpetrated for centuries and is increasing in a number of communities (Mannan et al., 2007; Ahmed et al., 2017). Recent high-profile attacks and advocacy campaigns have brought chemical assault to public attention, and survivors are seeking systematic change to the way chemical assault is viewed and controlled through legislation (Kay, 2013; Bagcchi, 2015b; Ahmed et al., 2017). Although such attacks are thought to be more prevalent in the developing countries, their incidence is on the rise in higher-income countries (Ahmed et al., 2017; Sugrue et al., 2018). However, little is known about the incidence and distribution of these events globally or the ways in which governments are addressing chemical assault (Mannan et al., 2007). Chemical burns can rapidly cause grave and permanent skin and soft tissue damage and potentially result in substantial morbidity. Victims frequently require significant medical, social and financial resources to deal with the wounds, psychological trauma, loss of prior social role and stigma. In many cases, chemical assaults occur as a manifestation of gender-based violence (GBV) or intimate partner violence. These motives in particular have inspired local, national and global campaigns to address acid attacks and empower women and girls in communities with particularly significant gender inequity (Alishahi Tabriz et al., 2016; Thomson Reuters Foundation et al., 2017). Reports from several countries have described local efforts to limit access to corrosive substances, facilitate the prosecution of assailants and increase support for victims (Mannan et al., 2007; Kay, 2013; Bagcchi, 2015a; Ahmed et al., 2017). While these reports are illustrative, a compilation of legal efforts directed towards the comprehensive control of chemical assaults does not exist. Resultantly, policymakers lack the tools necessary to make informed decisions regarding potential legal avenues to address endemic and epidemic chemical assaults. To fill this gap, we aimed to systematically search the health, lay, and legal literatures for reports related to the legislative action around chemical assaults and gaps in legal protections. By compiling these reports, the findings could be used to classify and catalogue current legislative strategies designed to aid chemical assault control initiatives. In addition, we hope that the findings from this study spark further discussion within civil society and governments regarding a useful framework for chemical assault control globally. Methods Systematic search of health, lay and legal literatures We defined chemical assault as the use of acid or another caustic or corrosive substance or vitriol by one person against another with the intent to injure or disfigure. A systematic search strategy was designed to capture records related to chemical assault from the academic, lay and legal literatures using the following databases: PubMed (academic), World Health Organization (WHO) Catalog (academic), LexisNexis Academic (lay, legal), and LexisNexis Advance (legal). Each database was searched with inclusive database-specific language and relevant MeSH headings designed to generally represent: (acid OR chemical OR corrosive OR caustic OR vitriol) AND (attack OR assault) NOT (gout OR uric OR nucleic OR protein OR peptide OR plant OR amino OR migraine OR porphyrin OR porphyria OR valproate OR enamel) (see Supplementary Material). The search was limited to reports published since 2008 to give a contemporary perspective of the epidemiology of chemical assaults. Reports that described the use of chemical weapons in warfare or assault with hot or cold implements were excluded. We also excluded reports of immolation with or without the use of flammable substances (e.g. gasoline, alcohol and other volatile fluids) to focus on the use and control of non-petroleum-based chemicals and strictly chemical assault without immolation. Records and reports in non-English were translated with Google Translate (Google, USA). Screening and data management Records were screened by two authors (YK, BM) for relevance and duplicates were removed. A third author (BTS) provided arbitration for differences in criteria selection. Covidence software was used to manage record screening (Covidence, Australia). Full-text reports of screened records were reviewed by the same three authors. Given significant paucity of reports and significant heterogeneity of reporting, a narrative analysis was planned a priori without exclusion for reports with high risk of bias. The search, data management and reporting followed preferred reporting items for systematic reviews and meta-analyses guidelines (Moher et al., 2009). Data analysis and creation of a legislative framework In addition to retaining records that described one or more chemical assault(s) to map the global distribution of attacks, a list of records that pertained to legislation (i.e. regulations, polices, decrees, resolutions, codes, and laws) was created. Relevant full-text reports from this list were reviewed, and data regarding control initiatives were extracted. A content analysis framework was used to identify dimensions central to chemical assault control (Stergiou-Kita and Grigorovich, 2013). After, keywords that represented each dimension were used to back search within the legislations of each country identified by the initial search using LexisNexis Advance. By doing so, legislative initiatives from countries with documented cases of chemical assaults could be reviewed for chemical assault control initiatives and legal gaps. Thematic and particularly illustrative examples of legislative efforts were identified. Chemical control efforts were grouped into categories that represented similar legislative initiatives. Then, categories were refined into useful legislative priorities and described. Efforts that were associated with a report(s) that described the effect of a legal case, policy, regulation or law on the respective target(s) were prioritized. Results Systematic search Search of WHO Global Health Library, PubMed and LexisNexis resulted in 1406, 1402 and 961 records, respectively (Figure 1). After duplicates were removed, 3769 articles were screened for relevance; 1896 records were irrelevant. Of the remaining 471 full-text reports, 386 records met criteria for analysis. Of the 85 records excluded, 79 described chemical weapons or unintentional chemical burns and 6 discussed medical management or prognosis of chemical burns generally. The included reports contained data from 104 countries (Figure 2). Figure 1 Open in new tabDownload slide Preferred reporting items for systematic reviews and meta-analyses flow diagram. Figure 1 Open in new tabDownload slide Preferred reporting items for systematic reviews and meta-analyses flow diagram. Figure 2 Open in new tabDownload slide Global distribution of reports that describe incidences of chemical assaults. Figure 2 Open in new tabDownload slide Global distribution of reports that describe incidences of chemical assaults. A comprehensive legislative framework Compiling related policies, laws and regulations and legal gaps from countries with reported chemical assaults, a framework consisting of five legislative priorities was developed: (a) apply a public health approach; (b) adopt legal definitions specific to chemical assault; (c) control chemical supply, sales and procurement; (d) facilitate justice; and (e) support survivors (Figure 3). While a number of countries have incorporated one or more of these priorities into chemical assault control legislation, no country has implemented all of these legislative priorities. Figure 3 Open in new tabDownload slide Comprehensive legislative framework to control chemical assault. Figure 3 Open in new tabDownload slide Comprehensive legislative framework to control chemical assault. Apply a public health approach The grave impact of chemical assaults on individuals and communities has prompted many efforts in public education and empowerment in preventing, coping with or overcoming such attacks. For instance, in the UK, the ‘Report, Remove, Rinse’ campaign released by the National Health Service England aims to educate the public about what to do in the event of an acid attack, and by virtue of popular discussion, raises awareness about acid attacks more broadly. Local news agencies in the UK and India have employed undercover surveillance to raise awareness about the easy accessibility of the substances mostly used in chemical assaults and the deficiencies in government oversight of their sale and distribution. The Colombian government ratified the Natalia Ponce de León law of 2010 to raise awareness about the gravity of chemical assaults, which calls for public education and harsher punishment of chemical assault offenders. Lastly, the government of Nepal attempted to bring awareness to the country’s rampant gender violence by dedicating the year 2010 to ending gender violence. Adopt legal definitions specific to chemical assault Chemical assaults are unique compared to other types of assault in that they are typically meant to disfigure rather than kill and have a devastating long-lasting impact on many aspects of the victim’s life. Rehabilitation after such attacks requires prolonged multidisciplinary efforts from not only the medical and surgical staff but also social work to assist with re-integration into society, legal aid and law enforcement for the facilitation of justice. Consequently, some countries have created specific legislature to define chemical assaults. As example, the UK defined corrosive substances as a weapon in 2018 after a continuous annual rise in chemical assaults, which allowed for better regulation of sale and penalization of persons carrying such substances with intent to harm. Only nine countries that had documented reports of chemical assault had specified legislature regarding the use of acid or caustic or corrosive substances in acts of violence; the remaining countries penalized acid attacks case by case under legislation that define simple or aggravated assault, (attempted) murder or domestic violence. Lack of legislation specific to chemical assaults has been associated with highly variable penalties and room for bias or outside influence. For instance, offenders in the UK received anywhere between 21 months to life in prison (median 72 month) for committing a chemical assault. In Columbia, prior to passage of the Natalia Ponce de León law (2016), which provides definition to chemical assaults as unique forms of assault, such events were treated as minor forms of assault such as ‘beating’. In 2002, Bangladesh was the first country that provided specific legislature regarding chemical assaults and eventually saw a significant drop in the incidence of such cases. Although there are shortcomings in implementation of such legislature, having specific and clear laws regarding chemical assaults can lead to more effective allocation of medical and legal resources to victims, as well as potentially less biased penalization of the assailants. Control chemical supply, sales and procurement With the global rise in chemical assaults, many countries have attempted to regulate the sale and handling of acidic, caustic, corrosive and flammable substances. For example, in Bangladesh, producing, importing or carrying strong acids without a licence could be punishable by a 3–10 year prison sentence. Cambodia and India require vendors to sell only to adults over the age of 18 years who present government-issued identification and a professional reason for buying the substances. Buyers’ names must be entered into a registry. In the UK, the 2018 amendment to the 1953 Prevention of Crime Act recognized caustic and corrosive substances as a weapon similar to a firearm or knife. As such, possession of such chemicals with the intent to harm carries a prison sentence of up to 4 years. In India, the Model Rules of 2013 directed by the Supreme Court to all States and Union Territories includes a number of stipulations regarding acid sales and procurement, including: a clear system of renewable licencing, inspection mandates for local magistrates, policies regarding sales to individuals with valid documentation and of sane mind and a broader definition of caustic and corrosive substances and acid. Other countries enforce the international safety guidelines for transporting and handling of these substances. However, undercover reporting and review of national trends show that in countries of all income levels, such guidelines are rarely followed. In many cases, sellers of strong acids are part of the informal economy and are not aware of guidelines or laws, or do not feel compelled to follow them due to the lack of official enforcement. Facilitate justice Countries that created specific laws against chemical assault often included harsher penalties for assailants as a deterrent. For example, under the Acid Crime Prevention Act of 2002, offenders in Bangladesh could face the death penalty for their offence. Per Cambodia’s 2012 Acid Law, the intentional killing of the victim is punishable by a 15–30-year jail sentence, or a life sentence if the victim was proven to be ambushed or tortured prior to being killed. Offenders of attacks that did not result in a victim’s death are subject to 2–5-year prison sentences, which can be doubled if the attack led to permanent disability. Under Colombia’s Natalia Ponce de León Law in 2016, offenders could face anywhere between 12 and 50 years in prison. Despite these possible sentences, many attacks go unreported or uninvestigated, in part, due to the legal systems’ deficiencies in pursuit of such crimes. For instance, in Cambodia, many victims opt for out-of-court settlements that lead to immediate return of some financial compensation without having to go through long, complicated, expensive and often unreliable legal proceedings. In Uganda, cases involving chemical assault are often met with lengthy investigative and legal processes. Per a 2009 report from Nepal, few women file suits against their attackers and most are faced with resistance from the law enforcement who urge victims to rely on informal community settlements instead. In the UK, offenders have repeatedly been charged with minor assault and given prison sentences of only a few months to years. To overcome some of these barriers, some countries such as Bangladesh have dedicated national funds for survivor’s legal aid to incentivize reporting and facilitate prosecution, as well as creation of a special tribunal to hear survivor’s testimony in chemical assault cases in the absence of the assailant. Support survivors The potentially severe impact of chemical assaults on victims’ physical, psychological and financial well-being has led many countries to create legislature to provide victims with access to immediate medical care, rehabilitation and resources for community re-integration. In Cambodia, the Law on the Management of Strong Acid allows for free medical care, legal support and protection to chemical assault survivors and encourages the participation of donors, associations, non-governmental organizations (NGOs) and the private sector to support survivorship programmes. The 2002 Bangladesh Acid Law allocated national funds to support survivors of chemical assault through creation of free rehabilitation centres and dedicated legal aid to victims to incentivize reporting. India has allotted financial aid to victims and has urged both public and private hospitals to treat chemical assault survivors free of charge. Colombian Decree 1033 from 2014 stipulates the pathway for the comprehensive attention of chemical assault survivors to include: comprehensive medical care, protection of the survivor and their family and continued and adapted employment. Several countries with critical healthcare deficiencies, such as Nepal, have dedicated funding to improve the availability of medical resources for survivors around the country by training more burn care and reconstruction practitioners. However, the rising incidence and the high costs associated with chemical assaults have caused several countries to fall short of their intent. For instance, in 2013 and 2014, India approved 46% of chemical assault cases for financial support and, of those, only 33% of the survivors received financial compensation, which often came months after the attack. Moreover, in many Indian states, hospitals are not compensated by the government for the services they provide, thereby limiting care delivery on the supply side. Discussion This systematic review aimed to broadly describe the global distribution of chemical assault and legislative actions to address it. The findings were then used to classify and catalogue current legislative strategies that aim to control chemical assaults into a legislative framework. Five legislative priorities are outlined: (a) apply a public health approach; (b) adopt legal definitions specific to chemical assault; (c) control chemical supply, sales and procurement; (d) facilitate justice; and (e) support survivors. Table 1 Legislation that applies a public health approach to chemical assault controls Country . Organization (years) . Public awareness/education . Effects . Afghanistan Afghanistan Independent Human Rights Commission (2017/2018) Report of violence against thousands of women and children. Girls seeking education are persistently targeted. Of the 237 cases of women seeking justice majority were turned away and were encouraged to accept mediation. Bangladesh Prothom Alo Newspaper (2009) Major daily newspaper fundraising for the survivors and raising awareness. India Deli University Students (2013) Freeze mob in metro-station. Stop Acid Attacks (2014) Acid Attack Survivors (2014) Make Love Not Scars (2015) Protest at metro-station, part of the ‘Spot of Shame’ campaign to raise awareness at the locations women have been attacks. 3-day protest in the city requesting regulation of acid sale, improved victim rehabilitation, harsher penalization of acid attacks. Make up tutorial by acid attack survivor to raise awareness. Iran Thousands protest string of acid attacks thought to be motivated by hardliners against women whose outfits were thought to be not Islamic (2014). Multiple journalists who reported the attacks were detained. The police temporarily detained a high-profile activist at the demonstrations. The president called for immediate identification and arrest of the offenders. Nepal 2010 was declared as the year to end gender-based violence and an action committee was created UK Emergency Medics, NHS (2017) The Sun (2015, 2017) The Police and the London Fire Brigade (2017, 2018) Emergency medics as well as the NHS have released public guidelines on what to do in the event of an acid attack: ‘Report, Remove, Rinse’ campaign. Renowned news organization, urged the Government to make it illegal for minors to buy household products containing acid or ammonia, as well as invoke tougher penalties for the offenders. The conservative MSP, Rachael Hamilton, has called for more record keeping of the acid attack incidents. The police is going to use stop-and-search powers to target people who carry acid in a crackdown on the possession of corrosive substances. In London Police are being equipped with 1000 acid response kits to combat the surge in attacks. Since the LFB has gallons of water available to them, per a recent protocol, if an incident is flagged as an acid attack, it will also be flagged to the LFB. After a tragic acid attack in a night club, nightclub staff in east London are set to undergo training on providing immediate support to patients. Hospitals are being urged to share data on acid attacks in London due to the concern of high number of unreported crimes. The police are using anonymous details that have been provided by the victims to the Emergency Department staff to identify previously unknown hotspots for acid attacks. Twenty-eight of 30 Emergency Departments have also joined the movement. Country . Organization (years) . Public awareness/education . Effects . Afghanistan Afghanistan Independent Human Rights Commission (2017/2018) Report of violence against thousands of women and children. Girls seeking education are persistently targeted. Of the 237 cases of women seeking justice majority were turned away and were encouraged to accept mediation. Bangladesh Prothom Alo Newspaper (2009) Major daily newspaper fundraising for the survivors and raising awareness. India Deli University Students (2013) Freeze mob in metro-station. Stop Acid Attacks (2014) Acid Attack Survivors (2014) Make Love Not Scars (2015) Protest at metro-station, part of the ‘Spot of Shame’ campaign to raise awareness at the locations women have been attacks. 3-day protest in the city requesting regulation of acid sale, improved victim rehabilitation, harsher penalization of acid attacks. Make up tutorial by acid attack survivor to raise awareness. Iran Thousands protest string of acid attacks thought to be motivated by hardliners against women whose outfits were thought to be not Islamic (2014). Multiple journalists who reported the attacks were detained. The police temporarily detained a high-profile activist at the demonstrations. The president called for immediate identification and arrest of the offenders. Nepal 2010 was declared as the year to end gender-based violence and an action committee was created UK Emergency Medics, NHS (2017) The Sun (2015, 2017) The Police and the London Fire Brigade (2017, 2018) Emergency medics as well as the NHS have released public guidelines on what to do in the event of an acid attack: ‘Report, Remove, Rinse’ campaign. Renowned news organization, urged the Government to make it illegal for minors to buy household products containing acid or ammonia, as well as invoke tougher penalties for the offenders. The conservative MSP, Rachael Hamilton, has called for more record keeping of the acid attack incidents. The police is going to use stop-and-search powers to target people who carry acid in a crackdown on the possession of corrosive substances. In London Police are being equipped with 1000 acid response kits to combat the surge in attacks. Since the LFB has gallons of water available to them, per a recent protocol, if an incident is flagged as an acid attack, it will also be flagged to the LFB. After a tragic acid attack in a night club, nightclub staff in east London are set to undergo training on providing immediate support to patients. Hospitals are being urged to share data on acid attacks in London due to the concern of high number of unreported crimes. The police are using anonymous details that have been provided by the victims to the Emergency Department staff to identify previously unknown hotspots for acid attacks. Twenty-eight of 30 Emergency Departments have also joined the movement. LFB, London Fire Brigade; NHS, National Health Service. Open in new tab Table 1 Legislation that applies a public health approach to chemical assault controls Country . Organization (years) . Public awareness/education . Effects . Afghanistan Afghanistan Independent Human Rights Commission (2017/2018) Report of violence against thousands of women and children. Girls seeking education are persistently targeted. Of the 237 cases of women seeking justice majority were turned away and were encouraged to accept mediation. Bangladesh Prothom Alo Newspaper (2009) Major daily newspaper fundraising for the survivors and raising awareness. India Deli University Students (2013) Freeze mob in metro-station. Stop Acid Attacks (2014) Acid Attack Survivors (2014) Make Love Not Scars (2015) Protest at metro-station, part of the ‘Spot of Shame’ campaign to raise awareness at the locations women have been attacks. 3-day protest in the city requesting regulation of acid sale, improved victim rehabilitation, harsher penalization of acid attacks. Make up tutorial by acid attack survivor to raise awareness. Iran Thousands protest string of acid attacks thought to be motivated by hardliners against women whose outfits were thought to be not Islamic (2014). Multiple journalists who reported the attacks were detained. The police temporarily detained a high-profile activist at the demonstrations. The president called for immediate identification and arrest of the offenders. Nepal 2010 was declared as the year to end gender-based violence and an action committee was created UK Emergency Medics, NHS (2017) The Sun (2015, 2017) The Police and the London Fire Brigade (2017, 2018) Emergency medics as well as the NHS have released public guidelines on what to do in the event of an acid attack: ‘Report, Remove, Rinse’ campaign. Renowned news organization, urged the Government to make it illegal for minors to buy household products containing acid or ammonia, as well as invoke tougher penalties for the offenders. The conservative MSP, Rachael Hamilton, has called for more record keeping of the acid attack incidents. The police is going to use stop-and-search powers to target people who carry acid in a crackdown on the possession of corrosive substances. In London Police are being equipped with 1000 acid response kits to combat the surge in attacks. Since the LFB has gallons of water available to them, per a recent protocol, if an incident is flagged as an acid attack, it will also be flagged to the LFB. After a tragic acid attack in a night club, nightclub staff in east London are set to undergo training on providing immediate support to patients. Hospitals are being urged to share data on acid attacks in London due to the concern of high number of unreported crimes. The police are using anonymous details that have been provided by the victims to the Emergency Department staff to identify previously unknown hotspots for acid attacks. Twenty-eight of 30 Emergency Departments have also joined the movement. Country . Organization (years) . Public awareness/education . Effects . Afghanistan Afghanistan Independent Human Rights Commission (2017/2018) Report of violence against thousands of women and children. Girls seeking education are persistently targeted. Of the 237 cases of women seeking justice majority were turned away and were encouraged to accept mediation. Bangladesh Prothom Alo Newspaper (2009) Major daily newspaper fundraising for the survivors and raising awareness. India Deli University Students (2013) Freeze mob in metro-station. Stop Acid Attacks (2014) Acid Attack Survivors (2014) Make Love Not Scars (2015) Protest at metro-station, part of the ‘Spot of Shame’ campaign to raise awareness at the locations women have been attacks. 3-day protest in the city requesting regulation of acid sale, improved victim rehabilitation, harsher penalization of acid attacks. Make up tutorial by acid attack survivor to raise awareness. Iran Thousands protest string of acid attacks thought to be motivated by hardliners against women whose outfits were thought to be not Islamic (2014). Multiple journalists who reported the attacks were detained. The police temporarily detained a high-profile activist at the demonstrations. The president called for immediate identification and arrest of the offenders. Nepal 2010 was declared as the year to end gender-based violence and an action committee was created UK Emergency Medics, NHS (2017) The Sun (2015, 2017) The Police and the London Fire Brigade (2017, 2018) Emergency medics as well as the NHS have released public guidelines on what to do in the event of an acid attack: ‘Report, Remove, Rinse’ campaign. Renowned news organization, urged the Government to make it illegal for minors to buy household products containing acid or ammonia, as well as invoke tougher penalties for the offenders. The conservative MSP, Rachael Hamilton, has called for more record keeping of the acid attack incidents. The police is going to use stop-and-search powers to target people who carry acid in a crackdown on the possession of corrosive substances. In London Police are being equipped with 1000 acid response kits to combat the surge in attacks. Since the LFB has gallons of water available to them, per a recent protocol, if an incident is flagged as an acid attack, it will also be flagged to the LFB. After a tragic acid attack in a night club, nightclub staff in east London are set to undergo training on providing immediate support to patients. Hospitals are being urged to share data on acid attacks in London due to the concern of high number of unreported crimes. The police are using anonymous details that have been provided by the victims to the Emergency Department staff to identify previously unknown hotspots for acid attacks. Twenty-eight of 30 Emergency Departments have also joined the movement. LFB, London Fire Brigade; NHS, National Health Service. Open in new tab Table 2 Legislation that adopts legal definitions specific to chemical assault Country . Bill name (year) . Bill content . Effects . Afghanistan Elimination of Violence Against Women (2009) The EVAW bill which was signed by the president in 2009, specified causing harm to women by spraying or forced ingestion of harmful chemicals is a crime. Per UN study of 233 events, in 2017–18, majority of complaints by victims was deferred to out-of-court settlements. Afghanistan Independent Human Rights Commission reported that incidents of violence against women remain severely under-reported due to the traditional value system, stigmatization and fear of retaliation of the victim. In 2013, the parliament refused to discuss the new law, condemning it as un-Islamic. Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) Bangladesh was the first country to pass a law banning acid violence. With a drastic drop in the frequency of acid assaults, Bangladesh is no longer the country with the highest number of such attacks and the highest incident rates for women. Some claim the lower numbers cannot be directly attributed to the law, but Selina Ahmed, the executive director of the Acid Survivors Foundation in Dhaka, believes that it had an effect. According to Acid Survivors Foundation, the number of known victims dropped by 45% between 2001 and 2007. Acid attacks continued to decrease to 59 occurrences in 2015 from 494 in 2002, according to Ahmed’s foundation. Cambodia Acid law (2012) Cambodia’s National Assembly in 2002 declined to enact legislation to regulate acid sales or punishment. After acid attacks on a young woman in Phnom Penh and a man in Kompong Cham province, The Interior Ministry claimed the government had formed a committee to study ways to suppressing acid crimes. Prior to the Acid law, acid attacks were prosecuted as ordinary assaults or attempted murders. The Acid Law, criminalized acid violence and established regulations governing the criminal liability of individuals and legal entities. The new law has been hard to implement due to Cambodia’s defective legal system. According to a Cambodia base NGO, the legislation has led to a significant drop in the number of reported attacks, from 36 survivors in 2010 to 6 in 2014, 3 in 2016 and 3 in the first 4 months of 2017. Colombia Natalia Ponce de León law (2016) Prior to this law, acid attacks were treated as lesser physical aggressions, like beatings, and few attackers were convicted and jailed. According to a 2015 study by National Institute of Legal Medicine, from 2004–14, there were 926 acid attack victims reported in Colombia, of which 565 (61%) were women.In 2017, there were 91 reported cases of attacks with corrosive substances. Fifty-seven of the victims were women, and four out of the six victims who died were women. Finland Criminal Code, Chapter 21 Sections 5 (1996), 6 (2001) No specific acid laws. Section 5: Assault is punishable by a fine and prison sentence of up to 2 years. Attempt is also punishable. Section 6: Aggravated assault is defined as grievous bodily injury or serious illness is caused to another or another is placed in mortal danger, (2) the offence is committed in a particularly brutal or cruel manner, or (3) a comparable lethal instrument is used and the offence is aggravated also when assessed as a whole. Attempt is punishable Section 6(a) recognizes possession of a lethal weapon as ‘Preparation of an aggravated offence against life or health’. India The Criminal Law (Amendment) Ordinance/Manodhairya Scheme (2013) Section 326A, 326B Prior to the 2013 ruling there were no laws dealing specifically with acid attacks, and acid attacks were a bailable offence but now offenders should not get bail. The Law Commission has recommended classifying acid attacks as an extremely violent crime. Section 326A covers harming via acid attack and 326B covers attempted acid attack. Madhya Pradesh promised that his government would book culprits in acid attack cases for attempt to murder (Section 307 of the Indian Penal Code). Union home minister Rajnath Singh has set to strengthen the legal and administrative framework against acid attacks by treating acid attacks that cause grievous hurt as ‘heinous’ crimes. Under Section 326A, a total of 203 cases were reported in 2014, 222 in 2015 (increase of 9.4%), 260 in 2016 (increase of 17%) involving 281 victims. In 2016, a total of 233 people were arrested in crimes related to acid attacks, 90% of whom were charged. Italy Law no. 154 (2001) 4th titled: Measures against intra-family violence It specifically addresses domestic violence but not acid attacks. The law requires that women must report a domestic violence offence within 3 months of a ‘first-time’ event. Jordan General recommendation of no. 19 (1994) The committee considers violence against women in the family a violation of the convention, but does not specify the type of attack. The committee states that state parties should be held accountable if they do not act to prevent violence against women or punish acts of violence against women. It also specifically stated that ‘while state parties may argue many traditional practices such as dowry deaths, acid attacks, and forced marriage are justifiable on the basis of protecting or controlling women, such violence directly prohibits women from enjoying their fundamental rights and freedoms’. Kenya Act no. 12: Prevention of torture act (2017) Includes ‘burning by acid’ or rubbing acid on mucus membranes as a form of physical torture. Once a police officer receives a complaint of an offence under this act, it needs to be forwarded immediately to higher levels (the officer-in-charge of the police station and Directorate of Criminal Investigation), which shall investigate the crime immediately. The complaint shall be recorded in writing. Where a complaint regarding an offence under this act is reported to a court or raised in the process of a trial, the court shall record the complaint and order investigation immediately and order submission of an investigation report within 7 days of the court's order. An order for investigation by the court shall include medical and psychological assessment of harm suffered by the alleged victim. Victim impact statement can be included. Malaysia Act no. 357: Corrosive and Explosive Substances and Offensive Weapons Act (1958) The act defines ‘corrosive substance’ as any of the acids and substances specified in the first schedule and includes all substances, which are capable of application to the human body of causing hurt through corrosive action. It is presumed that any substance was intended to be used for the purpose of causing hurt if there is too much quantity, held in places other than their primary use, or held in a container other than the original container. This law was amended five times, last 2014. Nepal Clause 7.6 of the Comprehensive Peace Agreement (2006) There are no specific laws addressing acid attacks in Nepal. The local district court issued directives to parliament as well as the Ministry of law, justice and parliamentary affairs to formulate appropriate legal provisions to deal with acid attack cases, as existing legal provisions allow the courts to award the offenders a maximum of 3 years in prison. The Interim Constitution prohibits ‘physical, mental or any other form of violence’ against women. The 5-year strategic plan of the National Women’s Commission deals extensively with strategies to combat violence against women, although it does not address Acid and Burn Violence as such. Clause 7.6 of the agreement requires that all types of violence against women and children be immediately prohibited. Per a 2009 report, only few women had attempted to file complaints with police. They were met by the police either refusing to file their cases or failing to take proper legal action. Per Advocacy Forum-Nepal, police urge women to use informal community justice systems or to settle on a deal with the perpetrator. Nigeria Violence against Persons Prohibition Act (2015) The 2015 Act: considers acid baths a criminal and punishable offence. Attempt to commit an offence is an offence in itself. The major drawback in relation to the VAPP is its limited application to the Federal Capital Territory, Abuja. Pakistan Criminal Law Amendment Act (2011) The Acid and Burn Crime Bill (2017) Through the Act several changes were introduced in the PPC and Code of Criminal Procedure: the words ‘disfigure’ and ‘deface’ were added in Section 332 of the PPC, dealing with different kinds of hurt, stating ‘disfigure means disfigurement of face or disfigurement or dismemberment of any organ or any part of the organ of human body which impairs or injures or corrodes or deforms the symmetry or appearance of a person’. The Acid and Burn Crime Bill, 2014, which was once cleared for passage by a committee of the lower house but remained pending for over 3 years, was deferred again without being considered by the National Assembly. However, on 8 May 2018, the Pakistani National Assembly passed a revised version of the Acid Burn Crime Bill (2017). The bill not only has emphasized a short legal process for the victims but also mandates the government to pay for the health care fee of the victims, allowing them to receive care free of charge. The Criminal Law Amendment Act 2011 has already acknowledged acid throwing as a crime and, after that, not only has the number of cases been reduced, the conviction rate has also increased by 10%. According to Acid Survivors Foundation Pakistan, there were 71 reported acid attack cases in 2016 and 69 in 2015, down from 153 in 2014 and 109 in 2013. Philippines Anti-violence against Women and their Children Act 9262 (2004) The act criminalizes physical, sexual and psychological violence against women and children. It also criminalizes violence against any woman with whom a man has an intimate relationship, not only a wife. The law is gender specific and applies only to women and their children. The law does not specify chemical assaults. South Africa Domestic Violence Act 116 (1998) The act recognizes ‘burns’ as physical abuse but does not specify Acid attacks. It also considers threats as an offence. It is not gender specific. Officers at the scene should help the complainant and can make arrests. Complainant can apply for a court protection order. Application can be brought forth on behalf of the complainant by another person, such as social or health care worker, given they have consent to do so. Unless the complainant is a child, is developmentally disabled, unconscious, has court order. It does not mention anything about victim rehabilitation efforts. In a 2015 incident, the offender was charged with two counts of attempted murder for throwing acid at a parent and their child but the judge found that the victim was ‘no angel’ and convicted the offender on a lesser charge. Tanzania There are no laws regarding ‘domestic violence’ unless it involves a married person. violence against married women is addressed in the Law of Marriage Act (2002). Despite the 2013 attack on two volunteer tourists that caught the international media’s attention, no specific laws against acid attacks were implemented. Thailand Domestic Violence Victim Protection Act, B.E. 2550 (2007) Defines domestic violence broadly to include: any act done either intentionally or in a manner likely to cause harm to body, mind, or health of a family member Does not specify acid attacks. Domestic violence is counted as a ‘compoundable offence’: the offender and victim are able to compromise rather than to go to trial. Turkey Law No. 6284, Prevention of Violence against Women and the Protection of the Family (2012) Includes all women. Includes physical abuse but does not specify acid attacks. Uganda Penal Code Cap 120 (1950) Recognizes acid violence as an offence and describes sentencing for perpetrators: should an acid attack committed with ‘malice aforethought’ lead to death and then it is considered murder (Section188). A government chemist must also give evidence to show a corrosive chemical was used. Prosecutions rarely entail the murder or attempted murder charge due difficulties in proving intent when an assault is premediated—requiring the procurement of acid, hiring of the assailant and planning the attack. Instead, suspects are often charged with the lesser crime of grievous harm (Section 219), or actual bodily harm (Section 236). There is only one government chemist, based in Kampala, serving the entire country, leading to long delays in court hearings, and costly transport of exhibits. UAE There are no laws specific to acid attacks. In 2010, the UAE Federal Supreme Court affirmed that UAE law permits a husband to physically chastise his wife leading to a UN Committee on women’s rights strongly criticizing the country for lack of a law on domestic violence. UK Prevention of Crime Act: Offensive Weapons (1953) Stephen Timms, a former Labor minister, said carrying sulphuric acid without justification should be made a crime—echoing the change in the law to make carrying knives an offence. Prior to 2018, there was no penalty for being caught with the chemicals commonly used in acid attacks. Per the new guidelines, acid is defined as a ‘highly dangerous weapon’ for the first time, allowing judges to, under the Prevention of Crime Act, impose harsher punishments on anyone found to be carrying it in public. USA Assault with acid or corrosive substances is considered a felony and can be prosecuted under the category Aggravated Assault—Other Dangerous Weapon (4c). Country . Bill name (year) . Bill content . Effects . Afghanistan Elimination of Violence Against Women (2009) The EVAW bill which was signed by the president in 2009, specified causing harm to women by spraying or forced ingestion of harmful chemicals is a crime. Per UN study of 233 events, in 2017–18, majority of complaints by victims was deferred to out-of-court settlements. Afghanistan Independent Human Rights Commission reported that incidents of violence against women remain severely under-reported due to the traditional value system, stigmatization and fear of retaliation of the victim. In 2013, the parliament refused to discuss the new law, condemning it as un-Islamic. Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) Bangladesh was the first country to pass a law banning acid violence. With a drastic drop in the frequency of acid assaults, Bangladesh is no longer the country with the highest number of such attacks and the highest incident rates for women. Some claim the lower numbers cannot be directly attributed to the law, but Selina Ahmed, the executive director of the Acid Survivors Foundation in Dhaka, believes that it had an effect. According to Acid Survivors Foundation, the number of known victims dropped by 45% between 2001 and 2007. Acid attacks continued to decrease to 59 occurrences in 2015 from 494 in 2002, according to Ahmed’s foundation. Cambodia Acid law (2012) Cambodia’s National Assembly in 2002 declined to enact legislation to regulate acid sales or punishment. After acid attacks on a young woman in Phnom Penh and a man in Kompong Cham province, The Interior Ministry claimed the government had formed a committee to study ways to suppressing acid crimes. Prior to the Acid law, acid attacks were prosecuted as ordinary assaults or attempted murders. The Acid Law, criminalized acid violence and established regulations governing the criminal liability of individuals and legal entities. The new law has been hard to implement due to Cambodia’s defective legal system. According to a Cambodia base NGO, the legislation has led to a significant drop in the number of reported attacks, from 36 survivors in 2010 to 6 in 2014, 3 in 2016 and 3 in the first 4 months of 2017. Colombia Natalia Ponce de León law (2016) Prior to this law, acid attacks were treated as lesser physical aggressions, like beatings, and few attackers were convicted and jailed. According to a 2015 study by National Institute of Legal Medicine, from 2004–14, there were 926 acid attack victims reported in Colombia, of which 565 (61%) were women.In 2017, there were 91 reported cases of attacks with corrosive substances. Fifty-seven of the victims were women, and four out of the six victims who died were women. Finland Criminal Code, Chapter 21 Sections 5 (1996), 6 (2001) No specific acid laws. Section 5: Assault is punishable by a fine and prison sentence of up to 2 years. Attempt is also punishable. Section 6: Aggravated assault is defined as grievous bodily injury or serious illness is caused to another or another is placed in mortal danger, (2) the offence is committed in a particularly brutal or cruel manner, or (3) a comparable lethal instrument is used and the offence is aggravated also when assessed as a whole. Attempt is punishable Section 6(a) recognizes possession of a lethal weapon as ‘Preparation of an aggravated offence against life or health’. India The Criminal Law (Amendment) Ordinance/Manodhairya Scheme (2013) Section 326A, 326B Prior to the 2013 ruling there were no laws dealing specifically with acid attacks, and acid attacks were a bailable offence but now offenders should not get bail. The Law Commission has recommended classifying acid attacks as an extremely violent crime. Section 326A covers harming via acid attack and 326B covers attempted acid attack. Madhya Pradesh promised that his government would book culprits in acid attack cases for attempt to murder (Section 307 of the Indian Penal Code). Union home minister Rajnath Singh has set to strengthen the legal and administrative framework against acid attacks by treating acid attacks that cause grievous hurt as ‘heinous’ crimes. Under Section 326A, a total of 203 cases were reported in 2014, 222 in 2015 (increase of 9.4%), 260 in 2016 (increase of 17%) involving 281 victims. In 2016, a total of 233 people were arrested in crimes related to acid attacks, 90% of whom were charged. Italy Law no. 154 (2001) 4th titled: Measures against intra-family violence It specifically addresses domestic violence but not acid attacks. The law requires that women must report a domestic violence offence within 3 months of a ‘first-time’ event. Jordan General recommendation of no. 19 (1994) The committee considers violence against women in the family a violation of the convention, but does not specify the type of attack. The committee states that state parties should be held accountable if they do not act to prevent violence against women or punish acts of violence against women. It also specifically stated that ‘while state parties may argue many traditional practices such as dowry deaths, acid attacks, and forced marriage are justifiable on the basis of protecting or controlling women, such violence directly prohibits women from enjoying their fundamental rights and freedoms’. Kenya Act no. 12: Prevention of torture act (2017) Includes ‘burning by acid’ or rubbing acid on mucus membranes as a form of physical torture. Once a police officer receives a complaint of an offence under this act, it needs to be forwarded immediately to higher levels (the officer-in-charge of the police station and Directorate of Criminal Investigation), which shall investigate the crime immediately. The complaint shall be recorded in writing. Where a complaint regarding an offence under this act is reported to a court or raised in the process of a trial, the court shall record the complaint and order investigation immediately and order submission of an investigation report within 7 days of the court's order. An order for investigation by the court shall include medical and psychological assessment of harm suffered by the alleged victim. Victim impact statement can be included. Malaysia Act no. 357: Corrosive and Explosive Substances and Offensive Weapons Act (1958) The act defines ‘corrosive substance’ as any of the acids and substances specified in the first schedule and includes all substances, which are capable of application to the human body of causing hurt through corrosive action. It is presumed that any substance was intended to be used for the purpose of causing hurt if there is too much quantity, held in places other than their primary use, or held in a container other than the original container. This law was amended five times, last 2014. Nepal Clause 7.6 of the Comprehensive Peace Agreement (2006) There are no specific laws addressing acid attacks in Nepal. The local district court issued directives to parliament as well as the Ministry of law, justice and parliamentary affairs to formulate appropriate legal provisions to deal with acid attack cases, as existing legal provisions allow the courts to award the offenders a maximum of 3 years in prison. The Interim Constitution prohibits ‘physical, mental or any other form of violence’ against women. The 5-year strategic plan of the National Women’s Commission deals extensively with strategies to combat violence against women, although it does not address Acid and Burn Violence as such. Clause 7.6 of the agreement requires that all types of violence against women and children be immediately prohibited. Per a 2009 report, only few women had attempted to file complaints with police. They were met by the police either refusing to file their cases or failing to take proper legal action. Per Advocacy Forum-Nepal, police urge women to use informal community justice systems or to settle on a deal with the perpetrator. Nigeria Violence against Persons Prohibition Act (2015) The 2015 Act: considers acid baths a criminal and punishable offence. Attempt to commit an offence is an offence in itself. The major drawback in relation to the VAPP is its limited application to the Federal Capital Territory, Abuja. Pakistan Criminal Law Amendment Act (2011) The Acid and Burn Crime Bill (2017) Through the Act several changes were introduced in the PPC and Code of Criminal Procedure: the words ‘disfigure’ and ‘deface’ were added in Section 332 of the PPC, dealing with different kinds of hurt, stating ‘disfigure means disfigurement of face or disfigurement or dismemberment of any organ or any part of the organ of human body which impairs or injures or corrodes or deforms the symmetry or appearance of a person’. The Acid and Burn Crime Bill, 2014, which was once cleared for passage by a committee of the lower house but remained pending for over 3 years, was deferred again without being considered by the National Assembly. However, on 8 May 2018, the Pakistani National Assembly passed a revised version of the Acid Burn Crime Bill (2017). The bill not only has emphasized a short legal process for the victims but also mandates the government to pay for the health care fee of the victims, allowing them to receive care free of charge. The Criminal Law Amendment Act 2011 has already acknowledged acid throwing as a crime and, after that, not only has the number of cases been reduced, the conviction rate has also increased by 10%. According to Acid Survivors Foundation Pakistan, there were 71 reported acid attack cases in 2016 and 69 in 2015, down from 153 in 2014 and 109 in 2013. Philippines Anti-violence against Women and their Children Act 9262 (2004) The act criminalizes physical, sexual and psychological violence against women and children. It also criminalizes violence against any woman with whom a man has an intimate relationship, not only a wife. The law is gender specific and applies only to women and their children. The law does not specify chemical assaults. South Africa Domestic Violence Act 116 (1998) The act recognizes ‘burns’ as physical abuse but does not specify Acid attacks. It also considers threats as an offence. It is not gender specific. Officers at the scene should help the complainant and can make arrests. Complainant can apply for a court protection order. Application can be brought forth on behalf of the complainant by another person, such as social or health care worker, given they have consent to do so. Unless the complainant is a child, is developmentally disabled, unconscious, has court order. It does not mention anything about victim rehabilitation efforts. In a 2015 incident, the offender was charged with two counts of attempted murder for throwing acid at a parent and their child but the judge found that the victim was ‘no angel’ and convicted the offender on a lesser charge. Tanzania There are no laws regarding ‘domestic violence’ unless it involves a married person. violence against married women is addressed in the Law of Marriage Act (2002). Despite the 2013 attack on two volunteer tourists that caught the international media’s attention, no specific laws against acid attacks were implemented. Thailand Domestic Violence Victim Protection Act, B.E. 2550 (2007) Defines domestic violence broadly to include: any act done either intentionally or in a manner likely to cause harm to body, mind, or health of a family member Does not specify acid attacks. Domestic violence is counted as a ‘compoundable offence’: the offender and victim are able to compromise rather than to go to trial. Turkey Law No. 6284, Prevention of Violence against Women and the Protection of the Family (2012) Includes all women. Includes physical abuse but does not specify acid attacks. Uganda Penal Code Cap 120 (1950) Recognizes acid violence as an offence and describes sentencing for perpetrators: should an acid attack committed with ‘malice aforethought’ lead to death and then it is considered murder (Section188). A government chemist must also give evidence to show a corrosive chemical was used. Prosecutions rarely entail the murder or attempted murder charge due difficulties in proving intent when an assault is premediated—requiring the procurement of acid, hiring of the assailant and planning the attack. Instead, suspects are often charged with the lesser crime of grievous harm (Section 219), or actual bodily harm (Section 236). There is only one government chemist, based in Kampala, serving the entire country, leading to long delays in court hearings, and costly transport of exhibits. UAE There are no laws specific to acid attacks. In 2010, the UAE Federal Supreme Court affirmed that UAE law permits a husband to physically chastise his wife leading to a UN Committee on women’s rights strongly criticizing the country for lack of a law on domestic violence. UK Prevention of Crime Act: Offensive Weapons (1953) Stephen Timms, a former Labor minister, said carrying sulphuric acid without justification should be made a crime—echoing the change in the law to make carrying knives an offence. Prior to 2018, there was no penalty for being caught with the chemicals commonly used in acid attacks. Per the new guidelines, acid is defined as a ‘highly dangerous weapon’ for the first time, allowing judges to, under the Prevention of Crime Act, impose harsher punishments on anyone found to be carrying it in public. USA Assault with acid or corrosive substances is considered a felony and can be prosecuted under the category Aggravated Assault—Other Dangerous Weapon (4c). PPC, Pakistan Penal Code; NGO, non-governmental organization; VAPP, Violence against Persons Prohibition Act. Open in new tab Table 2 Legislation that adopts legal definitions specific to chemical assault Country . Bill name (year) . Bill content . Effects . Afghanistan Elimination of Violence Against Women (2009) The EVAW bill which was signed by the president in 2009, specified causing harm to women by spraying or forced ingestion of harmful chemicals is a crime. Per UN study of 233 events, in 2017–18, majority of complaints by victims was deferred to out-of-court settlements. Afghanistan Independent Human Rights Commission reported that incidents of violence against women remain severely under-reported due to the traditional value system, stigmatization and fear of retaliation of the victim. In 2013, the parliament refused to discuss the new law, condemning it as un-Islamic. Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) Bangladesh was the first country to pass a law banning acid violence. With a drastic drop in the frequency of acid assaults, Bangladesh is no longer the country with the highest number of such attacks and the highest incident rates for women. Some claim the lower numbers cannot be directly attributed to the law, but Selina Ahmed, the executive director of the Acid Survivors Foundation in Dhaka, believes that it had an effect. According to Acid Survivors Foundation, the number of known victims dropped by 45% between 2001 and 2007. Acid attacks continued to decrease to 59 occurrences in 2015 from 494 in 2002, according to Ahmed’s foundation. Cambodia Acid law (2012) Cambodia’s National Assembly in 2002 declined to enact legislation to regulate acid sales or punishment. After acid attacks on a young woman in Phnom Penh and a man in Kompong Cham province, The Interior Ministry claimed the government had formed a committee to study ways to suppressing acid crimes. Prior to the Acid law, acid attacks were prosecuted as ordinary assaults or attempted murders. The Acid Law, criminalized acid violence and established regulations governing the criminal liability of individuals and legal entities. The new law has been hard to implement due to Cambodia’s defective legal system. According to a Cambodia base NGO, the legislation has led to a significant drop in the number of reported attacks, from 36 survivors in 2010 to 6 in 2014, 3 in 2016 and 3 in the first 4 months of 2017. Colombia Natalia Ponce de León law (2016) Prior to this law, acid attacks were treated as lesser physical aggressions, like beatings, and few attackers were convicted and jailed. According to a 2015 study by National Institute of Legal Medicine, from 2004–14, there were 926 acid attack victims reported in Colombia, of which 565 (61%) were women.In 2017, there were 91 reported cases of attacks with corrosive substances. Fifty-seven of the victims were women, and four out of the six victims who died were women. Finland Criminal Code, Chapter 21 Sections 5 (1996), 6 (2001) No specific acid laws. Section 5: Assault is punishable by a fine and prison sentence of up to 2 years. Attempt is also punishable. Section 6: Aggravated assault is defined as grievous bodily injury or serious illness is caused to another or another is placed in mortal danger, (2) the offence is committed in a particularly brutal or cruel manner, or (3) a comparable lethal instrument is used and the offence is aggravated also when assessed as a whole. Attempt is punishable Section 6(a) recognizes possession of a lethal weapon as ‘Preparation of an aggravated offence against life or health’. India The Criminal Law (Amendment) Ordinance/Manodhairya Scheme (2013) Section 326A, 326B Prior to the 2013 ruling there were no laws dealing specifically with acid attacks, and acid attacks were a bailable offence but now offenders should not get bail. The Law Commission has recommended classifying acid attacks as an extremely violent crime. Section 326A covers harming via acid attack and 326B covers attempted acid attack. Madhya Pradesh promised that his government would book culprits in acid attack cases for attempt to murder (Section 307 of the Indian Penal Code). Union home minister Rajnath Singh has set to strengthen the legal and administrative framework against acid attacks by treating acid attacks that cause grievous hurt as ‘heinous’ crimes. Under Section 326A, a total of 203 cases were reported in 2014, 222 in 2015 (increase of 9.4%), 260 in 2016 (increase of 17%) involving 281 victims. In 2016, a total of 233 people were arrested in crimes related to acid attacks, 90% of whom were charged. Italy Law no. 154 (2001) 4th titled: Measures against intra-family violence It specifically addresses domestic violence but not acid attacks. The law requires that women must report a domestic violence offence within 3 months of a ‘first-time’ event. Jordan General recommendation of no. 19 (1994) The committee considers violence against women in the family a violation of the convention, but does not specify the type of attack. The committee states that state parties should be held accountable if they do not act to prevent violence against women or punish acts of violence against women. It also specifically stated that ‘while state parties may argue many traditional practices such as dowry deaths, acid attacks, and forced marriage are justifiable on the basis of protecting or controlling women, such violence directly prohibits women from enjoying their fundamental rights and freedoms’. Kenya Act no. 12: Prevention of torture act (2017) Includes ‘burning by acid’ or rubbing acid on mucus membranes as a form of physical torture. Once a police officer receives a complaint of an offence under this act, it needs to be forwarded immediately to higher levels (the officer-in-charge of the police station and Directorate of Criminal Investigation), which shall investigate the crime immediately. The complaint shall be recorded in writing. Where a complaint regarding an offence under this act is reported to a court or raised in the process of a trial, the court shall record the complaint and order investigation immediately and order submission of an investigation report within 7 days of the court's order. An order for investigation by the court shall include medical and psychological assessment of harm suffered by the alleged victim. Victim impact statement can be included. Malaysia Act no. 357: Corrosive and Explosive Substances and Offensive Weapons Act (1958) The act defines ‘corrosive substance’ as any of the acids and substances specified in the first schedule and includes all substances, which are capable of application to the human body of causing hurt through corrosive action. It is presumed that any substance was intended to be used for the purpose of causing hurt if there is too much quantity, held in places other than their primary use, or held in a container other than the original container. This law was amended five times, last 2014. Nepal Clause 7.6 of the Comprehensive Peace Agreement (2006) There are no specific laws addressing acid attacks in Nepal. The local district court issued directives to parliament as well as the Ministry of law, justice and parliamentary affairs to formulate appropriate legal provisions to deal with acid attack cases, as existing legal provisions allow the courts to award the offenders a maximum of 3 years in prison. The Interim Constitution prohibits ‘physical, mental or any other form of violence’ against women. The 5-year strategic plan of the National Women’s Commission deals extensively with strategies to combat violence against women, although it does not address Acid and Burn Violence as such. Clause 7.6 of the agreement requires that all types of violence against women and children be immediately prohibited. Per a 2009 report, only few women had attempted to file complaints with police. They were met by the police either refusing to file their cases or failing to take proper legal action. Per Advocacy Forum-Nepal, police urge women to use informal community justice systems or to settle on a deal with the perpetrator. Nigeria Violence against Persons Prohibition Act (2015) The 2015 Act: considers acid baths a criminal and punishable offence. Attempt to commit an offence is an offence in itself. The major drawback in relation to the VAPP is its limited application to the Federal Capital Territory, Abuja. Pakistan Criminal Law Amendment Act (2011) The Acid and Burn Crime Bill (2017) Through the Act several changes were introduced in the PPC and Code of Criminal Procedure: the words ‘disfigure’ and ‘deface’ were added in Section 332 of the PPC, dealing with different kinds of hurt, stating ‘disfigure means disfigurement of face or disfigurement or dismemberment of any organ or any part of the organ of human body which impairs or injures or corrodes or deforms the symmetry or appearance of a person’. The Acid and Burn Crime Bill, 2014, which was once cleared for passage by a committee of the lower house but remained pending for over 3 years, was deferred again without being considered by the National Assembly. However, on 8 May 2018, the Pakistani National Assembly passed a revised version of the Acid Burn Crime Bill (2017). The bill not only has emphasized a short legal process for the victims but also mandates the government to pay for the health care fee of the victims, allowing them to receive care free of charge. The Criminal Law Amendment Act 2011 has already acknowledged acid throwing as a crime and, after that, not only has the number of cases been reduced, the conviction rate has also increased by 10%. According to Acid Survivors Foundation Pakistan, there were 71 reported acid attack cases in 2016 and 69 in 2015, down from 153 in 2014 and 109 in 2013. Philippines Anti-violence against Women and their Children Act 9262 (2004) The act criminalizes physical, sexual and psychological violence against women and children. It also criminalizes violence against any woman with whom a man has an intimate relationship, not only a wife. The law is gender specific and applies only to women and their children. The law does not specify chemical assaults. South Africa Domestic Violence Act 116 (1998) The act recognizes ‘burns’ as physical abuse but does not specify Acid attacks. It also considers threats as an offence. It is not gender specific. Officers at the scene should help the complainant and can make arrests. Complainant can apply for a court protection order. Application can be brought forth on behalf of the complainant by another person, such as social or health care worker, given they have consent to do so. Unless the complainant is a child, is developmentally disabled, unconscious, has court order. It does not mention anything about victim rehabilitation efforts. In a 2015 incident, the offender was charged with two counts of attempted murder for throwing acid at a parent and their child but the judge found that the victim was ‘no angel’ and convicted the offender on a lesser charge. Tanzania There are no laws regarding ‘domestic violence’ unless it involves a married person. violence against married women is addressed in the Law of Marriage Act (2002). Despite the 2013 attack on two volunteer tourists that caught the international media’s attention, no specific laws against acid attacks were implemented. Thailand Domestic Violence Victim Protection Act, B.E. 2550 (2007) Defines domestic violence broadly to include: any act done either intentionally or in a manner likely to cause harm to body, mind, or health of a family member Does not specify acid attacks. Domestic violence is counted as a ‘compoundable offence’: the offender and victim are able to compromise rather than to go to trial. Turkey Law No. 6284, Prevention of Violence against Women and the Protection of the Family (2012) Includes all women. Includes physical abuse but does not specify acid attacks. Uganda Penal Code Cap 120 (1950) Recognizes acid violence as an offence and describes sentencing for perpetrators: should an acid attack committed with ‘malice aforethought’ lead to death and then it is considered murder (Section188). A government chemist must also give evidence to show a corrosive chemical was used. Prosecutions rarely entail the murder or attempted murder charge due difficulties in proving intent when an assault is premediated—requiring the procurement of acid, hiring of the assailant and planning the attack. Instead, suspects are often charged with the lesser crime of grievous harm (Section 219), or actual bodily harm (Section 236). There is only one government chemist, based in Kampala, serving the entire country, leading to long delays in court hearings, and costly transport of exhibits. UAE There are no laws specific to acid attacks. In 2010, the UAE Federal Supreme Court affirmed that UAE law permits a husband to physically chastise his wife leading to a UN Committee on women’s rights strongly criticizing the country for lack of a law on domestic violence. UK Prevention of Crime Act: Offensive Weapons (1953) Stephen Timms, a former Labor minister, said carrying sulphuric acid without justification should be made a crime—echoing the change in the law to make carrying knives an offence. Prior to 2018, there was no penalty for being caught with the chemicals commonly used in acid attacks. Per the new guidelines, acid is defined as a ‘highly dangerous weapon’ for the first time, allowing judges to, under the Prevention of Crime Act, impose harsher punishments on anyone found to be carrying it in public. USA Assault with acid or corrosive substances is considered a felony and can be prosecuted under the category Aggravated Assault—Other Dangerous Weapon (4c). Country . Bill name (year) . Bill content . Effects . Afghanistan Elimination of Violence Against Women (2009) The EVAW bill which was signed by the president in 2009, specified causing harm to women by spraying or forced ingestion of harmful chemicals is a crime. Per UN study of 233 events, in 2017–18, majority of complaints by victims was deferred to out-of-court settlements. Afghanistan Independent Human Rights Commission reported that incidents of violence against women remain severely under-reported due to the traditional value system, stigmatization and fear of retaliation of the victim. In 2013, the parliament refused to discuss the new law, condemning it as un-Islamic. Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) Bangladesh was the first country to pass a law banning acid violence. With a drastic drop in the frequency of acid assaults, Bangladesh is no longer the country with the highest number of such attacks and the highest incident rates for women. Some claim the lower numbers cannot be directly attributed to the law, but Selina Ahmed, the executive director of the Acid Survivors Foundation in Dhaka, believes that it had an effect. According to Acid Survivors Foundation, the number of known victims dropped by 45% between 2001 and 2007. Acid attacks continued to decrease to 59 occurrences in 2015 from 494 in 2002, according to Ahmed’s foundation. Cambodia Acid law (2012) Cambodia’s National Assembly in 2002 declined to enact legislation to regulate acid sales or punishment. After acid attacks on a young woman in Phnom Penh and a man in Kompong Cham province, The Interior Ministry claimed the government had formed a committee to study ways to suppressing acid crimes. Prior to the Acid law, acid attacks were prosecuted as ordinary assaults or attempted murders. The Acid Law, criminalized acid violence and established regulations governing the criminal liability of individuals and legal entities. The new law has been hard to implement due to Cambodia’s defective legal system. According to a Cambodia base NGO, the legislation has led to a significant drop in the number of reported attacks, from 36 survivors in 2010 to 6 in 2014, 3 in 2016 and 3 in the first 4 months of 2017. Colombia Natalia Ponce de León law (2016) Prior to this law, acid attacks were treated as lesser physical aggressions, like beatings, and few attackers were convicted and jailed. According to a 2015 study by National Institute of Legal Medicine, from 2004–14, there were 926 acid attack victims reported in Colombia, of which 565 (61%) were women.In 2017, there were 91 reported cases of attacks with corrosive substances. Fifty-seven of the victims were women, and four out of the six victims who died were women. Finland Criminal Code, Chapter 21 Sections 5 (1996), 6 (2001) No specific acid laws. Section 5: Assault is punishable by a fine and prison sentence of up to 2 years. Attempt is also punishable. Section 6: Aggravated assault is defined as grievous bodily injury or serious illness is caused to another or another is placed in mortal danger, (2) the offence is committed in a particularly brutal or cruel manner, or (3) a comparable lethal instrument is used and the offence is aggravated also when assessed as a whole. Attempt is punishable Section 6(a) recognizes possession of a lethal weapon as ‘Preparation of an aggravated offence against life or health’. India The Criminal Law (Amendment) Ordinance/Manodhairya Scheme (2013) Section 326A, 326B Prior to the 2013 ruling there were no laws dealing specifically with acid attacks, and acid attacks were a bailable offence but now offenders should not get bail. The Law Commission has recommended classifying acid attacks as an extremely violent crime. Section 326A covers harming via acid attack and 326B covers attempted acid attack. Madhya Pradesh promised that his government would book culprits in acid attack cases for attempt to murder (Section 307 of the Indian Penal Code). Union home minister Rajnath Singh has set to strengthen the legal and administrative framework against acid attacks by treating acid attacks that cause grievous hurt as ‘heinous’ crimes. Under Section 326A, a total of 203 cases were reported in 2014, 222 in 2015 (increase of 9.4%), 260 in 2016 (increase of 17%) involving 281 victims. In 2016, a total of 233 people were arrested in crimes related to acid attacks, 90% of whom were charged. Italy Law no. 154 (2001) 4th titled: Measures against intra-family violence It specifically addresses domestic violence but not acid attacks. The law requires that women must report a domestic violence offence within 3 months of a ‘first-time’ event. Jordan General recommendation of no. 19 (1994) The committee considers violence against women in the family a violation of the convention, but does not specify the type of attack. The committee states that state parties should be held accountable if they do not act to prevent violence against women or punish acts of violence against women. It also specifically stated that ‘while state parties may argue many traditional practices such as dowry deaths, acid attacks, and forced marriage are justifiable on the basis of protecting or controlling women, such violence directly prohibits women from enjoying their fundamental rights and freedoms’. Kenya Act no. 12: Prevention of torture act (2017) Includes ‘burning by acid’ or rubbing acid on mucus membranes as a form of physical torture. Once a police officer receives a complaint of an offence under this act, it needs to be forwarded immediately to higher levels (the officer-in-charge of the police station and Directorate of Criminal Investigation), which shall investigate the crime immediately. The complaint shall be recorded in writing. Where a complaint regarding an offence under this act is reported to a court or raised in the process of a trial, the court shall record the complaint and order investigation immediately and order submission of an investigation report within 7 days of the court's order. An order for investigation by the court shall include medical and psychological assessment of harm suffered by the alleged victim. Victim impact statement can be included. Malaysia Act no. 357: Corrosive and Explosive Substances and Offensive Weapons Act (1958) The act defines ‘corrosive substance’ as any of the acids and substances specified in the first schedule and includes all substances, which are capable of application to the human body of causing hurt through corrosive action. It is presumed that any substance was intended to be used for the purpose of causing hurt if there is too much quantity, held in places other than their primary use, or held in a container other than the original container. This law was amended five times, last 2014. Nepal Clause 7.6 of the Comprehensive Peace Agreement (2006) There are no specific laws addressing acid attacks in Nepal. The local district court issued directives to parliament as well as the Ministry of law, justice and parliamentary affairs to formulate appropriate legal provisions to deal with acid attack cases, as existing legal provisions allow the courts to award the offenders a maximum of 3 years in prison. The Interim Constitution prohibits ‘physical, mental or any other form of violence’ against women. The 5-year strategic plan of the National Women’s Commission deals extensively with strategies to combat violence against women, although it does not address Acid and Burn Violence as such. Clause 7.6 of the agreement requires that all types of violence against women and children be immediately prohibited. Per a 2009 report, only few women had attempted to file complaints with police. They were met by the police either refusing to file their cases or failing to take proper legal action. Per Advocacy Forum-Nepal, police urge women to use informal community justice systems or to settle on a deal with the perpetrator. Nigeria Violence against Persons Prohibition Act (2015) The 2015 Act: considers acid baths a criminal and punishable offence. Attempt to commit an offence is an offence in itself. The major drawback in relation to the VAPP is its limited application to the Federal Capital Territory, Abuja. Pakistan Criminal Law Amendment Act (2011) The Acid and Burn Crime Bill (2017) Through the Act several changes were introduced in the PPC and Code of Criminal Procedure: the words ‘disfigure’ and ‘deface’ were added in Section 332 of the PPC, dealing with different kinds of hurt, stating ‘disfigure means disfigurement of face or disfigurement or dismemberment of any organ or any part of the organ of human body which impairs or injures or corrodes or deforms the symmetry or appearance of a person’. The Acid and Burn Crime Bill, 2014, which was once cleared for passage by a committee of the lower house but remained pending for over 3 years, was deferred again without being considered by the National Assembly. However, on 8 May 2018, the Pakistani National Assembly passed a revised version of the Acid Burn Crime Bill (2017). The bill not only has emphasized a short legal process for the victims but also mandates the government to pay for the health care fee of the victims, allowing them to receive care free of charge. The Criminal Law Amendment Act 2011 has already acknowledged acid throwing as a crime and, after that, not only has the number of cases been reduced, the conviction rate has also increased by 10%. According to Acid Survivors Foundation Pakistan, there were 71 reported acid attack cases in 2016 and 69 in 2015, down from 153 in 2014 and 109 in 2013. Philippines Anti-violence against Women and their Children Act 9262 (2004) The act criminalizes physical, sexual and psychological violence against women and children. It also criminalizes violence against any woman with whom a man has an intimate relationship, not only a wife. The law is gender specific and applies only to women and their children. The law does not specify chemical assaults. South Africa Domestic Violence Act 116 (1998) The act recognizes ‘burns’ as physical abuse but does not specify Acid attacks. It also considers threats as an offence. It is not gender specific. Officers at the scene should help the complainant and can make arrests. Complainant can apply for a court protection order. Application can be brought forth on behalf of the complainant by another person, such as social or health care worker, given they have consent to do so. Unless the complainant is a child, is developmentally disabled, unconscious, has court order. It does not mention anything about victim rehabilitation efforts. In a 2015 incident, the offender was charged with two counts of attempted murder for throwing acid at a parent and their child but the judge found that the victim was ‘no angel’ and convicted the offender on a lesser charge. Tanzania There are no laws regarding ‘domestic violence’ unless it involves a married person. violence against married women is addressed in the Law of Marriage Act (2002). Despite the 2013 attack on two volunteer tourists that caught the international media’s attention, no specific laws against acid attacks were implemented. Thailand Domestic Violence Victim Protection Act, B.E. 2550 (2007) Defines domestic violence broadly to include: any act done either intentionally or in a manner likely to cause harm to body, mind, or health of a family member Does not specify acid attacks. Domestic violence is counted as a ‘compoundable offence’: the offender and victim are able to compromise rather than to go to trial. Turkey Law No. 6284, Prevention of Violence against Women and the Protection of the Family (2012) Includes all women. Includes physical abuse but does not specify acid attacks. Uganda Penal Code Cap 120 (1950) Recognizes acid violence as an offence and describes sentencing for perpetrators: should an acid attack committed with ‘malice aforethought’ lead to death and then it is considered murder (Section188). A government chemist must also give evidence to show a corrosive chemical was used. Prosecutions rarely entail the murder or attempted murder charge due difficulties in proving intent when an assault is premediated—requiring the procurement of acid, hiring of the assailant and planning the attack. Instead, suspects are often charged with the lesser crime of grievous harm (Section 219), or actual bodily harm (Section 236). There is only one government chemist, based in Kampala, serving the entire country, leading to long delays in court hearings, and costly transport of exhibits. UAE There are no laws specific to acid attacks. In 2010, the UAE Federal Supreme Court affirmed that UAE law permits a husband to physically chastise his wife leading to a UN Committee on women’s rights strongly criticizing the country for lack of a law on domestic violence. UK Prevention of Crime Act: Offensive Weapons (1953) Stephen Timms, a former Labor minister, said carrying sulphuric acid without justification should be made a crime—echoing the change in the law to make carrying knives an offence. Prior to 2018, there was no penalty for being caught with the chemicals commonly used in acid attacks. Per the new guidelines, acid is defined as a ‘highly dangerous weapon’ for the first time, allowing judges to, under the Prevention of Crime Act, impose harsher punishments on anyone found to be carrying it in public. USA Assault with acid or corrosive substances is considered a felony and can be prosecuted under the category Aggravated Assault—Other Dangerous Weapon (4c). PPC, Pakistan Penal Code; NGO, non-governmental organization; VAPP, Violence against Persons Prohibition Act. Open in new tab Table 3 Legislation that aims to control chemical supply, sales and procurement Country . Bill name (year enacted) . Bill content . Effects . Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) The unlicenced production, import, transport, storage, sale and use of acid can result in a prison term of 3–10 years. Those who possess chemicals and equipment for the unlicenced production of acid can get the same prison term. Shops that did not follow the guidelines were closed. Acid-selling licences were temporarily cancelled. Cambodia Acid Law (2012) Sub-decree on the Formalities and Conditions for Strong Acid Control (2013) Sales and distribution must occur at a location authorized by administrative authorities. Sales must be limited to customers who provide appropriate documentation. Sellers must keep record of the purchases and customers. Sellers must correctly pack and store acid in safe bottles or containers with warning labels in Khmer, and a certificate of origin. Purchasers of acid must be at least 18 years old, have proof of professional occupation requiring use of acid and the purpose of buying acid and keep the invoice ready to be inspected during transport. Finland European Union Classification, Labelling and Packaging of Substances and Mixtures (2009) Dangerous substances need to be labelled and registered. Sulphuric acid is considered hazardous with CLP. Hong Kong Dangerous Goods Ordinance Cap. 295, Laws of Hong Kong Section 6: Licence is required to manufacture, store, convey or use any dangerous goods. India Supreme Court Oder (2013) Tamil Nadu Poisons (Regulation of Possession and Sale) Act (2015) (A) Acid should be sold only to adults, (B) government-issued identity cards needed and (C) purchasers must declare the reason for the purchase. Under the Tamil Nadu Poisons (Regulation of Possession and Sale) Act, traders and sellers who deal with acid must have a proper licence. The Supreme Court has prohibited the over-the-counter sale of acid unless the seller maintains a log recording the address and other details of the buyer and the quantum. Developing web application to regulate the sale of acid: registration of stockists and retailers, issue of licences by the district administration and limiting sale of acid. Obtaining acid from stores remain easy years after the 2013 Supreme Court Order. Some attribute this to the loopholes that make the implementation of the law difficult. Few shops were sealed due to illegal sale of acid. Some experts claim the ban has not helped to reduce the incidence of acid attacks. Israel Hazardous Substances Law 5753 (1993) Specifies, hydrochloric acid, sulphuric acid, nitric acid and other caustic agents as hazardous materials whose import, storage, sale and distribution is regulated and offenders are subjected to monetary fines as well as prison sentences. Sellers need Hazardous Substance licence to buy and sell hazardous material. Japan Chemical Substance Control Law (1973, amended 2009, implemented 2011) Sulphuric acid is included in the ‘existing substances’ list, which involves regulations for the manufacturers but not distributors. Malaysia Globally Harmonized System CLASS Regulations (issued in 2013) Industry Code of Practice on Chemicals Classification and Hazard Communication (ICOP) (issued in 2014) CLASS regulations require chemical manufacturers, importers, formulators, and distributors to classify, label and package chemicals and compile safety data sheets according to the ICOP. Per Inspector-General of Police current regulations for the sale of acids are sufficient; however, sale of other corrosive liquids needs to be better regulated.Some deem the current regulations insufficient as acid and other corrosive substances can be easily bought from stores. Nepal Per the Ministry of Commerce and Supplies, ‘the Department of Commerce and Supplies will direct dealers and retailers to sell and distribute the chemical substance only to the right person or organization to ensure that it does not land on wrong hands’. New Zealand Hazardous Substances and New Organisms Act (1996) Implements regulations for production, import, storage, sale, distribution and work place handling of hazardous substances. Nigeria Criminal Code Act, Section 350 (1990) One can be charged by law for any duty regarding the transportation, distribution and storage of any acid from any vessel, fails to perform that duty or being concerned in the transportation, distribution and storage of any such substance. Such person is guilty of a felony and is liable to imprisonment for 3 years. Pakistan Control of Narcotics Substance Act (1997)Criminal Law Act (2011) Some provinces lack proper oversight and enforcement of legal acid sale. Peshawar High Court directed the federal government to make amendments to the law that includes regulations of sale and purchase of acids, as the Criminal Law Act did not address sale and purchase of corrosive substances. Philippines Comprehensive Dangerous Drugs Act (2002) Both sulphuric acid and hydrochloric acid are under Table II of the controlled substances. Their salts are exempt from the table. They are not included in the ‘dangerous drugs’ section whose sale and distribution are heavily regulated and penalized. Poland REACH Regulation (2006) CLP Regulation (2008) Poland follows the European Nation’s CLP guidelines and other regulations set for production, import, storage, sale and distribution of hazardous substances. South Africa Hazardous Substances Act No. 15, (1973) Does not consider sulphuric and hydrochloric acids as hazardous materials and they are not heavily regulated. South Korea Buyers must provide, and sellers must keep the purchasers personal information. There are no regulations for the intended use of acids. Buyers can purchase a bottle of solution with a concentration of acid that is ˂10% from pharmacies in Seoul without having to present identification. Acids ˂10% concentration are regarded ‘less harmful’ and available at normal stores. Taiwan Toxic Chemical Substances Control Act 1986, last amendment 2014 Mandates permits for import, production, storage, sale and distribution of hazardous substances. Thailand Hazardous Substance Act B.E. 2535 (1992) High concentration sulphuric acid (>50%) and hydrochloric acid >15% are considered hazardous. Turkey KKDIK (2017) Similar to EU’s REACH legislature, KKDIK was implemented for the registration and regulation of hazardous material. Uganda Control Agricultural Chemicals Act (2007) Toxic Chemicals Prohibition and Control Bill (2015) There are no specific provisions, regulations or guidelines related solely to the importation, transportation, storage, distribution and use of industrial chemicals in Uganda applicable for all dealers, unlike for agricultural chemicals. Toxic Chemicals Prohibition and Control Bill (2015) focuses on chemical weapons and does not specify acid attacks. UK Poisons Act (1972) Deregulation Act (2015) Per a report prepared for the home affairs committee: the liquids used in the attacks should be made too thick to throw easily and calls for shops selling highly acidic products to hold a licence requiring them to log the names and addresses of customers. Buyers would have to prove they are over 18. Under the 1972 Poisons Act, any business selling dangerous substances was required to register annually with their local council, ensuring that there was a record of companies selling hazardous chemicals. Those selling the most lethal chemicals also needed a licence from the Home Office. Per the Deregulation Act of 2015, only the most potent toxins, including those used to make explosives, are deemed ‘regulated substances’ and require a Licence to sell. Sulphuric acid and hydrochloric acid are included on a separate list of ‘reportable substances’ that do not require a licence. On 17 July 2017, the House of Commons held a debate to consider licencing the sale of corrosive liquids. Reportedly, buying chemicals used in attacks were so easy that the Home Secretary Amber Rudd urged business owners to help tackle the ‘scourge’ of acid attacks. Undercover video footage shows teenagers buying highly concentrated acid at markets. Country . Bill name (year enacted) . Bill content . Effects . Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) The unlicenced production, import, transport, storage, sale and use of acid can result in a prison term of 3–10 years. Those who possess chemicals and equipment for the unlicenced production of acid can get the same prison term. Shops that did not follow the guidelines were closed. Acid-selling licences were temporarily cancelled. Cambodia Acid Law (2012) Sub-decree on the Formalities and Conditions for Strong Acid Control (2013) Sales and distribution must occur at a location authorized by administrative authorities. Sales must be limited to customers who provide appropriate documentation. Sellers must keep record of the purchases and customers. Sellers must correctly pack and store acid in safe bottles or containers with warning labels in Khmer, and a certificate of origin. Purchasers of acid must be at least 18 years old, have proof of professional occupation requiring use of acid and the purpose of buying acid and keep the invoice ready to be inspected during transport. Finland European Union Classification, Labelling and Packaging of Substances and Mixtures (2009) Dangerous substances need to be labelled and registered. Sulphuric acid is considered hazardous with CLP. Hong Kong Dangerous Goods Ordinance Cap. 295, Laws of Hong Kong Section 6: Licence is required to manufacture, store, convey or use any dangerous goods. India Supreme Court Oder (2013) Tamil Nadu Poisons (Regulation of Possession and Sale) Act (2015) (A) Acid should be sold only to adults, (B) government-issued identity cards needed and (C) purchasers must declare the reason for the purchase. Under the Tamil Nadu Poisons (Regulation of Possession and Sale) Act, traders and sellers who deal with acid must have a proper licence. The Supreme Court has prohibited the over-the-counter sale of acid unless the seller maintains a log recording the address and other details of the buyer and the quantum. Developing web application to regulate the sale of acid: registration of stockists and retailers, issue of licences by the district administration and limiting sale of acid. Obtaining acid from stores remain easy years after the 2013 Supreme Court Order. Some attribute this to the loopholes that make the implementation of the law difficult. Few shops were sealed due to illegal sale of acid. Some experts claim the ban has not helped to reduce the incidence of acid attacks. Israel Hazardous Substances Law 5753 (1993) Specifies, hydrochloric acid, sulphuric acid, nitric acid and other caustic agents as hazardous materials whose import, storage, sale and distribution is regulated and offenders are subjected to monetary fines as well as prison sentences. Sellers need Hazardous Substance licence to buy and sell hazardous material. Japan Chemical Substance Control Law (1973, amended 2009, implemented 2011) Sulphuric acid is included in the ‘existing substances’ list, which involves regulations for the manufacturers but not distributors. Malaysia Globally Harmonized System CLASS Regulations (issued in 2013) Industry Code of Practice on Chemicals Classification and Hazard Communication (ICOP) (issued in 2014) CLASS regulations require chemical manufacturers, importers, formulators, and distributors to classify, label and package chemicals and compile safety data sheets according to the ICOP. Per Inspector-General of Police current regulations for the sale of acids are sufficient; however, sale of other corrosive liquids needs to be better regulated.Some deem the current regulations insufficient as acid and other corrosive substances can be easily bought from stores. Nepal Per the Ministry of Commerce and Supplies, ‘the Department of Commerce and Supplies will direct dealers and retailers to sell and distribute the chemical substance only to the right person or organization to ensure that it does not land on wrong hands’. New Zealand Hazardous Substances and New Organisms Act (1996) Implements regulations for production, import, storage, sale, distribution and work place handling of hazardous substances. Nigeria Criminal Code Act, Section 350 (1990) One can be charged by law for any duty regarding the transportation, distribution and storage of any acid from any vessel, fails to perform that duty or being concerned in the transportation, distribution and storage of any such substance. Such person is guilty of a felony and is liable to imprisonment for 3 years. Pakistan Control of Narcotics Substance Act (1997)Criminal Law Act (2011) Some provinces lack proper oversight and enforcement of legal acid sale. Peshawar High Court directed the federal government to make amendments to the law that includes regulations of sale and purchase of acids, as the Criminal Law Act did not address sale and purchase of corrosive substances. Philippines Comprehensive Dangerous Drugs Act (2002) Both sulphuric acid and hydrochloric acid are under Table II of the controlled substances. Their salts are exempt from the table. They are not included in the ‘dangerous drugs’ section whose sale and distribution are heavily regulated and penalized. Poland REACH Regulation (2006) CLP Regulation (2008) Poland follows the European Nation’s CLP guidelines and other regulations set for production, import, storage, sale and distribution of hazardous substances. South Africa Hazardous Substances Act No. 15, (1973) Does not consider sulphuric and hydrochloric acids as hazardous materials and they are not heavily regulated. South Korea Buyers must provide, and sellers must keep the purchasers personal information. There are no regulations for the intended use of acids. Buyers can purchase a bottle of solution with a concentration of acid that is ˂10% from pharmacies in Seoul without having to present identification. Acids ˂10% concentration are regarded ‘less harmful’ and available at normal stores. Taiwan Toxic Chemical Substances Control Act 1986, last amendment 2014 Mandates permits for import, production, storage, sale and distribution of hazardous substances. Thailand Hazardous Substance Act B.E. 2535 (1992) High concentration sulphuric acid (>50%) and hydrochloric acid >15% are considered hazardous. Turkey KKDIK (2017) Similar to EU’s REACH legislature, KKDIK was implemented for the registration and regulation of hazardous material. Uganda Control Agricultural Chemicals Act (2007) Toxic Chemicals Prohibition and Control Bill (2015) There are no specific provisions, regulations or guidelines related solely to the importation, transportation, storage, distribution and use of industrial chemicals in Uganda applicable for all dealers, unlike for agricultural chemicals. Toxic Chemicals Prohibition and Control Bill (2015) focuses on chemical weapons and does not specify acid attacks. UK Poisons Act (1972) Deregulation Act (2015) Per a report prepared for the home affairs committee: the liquids used in the attacks should be made too thick to throw easily and calls for shops selling highly acidic products to hold a licence requiring them to log the names and addresses of customers. Buyers would have to prove they are over 18. Under the 1972 Poisons Act, any business selling dangerous substances was required to register annually with their local council, ensuring that there was a record of companies selling hazardous chemicals. Those selling the most lethal chemicals also needed a licence from the Home Office. Per the Deregulation Act of 2015, only the most potent toxins, including those used to make explosives, are deemed ‘regulated substances’ and require a Licence to sell. Sulphuric acid and hydrochloric acid are included on a separate list of ‘reportable substances’ that do not require a licence. On 17 July 2017, the House of Commons held a debate to consider licencing the sale of corrosive liquids. Reportedly, buying chemicals used in attacks were so easy that the Home Secretary Amber Rudd urged business owners to help tackle the ‘scourge’ of acid attacks. Undercover video footage shows teenagers buying highly concentrated acid at markets. CLP, Classification Labelling Packaging. Open in new tab Table 3 Legislation that aims to control chemical supply, sales and procurement Country . Bill name (year enacted) . Bill content . Effects . Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) The unlicenced production, import, transport, storage, sale and use of acid can result in a prison term of 3–10 years. Those who possess chemicals and equipment for the unlicenced production of acid can get the same prison term. Shops that did not follow the guidelines were closed. Acid-selling licences were temporarily cancelled. Cambodia Acid Law (2012) Sub-decree on the Formalities and Conditions for Strong Acid Control (2013) Sales and distribution must occur at a location authorized by administrative authorities. Sales must be limited to customers who provide appropriate documentation. Sellers must keep record of the purchases and customers. Sellers must correctly pack and store acid in safe bottles or containers with warning labels in Khmer, and a certificate of origin. Purchasers of acid must be at least 18 years old, have proof of professional occupation requiring use of acid and the purpose of buying acid and keep the invoice ready to be inspected during transport. Finland European Union Classification, Labelling and Packaging of Substances and Mixtures (2009) Dangerous substances need to be labelled and registered. Sulphuric acid is considered hazardous with CLP. Hong Kong Dangerous Goods Ordinance Cap. 295, Laws of Hong Kong Section 6: Licence is required to manufacture, store, convey or use any dangerous goods. India Supreme Court Oder (2013) Tamil Nadu Poisons (Regulation of Possession and Sale) Act (2015) (A) Acid should be sold only to adults, (B) government-issued identity cards needed and (C) purchasers must declare the reason for the purchase. Under the Tamil Nadu Poisons (Regulation of Possession and Sale) Act, traders and sellers who deal with acid must have a proper licence. The Supreme Court has prohibited the over-the-counter sale of acid unless the seller maintains a log recording the address and other details of the buyer and the quantum. Developing web application to regulate the sale of acid: registration of stockists and retailers, issue of licences by the district administration and limiting sale of acid. Obtaining acid from stores remain easy years after the 2013 Supreme Court Order. Some attribute this to the loopholes that make the implementation of the law difficult. Few shops were sealed due to illegal sale of acid. Some experts claim the ban has not helped to reduce the incidence of acid attacks. Israel Hazardous Substances Law 5753 (1993) Specifies, hydrochloric acid, sulphuric acid, nitric acid and other caustic agents as hazardous materials whose import, storage, sale and distribution is regulated and offenders are subjected to monetary fines as well as prison sentences. Sellers need Hazardous Substance licence to buy and sell hazardous material. Japan Chemical Substance Control Law (1973, amended 2009, implemented 2011) Sulphuric acid is included in the ‘existing substances’ list, which involves regulations for the manufacturers but not distributors. Malaysia Globally Harmonized System CLASS Regulations (issued in 2013) Industry Code of Practice on Chemicals Classification and Hazard Communication (ICOP) (issued in 2014) CLASS regulations require chemical manufacturers, importers, formulators, and distributors to classify, label and package chemicals and compile safety data sheets according to the ICOP. Per Inspector-General of Police current regulations for the sale of acids are sufficient; however, sale of other corrosive liquids needs to be better regulated.Some deem the current regulations insufficient as acid and other corrosive substances can be easily bought from stores. Nepal Per the Ministry of Commerce and Supplies, ‘the Department of Commerce and Supplies will direct dealers and retailers to sell and distribute the chemical substance only to the right person or organization to ensure that it does not land on wrong hands’. New Zealand Hazardous Substances and New Organisms Act (1996) Implements regulations for production, import, storage, sale, distribution and work place handling of hazardous substances. Nigeria Criminal Code Act, Section 350 (1990) One can be charged by law for any duty regarding the transportation, distribution and storage of any acid from any vessel, fails to perform that duty or being concerned in the transportation, distribution and storage of any such substance. Such person is guilty of a felony and is liable to imprisonment for 3 years. Pakistan Control of Narcotics Substance Act (1997)Criminal Law Act (2011) Some provinces lack proper oversight and enforcement of legal acid sale. Peshawar High Court directed the federal government to make amendments to the law that includes regulations of sale and purchase of acids, as the Criminal Law Act did not address sale and purchase of corrosive substances. Philippines Comprehensive Dangerous Drugs Act (2002) Both sulphuric acid and hydrochloric acid are under Table II of the controlled substances. Their salts are exempt from the table. They are not included in the ‘dangerous drugs’ section whose sale and distribution are heavily regulated and penalized. Poland REACH Regulation (2006) CLP Regulation (2008) Poland follows the European Nation’s CLP guidelines and other regulations set for production, import, storage, sale and distribution of hazardous substances. South Africa Hazardous Substances Act No. 15, (1973) Does not consider sulphuric and hydrochloric acids as hazardous materials and they are not heavily regulated. South Korea Buyers must provide, and sellers must keep the purchasers personal information. There are no regulations for the intended use of acids. Buyers can purchase a bottle of solution with a concentration of acid that is ˂10% from pharmacies in Seoul without having to present identification. Acids ˂10% concentration are regarded ‘less harmful’ and available at normal stores. Taiwan Toxic Chemical Substances Control Act 1986, last amendment 2014 Mandates permits for import, production, storage, sale and distribution of hazardous substances. Thailand Hazardous Substance Act B.E. 2535 (1992) High concentration sulphuric acid (>50%) and hydrochloric acid >15% are considered hazardous. Turkey KKDIK (2017) Similar to EU’s REACH legislature, KKDIK was implemented for the registration and regulation of hazardous material. Uganda Control Agricultural Chemicals Act (2007) Toxic Chemicals Prohibition and Control Bill (2015) There are no specific provisions, regulations or guidelines related solely to the importation, transportation, storage, distribution and use of industrial chemicals in Uganda applicable for all dealers, unlike for agricultural chemicals. Toxic Chemicals Prohibition and Control Bill (2015) focuses on chemical weapons and does not specify acid attacks. UK Poisons Act (1972) Deregulation Act (2015) Per a report prepared for the home affairs committee: the liquids used in the attacks should be made too thick to throw easily and calls for shops selling highly acidic products to hold a licence requiring them to log the names and addresses of customers. Buyers would have to prove they are over 18. Under the 1972 Poisons Act, any business selling dangerous substances was required to register annually with their local council, ensuring that there was a record of companies selling hazardous chemicals. Those selling the most lethal chemicals also needed a licence from the Home Office. Per the Deregulation Act of 2015, only the most potent toxins, including those used to make explosives, are deemed ‘regulated substances’ and require a Licence to sell. Sulphuric acid and hydrochloric acid are included on a separate list of ‘reportable substances’ that do not require a licence. On 17 July 2017, the House of Commons held a debate to consider licencing the sale of corrosive liquids. Reportedly, buying chemicals used in attacks were so easy that the Home Secretary Amber Rudd urged business owners to help tackle the ‘scourge’ of acid attacks. Undercover video footage shows teenagers buying highly concentrated acid at markets. Country . Bill name (year enacted) . Bill content . Effects . Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) The unlicenced production, import, transport, storage, sale and use of acid can result in a prison term of 3–10 years. Those who possess chemicals and equipment for the unlicenced production of acid can get the same prison term. Shops that did not follow the guidelines were closed. Acid-selling licences were temporarily cancelled. Cambodia Acid Law (2012) Sub-decree on the Formalities and Conditions for Strong Acid Control (2013) Sales and distribution must occur at a location authorized by administrative authorities. Sales must be limited to customers who provide appropriate documentation. Sellers must keep record of the purchases and customers. Sellers must correctly pack and store acid in safe bottles or containers with warning labels in Khmer, and a certificate of origin. Purchasers of acid must be at least 18 years old, have proof of professional occupation requiring use of acid and the purpose of buying acid and keep the invoice ready to be inspected during transport. Finland European Union Classification, Labelling and Packaging of Substances and Mixtures (2009) Dangerous substances need to be labelled and registered. Sulphuric acid is considered hazardous with CLP. Hong Kong Dangerous Goods Ordinance Cap. 295, Laws of Hong Kong Section 6: Licence is required to manufacture, store, convey or use any dangerous goods. India Supreme Court Oder (2013) Tamil Nadu Poisons (Regulation of Possession and Sale) Act (2015) (A) Acid should be sold only to adults, (B) government-issued identity cards needed and (C) purchasers must declare the reason for the purchase. Under the Tamil Nadu Poisons (Regulation of Possession and Sale) Act, traders and sellers who deal with acid must have a proper licence. The Supreme Court has prohibited the over-the-counter sale of acid unless the seller maintains a log recording the address and other details of the buyer and the quantum. Developing web application to regulate the sale of acid: registration of stockists and retailers, issue of licences by the district administration and limiting sale of acid. Obtaining acid from stores remain easy years after the 2013 Supreme Court Order. Some attribute this to the loopholes that make the implementation of the law difficult. Few shops were sealed due to illegal sale of acid. Some experts claim the ban has not helped to reduce the incidence of acid attacks. Israel Hazardous Substances Law 5753 (1993) Specifies, hydrochloric acid, sulphuric acid, nitric acid and other caustic agents as hazardous materials whose import, storage, sale and distribution is regulated and offenders are subjected to monetary fines as well as prison sentences. Sellers need Hazardous Substance licence to buy and sell hazardous material. Japan Chemical Substance Control Law (1973, amended 2009, implemented 2011) Sulphuric acid is included in the ‘existing substances’ list, which involves regulations for the manufacturers but not distributors. Malaysia Globally Harmonized System CLASS Regulations (issued in 2013) Industry Code of Practice on Chemicals Classification and Hazard Communication (ICOP) (issued in 2014) CLASS regulations require chemical manufacturers, importers, formulators, and distributors to classify, label and package chemicals and compile safety data sheets according to the ICOP. Per Inspector-General of Police current regulations for the sale of acids are sufficient; however, sale of other corrosive liquids needs to be better regulated.Some deem the current regulations insufficient as acid and other corrosive substances can be easily bought from stores. Nepal Per the Ministry of Commerce and Supplies, ‘the Department of Commerce and Supplies will direct dealers and retailers to sell and distribute the chemical substance only to the right person or organization to ensure that it does not land on wrong hands’. New Zealand Hazardous Substances and New Organisms Act (1996) Implements regulations for production, import, storage, sale, distribution and work place handling of hazardous substances. Nigeria Criminal Code Act, Section 350 (1990) One can be charged by law for any duty regarding the transportation, distribution and storage of any acid from any vessel, fails to perform that duty or being concerned in the transportation, distribution and storage of any such substance. Such person is guilty of a felony and is liable to imprisonment for 3 years. Pakistan Control of Narcotics Substance Act (1997)Criminal Law Act (2011) Some provinces lack proper oversight and enforcement of legal acid sale. Peshawar High Court directed the federal government to make amendments to the law that includes regulations of sale and purchase of acids, as the Criminal Law Act did not address sale and purchase of corrosive substances. Philippines Comprehensive Dangerous Drugs Act (2002) Both sulphuric acid and hydrochloric acid are under Table II of the controlled substances. Their salts are exempt from the table. They are not included in the ‘dangerous drugs’ section whose sale and distribution are heavily regulated and penalized. Poland REACH Regulation (2006) CLP Regulation (2008) Poland follows the European Nation’s CLP guidelines and other regulations set for production, import, storage, sale and distribution of hazardous substances. South Africa Hazardous Substances Act No. 15, (1973) Does not consider sulphuric and hydrochloric acids as hazardous materials and they are not heavily regulated. South Korea Buyers must provide, and sellers must keep the purchasers personal information. There are no regulations for the intended use of acids. Buyers can purchase a bottle of solution with a concentration of acid that is ˂10% from pharmacies in Seoul without having to present identification. Acids ˂10% concentration are regarded ‘less harmful’ and available at normal stores. Taiwan Toxic Chemical Substances Control Act 1986, last amendment 2014 Mandates permits for import, production, storage, sale and distribution of hazardous substances. Thailand Hazardous Substance Act B.E. 2535 (1992) High concentration sulphuric acid (>50%) and hydrochloric acid >15% are considered hazardous. Turkey KKDIK (2017) Similar to EU’s REACH legislature, KKDIK was implemented for the registration and regulation of hazardous material. Uganda Control Agricultural Chemicals Act (2007) Toxic Chemicals Prohibition and Control Bill (2015) There are no specific provisions, regulations or guidelines related solely to the importation, transportation, storage, distribution and use of industrial chemicals in Uganda applicable for all dealers, unlike for agricultural chemicals. Toxic Chemicals Prohibition and Control Bill (2015) focuses on chemical weapons and does not specify acid attacks. UK Poisons Act (1972) Deregulation Act (2015) Per a report prepared for the home affairs committee: the liquids used in the attacks should be made too thick to throw easily and calls for shops selling highly acidic products to hold a licence requiring them to log the names and addresses of customers. Buyers would have to prove they are over 18. Under the 1972 Poisons Act, any business selling dangerous substances was required to register annually with their local council, ensuring that there was a record of companies selling hazardous chemicals. Those selling the most lethal chemicals also needed a licence from the Home Office. Per the Deregulation Act of 2015, only the most potent toxins, including those used to make explosives, are deemed ‘regulated substances’ and require a Licence to sell. Sulphuric acid and hydrochloric acid are included on a separate list of ‘reportable substances’ that do not require a licence. On 17 July 2017, the House of Commons held a debate to consider licencing the sale of corrosive liquids. Reportedly, buying chemicals used in attacks were so easy that the Home Secretary Amber Rudd urged business owners to help tackle the ‘scourge’ of acid attacks. Undercover video footage shows teenagers buying highly concentrated acid at markets. CLP, Classification Labelling Packaging. Open in new tab Chemical assaults are unique in that they are predominately meant to cause morbidity rather than mortality and have detrimental effects on all aspects of the victim’s life. Once, thought of as a primarily a GBV in developing countries, recent rising incidents of chemical assaults in developed nations such as the UK has given a new perspective to this global health problem. In countries, such as Bangladesh, strict national guidelines defining chemical assaults, regulating its distribution, harsh penalization of the assailants and implementation of resources for victim rehabilitation have shown to be promising in decreasing the incidents and the societal burden of such crimes. Used as a means to threaten or punish women and girls, chemical assaults appear to remain primarily a gender-based crime in most developing countries. However, over the past decade, nations such as the UK have seen a significant rise in the utilization of chemical substances as weapons by persons of various backgrounds, such as school children and gangs, against others. London has become one of the epicentres of chemical assaults, affecting victims of all backgrounds. Further differentiating the rise in chemical assaults in the UK from countries with limited resources is the availability and accessibility of data regarding such assaults. There are specific detailed reports of trends in chemical assaults in each region of the country that is accessible to public. Hospitals in London have been urged to share their data with the London Police to account for the unreported crimes. Such data have been invaluable in the utilization of resources and urging for more legislature combatting chemical assaults and violence more broadly. There are limited data regarding each country’s true incidence of chemical assaults. Such information is critical for (a) targeted public education (e.g. teenagers and gang members in the UK vs adult men in India); (b) establishing a legal framework based on a country’s capacity to effectively control chemical substances commonly used in assaults, (c) effective and ethical penalization of such crimes, and (d) adequate allocation of resources to victim rehabilitation. To date, NGOs working in local communities have been of outmost importance in not only gathering crucial information but also utilizing data to advocate for specific legislature or amendments to current ones while providing aid to the victims (Peck et al., 2009). In countries, such as Iran, where raising awareness about chemical assaults is met with government sanctioned arrests and prosecution of campaigners, true incidence and cause of acid attacks remain undetected, and victims do not receive adequate care or justice (Farhad et al., 2011; Rasouli et al., 2015). The paucity of epidemiological and legal data around chemical assault limits understanding of the drivers (e.g. motives, social construct, legal deterrence) and means (e.g. methods of chemical acquisition, delivery mechanisms) of attacks and ways to prevent them and protect potential victims. Although reports suggest that chemical assaults are not confined to a particular region, religion or culture, little is known about the impacts of chemical assaults on victims across their life courses (Sabzi Khoshnami et al., 2017). It is difficult to ascertain the true scope of the problem for several reasons (Thomson Reuters Foundation et al., 2017). First, victims are often marginalized, disenfranchised and unable to report their assault or do not do so due to shame or fear. Second, most countries do not have a system for recording and monitoring chemical assaults or burn injuries generally. Lastly, when resources are available to facilitate reporting, monitoring and care (e.g. advocacy groups, hospitals capable of burn care), they are often concentrated in urban centres, which leads to underreporting in more rural areas (Dissanaike and Rahimi, 2009). Victims may never present to care for a multitude of reasons so their cases go unrecorded or live countries in conflict (e.g. northern Uganda, South Sudan, Somalia, Syria, Afghanistan) where there may not be medical facilities to participate in data collection and provide adequate care of these often complex injuries (Sankari et al., 2013; Gupta et al., 2014b). Research that aims to improve our understanding of the causes and effects of chemical assaults is needed. In addition, research can aid the identification of individual and social interventions, as well as practical and feasible policy and legal solutions to control chemical assaults. Academia, non-governmental organizations, domestic and international health agencies, advocacy groups, survivor support organizations and legal think tanks can support research around chemical assault to improve targeted interventions and lower the burdens associated with this type of violence. Comprehensive legislative frameworks have been successful in controlling injury and violence (e.g. prevention of road traffic injuries, firearm violence, sexual assault, occupational injuries) when fully implemented and enforced (Swahn et al., 2011). For example, the United Nations and WHO proposed the following legislative framework for controlling road traffic injuries: support an institutional agency and strategy; make safer roads and mobility; make safer vehicles; make safer road users; improve post-crash care; and collect and use data (Toroyan et al., 2015). The WHO’s Global Status Report on Road Safety reported increasing adoption of this framework by national legislatures globally. As a result, there has been a plateau in road traffic deaths in the last few years. Set against a 4% increase in global population and a 16% increase in motorization, current estimates suggest that road safety legislative efforts have saved lives. In the last 3 years alone, 17 countries representing 409 million people have amended their laws on one or more key risk factors for road traffic injuries to bring them into line with the framework (Toroyan et al., 2015). Road safety gains have been particularly apparent in countries with comprehensive implementation and high enforcement rates (Ning et al., 2016). Although chemical attacks represent smaller health, economic and social burdens than road traffic injuries, numerous lessons can be learned from the creation and implementation of road safety legislation and applied to chemical assault control, including the importance of comprehensive vs piece-meal legal framework implementation, enforcement and data sharing and harmonization. However, more evidence is needed regarding the effects of these legislative priorities on the incidence and impacts of chemical assault to inform projections after implementation. In addition, there are several anticipated challenges, including the unexpected impacts on the commerce of acids and corrosive substances for appropriate purposes, creation of special legal pathways for specific victim types and lack of popular support in communities that promote or maintain gender inequity. Table 4 Legislation that facilitates justice in cases of chemical assault Country . Bill name (year) . Bill content . Effects . Afghanistan Elimination Violence against Women (2009) The offender of violence against women by spraying or forced ingestion of chemical or poisonous substances is subject to ‘long term imprisonment not ˂10 years’. If the attack is meant to spread terror in the society to prevent women from exercising their civil rights or results in the death of the victim the offender is subject to life sentence or the death penalty. Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) The new act imposes the death penalty for perpetrators of acid violence and penalty of up to Tk 1 lakh (∼$1709). The police sometimes takes the side of the offenders for a bribe and protect them from law. Cambodia Acid Law (2012) Intentional killing with acid is punishable by 15–30 years imprisonment, which can be increased to life if the attack was an ambush or the perpetrator tortured the victim before or during the killing. Torture and other cruel acts while using acid is punishable by 10–20 years, increased to 15–25 years if the act leads to permanent disability, and 20–30 years if the torture leads to the unintentional death or suicide of the victim. Intentional violence using acid is punishable by 2–5 years, increased to 5–10 years if the intentional violence leads to permanent disability, and 15–25 years if the intentional violence leads to the unintentional death or suicide of the victim. Lack of prosecution of perpetrators is primarily due to the decision of survivors not to file a complaint after they have been attacked. Survivors’ reasons for not filing a complaint include fear of reprisals by the perpetrator, fear of costs related to court proceedings and legal aid, a lack of confidence in the legal system and not wanting to ‘waste’ time on it and agreements to settle the matter out of court, which gives survivors immediate access to the funds. Frequently, the legal process suddenly stops and legal cases unofficially end. There is low sentencing rate even when a trial is held. Even when a court has sentenced a perpetrator of acid violence to imprisonment and/or a monetary fine, it is common that these sentences are never executed. Colombia Natalia Ponce de León law (2016) Prior to this law, acid attacks were treated as lesser physical aggressions, like beatings, and few attackers were convicted and jailed. The 2016 law states that those who ‘use any type of chemical agent’ to hurt someone will be jailed for between 12 and 20 years, but the sentence can increase to up to 50 years in prison for those who permanently disfigure their victim. Finland Criminal Code, Chapter 21 Sections 5 (1996), 6 (2001) Chemical assault offenders can be charged either with Assault, maximum 2 years in prison, or Aggravated Assault, with 1–10 years in prison. India The Criminal Law (Amendment) Ordinance/Manodhairya Scheme (2013) Sections 326A, 326B The Law Commission of India has recommended that the offender committing acid attacks be sent to 10-year imprisonment extendible up to life term and fined Rs 10 lakhs (∼$15K) by including the offence in the Penal Code. The National Commission for Women has also asked to include a separate clause for acid attack with punishment not less than imprisonment for 10 years. Under Section 326A, chemical assaults are punishable by up to life in prison and under 326B, attempted chemical assaults are punishable by 10 years in prison. In one incident the offender was sentenced to death. Iran Assaults, not specific to corrosive substances, are punishable by monetary compensation, imprisonment, or retribution ruling. In two instances, the offenders who had caused blindness were also blinded. Kenya Act no. 12: Prevention of Torture Act. (2017) Violence Against Persons Prohibition Act (2015) In sentencing a person convicted for an offence under this Act, severity of the act of torture should be considered. This depends on all the circumstances of the case, including: Duration of the treatment or punishment, duration of the physical and mental effects or the treatment or punishment and the victim's sex, religion, age and state of health. A person who commits an offence under this Act for which a penalty is not prescribed, shall be liable, on conviction, to a fine not exceeding one million shillings (∼$10K) or to imprisonment for a term not exceeding 5 years or to both. Per the VAPP, any person who unlawfully does grievous harm to another is guilty of a felony and is liable to imprisonment for 7 years. Malaysia Act no. 357: Corrosive and Explosive Substances and Offensive Weapons Act (1958) Per the Act: any person who carries or has in his possession or under his control any corrosive in circumstances which raise a reasonable presumption that he intends to use or intends to enable some other person to use such substance for the purpose of causing hurt shall on conviction be liable to imprisonment for a term of maximum 3 years, and to whipping. If found with someone who had acid, while knowing about it, the person will be found ‘guilty of an offence and shall be liable to the like punishment as that other person with whom he was consorting or in whose company he was found.' Nepal Criminal Code Bill (August, 2018) Per Section 193 of the bill ‘If a person throws acid at a person or uses any chemical substance to attack a person in a manner to cause serious harm or disfigure the victim, the perpetrator shall be sentenced to 5–8 years in jail depending on the gravity of the offence and fined Rs 100 000 to Rs 300 000 (∼$100–300). The fine imposed on the perpetrator shall be provided to the survivor.’ In case of death of the victim in such attack, the perpetrator will face murder charge. The bill will take the shape of act and come into effect on 17 August 2018. Pakistan Criminal Law Act (2011) The Act provides for maximum sentence of life imprisonment and minimum of up to 14 years with fine of Rs1 million (∼$8.7K) for the offence of hurting a person with corrosive substances. Thailand Domestic Violence Victim Protection Act, B.E. 2550 (2007) The maximum penalty for committing an act of domestic violence is up to 6 months imprisonment or a 6000 baht (∼$186) fine or both, excluding ‘relief money’ that a Court may order an offender to pay to the victim. If the offender is found to be guilty of committing domestic violence, instead of punishing the offender, the Court may ‘apply certain measures of reformation, treatment or correction to him, may direct him to pay an amount of relief money, carry out community service or refrain from the acts giving rise to domestic violence, or may place him under a peace bond, subject to the procedure and period of time determined by the court’. Uganda Penal Code Cap 120 (1950) Per Section 216: perpetrators of acid or burn violence intending to maim are liable to life imprisonment. Should an acid attack committed with ‘malice aforethought’ lead to death, then it is considered murder (Section188) and is punishable by death (Section189). The lesser crime of attempted murder, covered by Section 204, provides for up to life imprisonment. Acid assaults committed during an actual or attempted robbery are liable to 10 years’ imprisonment if convicted by a magistrate’s court or to life imprisonment if convicted by the High Court (Section 286(3)(ii)). The individuals who do not actually pour the acid but are complicit in the crime are charged under Section 208 and liable to 14 years in prison if they intended to murder the victim, or 7 years if such intent cannot be proved (Section 390). Chronic under-funding of the police’s Criminal Investigation Department causes slow investigations and consequently delays in prosecution. While it is the state’s responsibility to ensure access to justice for all, Acid Survivors Fund Uganda has sometimes paid the transport costs for investigating officers to enable cases to proceed. Sentences applied to those convicted of acid attacks vary widely, ranging from a few months to life imprisonment. UK Prevention of Crime Act: Offensive Weapons (1953) Per the Home Secretary, acid attack convictions could soon carry life sentences. Possession of acid or other corrosive substances with the intention to do harm can now be treated as possession of an offensive weapon under the Prevention of Crime Act, which carries a 4-year maximum penalty. Adults convicted of carrying a corrosive substance in public for a second time will be given a minimum 6-month jail term, and minors handed a 4-month detention and training order. Statistics released by the Met Police Force show 465 violent ‘corrosive liquid’ offences recorded in London alone in 2017, up from 397 incidents in 2016 (17% increase) and 260 in 2015 (78% increase). Only 19% of the offences resulted in a charge in 2017, down from over 25% in 2015—though the police claims to be still investigating many of the inquiries. USA Aggravated assault is usually a felony punishable by ∼1–20 years in prison, depending on the specific provisions of each state’s sentencing statute or sentencing guidelines. In sentencing the judge considers the following criteria: whether the perpetrator has taken responsibility for the crime and shows remorse, circumstances surrounding the crime, the extent of any injuries to the victim the type of weapon used, the accused’s prior criminal record and, sometimes, the victim's background or relationship to the defendant. In a 2016 domestic violence with corrosive substances, the offender was sentenced to 40 years in prison for two counts of aggravated battery and one count of aggravated assault. Country . Bill name (year) . Bill content . Effects . Afghanistan Elimination Violence against Women (2009) The offender of violence against women by spraying or forced ingestion of chemical or poisonous substances is subject to ‘long term imprisonment not ˂10 years’. If the attack is meant to spread terror in the society to prevent women from exercising their civil rights or results in the death of the victim the offender is subject to life sentence or the death penalty. Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) The new act imposes the death penalty for perpetrators of acid violence and penalty of up to Tk 1 lakh (∼$1709). The police sometimes takes the side of the offenders for a bribe and protect them from law. Cambodia Acid Law (2012) Intentional killing with acid is punishable by 15–30 years imprisonment, which can be increased to life if the attack was an ambush or the perpetrator tortured the victim before or during the killing. Torture and other cruel acts while using acid is punishable by 10–20 years, increased to 15–25 years if the act leads to permanent disability, and 20–30 years if the torture leads to the unintentional death or suicide of the victim. Intentional violence using acid is punishable by 2–5 years, increased to 5–10 years if the intentional violence leads to permanent disability, and 15–25 years if the intentional violence leads to the unintentional death or suicide of the victim. Lack of prosecution of perpetrators is primarily due to the decision of survivors not to file a complaint after they have been attacked. Survivors’ reasons for not filing a complaint include fear of reprisals by the perpetrator, fear of costs related to court proceedings and legal aid, a lack of confidence in the legal system and not wanting to ‘waste’ time on it and agreements to settle the matter out of court, which gives survivors immediate access to the funds. Frequently, the legal process suddenly stops and legal cases unofficially end. There is low sentencing rate even when a trial is held. Even when a court has sentenced a perpetrator of acid violence to imprisonment and/or a monetary fine, it is common that these sentences are never executed. Colombia Natalia Ponce de León law (2016) Prior to this law, acid attacks were treated as lesser physical aggressions, like beatings, and few attackers were convicted and jailed. The 2016 law states that those who ‘use any type of chemical agent’ to hurt someone will be jailed for between 12 and 20 years, but the sentence can increase to up to 50 years in prison for those who permanently disfigure their victim. Finland Criminal Code, Chapter 21 Sections 5 (1996), 6 (2001) Chemical assault offenders can be charged either with Assault, maximum 2 years in prison, or Aggravated Assault, with 1–10 years in prison. India The Criminal Law (Amendment) Ordinance/Manodhairya Scheme (2013) Sections 326A, 326B The Law Commission of India has recommended that the offender committing acid attacks be sent to 10-year imprisonment extendible up to life term and fined Rs 10 lakhs (∼$15K) by including the offence in the Penal Code. The National Commission for Women has also asked to include a separate clause for acid attack with punishment not less than imprisonment for 10 years. Under Section 326A, chemical assaults are punishable by up to life in prison and under 326B, attempted chemical assaults are punishable by 10 years in prison. In one incident the offender was sentenced to death. Iran Assaults, not specific to corrosive substances, are punishable by monetary compensation, imprisonment, or retribution ruling. In two instances, the offenders who had caused blindness were also blinded. Kenya Act no. 12: Prevention of Torture Act. (2017) Violence Against Persons Prohibition Act (2015) In sentencing a person convicted for an offence under this Act, severity of the act of torture should be considered. This depends on all the circumstances of the case, including: Duration of the treatment or punishment, duration of the physical and mental effects or the treatment or punishment and the victim's sex, religion, age and state of health. A person who commits an offence under this Act for which a penalty is not prescribed, shall be liable, on conviction, to a fine not exceeding one million shillings (∼$10K) or to imprisonment for a term not exceeding 5 years or to both. Per the VAPP, any person who unlawfully does grievous harm to another is guilty of a felony and is liable to imprisonment for 7 years. Malaysia Act no. 357: Corrosive and Explosive Substances and Offensive Weapons Act (1958) Per the Act: any person who carries or has in his possession or under his control any corrosive in circumstances which raise a reasonable presumption that he intends to use or intends to enable some other person to use such substance for the purpose of causing hurt shall on conviction be liable to imprisonment for a term of maximum 3 years, and to whipping. If found with someone who had acid, while knowing about it, the person will be found ‘guilty of an offence and shall be liable to the like punishment as that other person with whom he was consorting or in whose company he was found.' Nepal Criminal Code Bill (August, 2018) Per Section 193 of the bill ‘If a person throws acid at a person or uses any chemical substance to attack a person in a manner to cause serious harm or disfigure the victim, the perpetrator shall be sentenced to 5–8 years in jail depending on the gravity of the offence and fined Rs 100 000 to Rs 300 000 (∼$100–300). The fine imposed on the perpetrator shall be provided to the survivor.’ In case of death of the victim in such attack, the perpetrator will face murder charge. The bill will take the shape of act and come into effect on 17 August 2018. Pakistan Criminal Law Act (2011) The Act provides for maximum sentence of life imprisonment and minimum of up to 14 years with fine of Rs1 million (∼$8.7K) for the offence of hurting a person with corrosive substances. Thailand Domestic Violence Victim Protection Act, B.E. 2550 (2007) The maximum penalty for committing an act of domestic violence is up to 6 months imprisonment or a 6000 baht (∼$186) fine or both, excluding ‘relief money’ that a Court may order an offender to pay to the victim. If the offender is found to be guilty of committing domestic violence, instead of punishing the offender, the Court may ‘apply certain measures of reformation, treatment or correction to him, may direct him to pay an amount of relief money, carry out community service or refrain from the acts giving rise to domestic violence, or may place him under a peace bond, subject to the procedure and period of time determined by the court’. Uganda Penal Code Cap 120 (1950) Per Section 216: perpetrators of acid or burn violence intending to maim are liable to life imprisonment. Should an acid attack committed with ‘malice aforethought’ lead to death, then it is considered murder (Section188) and is punishable by death (Section189). The lesser crime of attempted murder, covered by Section 204, provides for up to life imprisonment. Acid assaults committed during an actual or attempted robbery are liable to 10 years’ imprisonment if convicted by a magistrate’s court or to life imprisonment if convicted by the High Court (Section 286(3)(ii)). The individuals who do not actually pour the acid but are complicit in the crime are charged under Section 208 and liable to 14 years in prison if they intended to murder the victim, or 7 years if such intent cannot be proved (Section 390). Chronic under-funding of the police’s Criminal Investigation Department causes slow investigations and consequently delays in prosecution. While it is the state’s responsibility to ensure access to justice for all, Acid Survivors Fund Uganda has sometimes paid the transport costs for investigating officers to enable cases to proceed. Sentences applied to those convicted of acid attacks vary widely, ranging from a few months to life imprisonment. UK Prevention of Crime Act: Offensive Weapons (1953) Per the Home Secretary, acid attack convictions could soon carry life sentences. Possession of acid or other corrosive substances with the intention to do harm can now be treated as possession of an offensive weapon under the Prevention of Crime Act, which carries a 4-year maximum penalty. Adults convicted of carrying a corrosive substance in public for a second time will be given a minimum 6-month jail term, and minors handed a 4-month detention and training order. Statistics released by the Met Police Force show 465 violent ‘corrosive liquid’ offences recorded in London alone in 2017, up from 397 incidents in 2016 (17% increase) and 260 in 2015 (78% increase). Only 19% of the offences resulted in a charge in 2017, down from over 25% in 2015—though the police claims to be still investigating many of the inquiries. USA Aggravated assault is usually a felony punishable by ∼1–20 years in prison, depending on the specific provisions of each state’s sentencing statute or sentencing guidelines. In sentencing the judge considers the following criteria: whether the perpetrator has taken responsibility for the crime and shows remorse, circumstances surrounding the crime, the extent of any injuries to the victim the type of weapon used, the accused’s prior criminal record and, sometimes, the victim's background or relationship to the defendant. In a 2016 domestic violence with corrosive substances, the offender was sentenced to 40 years in prison for two counts of aggravated battery and one count of aggravated assault. VAPP, Violence against Persons Prohibition Act. Open in new tab Table 4 Legislation that facilitates justice in cases of chemical assault Country . Bill name (year) . Bill content . Effects . Afghanistan Elimination Violence against Women (2009) The offender of violence against women by spraying or forced ingestion of chemical or poisonous substances is subject to ‘long term imprisonment not ˂10 years’. If the attack is meant to spread terror in the society to prevent women from exercising their civil rights or results in the death of the victim the offender is subject to life sentence or the death penalty. Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) The new act imposes the death penalty for perpetrators of acid violence and penalty of up to Tk 1 lakh (∼$1709). The police sometimes takes the side of the offenders for a bribe and protect them from law. Cambodia Acid Law (2012) Intentional killing with acid is punishable by 15–30 years imprisonment, which can be increased to life if the attack was an ambush or the perpetrator tortured the victim before or during the killing. Torture and other cruel acts while using acid is punishable by 10–20 years, increased to 15–25 years if the act leads to permanent disability, and 20–30 years if the torture leads to the unintentional death or suicide of the victim. Intentional violence using acid is punishable by 2–5 years, increased to 5–10 years if the intentional violence leads to permanent disability, and 15–25 years if the intentional violence leads to the unintentional death or suicide of the victim. Lack of prosecution of perpetrators is primarily due to the decision of survivors not to file a complaint after they have been attacked. Survivors’ reasons for not filing a complaint include fear of reprisals by the perpetrator, fear of costs related to court proceedings and legal aid, a lack of confidence in the legal system and not wanting to ‘waste’ time on it and agreements to settle the matter out of court, which gives survivors immediate access to the funds. Frequently, the legal process suddenly stops and legal cases unofficially end. There is low sentencing rate even when a trial is held. Even when a court has sentenced a perpetrator of acid violence to imprisonment and/or a monetary fine, it is common that these sentences are never executed. Colombia Natalia Ponce de León law (2016) Prior to this law, acid attacks were treated as lesser physical aggressions, like beatings, and few attackers were convicted and jailed. The 2016 law states that those who ‘use any type of chemical agent’ to hurt someone will be jailed for between 12 and 20 years, but the sentence can increase to up to 50 years in prison for those who permanently disfigure their victim. Finland Criminal Code, Chapter 21 Sections 5 (1996), 6 (2001) Chemical assault offenders can be charged either with Assault, maximum 2 years in prison, or Aggravated Assault, with 1–10 years in prison. India The Criminal Law (Amendment) Ordinance/Manodhairya Scheme (2013) Sections 326A, 326B The Law Commission of India has recommended that the offender committing acid attacks be sent to 10-year imprisonment extendible up to life term and fined Rs 10 lakhs (∼$15K) by including the offence in the Penal Code. The National Commission for Women has also asked to include a separate clause for acid attack with punishment not less than imprisonment for 10 years. Under Section 326A, chemical assaults are punishable by up to life in prison and under 326B, attempted chemical assaults are punishable by 10 years in prison. In one incident the offender was sentenced to death. Iran Assaults, not specific to corrosive substances, are punishable by monetary compensation, imprisonment, or retribution ruling. In two instances, the offenders who had caused blindness were also blinded. Kenya Act no. 12: Prevention of Torture Act. (2017) Violence Against Persons Prohibition Act (2015) In sentencing a person convicted for an offence under this Act, severity of the act of torture should be considered. This depends on all the circumstances of the case, including: Duration of the treatment or punishment, duration of the physical and mental effects or the treatment or punishment and the victim's sex, religion, age and state of health. A person who commits an offence under this Act for which a penalty is not prescribed, shall be liable, on conviction, to a fine not exceeding one million shillings (∼$10K) or to imprisonment for a term not exceeding 5 years or to both. Per the VAPP, any person who unlawfully does grievous harm to another is guilty of a felony and is liable to imprisonment for 7 years. Malaysia Act no. 357: Corrosive and Explosive Substances and Offensive Weapons Act (1958) Per the Act: any person who carries or has in his possession or under his control any corrosive in circumstances which raise a reasonable presumption that he intends to use or intends to enable some other person to use such substance for the purpose of causing hurt shall on conviction be liable to imprisonment for a term of maximum 3 years, and to whipping. If found with someone who had acid, while knowing about it, the person will be found ‘guilty of an offence and shall be liable to the like punishment as that other person with whom he was consorting or in whose company he was found.' Nepal Criminal Code Bill (August, 2018) Per Section 193 of the bill ‘If a person throws acid at a person or uses any chemical substance to attack a person in a manner to cause serious harm or disfigure the victim, the perpetrator shall be sentenced to 5–8 years in jail depending on the gravity of the offence and fined Rs 100 000 to Rs 300 000 (∼$100–300). The fine imposed on the perpetrator shall be provided to the survivor.’ In case of death of the victim in such attack, the perpetrator will face murder charge. The bill will take the shape of act and come into effect on 17 August 2018. Pakistan Criminal Law Act (2011) The Act provides for maximum sentence of life imprisonment and minimum of up to 14 years with fine of Rs1 million (∼$8.7K) for the offence of hurting a person with corrosive substances. Thailand Domestic Violence Victim Protection Act, B.E. 2550 (2007) The maximum penalty for committing an act of domestic violence is up to 6 months imprisonment or a 6000 baht (∼$186) fine or both, excluding ‘relief money’ that a Court may order an offender to pay to the victim. If the offender is found to be guilty of committing domestic violence, instead of punishing the offender, the Court may ‘apply certain measures of reformation, treatment or correction to him, may direct him to pay an amount of relief money, carry out community service or refrain from the acts giving rise to domestic violence, or may place him under a peace bond, subject to the procedure and period of time determined by the court’. Uganda Penal Code Cap 120 (1950) Per Section 216: perpetrators of acid or burn violence intending to maim are liable to life imprisonment. Should an acid attack committed with ‘malice aforethought’ lead to death, then it is considered murder (Section188) and is punishable by death (Section189). The lesser crime of attempted murder, covered by Section 204, provides for up to life imprisonment. Acid assaults committed during an actual or attempted robbery are liable to 10 years’ imprisonment if convicted by a magistrate’s court or to life imprisonment if convicted by the High Court (Section 286(3)(ii)). The individuals who do not actually pour the acid but are complicit in the crime are charged under Section 208 and liable to 14 years in prison if they intended to murder the victim, or 7 years if such intent cannot be proved (Section 390). Chronic under-funding of the police’s Criminal Investigation Department causes slow investigations and consequently delays in prosecution. While it is the state’s responsibility to ensure access to justice for all, Acid Survivors Fund Uganda has sometimes paid the transport costs for investigating officers to enable cases to proceed. Sentences applied to those convicted of acid attacks vary widely, ranging from a few months to life imprisonment. UK Prevention of Crime Act: Offensive Weapons (1953) Per the Home Secretary, acid attack convictions could soon carry life sentences. Possession of acid or other corrosive substances with the intention to do harm can now be treated as possession of an offensive weapon under the Prevention of Crime Act, which carries a 4-year maximum penalty. Adults convicted of carrying a corrosive substance in public for a second time will be given a minimum 6-month jail term, and minors handed a 4-month detention and training order. Statistics released by the Met Police Force show 465 violent ‘corrosive liquid’ offences recorded in London alone in 2017, up from 397 incidents in 2016 (17% increase) and 260 in 2015 (78% increase). Only 19% of the offences resulted in a charge in 2017, down from over 25% in 2015—though the police claims to be still investigating many of the inquiries. USA Aggravated assault is usually a felony punishable by ∼1–20 years in prison, depending on the specific provisions of each state’s sentencing statute or sentencing guidelines. In sentencing the judge considers the following criteria: whether the perpetrator has taken responsibility for the crime and shows remorse, circumstances surrounding the crime, the extent of any injuries to the victim the type of weapon used, the accused’s prior criminal record and, sometimes, the victim's background or relationship to the defendant. In a 2016 domestic violence with corrosive substances, the offender was sentenced to 40 years in prison for two counts of aggravated battery and one count of aggravated assault. Country . Bill name (year) . Bill content . Effects . Afghanistan Elimination Violence against Women (2009) The offender of violence against women by spraying or forced ingestion of chemical or poisonous substances is subject to ‘long term imprisonment not ˂10 years’. If the attack is meant to spread terror in the society to prevent women from exercising their civil rights or results in the death of the victim the offender is subject to life sentence or the death penalty. Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) The new act imposes the death penalty for perpetrators of acid violence and penalty of up to Tk 1 lakh (∼$1709). The police sometimes takes the side of the offenders for a bribe and protect them from law. Cambodia Acid Law (2012) Intentional killing with acid is punishable by 15–30 years imprisonment, which can be increased to life if the attack was an ambush or the perpetrator tortured the victim before or during the killing. Torture and other cruel acts while using acid is punishable by 10–20 years, increased to 15–25 years if the act leads to permanent disability, and 20–30 years if the torture leads to the unintentional death or suicide of the victim. Intentional violence using acid is punishable by 2–5 years, increased to 5–10 years if the intentional violence leads to permanent disability, and 15–25 years if the intentional violence leads to the unintentional death or suicide of the victim. Lack of prosecution of perpetrators is primarily due to the decision of survivors not to file a complaint after they have been attacked. Survivors’ reasons for not filing a complaint include fear of reprisals by the perpetrator, fear of costs related to court proceedings and legal aid, a lack of confidence in the legal system and not wanting to ‘waste’ time on it and agreements to settle the matter out of court, which gives survivors immediate access to the funds. Frequently, the legal process suddenly stops and legal cases unofficially end. There is low sentencing rate even when a trial is held. Even when a court has sentenced a perpetrator of acid violence to imprisonment and/or a monetary fine, it is common that these sentences are never executed. Colombia Natalia Ponce de León law (2016) Prior to this law, acid attacks were treated as lesser physical aggressions, like beatings, and few attackers were convicted and jailed. The 2016 law states that those who ‘use any type of chemical agent’ to hurt someone will be jailed for between 12 and 20 years, but the sentence can increase to up to 50 years in prison for those who permanently disfigure their victim. Finland Criminal Code, Chapter 21 Sections 5 (1996), 6 (2001) Chemical assault offenders can be charged either with Assault, maximum 2 years in prison, or Aggravated Assault, with 1–10 years in prison. India The Criminal Law (Amendment) Ordinance/Manodhairya Scheme (2013) Sections 326A, 326B The Law Commission of India has recommended that the offender committing acid attacks be sent to 10-year imprisonment extendible up to life term and fined Rs 10 lakhs (∼$15K) by including the offence in the Penal Code. The National Commission for Women has also asked to include a separate clause for acid attack with punishment not less than imprisonment for 10 years. Under Section 326A, chemical assaults are punishable by up to life in prison and under 326B, attempted chemical assaults are punishable by 10 years in prison. In one incident the offender was sentenced to death. Iran Assaults, not specific to corrosive substances, are punishable by monetary compensation, imprisonment, or retribution ruling. In two instances, the offenders who had caused blindness were also blinded. Kenya Act no. 12: Prevention of Torture Act. (2017) Violence Against Persons Prohibition Act (2015) In sentencing a person convicted for an offence under this Act, severity of the act of torture should be considered. This depends on all the circumstances of the case, including: Duration of the treatment or punishment, duration of the physical and mental effects or the treatment or punishment and the victim's sex, religion, age and state of health. A person who commits an offence under this Act for which a penalty is not prescribed, shall be liable, on conviction, to a fine not exceeding one million shillings (∼$10K) or to imprisonment for a term not exceeding 5 years or to both. Per the VAPP, any person who unlawfully does grievous harm to another is guilty of a felony and is liable to imprisonment for 7 years. Malaysia Act no. 357: Corrosive and Explosive Substances and Offensive Weapons Act (1958) Per the Act: any person who carries or has in his possession or under his control any corrosive in circumstances which raise a reasonable presumption that he intends to use or intends to enable some other person to use such substance for the purpose of causing hurt shall on conviction be liable to imprisonment for a term of maximum 3 years, and to whipping. If found with someone who had acid, while knowing about it, the person will be found ‘guilty of an offence and shall be liable to the like punishment as that other person with whom he was consorting or in whose company he was found.' Nepal Criminal Code Bill (August, 2018) Per Section 193 of the bill ‘If a person throws acid at a person or uses any chemical substance to attack a person in a manner to cause serious harm or disfigure the victim, the perpetrator shall be sentenced to 5–8 years in jail depending on the gravity of the offence and fined Rs 100 000 to Rs 300 000 (∼$100–300). The fine imposed on the perpetrator shall be provided to the survivor.’ In case of death of the victim in such attack, the perpetrator will face murder charge. The bill will take the shape of act and come into effect on 17 August 2018. Pakistan Criminal Law Act (2011) The Act provides for maximum sentence of life imprisonment and minimum of up to 14 years with fine of Rs1 million (∼$8.7K) for the offence of hurting a person with corrosive substances. Thailand Domestic Violence Victim Protection Act, B.E. 2550 (2007) The maximum penalty for committing an act of domestic violence is up to 6 months imprisonment or a 6000 baht (∼$186) fine or both, excluding ‘relief money’ that a Court may order an offender to pay to the victim. If the offender is found to be guilty of committing domestic violence, instead of punishing the offender, the Court may ‘apply certain measures of reformation, treatment or correction to him, may direct him to pay an amount of relief money, carry out community service or refrain from the acts giving rise to domestic violence, or may place him under a peace bond, subject to the procedure and period of time determined by the court’. Uganda Penal Code Cap 120 (1950) Per Section 216: perpetrators of acid or burn violence intending to maim are liable to life imprisonment. Should an acid attack committed with ‘malice aforethought’ lead to death, then it is considered murder (Section188) and is punishable by death (Section189). The lesser crime of attempted murder, covered by Section 204, provides for up to life imprisonment. Acid assaults committed during an actual or attempted robbery are liable to 10 years’ imprisonment if convicted by a magistrate’s court or to life imprisonment if convicted by the High Court (Section 286(3)(ii)). The individuals who do not actually pour the acid but are complicit in the crime are charged under Section 208 and liable to 14 years in prison if they intended to murder the victim, or 7 years if such intent cannot be proved (Section 390). Chronic under-funding of the police’s Criminal Investigation Department causes slow investigations and consequently delays in prosecution. While it is the state’s responsibility to ensure access to justice for all, Acid Survivors Fund Uganda has sometimes paid the transport costs for investigating officers to enable cases to proceed. Sentences applied to those convicted of acid attacks vary widely, ranging from a few months to life imprisonment. UK Prevention of Crime Act: Offensive Weapons (1953) Per the Home Secretary, acid attack convictions could soon carry life sentences. Possession of acid or other corrosive substances with the intention to do harm can now be treated as possession of an offensive weapon under the Prevention of Crime Act, which carries a 4-year maximum penalty. Adults convicted of carrying a corrosive substance in public for a second time will be given a minimum 6-month jail term, and minors handed a 4-month detention and training order. Statistics released by the Met Police Force show 465 violent ‘corrosive liquid’ offences recorded in London alone in 2017, up from 397 incidents in 2016 (17% increase) and 260 in 2015 (78% increase). Only 19% of the offences resulted in a charge in 2017, down from over 25% in 2015—though the police claims to be still investigating many of the inquiries. USA Aggravated assault is usually a felony punishable by ∼1–20 years in prison, depending on the specific provisions of each state’s sentencing statute or sentencing guidelines. In sentencing the judge considers the following criteria: whether the perpetrator has taken responsibility for the crime and shows remorse, circumstances surrounding the crime, the extent of any injuries to the victim the type of weapon used, the accused’s prior criminal record and, sometimes, the victim's background or relationship to the defendant. In a 2016 domestic violence with corrosive substances, the offender was sentenced to 40 years in prison for two counts of aggravated battery and one count of aggravated assault. VAPP, Violence against Persons Prohibition Act. Open in new tab Although implementation and enforcement of a comprehensive legislative framework may prevent a proportion of chemical assaults and reduce their impacts on victims and their communities, the social constructs that allow chemical assaults must be addressed (Ellsberg et al., 2015). In much of the world, chemical assaults reflect and perpetuate discrimination of women and girls in societies and can only be eliminated through active promotion of gender equity and education by community organizations, religious institutions and governments (Jayatilleke et al., 2015). Violence against women and girls is the most widespread form of systematic abuse worldwide, affecting one third of all women in their lifetime (Garcia-Moreno and Watts, 2011). Preventing chemical assaults and violence against women and girls (VAWGs) more broadly is a development goal in its own right and central to achieving better lives for individual women and their families and greater development progress for communities, nations and the world (Kusuma and Babu, 2017). Beyond a moral imperative, states have a due diligence obligation to prevent chemical assault from occurring, to protect victims, to punish perpetrators and to provide recompense to those who have suffered from it. United Nations Special Rapporteurs on Torture and Cruel, Inhuman or Degrading Treatment or Punishment and Violence against Women, Its Causes and Consequences have defined the linkages between chemical assault and torture and VAWG (ASTI, 2018). As such, there is emerging attention to this crime on the international legal agenda. Table 5 Legislation that supports chemical assault survivors Country . Bill name (year) . Bill content . Effects . Afghanistan Elimination Violence Against Women (2009) The bill mandates providing victims with shelter or a safe space, legal aid, monetary compensation and free access to emergency health services. Victims can obtain prompt free treatment at the Ministry of Public Health and the crimes shall be reported accordingly to the Ministry of Women Affairs. Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) The Act has led to: Establishment of a National Acid Control Council Fund; Establishment of a Rehabilitation Centre for victims of acid crimes; Treatment for victims of acid crimes; Provision of Legal Aid for victims of acid crimes; Judgment in special tribunals; Judgment in the absence of the criminal; Power of the Magistrate to take record of witnesses anywhere. Providing medical care to victims has been limited since in 2009, there was only one Burn unit in a public hospital for the entire country. However, the Bangladesh Army is working to construct a 500-bed Burn institute, world’s largest burn and plastic surgery institute, in Dhaka, by September 2018. Cambodia Acid Law (2012) Requires the government to provide for acid victims’ medical care, legal support and protection. However, it is not clear how this will take place in practice. Per 2017 report by a Cambodia based NGO, the government has failed to provide medical and social support for the victims. India Criminal Law (Amended) Act, (2013) Code of Criminal Procedure Amendment (2009) Rights of Person with Disabilities Act (2016) Medical: All private and public hospitals will have to provide completely free treatment to victims of acid attacks. Doctor or hospital authorities found refusing such treatment will face up to 1 year of imprisonment. Section 166(B) of Criminal Law (Amended) Act, 2013. Karnataka State’s private hospitals or nursing homes, which admit the victims, must treat them on priority, extend scaled up security and, if needed, must counsel them with the help of psychiatrists. Per the Supreme Court, the states are responsible for the medical treatment and rehabilitation of the victims. The Bombay HC directed the State government to install trauma centres across the State for victims of acid attacks. The state government introduced a time-bound response system to ensure counselling and compensation for acid attack victims within 15 days of incidents. Police is responsible to immediately report acid attack cases to the local counselling ‘trauma team’ of the women and child development department, which will then start counselling of the survivor. Legal inquiries: Within 1 h of registering an FIR, police officials have to either message or email details of the incident and the victim to the welfare officer in the ‘trauma team’. Within 7 days of the FIR, a meeting of the ‘trauma team’ will have to be convened to examine whether the case is genuine and decide compensation. The compensation will have to be disbursed within 15 days of the incident. A female police officer should register and probe cases of crime against women. Compensation: CVCF scheme was created with an initial corpus of Rs 200 crores (∼$29.5 million). The state cabinet approved a scheme for paying compensation of Rs 3–10 lakhs ($4.4–15K) to acid attack victims. the scheme provides for financial, medical and other assistances to these victims. Previously victims were ineligible for compensation if attacked before February 10 2013. A separate policy decision will be taken on the cut-off date for the scheme. The policy decision will also decide if the cases lodged after 31 December 2009, would be covered under the October 2013 scheme or the revised scheme of December 2017 (Rs 10 lakhs vs Rs 30 lakhs). If the victims are ˂4 years old, there will be another 50% increase in ex gratia amounts. The Bombay HC has directed the state government to release funds under the compensation scheme for acid attack victims on a case-to-case basis, without limiting the amount to Rs 3 lakhs. Rs 30 000 ($442) must be disbursed as immediate relief to the victim within 7 days. The Maharashtra government was ordered by the Bombay HC to reduce procedural technicalities while awarding monetary compensation to victims of acid attacks. Acid attack survivors, who are now part of the list of those with disabilities under the new law, will be eligible for government jobs under the 1% quota for people with physical disabilities. The Rights of Person with Disabilities Act, 2016, was notified on December 28. The ministry of social justice and empowerment plans to notify the rules to implement the Act by mid-April. Per the 2013 law the person needed to be 40 % or more handicapped to be eligible for a reservation for government positions. It also did not consider acid attack victims as handicapped. In 2009, the Code of Criminal Procedure was amended, making it mandatory for the States to set up a ‘ victim compensation relief fund’ for victims or rape, sexual assault, and acid attacks. But state took almost 4 years to implement this. More than 80 acid attack survivors from across UP may have finally gotten reconstructive surgery with the state government extending them financial scheme of up to Rs 10 lakhs ($15K), funded under the UP women’s empowerment mission. A week after the state minister for women and child assured financial assistance for an acid attack victim, no aid has been forthcoming. A survivor of an acid attack, which occurred a year prior to October 2013, petitioned to receive compensation per the Manodhairya scheme. Her petition was originally declined. However, on 19 March, the Bombay HC had directed the state government to pay her medical bills and reimburse the hospital. The surgery, involving scar revision and laser treatment on her face, is estimated to be ∼Rs 4 lakhs (∼$6K). During 2013–14, the Women and Child Welfare Department dedicated Rs 5.10 crores for the scheme. Between October 2013 and March 2014, 876 FIRs were made. The state's committee approved of 410 cases (46%). The payment of Rs 5.04 crores was made in 296 cases (33% of FIRs). However, in 115 cases (13% of FIRs), payments could not be made because of the paucity of funds. Some victims had to wait for nearly 7 weeks after being approved to receive funds. Acid survivor, Reshma Qureshi should have received swift state aid after India's top court ruled that victims were entitled to 100 000 rupees ($1600) within 15 days, but she had not received any compensation 5 months later. With the support of one of the three Supreme Court judges and being recognized as ‘handicapped’ an acid attack survivor became an employee of the Supreme Court. Kenya Act no. 12: Prevention of Torture Act. (2017) Act no. 17: Victim Protection Act (2014) The act states that the victim shall obtain redress and have an enforceable right to restitution; adequate compensation; and rehabilitation. In the event of death, victim’s dependents should receive reparation Per the court’s order or victim’s request, the convicted should pay for the costs of medical and psychological treatment; and any other relief that the court may consider just. A court may at any time, at the request of the victim or an intermediary, grant an order for the treatment or counselling of the victim. The expenses incurred for the treatment or professional counselling of a victim granted under this section shall be charged on the Victim Protection Trust Fund established under Section 27 of the Victim Protection Act. All medical records relating to treatment are the property of the victim and shall only be used as evidence before any court with regard to any offence under informed consent of the victim. Nepal The current law states that anyone who commits a crime of domestic violence can pay victim compensation of up to 200 000 rupees (∼$1835). The Supreme Court, in response to a petition, ordered the government to amend the law to ensure victims of acid attacks and burning receive immediate compensation and critical care. Practically, only a small amount of compensation is ever given to a victim for treatment and nothing at all before any final decision is made in the case. There is shortage of specialized burns units and plastic surgeons, especially outside of the capital but the director of the National Health Training is hopeful that a basic burns management course could be incorporated into the curriculum for training health practitioners. Nigeria Violence against Persons Prohibition Act (2015) The act mandates that the identities of victims of offences remain protected. Through a protection order, he act restrains an individual or state actors from further abusive behaviour towards a victim The protection order is accompanied by a warrant for the arrest of the respondent, and it remains valid throughout the period of the protection order. Pakistan The latest version of the law directed medical officers to start treating the victim as soon as he/she is brought to the hospital. In the past, medical officers used to wait for police before starting treatment. The parliament once again rejected the Acid Burn Crime Bill (2014) that mandates completion of the investigation within 60 days, treatment of victim without waiting for police go-ahead, and for the government to bear the cost of the treatment of patient as well as compensation to the victim. Country . Bill name (year) . Bill content . Effects . Afghanistan Elimination Violence Against Women (2009) The bill mandates providing victims with shelter or a safe space, legal aid, monetary compensation and free access to emergency health services. Victims can obtain prompt free treatment at the Ministry of Public Health and the crimes shall be reported accordingly to the Ministry of Women Affairs. Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) The Act has led to: Establishment of a National Acid Control Council Fund; Establishment of a Rehabilitation Centre for victims of acid crimes; Treatment for victims of acid crimes; Provision of Legal Aid for victims of acid crimes; Judgment in special tribunals; Judgment in the absence of the criminal; Power of the Magistrate to take record of witnesses anywhere. Providing medical care to victims has been limited since in 2009, there was only one Burn unit in a public hospital for the entire country. However, the Bangladesh Army is working to construct a 500-bed Burn institute, world’s largest burn and plastic surgery institute, in Dhaka, by September 2018. Cambodia Acid Law (2012) Requires the government to provide for acid victims’ medical care, legal support and protection. However, it is not clear how this will take place in practice. Per 2017 report by a Cambodia based NGO, the government has failed to provide medical and social support for the victims. India Criminal Law (Amended) Act, (2013) Code of Criminal Procedure Amendment (2009) Rights of Person with Disabilities Act (2016) Medical: All private and public hospitals will have to provide completely free treatment to victims of acid attacks. Doctor or hospital authorities found refusing such treatment will face up to 1 year of imprisonment. Section 166(B) of Criminal Law (Amended) Act, 2013. Karnataka State’s private hospitals or nursing homes, which admit the victims, must treat them on priority, extend scaled up security and, if needed, must counsel them with the help of psychiatrists. Per the Supreme Court, the states are responsible for the medical treatment and rehabilitation of the victims. The Bombay HC directed the State government to install trauma centres across the State for victims of acid attacks. The state government introduced a time-bound response system to ensure counselling and compensation for acid attack victims within 15 days of incidents. Police is responsible to immediately report acid attack cases to the local counselling ‘trauma team’ of the women and child development department, which will then start counselling of the survivor. Legal inquiries: Within 1 h of registering an FIR, police officials have to either message or email details of the incident and the victim to the welfare officer in the ‘trauma team’. Within 7 days of the FIR, a meeting of the ‘trauma team’ will have to be convened to examine whether the case is genuine and decide compensation. The compensation will have to be disbursed within 15 days of the incident. A female police officer should register and probe cases of crime against women. Compensation: CVCF scheme was created with an initial corpus of Rs 200 crores (∼$29.5 million). The state cabinet approved a scheme for paying compensation of Rs 3–10 lakhs ($4.4–15K) to acid attack victims. the scheme provides for financial, medical and other assistances to these victims. Previously victims were ineligible for compensation if attacked before February 10 2013. A separate policy decision will be taken on the cut-off date for the scheme. The policy decision will also decide if the cases lodged after 31 December 2009, would be covered under the October 2013 scheme or the revised scheme of December 2017 (Rs 10 lakhs vs Rs 30 lakhs). If the victims are ˂4 years old, there will be another 50% increase in ex gratia amounts. The Bombay HC has directed the state government to release funds under the compensation scheme for acid attack victims on a case-to-case basis, without limiting the amount to Rs 3 lakhs. Rs 30 000 ($442) must be disbursed as immediate relief to the victim within 7 days. The Maharashtra government was ordered by the Bombay HC to reduce procedural technicalities while awarding monetary compensation to victims of acid attacks. Acid attack survivors, who are now part of the list of those with disabilities under the new law, will be eligible for government jobs under the 1% quota for people with physical disabilities. The Rights of Person with Disabilities Act, 2016, was notified on December 28. The ministry of social justice and empowerment plans to notify the rules to implement the Act by mid-April. Per the 2013 law the person needed to be 40 % or more handicapped to be eligible for a reservation for government positions. It also did not consider acid attack victims as handicapped. In 2009, the Code of Criminal Procedure was amended, making it mandatory for the States to set up a ‘ victim compensation relief fund’ for victims or rape, sexual assault, and acid attacks. But state took almost 4 years to implement this. More than 80 acid attack survivors from across UP may have finally gotten reconstructive surgery with the state government extending them financial scheme of up to Rs 10 lakhs ($15K), funded under the UP women’s empowerment mission. A week after the state minister for women and child assured financial assistance for an acid attack victim, no aid has been forthcoming. A survivor of an acid attack, which occurred a year prior to October 2013, petitioned to receive compensation per the Manodhairya scheme. Her petition was originally declined. However, on 19 March, the Bombay HC had directed the state government to pay her medical bills and reimburse the hospital. The surgery, involving scar revision and laser treatment on her face, is estimated to be ∼Rs 4 lakhs (∼$6K). During 2013–14, the Women and Child Welfare Department dedicated Rs 5.10 crores for the scheme. Between October 2013 and March 2014, 876 FIRs were made. The state's committee approved of 410 cases (46%). The payment of Rs 5.04 crores was made in 296 cases (33% of FIRs). However, in 115 cases (13% of FIRs), payments could not be made because of the paucity of funds. Some victims had to wait for nearly 7 weeks after being approved to receive funds. Acid survivor, Reshma Qureshi should have received swift state aid after India's top court ruled that victims were entitled to 100 000 rupees ($1600) within 15 days, but she had not received any compensation 5 months later. With the support of one of the three Supreme Court judges and being recognized as ‘handicapped’ an acid attack survivor became an employee of the Supreme Court. Kenya Act no. 12: Prevention of Torture Act. (2017) Act no. 17: Victim Protection Act (2014) The act states that the victim shall obtain redress and have an enforceable right to restitution; adequate compensation; and rehabilitation. In the event of death, victim’s dependents should receive reparation Per the court’s order or victim’s request, the convicted should pay for the costs of medical and psychological treatment; and any other relief that the court may consider just. A court may at any time, at the request of the victim or an intermediary, grant an order for the treatment or counselling of the victim. The expenses incurred for the treatment or professional counselling of a victim granted under this section shall be charged on the Victim Protection Trust Fund established under Section 27 of the Victim Protection Act. All medical records relating to treatment are the property of the victim and shall only be used as evidence before any court with regard to any offence under informed consent of the victim. Nepal The current law states that anyone who commits a crime of domestic violence can pay victim compensation of up to 200 000 rupees (∼$1835). The Supreme Court, in response to a petition, ordered the government to amend the law to ensure victims of acid attacks and burning receive immediate compensation and critical care. Practically, only a small amount of compensation is ever given to a victim for treatment and nothing at all before any final decision is made in the case. There is shortage of specialized burns units and plastic surgeons, especially outside of the capital but the director of the National Health Training is hopeful that a basic burns management course could be incorporated into the curriculum for training health practitioners. Nigeria Violence against Persons Prohibition Act (2015) The act mandates that the identities of victims of offences remain protected. Through a protection order, he act restrains an individual or state actors from further abusive behaviour towards a victim The protection order is accompanied by a warrant for the arrest of the respondent, and it remains valid throughout the period of the protection order. Pakistan The latest version of the law directed medical officers to start treating the victim as soon as he/she is brought to the hospital. In the past, medical officers used to wait for police before starting treatment. The parliament once again rejected the Acid Burn Crime Bill (2014) that mandates completion of the investigation within 60 days, treatment of victim without waiting for police go-ahead, and for the government to bear the cost of the treatment of patient as well as compensation to the victim. HC, high court; FIR, first information report; CVCF, Central Victim Compensation Fund; NGO, non-governmental organization. Open in new tab Table 5 Legislation that supports chemical assault survivors Country . Bill name (year) . Bill content . Effects . Afghanistan Elimination Violence Against Women (2009) The bill mandates providing victims with shelter or a safe space, legal aid, monetary compensation and free access to emergency health services. Victims can obtain prompt free treatment at the Ministry of Public Health and the crimes shall be reported accordingly to the Ministry of Women Affairs. Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) The Act has led to: Establishment of a National Acid Control Council Fund; Establishment of a Rehabilitation Centre for victims of acid crimes; Treatment for victims of acid crimes; Provision of Legal Aid for victims of acid crimes; Judgment in special tribunals; Judgment in the absence of the criminal; Power of the Magistrate to take record of witnesses anywhere. Providing medical care to victims has been limited since in 2009, there was only one Burn unit in a public hospital for the entire country. However, the Bangladesh Army is working to construct a 500-bed Burn institute, world’s largest burn and plastic surgery institute, in Dhaka, by September 2018. Cambodia Acid Law (2012) Requires the government to provide for acid victims’ medical care, legal support and protection. However, it is not clear how this will take place in practice. Per 2017 report by a Cambodia based NGO, the government has failed to provide medical and social support for the victims. India Criminal Law (Amended) Act, (2013) Code of Criminal Procedure Amendment (2009) Rights of Person with Disabilities Act (2016) Medical: All private and public hospitals will have to provide completely free treatment to victims of acid attacks. Doctor or hospital authorities found refusing such treatment will face up to 1 year of imprisonment. Section 166(B) of Criminal Law (Amended) Act, 2013. Karnataka State’s private hospitals or nursing homes, which admit the victims, must treat them on priority, extend scaled up security and, if needed, must counsel them with the help of psychiatrists. Per the Supreme Court, the states are responsible for the medical treatment and rehabilitation of the victims. The Bombay HC directed the State government to install trauma centres across the State for victims of acid attacks. The state government introduced a time-bound response system to ensure counselling and compensation for acid attack victims within 15 days of incidents. Police is responsible to immediately report acid attack cases to the local counselling ‘trauma team’ of the women and child development department, which will then start counselling of the survivor. Legal inquiries: Within 1 h of registering an FIR, police officials have to either message or email details of the incident and the victim to the welfare officer in the ‘trauma team’. Within 7 days of the FIR, a meeting of the ‘trauma team’ will have to be convened to examine whether the case is genuine and decide compensation. The compensation will have to be disbursed within 15 days of the incident. A female police officer should register and probe cases of crime against women. Compensation: CVCF scheme was created with an initial corpus of Rs 200 crores (∼$29.5 million). The state cabinet approved a scheme for paying compensation of Rs 3–10 lakhs ($4.4–15K) to acid attack victims. the scheme provides for financial, medical and other assistances to these victims. Previously victims were ineligible for compensation if attacked before February 10 2013. A separate policy decision will be taken on the cut-off date for the scheme. The policy decision will also decide if the cases lodged after 31 December 2009, would be covered under the October 2013 scheme or the revised scheme of December 2017 (Rs 10 lakhs vs Rs 30 lakhs). If the victims are ˂4 years old, there will be another 50% increase in ex gratia amounts. The Bombay HC has directed the state government to release funds under the compensation scheme for acid attack victims on a case-to-case basis, without limiting the amount to Rs 3 lakhs. Rs 30 000 ($442) must be disbursed as immediate relief to the victim within 7 days. The Maharashtra government was ordered by the Bombay HC to reduce procedural technicalities while awarding monetary compensation to victims of acid attacks. Acid attack survivors, who are now part of the list of those with disabilities under the new law, will be eligible for government jobs under the 1% quota for people with physical disabilities. The Rights of Person with Disabilities Act, 2016, was notified on December 28. The ministry of social justice and empowerment plans to notify the rules to implement the Act by mid-April. Per the 2013 law the person needed to be 40 % or more handicapped to be eligible for a reservation for government positions. It also did not consider acid attack victims as handicapped. In 2009, the Code of Criminal Procedure was amended, making it mandatory for the States to set up a ‘ victim compensation relief fund’ for victims or rape, sexual assault, and acid attacks. But state took almost 4 years to implement this. More than 80 acid attack survivors from across UP may have finally gotten reconstructive surgery with the state government extending them financial scheme of up to Rs 10 lakhs ($15K), funded under the UP women’s empowerment mission. A week after the state minister for women and child assured financial assistance for an acid attack victim, no aid has been forthcoming. A survivor of an acid attack, which occurred a year prior to October 2013, petitioned to receive compensation per the Manodhairya scheme. Her petition was originally declined. However, on 19 March, the Bombay HC had directed the state government to pay her medical bills and reimburse the hospital. The surgery, involving scar revision and laser treatment on her face, is estimated to be ∼Rs 4 lakhs (∼$6K). During 2013–14, the Women and Child Welfare Department dedicated Rs 5.10 crores for the scheme. Between October 2013 and March 2014, 876 FIRs were made. The state's committee approved of 410 cases (46%). The payment of Rs 5.04 crores was made in 296 cases (33% of FIRs). However, in 115 cases (13% of FIRs), payments could not be made because of the paucity of funds. Some victims had to wait for nearly 7 weeks after being approved to receive funds. Acid survivor, Reshma Qureshi should have received swift state aid after India's top court ruled that victims were entitled to 100 000 rupees ($1600) within 15 days, but she had not received any compensation 5 months later. With the support of one of the three Supreme Court judges and being recognized as ‘handicapped’ an acid attack survivor became an employee of the Supreme Court. Kenya Act no. 12: Prevention of Torture Act. (2017) Act no. 17: Victim Protection Act (2014) The act states that the victim shall obtain redress and have an enforceable right to restitution; adequate compensation; and rehabilitation. In the event of death, victim’s dependents should receive reparation Per the court’s order or victim’s request, the convicted should pay for the costs of medical and psychological treatment; and any other relief that the court may consider just. A court may at any time, at the request of the victim or an intermediary, grant an order for the treatment or counselling of the victim. The expenses incurred for the treatment or professional counselling of a victim granted under this section shall be charged on the Victim Protection Trust Fund established under Section 27 of the Victim Protection Act. All medical records relating to treatment are the property of the victim and shall only be used as evidence before any court with regard to any offence under informed consent of the victim. Nepal The current law states that anyone who commits a crime of domestic violence can pay victim compensation of up to 200 000 rupees (∼$1835). The Supreme Court, in response to a petition, ordered the government to amend the law to ensure victims of acid attacks and burning receive immediate compensation and critical care. Practically, only a small amount of compensation is ever given to a victim for treatment and nothing at all before any final decision is made in the case. There is shortage of specialized burns units and plastic surgeons, especially outside of the capital but the director of the National Health Training is hopeful that a basic burns management course could be incorporated into the curriculum for training health practitioners. Nigeria Violence against Persons Prohibition Act (2015) The act mandates that the identities of victims of offences remain protected. Through a protection order, he act restrains an individual or state actors from further abusive behaviour towards a victim The protection order is accompanied by a warrant for the arrest of the respondent, and it remains valid throughout the period of the protection order. Pakistan The latest version of the law directed medical officers to start treating the victim as soon as he/she is brought to the hospital. In the past, medical officers used to wait for police before starting treatment. The parliament once again rejected the Acid Burn Crime Bill (2014) that mandates completion of the investigation within 60 days, treatment of victim without waiting for police go-ahead, and for the government to bear the cost of the treatment of patient as well as compensation to the victim. Country . Bill name (year) . Bill content . Effects . Afghanistan Elimination Violence Against Women (2009) The bill mandates providing victims with shelter or a safe space, legal aid, monetary compensation and free access to emergency health services. Victims can obtain prompt free treatment at the Ministry of Public Health and the crimes shall be reported accordingly to the Ministry of Women Affairs. Bangladesh Acid Control Act (2002) Acid Crime Prevention Act (2002) The Act has led to: Establishment of a National Acid Control Council Fund; Establishment of a Rehabilitation Centre for victims of acid crimes; Treatment for victims of acid crimes; Provision of Legal Aid for victims of acid crimes; Judgment in special tribunals; Judgment in the absence of the criminal; Power of the Magistrate to take record of witnesses anywhere. Providing medical care to victims has been limited since in 2009, there was only one Burn unit in a public hospital for the entire country. However, the Bangladesh Army is working to construct a 500-bed Burn institute, world’s largest burn and plastic surgery institute, in Dhaka, by September 2018. Cambodia Acid Law (2012) Requires the government to provide for acid victims’ medical care, legal support and protection. However, it is not clear how this will take place in practice. Per 2017 report by a Cambodia based NGO, the government has failed to provide medical and social support for the victims. India Criminal Law (Amended) Act, (2013) Code of Criminal Procedure Amendment (2009) Rights of Person with Disabilities Act (2016) Medical: All private and public hospitals will have to provide completely free treatment to victims of acid attacks. Doctor or hospital authorities found refusing such treatment will face up to 1 year of imprisonment. Section 166(B) of Criminal Law (Amended) Act, 2013. Karnataka State’s private hospitals or nursing homes, which admit the victims, must treat them on priority, extend scaled up security and, if needed, must counsel them with the help of psychiatrists. Per the Supreme Court, the states are responsible for the medical treatment and rehabilitation of the victims. The Bombay HC directed the State government to install trauma centres across the State for victims of acid attacks. The state government introduced a time-bound response system to ensure counselling and compensation for acid attack victims within 15 days of incidents. Police is responsible to immediately report acid attack cases to the local counselling ‘trauma team’ of the women and child development department, which will then start counselling of the survivor. Legal inquiries: Within 1 h of registering an FIR, police officials have to either message or email details of the incident and the victim to the welfare officer in the ‘trauma team’. Within 7 days of the FIR, a meeting of the ‘trauma team’ will have to be convened to examine whether the case is genuine and decide compensation. The compensation will have to be disbursed within 15 days of the incident. A female police officer should register and probe cases of crime against women. Compensation: CVCF scheme was created with an initial corpus of Rs 200 crores (∼$29.5 million). The state cabinet approved a scheme for paying compensation of Rs 3–10 lakhs ($4.4–15K) to acid attack victims. the scheme provides for financial, medical and other assistances to these victims. Previously victims were ineligible for compensation if attacked before February 10 2013. A separate policy decision will be taken on the cut-off date for the scheme. The policy decision will also decide if the cases lodged after 31 December 2009, would be covered under the October 2013 scheme or the revised scheme of December 2017 (Rs 10 lakhs vs Rs 30 lakhs). If the victims are ˂4 years old, there will be another 50% increase in ex gratia amounts. The Bombay HC has directed the state government to release funds under the compensation scheme for acid attack victims on a case-to-case basis, without limiting the amount to Rs 3 lakhs. Rs 30 000 ($442) must be disbursed as immediate relief to the victim within 7 days. The Maharashtra government was ordered by the Bombay HC to reduce procedural technicalities while awarding monetary compensation to victims of acid attacks. Acid attack survivors, who are now part of the list of those with disabilities under the new law, will be eligible for government jobs under the 1% quota for people with physical disabilities. The Rights of Person with Disabilities Act, 2016, was notified on December 28. The ministry of social justice and empowerment plans to notify the rules to implement the Act by mid-April. Per the 2013 law the person needed to be 40 % or more handicapped to be eligible for a reservation for government positions. It also did not consider acid attack victims as handicapped. In 2009, the Code of Criminal Procedure was amended, making it mandatory for the States to set up a ‘ victim compensation relief fund’ for victims or rape, sexual assault, and acid attacks. But state took almost 4 years to implement this. More than 80 acid attack survivors from across UP may have finally gotten reconstructive surgery with the state government extending them financial scheme of up to Rs 10 lakhs ($15K), funded under the UP women’s empowerment mission. A week after the state minister for women and child assured financial assistance for an acid attack victim, no aid has been forthcoming. A survivor of an acid attack, which occurred a year prior to October 2013, petitioned to receive compensation per the Manodhairya scheme. Her petition was originally declined. However, on 19 March, the Bombay HC had directed the state government to pay her medical bills and reimburse the hospital. The surgery, involving scar revision and laser treatment on her face, is estimated to be ∼Rs 4 lakhs (∼$6K). During 2013–14, the Women and Child Welfare Department dedicated Rs 5.10 crores for the scheme. Between October 2013 and March 2014, 876 FIRs were made. The state's committee approved of 410 cases (46%). The payment of Rs 5.04 crores was made in 296 cases (33% of FIRs). However, in 115 cases (13% of FIRs), payments could not be made because of the paucity of funds. Some victims had to wait for nearly 7 weeks after being approved to receive funds. Acid survivor, Reshma Qureshi should have received swift state aid after India's top court ruled that victims were entitled to 100 000 rupees ($1600) within 15 days, but she had not received any compensation 5 months later. With the support of one of the three Supreme Court judges and being recognized as ‘handicapped’ an acid attack survivor became an employee of the Supreme Court. Kenya Act no. 12: Prevention of Torture Act. (2017) Act no. 17: Victim Protection Act (2014) The act states that the victim shall obtain redress and have an enforceable right to restitution; adequate compensation; and rehabilitation. In the event of death, victim’s dependents should receive reparation Per the court’s order or victim’s request, the convicted should pay for the costs of medical and psychological treatment; and any other relief that the court may consider just. A court may at any time, at the request of the victim or an intermediary, grant an order for the treatment or counselling of the victim. The expenses incurred for the treatment or professional counselling of a victim granted under this section shall be charged on the Victim Protection Trust Fund established under Section 27 of the Victim Protection Act. All medical records relating to treatment are the property of the victim and shall only be used as evidence before any court with regard to any offence under informed consent of the victim. Nepal The current law states that anyone who commits a crime of domestic violence can pay victim compensation of up to 200 000 rupees (∼$1835). The Supreme Court, in response to a petition, ordered the government to amend the law to ensure victims of acid attacks and burning receive immediate compensation and critical care. Practically, only a small amount of compensation is ever given to a victim for treatment and nothing at all before any final decision is made in the case. There is shortage of specialized burns units and plastic surgeons, especially outside of the capital but the director of the National Health Training is hopeful that a basic burns management course could be incorporated into the curriculum for training health practitioners. Nigeria Violence against Persons Prohibition Act (2015) The act mandates that the identities of victims of offences remain protected. Through a protection order, he act restrains an individual or state actors from further abusive behaviour towards a victim The protection order is accompanied by a warrant for the arrest of the respondent, and it remains valid throughout the period of the protection order. Pakistan The latest version of the law directed medical officers to start treating the victim as soon as he/she is brought to the hospital. In the past, medical officers used to wait for police before starting treatment. The parliament once again rejected the Acid Burn Crime Bill (2014) that mandates completion of the investigation within 60 days, treatment of victim without waiting for police go-ahead, and for the government to bear the cost of the treatment of patient as well as compensation to the victim. HC, high court; FIR, first information report; CVCF, Central Victim Compensation Fund; NGO, non-governmental organization. Open in new tab Surveillance, legal clarity, prevention, punishment and recompense are clearly important in controlling chemical assault. However, victims also need access to timely, free, safe and comprehensive burn care that exemplifies dignity, privacy and confidentiality. More than 90% of chemical assaults occur in low- and middle-income countries that have critical deficiencies in burn care capacity ( Calland et al., 2014; Gupta et al., 2014a,b). Therefore, more must be done to empower health systems to provide care for victims and incorporate burn care into capacity building efforts, task-sharing initiatives and national health plans. A number of low-cost opportunities and resources exist to improve the availability and quality of comprehensive burn care in low-resource settings, including sponsoring resource-appropriate training programmes, building on human and physical resource guidelines and using the WHO’s Community-Based Burns Surveillance Toolkit and Global Burn Registry for quality improvement purposes (Mock et al., 2004; Chamania, 2010; Ahuja et al., 2016; Young et al., 2017; WHO, 2018). Opportunely, quality burn care is created through the integration of pre-hospital, emergency, surgical, acute and critical care and rehabilitation. It involves a number of health professions (e.g. medics, physicians, surgeons, nurses, therapists, nutritionists, social workers, spiritual care providers). Therefore, health systems can leverage efforts to strengthen burn care to build horizontal capacity within emergency care, trauma and health systems broadly. Although this study provides insight into the global chemical assault situation and proposes a novel and comprehensive legislative framework for chemical assault control, several limitations must be considered. First, there are very few high-quality studies that describe chemical assault epidemiology and none are population based. Therefore, the description and estimates above should not be taken as true figures, but as minimum estimates of the problem solely to provide context and broad generalizations that are not context specific. Second, some of the countries that had reports of chemical assault did not have up-to-date or relevant legislation available through online databases or portals. Consequently, some legislative efforts may not have been captured. However, given the significant number of laws reviewed from countries of variable national income and legal development, as well as reaching saturation during content analysis, it is unlikely that novel laws, policies or regulations were missed. Third, the framework was built around existing legislative efforts and legal gaps with priority given to those that had a documented some effect. Since not all laws have undergone policy analysis or study after implementation, it is possible that a less common but particularly effective or detrimental law was overlooked. To mitigate this, the framework remained broad and all available laws that addressed chemical assaults were included in the tables for review. Despite these limitations, the findings and resultant framework may be used to spark further discussion within civil societies and governments regarding chemical assault control locally and globally. Conclusion Given their profound impact on individuals and often-marginalized groups within society, advocacy groups and governments must play active roles in the control of chemical assault. Using existing legislative efforts and legal gaps, the framework presented may be used to identify opportunities to better address chemical assault. Supplementary Data Supplementary data are available at Health Policy and Planning online. Acknowledgements No financial or material support was used for this project. This project was awarded the American Burn Association Peter Brigham Burn Epidemiology and Prevention Award in 2019. Conflict of interest statement. None declared. Ethical approval. No ethical approval was required for this study. References Acid Survivors Trust International (ASTI). 2018 . A Worldwide Problem . London : Acid Survivors Trust International . http://www.asti.org.uk/a-worldwide-problem.html, accessed 21 June 2020 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Ahmed F , Maroof H , Ahmed N , Sheridan R. 2017 . Acid attacks: a new public health pandemic ins the west? International Journal of Surgery 48 : 32 – 3 . Google Scholar Crossref Search ADS PubMed WorldCat Ahuja RB , Gibran N , Greenhalgh D et al. 2016 . ISBI practice guidelines for burn care . Burns 42 : 953 – 1021 . Google Scholar Crossref Search ADS PubMed WorldCat Alishahi Tabriz A , Dabbagh H , Koenig HG. 2016 . Medical ethics in Qiṣāṣ (eye-for-an-eye) punishment: an Islamic view; an examination of acid throwing . Journal of Religion and Health 55 : 1426 – 32 . Google Scholar Crossref Search ADS PubMed WorldCat Bagcchi S. 2015 a. Acid attack victims should have same rights as disabled people, Indian Supreme Court rules . BMJ 351 : h6787 . Google Scholar Crossref Search ADS PubMed WorldCat Bagcchi S. 2015 b. Private hospitals are told to treat acid attack victims free of charge . BMJ 350 : h2224 . Google Scholar Crossref Search ADS PubMed WorldCat Calland JF , Holland MC , Mwizerwa O et al. 2014 . Burn management in sub-Saharan Africa: opportunities for implementation of dedicated training and development of specialty centers . Burns 40 : 157 – 63 . Google Scholar Crossref Search ADS PubMed WorldCat Chamania S. 2010 . Training and burn care in rural India . Indian Journal of Plastic Surgery 43 : 126 – 30 . Google Scholar Crossref Search ADS WorldCat Dissanaike S , Rahimi M. 2009 . Epidemiology of burn injuries: highlighting cultural and socio-demographic aspects . International Review of Psychiatry 21 : 505 – 11 . Google Scholar Crossref Search ADS PubMed WorldCat Ellsberg M , Arango DJ , Morton M et al. 2015 . Prevention of violence against women and girls: what does the evidence say? Lancet 385 : 1555 – 66 . Google Scholar Crossref Search ADS PubMed WorldCat Farhad H , Naghibzadeh B , Nouhi AH , Rad HE. 2011 . Acid burn violence in Iran . Annals of Burns and Fire Disasters 24 : 138 – 40 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Garcia-Moreno C , Watts C. 2011 . Violence against women: an urgent public health priority . Bulletin of the World Health Organization 89 : 2 . Google Scholar Crossref Search ADS PubMed WorldCat Gupta S , Mahmood U , Gurung S et al. 2014 a. Burns in Nepal: a population based national assessment . Burns 41 : 1126 – 32 . Google Scholar Crossref Search ADS PubMed WorldCat Gupta S , Wong E , Mahmood U et al. 2014 b. Burn management capacity in low and middle-income countries: a systematic review of 458 hospitals across 14 countries . International Journal of Surgery12: 1070–3. Google Scholar OpenURL Placeholder Text WorldCat Jayatilleke AC , Jayatilleke AU , Yasuoka J , Jimba M. 2015 . Prevention of violence against women and girls at the community level . Lancet 386 : 427 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat Kay M. 2013 . Indian court restricts the sale of acid to try to curb attacks on women . BMJ 347 : f4762 . Google Scholar Crossref Search ADS PubMed WorldCat Kusuma YS , Babu BV. 2017 . Elimination of violence against women and girls as a global action agenda . Journal of Injury & Violence Research 9 : 117 – 121 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Mannan A , Ghani S , Clarke A , Butler PE. 2007 . Cases of chemical assault worldwide: a literature review . Burns 33 : 149 – 54 . Google Scholar Crossref Search ADS PubMed WorldCat Mock C , Lormand J , Goosen J , Joshipura M , Peden M. 2004 . Guidelines for Essential Trauma Care . Geneva : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Moher D , Liberati A , Tetzlaff J , Altman DG , Group P. ; The PRISMA Group. 2009 . Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement . PLoS Medicine 6 : e1000097 . Google Scholar Crossref Search ADS PubMed WorldCat Ning P , Schwebel DC , Huang H et al. 2016 . Global progress in road injury mortality since 2010 . PLoS One 11 : e0164560 . Google Scholar Crossref Search ADS PubMed WorldCat Peck M , Molnar J , Swart D. 2009 . A global plan for burn prevention and care . Bulletin of the World Health Organization 87 : 802 – 3 . Google Scholar Crossref Search ADS PubMed WorldCat Rasouli HR , Ebrahimi A , Motamedi MH. 2015 . Raising awareness against acid attacks . Lancet 385 : 772 – 3 . Google Scholar Crossref Search ADS PubMed WorldCat Sabzi Khoshnami M , Mohammadi E , Addelyan Rasi H , Khankeh HR , Arshi M. 2017 . Conceptual model of acid attacks based on survivor’s experiences: lessons from a qualitative exploration . Burns 43 : 608 – 618 . Google Scholar Crossref Search ADS PubMed WorldCat Sankari A , Atassi B , Sahloul MZ. 2013 . Syrian field hospitals: a creative solution in urban military conflict combat in Syria . Avicenna Journal of Medicine 3 : 84 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat Stergiou-Kita M , Grigorovich A. 2013 . Guidelines for vocational evaluation following burns: integrated review of relevant process and factors . Journal of Occupational Rehabilitation 23 : 476 – 503 . Google Scholar Crossref Search ADS PubMed WorldCat Sugrue R , Reilly F , Kelly J , Clover J. 2018 . The discordant relationship between acid attack incidence and advances in management . Burns 44 : 236 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat Swahn MH , Hankin A , Houry D. 2011 . Using policy to strengthen the reach and impact of injury prevention efforts . The Western Journal of Emergency Medicine 12 : 268 – 70 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Thomson Reuters Foundation, Acid Survivors Trust International, J. Sagar Associates. 2017 . Justice? What Justice?: Tackling Acid Violence and Ensuring Justice for Survivors. India: J. Sagar Associates, Acid Survivors Trust International and TrustLaw. Toroyan T , Iaych K , Altieri E et al. 2015 . Global Status Report on Road Safety 2015 . Geneva : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization (WHO). 2018 . Global Burn Registry . Geneva : World Health Organization . http://www.who.int/violence_injury_prevention/other_injury/burns/global_burns_registry/en/, accessed 21 June 2020 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Young AW , Graves C , Kowalske KJ et al. 2017 . Guideline for burn care under austere conditions: special care topics . Journal of Burn Care and Research 38 : e497 – 509 . OpenURL Placeholder Text WorldCat © The Author(s) 2020. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. All rights reserved. For permissions, please e-mail: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Ryan,, Nessa;Vieira,, Dorice;Goffman,, Dena;Bloch, Evan, M;Akaba, Godwin, O;D’mello, Brenda, S;Egekeze,, Chioma;Snyder,, Anya;Lyimo,, Magdalena;Nnodu,, Obiageli;Peprah,, Emmanuel

doi: 10.1093/heapol/czaa074pmid: 32995854

Abstract Globally, obstetric haemorrhage (OH) remains the leading cause of maternal mortality. Much of the associated mortality is ascribed to challenges surrounding deployment of innovations rather than lack of availability. In low- and middle-income countries (LMICs), where the burden is highest, there is a growing interest in implementation research as a means to bridge the ‘know–do’ gap between proven interventions and their reliable implementation at scale. In this systematic review, we identified and synthesized qualitative and quantitative data across the implementation outcomes of OH prevention innovations in LMICs using a taxonomy developed by Proctor et al. We also identified service outcomes for the included innovations, as well as implementation strategies and implementation facilitators and barriers. Eligible studies were empirical, focused on the implementation of OH prevention programmes or policies and occurred in an LMIC. Eight databases were searched. Two authors independently assessed studies for selection and extracted data; the first author resolved discrepancies. Narrative synthesis was used to analyse and interpret the findings. Studies were predominantly focused in Africa and on primary prevention. Interventions included prophylactic use of uterotonics (n = 7), clinical provider skills training (n = 4) and provision of clinical guidelines (n = 1); some (n = 3) were also part of a multi-component quality improvement bundle. Various barriers were reported, including challenges among intervention beneficiaries, providers and within the health system; however, studies reported the development and testing of practical implementation solutions. These included training and monitoring of implementers, community and stakeholder engagement and guidance by external mentors. Some studies linked successful delivery to implementation outcomes, most commonly adoption and acceptability, but also feasibility, penetration and sustainability. Findings suggest that innovations to prevent OH can be acceptable, appropriate and feasible in LMIC settings; however, more research is needed to better evaluate these and other under-reported implementation outcomes. Obstetrics, prevention, implementation, haemorrhage, intervention, LMIC KEY MESSAGES Greater focus on rigourous evaluation of programmes and policies for prevention (particularly secondary prevention) of obstetric haemorrhage (OH) in low- and middle-income countries (LMICs) is needed, particularly using implementation research methodology to adapt and scale up evidence-based interventions. Evidence gaps persist on how to develop appropriate policies for implementing effective programmes to prevent OH in LMICs in a context-specific manner. Addressing current knowledge gaps for implementing effective prevention strategies for OH in LMICs can decrease negative outcomes. Application of Proctor et al.’s implementation taxonomy provides a means to examine the know–do gap, guide reporting of implementation studies and establish implementation research priorities to improve health outcomes in LMICs. Introduction Obstetric haemorrhage (OH) remains the leading cause of preventable maternal mortality worldwide (Say et al., 2014), with an estimated 25% of all maternal deaths attributable to OH. Approximately 14 million cases occur annually, which equates to 5% of all women giving birth (Say et al., 2014). While OH refers to excess bleeding at any time during pregnancy and delivery, the majority of cases occurs in the postpartum period (Begley et al., 2019). Postpartum haemorrhage (PPH) is defined as a loss of blood of 500 ml or more within 24 h (known as primary PPH) following childbirth via normal vaginal delivery, or of 1000 ml or more for caesarean delivery (Evensen et al. 2017; WHO, 2018a). Secondary PPH can occur after the first day and up to 6 weeks after childbirth. In the absence of timely and effective intervention, PPH confers high risk of rapid exsanguination and death. Almost all (∼99%) of deaths due to OH occur in low- and middle-income countries (LMICs) (Mclintock and James, 2011), and the majority can be prevented. Most OH can be prevented via intervention throughout the various stages of prevention (i.e. primary, secondary or tertiary prevention). Primary prevention, including any intervention that is applied before there is evidence of disease or injury (i.e. OH), targets healthy individuals and is focused on limiting risk exposure and on reducing OH incidence. This includes antenatal care and active management of the third stage of labour (AMTSL) (Tanna, 2013). AMTSL includes prophylactic administration of one of the following uterotonics: oxytocin, carbetocin, misoprostol, ergometrine/methylergometrine or oxytocin and ergometrine fixed-dose combination during delivery; delivery of the placenta using controlled cord traction; and uterine massage. Secondary prevention, including any screening intervention that aims to identify disease before the onset of signs and symptoms, targets healthy-appearing individuals with subclinical forms of the disease and is focused on early disease detection. This includes tools for OH risk assessment and risk stratification, like what has been implemented in OH bundles in the USA (Goffman et al., 2019). Tertiary prevention, including any intervention that aims to manage disease post-diagnosis to slow or stop disease progression through measures such as screening for complications (i.e. quantifying blood loss during and after labour and delivery), targets symptomatic patients and is focused on both the clinical and outcome stages of a disease. While secondary OH prevention seeks to prevent the onset of OH, tertiary prevention seeks to reduce the severity of OH and its consequences and to improve quality of life for those living with its sequelae. Forms of tertiary prevention are commonly rehabilitation efforts. Different preventative efforts may be employed, depending on the setting and the availability of skilled birth attendants and supplies, and the World Health Organization (WHO) provides guidance on these evidence-based interventions. In 2012, the WHO published priority interventions for the prevention and management of OH (Sheldon et al., 2014), which were integrated into subsequent related recommendations for maternal and neonatal care (WHO, 2016, 2018b). After publication of a systematic review and network meta-analysis comparing all uterotonic options and placebo or no treatment (Gallos et al., 2019), the recommendations were updated in 2018 to incorporate the expanding evidence (WHO, 2018a) (Table 1). Administration of oxytocin remains the gold standard for the prevention of OH and the first line for treatment once bleeding has occurred. However, in settings where oxytocin is unavailable, administration of 600 μg of misoprostol given orally immediately after birth is a good alternative. As oxytocin is an injectable drug that requires a cold chain, misoprostol has a public health advantage as it can more easily be distributed by a health professional and even by a non-health professional outside a health facility. Misoprostol, however, should not be a replacement for skilled attendants or emergency obstetric care but rather should be a supplementary intervention for women served by health systems that are not yet able to make more comprehensive services widely accessible (Spangler et al., 2014). Despite some acknowledged overlap between the prevention and treatment of OH, prevention efforts, including programmes and policies, have greater potential to reduce the overall burden of OH in LMICs. Consistent implementation of these evidence-based prevention efforts has not yet been achieved in LMICs (Stanton et al., 2009). Table 1 Efficacy and safety of WHO-recommended uterotonics for OH prevention Uterotonic name . Dose/route . Delivery mode . Who administers? . Oxytocin 10 IU, IM/IV Caesarean section (consider dividing doses and avoid a rapid IV bolus); vaginal birth Skilled health personnel Carbetocin 100 µg, IM/IV Caesarean section; vaginal births (in contexts where its cost is comparable to other effective uterotonics) Skilled health personnel Misoprostol Either 400 or 600 µg PO (no clear evidence of which dose is superior, but higher doses have more side effects. Must inform women of side effects) Caesarean section (alternative routes may be needed at caesarean section, but oral route is preferred by women); vaginal births In hospital or community Ergometrine/methylergometrine 200 µg, IM/IV (in contexts where hypertensive disorders can be safely excluded prior to its use. Must inform women of side effects) Caesarean section; vaginal births Skilled health personnel Oxytocin and ergometrine fixed-dose combination 5 IU/500 µg IM (in contexts where hypertensive disorders can be safely excluded prior to its use) Caesarean section; vaginal births Skilled health personnel Injectable prostaglandins (carboprost or sulprostone) not recommended for the prevention of OH Uterotonic name . Dose/route . Delivery mode . Who administers? . Oxytocin 10 IU, IM/IV Caesarean section (consider dividing doses and avoid a rapid IV bolus); vaginal birth Skilled health personnel Carbetocin 100 µg, IM/IV Caesarean section; vaginal births (in contexts where its cost is comparable to other effective uterotonics) Skilled health personnel Misoprostol Either 400 or 600 µg PO (no clear evidence of which dose is superior, but higher doses have more side effects. Must inform women of side effects) Caesarean section (alternative routes may be needed at caesarean section, but oral route is preferred by women); vaginal births In hospital or community Ergometrine/methylergometrine 200 µg, IM/IV (in contexts where hypertensive disorders can be safely excluded prior to its use. Must inform women of side effects) Caesarean section; vaginal births Skilled health personnel Oxytocin and ergometrine fixed-dose combination 5 IU/500 µg IM (in contexts where hypertensive disorders can be safely excluded prior to its use) Caesarean section; vaginal births Skilled health personnel Injectable prostaglandins (carboprost or sulprostone) not recommended for the prevention of OH IM: intramuscular; IV: intravenous; PO: oral. Open in new tab Table 1 Efficacy and safety of WHO-recommended uterotonics for OH prevention Uterotonic name . Dose/route . Delivery mode . Who administers? . Oxytocin 10 IU, IM/IV Caesarean section (consider dividing doses and avoid a rapid IV bolus); vaginal birth Skilled health personnel Carbetocin 100 µg, IM/IV Caesarean section; vaginal births (in contexts where its cost is comparable to other effective uterotonics) Skilled health personnel Misoprostol Either 400 or 600 µg PO (no clear evidence of which dose is superior, but higher doses have more side effects. Must inform women of side effects) Caesarean section (alternative routes may be needed at caesarean section, but oral route is preferred by women); vaginal births In hospital or community Ergometrine/methylergometrine 200 µg, IM/IV (in contexts where hypertensive disorders can be safely excluded prior to its use. Must inform women of side effects) Caesarean section; vaginal births Skilled health personnel Oxytocin and ergometrine fixed-dose combination 5 IU/500 µg IM (in contexts where hypertensive disorders can be safely excluded prior to its use) Caesarean section; vaginal births Skilled health personnel Injectable prostaglandins (carboprost or sulprostone) not recommended for the prevention of OH Uterotonic name . Dose/route . Delivery mode . Who administers? . Oxytocin 10 IU, IM/IV Caesarean section (consider dividing doses and avoid a rapid IV bolus); vaginal birth Skilled health personnel Carbetocin 100 µg, IM/IV Caesarean section; vaginal births (in contexts where its cost is comparable to other effective uterotonics) Skilled health personnel Misoprostol Either 400 or 600 µg PO (no clear evidence of which dose is superior, but higher doses have more side effects. Must inform women of side effects) Caesarean section (alternative routes may be needed at caesarean section, but oral route is preferred by women); vaginal births In hospital or community Ergometrine/methylergometrine 200 µg, IM/IV (in contexts where hypertensive disorders can be safely excluded prior to its use. Must inform women of side effects) Caesarean section; vaginal births Skilled health personnel Oxytocin and ergometrine fixed-dose combination 5 IU/500 µg IM (in contexts where hypertensive disorders can be safely excluded prior to its use) Caesarean section; vaginal births Skilled health personnel Injectable prostaglandins (carboprost or sulprostone) not recommended for the prevention of OH IM: intramuscular; IV: intravenous; PO: oral. Open in new tab Effective implementation of evidence-based interventions for OH prevention depends on context-specific factors at the patient, provider and health system levels, and implementation research provides us with the tools to study these factors. Pertinent implementation barriers may include inadequate training or skills mix of providers, drug and therapeutics procurement or logistics problems (including poor storage of heat labile oxytocin, lack of blood and blood products), staff shortages, lack of equipment and low staff motivation (due in part to understaffing and overburdening of skilled personnel spanning midwives to obstetricians) (Bates et al., 2008; Knight et al., 2013; Prata et al., 2013). Implementation research allows us to conceptualize and evaluate successful implementation of programmes and policies, particularly via assessment of implementation outcomes. Implementation outcomes defined as the results of deliberate, purposive actions to implement new treatments, practices and services (Proctor et al., 2011). In the popular taxonomy developed by Proctor et al., implementation outcomes are differentiated from service or patient outcomes. As a poorly implemented innovation, which Proctor uses to refer broadly to evidence-based practices, will not be clinically effective, implementation outcomes are needed for desired impact in clinical or service outcomes. Identifying and synthesizing implementation outcomes, including acceptability, adoption, appropriateness, feasibility, fidelity, implementation cost, penetration and sustainability (defined in Table 2), will ultimately improve our understanding of implementation processes and allow comparison of the effectiveness of different implementation strategies. Feasibility, acceptability and appropriateness, more commonly assessed in the pre-implementation period, refer to the degree to which an innovation can be used, to which stakeholders find it agreeable and to which it is perceived to be relevant to address the issue or fit with a setting, provider or beneficiary, respectively. Adoption refers to the intention, initial decision or enacted employment of the innovation, and the degree to which the innovation can be adopted as intended describes its fidelity. Cost, which can be incremental or implementation cost, is defined as the cost impact of implementing the innovation. Penetration is the integration of the innovation within a given setting, and the extent to which the adopted innovation is maintained within a setting as part of ongoing operations describes its sustainability. Table 2 Definitions of implementation research outcomes Outcome . Definition . Acceptability The perception among implementation stakeholders (beneficiaries and implementers) that the innovation is agreeable, palatable or satisfactory Adoption The intention, initial decision or action to try or employ the innovation (i.e. uptake) Appropriateness The perceived fit, relevance or compatibility of the innovation for a given practice setting, provider or beneficiary; and/or perceived fit of the innovation to address a particular issue or problem (OH prevention). Cost (Incremental or implementation cost) is defined as the cost impact of an implementation effort Feasibility The extent to which the innovation can be successfully used or carried out within a given agency or setting Fidelity Degree to which the innovation can be implemented as it was prescribed in the original protocol or as it was intended by the programme developer Penetration The integration of a practice within a service setting and its subsystems Sustainability The extent to which a newly implemented innovation is maintained or institutionalized within a service setting’s ongoing, stable operations Outcome . Definition . Acceptability The perception among implementation stakeholders (beneficiaries and implementers) that the innovation is agreeable, palatable or satisfactory Adoption The intention, initial decision or action to try or employ the innovation (i.e. uptake) Appropriateness The perceived fit, relevance or compatibility of the innovation for a given practice setting, provider or beneficiary; and/or perceived fit of the innovation to address a particular issue or problem (OH prevention). Cost (Incremental or implementation cost) is defined as the cost impact of an implementation effort Feasibility The extent to which the innovation can be successfully used or carried out within a given agency or setting Fidelity Degree to which the innovation can be implemented as it was prescribed in the original protocol or as it was intended by the programme developer Penetration The integration of a practice within a service setting and its subsystems Sustainability The extent to which a newly implemented innovation is maintained or institutionalized within a service setting’s ongoing, stable operations Based on types of outcomes in implementation research reported within Proctor et al.’s (2011) taxonomy for implementation outcomes. Open in new tab Table 2 Definitions of implementation research outcomes Outcome . Definition . Acceptability The perception among implementation stakeholders (beneficiaries and implementers) that the innovation is agreeable, palatable or satisfactory Adoption The intention, initial decision or action to try or employ the innovation (i.e. uptake) Appropriateness The perceived fit, relevance or compatibility of the innovation for a given practice setting, provider or beneficiary; and/or perceived fit of the innovation to address a particular issue or problem (OH prevention). Cost (Incremental or implementation cost) is defined as the cost impact of an implementation effort Feasibility The extent to which the innovation can be successfully used or carried out within a given agency or setting Fidelity Degree to which the innovation can be implemented as it was prescribed in the original protocol or as it was intended by the programme developer Penetration The integration of a practice within a service setting and its subsystems Sustainability The extent to which a newly implemented innovation is maintained or institutionalized within a service setting’s ongoing, stable operations Outcome . Definition . Acceptability The perception among implementation stakeholders (beneficiaries and implementers) that the innovation is agreeable, palatable or satisfactory Adoption The intention, initial decision or action to try or employ the innovation (i.e. uptake) Appropriateness The perceived fit, relevance or compatibility of the innovation for a given practice setting, provider or beneficiary; and/or perceived fit of the innovation to address a particular issue or problem (OH prevention). Cost (Incremental or implementation cost) is defined as the cost impact of an implementation effort Feasibility The extent to which the innovation can be successfully used or carried out within a given agency or setting Fidelity Degree to which the innovation can be implemented as it was prescribed in the original protocol or as it was intended by the programme developer Penetration The integration of a practice within a service setting and its subsystems Sustainability The extent to which a newly implemented innovation is maintained or institutionalized within a service setting’s ongoing, stable operations Based on types of outcomes in implementation research reported within Proctor et al.’s (2011) taxonomy for implementation outcomes. Open in new tab Some recent reviews have systematically examined implementation outcomes for other innovations of global health significance (Kemp et al., 2019; Verhey et al., 2020); however, reviews of maternal health innovations, including those targeting OH, have lacked a comprehensive (Gaucher et al., 2019; Tiruneh et al., 2019) or systematic focus on implementation (Smith et al., 2016). Various systematic reviews have established the effectiveness of innovations for the prevention of OH in high-income countries and in LMICs, but none have synthesized the evidence on implementation for these programmes and policies. In addition, none have utilized Proctor et al.’s taxonomy to conceptualize all implementation outcomes (Figure 1). This conceptual framing shows how it is helpful to understand service outcomes in addition to implementation outcomes when examining diverse innovations for the prevention of OH, which can range from medications to guideline dissemination to training and could operate through different implementation pathways. Therefore, the primary objective of this review was to identify and synthesize the implementation outcomes reported within studies of programmes and policies regarding the primary, secondary or tertiary prevention of OH in LMICs. We wanted to report the quantitative and qualitative data for these implementation outcomes, to better understand how OH prevention and the implementation of relevant innovations are presented in the literature. Our secondary objectives were to report service outcomes for included innovations, assess the successes and challenges when implementing included innovations and identify the reported implementation strategies for these innovations. We highlight gaps in the research that can inform future implementation research aimed at addressing OH and provide opportunities to integrate clinicians and policymakers into the programme and policy development and evaluation process. Figure 1 Open in new tabDownload slide Conceptual framework for the review of implementation outcomes for innovations to prevent OH in LMICs. Adapted from Chaudoir et al. (2013) and Proctor et al. (2011). Figure 1 Open in new tabDownload slide Conceptual framework for the review of implementation outcomes for innovations to prevent OH in LMICs. Adapted from Chaudoir et al. (2013) and Proctor et al. (2011). Methods Search strategy We searched eight databases for studies assessing implementation outcomes for OH prevention intervention in LMICs. Per Cochrane Handbook recommendations, PubMed/Medline, Embase and the Cochrane database (including Central) were searched as the mandatory databases. To be comprehensive and minimize selection bias (Lefebvre et al., 2019), we also searched Web of Science Plus, including Biological Abstracts and Biosis Citation Index, Global Health (Ovid), PsycINFO (Ovid) and the Cumulative Index to Nursing and Allied Health. With each concept in our search strategy, we included the Medical Subject Headings terms relative to our study as outlined in the National Library of Medicine’s database. A sample search strategy is included in Appendix 1. Prior to study implementation, we developed a review protocol, which was registered on PROSPERO (#158196). Resulting articles were imported into Endnote v.8 and de-duplicated. The Endnote library was then imported into Covidence, an online systematic review management software. Selection criteria We used the PICO (population, intervention, comparison, outcome) format to identify a population (individuals in LMICs, as per the World Bank classification during the reported period of date collection), intervention (any intervention that the authors identified as targeting OH prevention), and outcome (one or more of the eight implementation outcomes defined in Table 2 and assessed either quantitatively or qualitatively). There was no comparison group identified. We included any empirical peer-reviewed articles and reports published in English. We conducted the search in October 2019 and did not limit by publication date, as we did not have justification to do so. We excluded review papers and any reports that did not clarify their evaluation methods. Conference presentations were used to search for subsequent peer-reviewed publications but were not themselves included. Screening, quality assessment and data extraction All screening occurred in Covidence. Screening of titles/abstracts and screening of full text were conducted independently (by NR and CE) and then discussed between reviewers at each stage. The first author ultimately resolved discrepant or discordant results. We reached out to request a full-text article from study investigators of one included article. Data extraction was carried out by two reviewers independently with a structured form. To enable identification and synthesis of reported implementation outcomes, we used the taxonomy for implementation research outcomes devised by Proctor et al. (2011), which included acceptability, appropriateness, feasibility, adoption, fidelity, cost, penetration and sustainability (defined in Table 1). Both quantitative and qualitative outcomes related to the examination of implementation were extracted. Proctor’s taxonomy, which has been utilized in various settings and disciplines, enables global assessment of implementation effectiveness. Service outcomes, as identified by Proctor et al. (2011) and defined within an Institute of Medicine report as the six aims of quality improvement [Institute of Medicine (US) Committee on Quality of Health Care in America, 2001], were also assessed, including efficiency, safety, effectiveness, equity, patient-centredness and timeliness (defined in Table 3). These were included as we estimated that our search would focus on various types of innovations with the shared target of prevention of OH, and, as intervention characteristics might be variable, an understanding of service outcomes could help us to understand the implementation process. Both quantitative and qualitative outcomes related to evaluation of the service system were extracted. Table 3 Definitions of service outcomes Outcome . Definition . Efficiency The avoidance of waste, including waste of equipment, supplies, ideas and energy Safety The avoidance of injuries to patients from the care that is intended to help them Effectiveness Provision of services based on scientific knowledge to all who could benefit and refraining from providing services to those not likely to benefit (avoiding underuse and overuse, respectively) Equity Provision of care that does not vary in quality because of personal characteristics such as gender, ethnicity, geographic location and socioeconomic status Patient-centredness Provision of care that is respectful of and responsive to individual patient preferences, needs and values and ensuring that patient values guide all clinical decisions Timeliness Reduction of waits and sometimes harmful delays for both those who receive and those who give care Outcome . Definition . Efficiency The avoidance of waste, including waste of equipment, supplies, ideas and energy Safety The avoidance of injuries to patients from the care that is intended to help them Effectiveness Provision of services based on scientific knowledge to all who could benefit and refraining from providing services to those not likely to benefit (avoiding underuse and overuse, respectively) Equity Provision of care that does not vary in quality because of personal characteristics such as gender, ethnicity, geographic location and socioeconomic status Patient-centredness Provision of care that is respectful of and responsive to individual patient preferences, needs and values and ensuring that patient values guide all clinical decisions Timeliness Reduction of waits and sometimes harmful delays for both those who receive and those who give care Based on types of outcomes in implementation research reported within Proctor et al.’s (2011) taxonomy for implementation outcomes and informed by Institute of Medicine ‘Crossing the Quality Chasm: A New Health System for the 21st Century’ [Institute of Medicine (US) Committee on Quality of Health Care in America, 2001]. Open in new tab Table 3 Definitions of service outcomes Outcome . Definition . Efficiency The avoidance of waste, including waste of equipment, supplies, ideas and energy Safety The avoidance of injuries to patients from the care that is intended to help them Effectiveness Provision of services based on scientific knowledge to all who could benefit and refraining from providing services to those not likely to benefit (avoiding underuse and overuse, respectively) Equity Provision of care that does not vary in quality because of personal characteristics such as gender, ethnicity, geographic location and socioeconomic status Patient-centredness Provision of care that is respectful of and responsive to individual patient preferences, needs and values and ensuring that patient values guide all clinical decisions Timeliness Reduction of waits and sometimes harmful delays for both those who receive and those who give care Outcome . Definition . Efficiency The avoidance of waste, including waste of equipment, supplies, ideas and energy Safety The avoidance of injuries to patients from the care that is intended to help them Effectiveness Provision of services based on scientific knowledge to all who could benefit and refraining from providing services to those not likely to benefit (avoiding underuse and overuse, respectively) Equity Provision of care that does not vary in quality because of personal characteristics such as gender, ethnicity, geographic location and socioeconomic status Patient-centredness Provision of care that is respectful of and responsive to individual patient preferences, needs and values and ensuring that patient values guide all clinical decisions Timeliness Reduction of waits and sometimes harmful delays for both those who receive and those who give care Based on types of outcomes in implementation research reported within Proctor et al.’s (2011) taxonomy for implementation outcomes and informed by Institute of Medicine ‘Crossing the Quality Chasm: A New Health System for the 21st Century’ [Institute of Medicine (US) Committee on Quality of Health Care in America, 2001]. Open in new tab Two reviewers also performed an independent assessment of quality using established checklists appropriate for the various included study designs, including cohorts, randomized control trials and qualitative assessments (Critical Appraisal Skills Programme, 2020a,b,c). Items from these checklists were integrated into the structured extraction form. The methodological quality of each study was assessed to determine the extent to which included studies had addressed the possibility of bias in their design, conduct and analysis. Reviewers presented and discussed this assessment and, as a team, came to consensus. Furthermore, the PRISMA checklist (Moher et al., 2009) was completed for this study, to establish minimum information that should be included when reviewing and reporting. Synthesis Narrative synthesis was used to analyse and interpret the findings, in which quantitative and qualitative studies are integrated. Using text and tables, we summarized descriptive information from included studies. Findings from the quantitative studies were narrated thematically, followed by findings of qualitative studies. The general characteristics of each study were summarized, and each was classified according to the innovation type (i.e. uterotonics, trainings, guidelines, bundles), prevention type (i.e. primary, secondary, tertiary) and then reported implementation outcomes. Reviewers discussed the generated themes, framing the discussion around potential research gaps within the literature and areas requiring more in-depth analysis. Lastly, we conducted multidisciplinary expert debriefing as a qualitative strategy for rigour, in which the initial findings of this review were shared for feedback among experts in the fields of Obstetrics and Gynecology, Hematology and Implementation Science. Their feedback included expansion of the description of various innovations and confirmation that service outcomes and implementation strategies should also be identified within results. Results An initial search of databases identified 90 articles; 71 remained after duplicates were removed. Appraisal of titles/abstracts excluded 45 articles. Twenty-six full-text articles were assessed for eligibility; ultimately, 12 were found to be relevant to the review topic. This process and the reasons for exclusion at each stage are presented in Figure 2. Study publication year ranged from 2013 to 2019. Among studies pertaining to the implementation of programmes and policies of OH prevention in LMICs, the types of interventions that were studied included prophylactic use of uterotonics (n = 7) (Tables 4 and 5), clinical provider skills training (n = 4) (Table 6) and provision of clinical guidelines (n = 1) (Table 7). In addition, we highlight among these interventions those that were also part of a multi-component quality improvement bundle (n = 3) (Table 8). The reviewed studies were conducted in various LMICs in Africa (e.g. Ethiopia, Niger, Nigeria, Malawi, Mali, Uganda) (n = 10), Asia (e.g. India) (n = 1) and South America (e.g. Nicaragua) (n = 1). Included studies reported on implementation outcomes within cohort designs (n = 4), randomized control trials or pre-/post-design trials (n = 3) and qualitative (n = 4) or mixed methods (n = 1) assessment. While experimental and quasi-experimental studies did provide adequate information about sampling and data collection for the primary aims of the studies, implementation outcomes frequently were not the main outcomes of interest and were reported as secondary outcomes but not interpreted from an implementation lens. For qualitative assessments, the design appropriately matched the study aims, recruitment and data collection were appropriate and the study results were valid; however, critical reflection of the researcher regarding their qualifications, interests, perspectives and relationship with the participants was often not reported adequately or at all, and thus, reporting bias was likely. Figure 2 Open in new tabDownload slide PRISMA flow diagram. Figure 2 Open in new tabDownload slide PRISMA flow diagram. Table 4 Service and implementation outcomes for interventions to prevent OH: uterotonics (misoprostol) Article . LMIC . Intervention . Implementation . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Uterotonics—misoprostol Cannon et al. (2017) Nigeria Distribution of misoprostol Effectiveness (Qual) Less bleeding than previous deliveries that used non-effective + non-medical remedies Advocacy Stakeholder engagement Community engagement During Post Shortage of CBHVs + drug keepers Sociocultural barriers among women Acceptability Adoption Appropriateness (All qual) Intervention received well (acceptability, appropriateness) Added drug to procurement list of maternal health facilities (adoption) Orobaton et al. (2017) Nigeria Community-based distribution of misoprostol (to mother–newborn dyads) Effectiveness Safety (All quant) Women reported that heavy bleeding persisted after misoprostol use + were promptly referred434 women (1.25%) who ingested misoprostol reported associated side effects During Post Feasibility Adoption Penetration (All quant) Successfully introduced (adoption) Reached mothers in labour in all wards (penetration) Established community-data collection systems to record maternal deaths (feasibility) Spangler et al. (2014) Ethiopia National policy for community-based distribution (by CHEWs, CHDAs, TBAs) Effectiveness Safety (All qual) High effectiveness in preventing OH in the community perceived by officials, but concerns about safety (mishandling by untrained persons) Distribution + administration by district health offices in line with regional regulations During Reconciling priorities involving facility-based services (quality, quantity, dem+) with widespread implementation of a more targeted, community-based intervention Acceptability Appropriateness Adoption (All qual) Officials in Amhara tended to adopt a strict interpretation that reflected fear of misuse + a deep concern for encouraging home birth (appropriateness, acceptability), thus deviating from the clear national goal to increase facility-based birth Officials in Oromiya framed policy in terms of the broader national goal to reduce maternal mortality (appropriateness, acceptability), which allowed adoption of multiple means of misoprostol distribution (adoption) Weeks et al. (2013) Uganda Misoprostol self-administered after homebirth (MamaMiso) Safety (Quant) No adverse effects among women who took med antenatally No significant difference between the study groups in the drop of maternal haemoglobin by >20% (misoprostol 9.4% vs placebo 7.5%, risk ratio 1.11, 95% CI 0.717–1.719) Significantly more fever (27% vs 14%) + shivering (51% vs 28%) in misoprostol group than placebo group During Compliance of women to adopt intervention at home Inadequate training of home-based intervention leading to complications otherwise not seen pre-intervention Acceptability Adoption (All quant) Satisfaction above 97% (acceptability) Participant compliance above 97% (adoption) Article . LMIC . Intervention . Implementation . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Uterotonics—misoprostol Cannon et al. (2017) Nigeria Distribution of misoprostol Effectiveness (Qual) Less bleeding than previous deliveries that used non-effective + non-medical remedies Advocacy Stakeholder engagement Community engagement During Post Shortage of CBHVs + drug keepers Sociocultural barriers among women Acceptability Adoption Appropriateness (All qual) Intervention received well (acceptability, appropriateness) Added drug to procurement list of maternal health facilities (adoption) Orobaton et al. (2017) Nigeria Community-based distribution of misoprostol (to mother–newborn dyads) Effectiveness Safety (All quant) Women reported that heavy bleeding persisted after misoprostol use + were promptly referred434 women (1.25%) who ingested misoprostol reported associated side effects During Post Feasibility Adoption Penetration (All quant) Successfully introduced (adoption) Reached mothers in labour in all wards (penetration) Established community-data collection systems to record maternal deaths (feasibility) Spangler et al. (2014) Ethiopia National policy for community-based distribution (by CHEWs, CHDAs, TBAs) Effectiveness Safety (All qual) High effectiveness in preventing OH in the community perceived by officials, but concerns about safety (mishandling by untrained persons) Distribution + administration by district health offices in line with regional regulations During Reconciling priorities involving facility-based services (quality, quantity, dem+) with widespread implementation of a more targeted, community-based intervention Acceptability Appropriateness Adoption (All qual) Officials in Amhara tended to adopt a strict interpretation that reflected fear of misuse + a deep concern for encouraging home birth (appropriateness, acceptability), thus deviating from the clear national goal to increase facility-based birth Officials in Oromiya framed policy in terms of the broader national goal to reduce maternal mortality (appropriateness, acceptability), which allowed adoption of multiple means of misoprostol distribution (adoption) Weeks et al. (2013) Uganda Misoprostol self-administered after homebirth (MamaMiso) Safety (Quant) No adverse effects among women who took med antenatally No significant difference between the study groups in the drop of maternal haemoglobin by >20% (misoprostol 9.4% vs placebo 7.5%, risk ratio 1.11, 95% CI 0.717–1.719) Significantly more fever (27% vs 14%) + shivering (51% vs 28%) in misoprostol group than placebo group During Compliance of women to adopt intervention at home Inadequate training of home-based intervention leading to complications otherwise not seen pre-intervention Acceptability Adoption (All quant) Satisfaction above 97% (acceptability) Participant compliance above 97% (adoption) CBHV: community-based health volunteer; CHDAs: community health development agents; CHEW: community health extension worker; TBA: traditional birth attendant. Open in new tab Table 4 Service and implementation outcomes for interventions to prevent OH: uterotonics (misoprostol) Article . LMIC . Intervention . Implementation . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Uterotonics—misoprostol Cannon et al. (2017) Nigeria Distribution of misoprostol Effectiveness (Qual) Less bleeding than previous deliveries that used non-effective + non-medical remedies Advocacy Stakeholder engagement Community engagement During Post Shortage of CBHVs + drug keepers Sociocultural barriers among women Acceptability Adoption Appropriateness (All qual) Intervention received well (acceptability, appropriateness) Added drug to procurement list of maternal health facilities (adoption) Orobaton et al. (2017) Nigeria Community-based distribution of misoprostol (to mother–newborn dyads) Effectiveness Safety (All quant) Women reported that heavy bleeding persisted after misoprostol use + were promptly referred434 women (1.25%) who ingested misoprostol reported associated side effects During Post Feasibility Adoption Penetration (All quant) Successfully introduced (adoption) Reached mothers in labour in all wards (penetration) Established community-data collection systems to record maternal deaths (feasibility) Spangler et al. (2014) Ethiopia National policy for community-based distribution (by CHEWs, CHDAs, TBAs) Effectiveness Safety (All qual) High effectiveness in preventing OH in the community perceived by officials, but concerns about safety (mishandling by untrained persons) Distribution + administration by district health offices in line with regional regulations During Reconciling priorities involving facility-based services (quality, quantity, dem+) with widespread implementation of a more targeted, community-based intervention Acceptability Appropriateness Adoption (All qual) Officials in Amhara tended to adopt a strict interpretation that reflected fear of misuse + a deep concern for encouraging home birth (appropriateness, acceptability), thus deviating from the clear national goal to increase facility-based birth Officials in Oromiya framed policy in terms of the broader national goal to reduce maternal mortality (appropriateness, acceptability), which allowed adoption of multiple means of misoprostol distribution (adoption) Weeks et al. (2013) Uganda Misoprostol self-administered after homebirth (MamaMiso) Safety (Quant) No adverse effects among women who took med antenatally No significant difference between the study groups in the drop of maternal haemoglobin by >20% (misoprostol 9.4% vs placebo 7.5%, risk ratio 1.11, 95% CI 0.717–1.719) Significantly more fever (27% vs 14%) + shivering (51% vs 28%) in misoprostol group than placebo group During Compliance of women to adopt intervention at home Inadequate training of home-based intervention leading to complications otherwise not seen pre-intervention Acceptability Adoption (All quant) Satisfaction above 97% (acceptability) Participant compliance above 97% (adoption) Article . LMIC . Intervention . Implementation . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Uterotonics—misoprostol Cannon et al. (2017) Nigeria Distribution of misoprostol Effectiveness (Qual) Less bleeding than previous deliveries that used non-effective + non-medical remedies Advocacy Stakeholder engagement Community engagement During Post Shortage of CBHVs + drug keepers Sociocultural barriers among women Acceptability Adoption Appropriateness (All qual) Intervention received well (acceptability, appropriateness) Added drug to procurement list of maternal health facilities (adoption) Orobaton et al. (2017) Nigeria Community-based distribution of misoprostol (to mother–newborn dyads) Effectiveness Safety (All quant) Women reported that heavy bleeding persisted after misoprostol use + were promptly referred434 women (1.25%) who ingested misoprostol reported associated side effects During Post Feasibility Adoption Penetration (All quant) Successfully introduced (adoption) Reached mothers in labour in all wards (penetration) Established community-data collection systems to record maternal deaths (feasibility) Spangler et al. (2014) Ethiopia National policy for community-based distribution (by CHEWs, CHDAs, TBAs) Effectiveness Safety (All qual) High effectiveness in preventing OH in the community perceived by officials, but concerns about safety (mishandling by untrained persons) Distribution + administration by district health offices in line with regional regulations During Reconciling priorities involving facility-based services (quality, quantity, dem+) with widespread implementation of a more targeted, community-based intervention Acceptability Appropriateness Adoption (All qual) Officials in Amhara tended to adopt a strict interpretation that reflected fear of misuse + a deep concern for encouraging home birth (appropriateness, acceptability), thus deviating from the clear national goal to increase facility-based birth Officials in Oromiya framed policy in terms of the broader national goal to reduce maternal mortality (appropriateness, acceptability), which allowed adoption of multiple means of misoprostol distribution (adoption) Weeks et al. (2013) Uganda Misoprostol self-administered after homebirth (MamaMiso) Safety (Quant) No adverse effects among women who took med antenatally No significant difference between the study groups in the drop of maternal haemoglobin by >20% (misoprostol 9.4% vs placebo 7.5%, risk ratio 1.11, 95% CI 0.717–1.719) Significantly more fever (27% vs 14%) + shivering (51% vs 28%) in misoprostol group than placebo group During Compliance of women to adopt intervention at home Inadequate training of home-based intervention leading to complications otherwise not seen pre-intervention Acceptability Adoption (All quant) Satisfaction above 97% (acceptability) Participant compliance above 97% (adoption) CBHV: community-based health volunteer; CHDAs: community health development agents; CHEW: community health extension worker; TBA: traditional birth attendant. Open in new tab Table 5 Service and implementation outcomes for interventions to prevent OH: uterotonics (oxytocin) Article . LMIC . Intervention . Implementation . Evidence grade (FAME)a . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Uterotonics—oxytocin Boucar et al. (2014)b Niger, Mali AMTSL (including use of oxytocin) among SBAs Effectiveness (Quant) PPH rate decreased by 90% from 2.2% to 0.2% of vaginal deliveries (Niger) Mean PPH rate decreased from 0.9% to <0.3% (Mali) Continuous shared learning across multiple facilities Engaging teams of healthcare providers, managers + staff in improving care Analysing how processes of care function within a system Changing processes + systems of care to implement best practices Continuous analysis of data to assess results During Provider buy-in to adherence of PPH best practices Sustaining intervention after cessation of study Adoption (Quant) % of births receiving AMTSL (of which oxytocin administration is one component) increased from 2% at baseline to >90% (Niger) + from 17% in the first month of reporting to 95–97% (Mali) AE Garcia-Elorrio et al. (2014) Nicaragua Prophylactic use of the oxytocin in Uniject (OiU) a timetemperature indicator among SBAs for AMTSL Effectiveness (Quant) 2% increase in the use of AMTSL Rate of correct oxytocin use was significantly increased, reaching 97.4% (P = 0.003) Intervention developed onsite and by staff Providers engaged and motivated to implement Prior During Use of opinion leaders, interactive workshops to develop guidelinesAcademic detailingUse of remindersFeedback on the rates of oxytocin use Adoption Acceptability (All quant) 86/104 (82.7%) reported that the use of oxytocin in a preloaded device was easier for administration of oxytocin than the traditional method (acceptability) 77 (74.0%) reported that OiU was preferred to standard oxytocin administration (acceptability) 84 (80.8%) reported that they would not stop using prophylactic oxytocin whether or not OiU was available Researchers noted 95% intervention uptake (adoption) AE Moucheraud et al. (2015) India Advance distribution of oxytocin Effectiveness Timeliness (All quant) Preparing injection prior to delivery + making it available at bedside before birth was associated with increased likelihood of its rapid postpartum use in a hospital During Adoption (Quant) Having the intervention on hand increased likelihood of uptake (adoption) Oxytocin was prepared + available at bedside for 38.8% of deliveries AE Article . LMIC . Intervention . Implementation . Evidence grade (FAME)a . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Uterotonics—oxytocin Boucar et al. (2014)b Niger, Mali AMTSL (including use of oxytocin) among SBAs Effectiveness (Quant) PPH rate decreased by 90% from 2.2% to 0.2% of vaginal deliveries (Niger) Mean PPH rate decreased from 0.9% to <0.3% (Mali) Continuous shared learning across multiple facilities Engaging teams of healthcare providers, managers + staff in improving care Analysing how processes of care function within a system Changing processes + systems of care to implement best practices Continuous analysis of data to assess results During Provider buy-in to adherence of PPH best practices Sustaining intervention after cessation of study Adoption (Quant) % of births receiving AMTSL (of which oxytocin administration is one component) increased from 2% at baseline to >90% (Niger) + from 17% in the first month of reporting to 95–97% (Mali) AE Garcia-Elorrio et al. (2014) Nicaragua Prophylactic use of the oxytocin in Uniject (OiU) a timetemperature indicator among SBAs for AMTSL Effectiveness (Quant) 2% increase in the use of AMTSL Rate of correct oxytocin use was significantly increased, reaching 97.4% (P = 0.003) Intervention developed onsite and by staff Providers engaged and motivated to implement Prior During Use of opinion leaders, interactive workshops to develop guidelinesAcademic detailingUse of remindersFeedback on the rates of oxytocin use Adoption Acceptability (All quant) 86/104 (82.7%) reported that the use of oxytocin in a preloaded device was easier for administration of oxytocin than the traditional method (acceptability) 77 (74.0%) reported that OiU was preferred to standard oxytocin administration (acceptability) 84 (80.8%) reported that they would not stop using prophylactic oxytocin whether or not OiU was available Researchers noted 95% intervention uptake (adoption) AE Moucheraud et al. (2015) India Advance distribution of oxytocin Effectiveness Timeliness (All quant) Preparing injection prior to delivery + making it available at bedside before birth was associated with increased likelihood of its rapid postpartum use in a hospital During Adoption (Quant) Having the intervention on hand increased likelihood of uptake (adoption) Oxytocin was prepared + available at bedside for 38.8% of deliveries AE AMTSL: active management of third stage of labour; SBA: skilled birth attendant. b Comprehensive intervention—intervening on multiple levels. Open in new tab Table 5 Service and implementation outcomes for interventions to prevent OH: uterotonics (oxytocin) Article . LMIC . Intervention . Implementation . Evidence grade (FAME)a . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Uterotonics—oxytocin Boucar et al. (2014)b Niger, Mali AMTSL (including use of oxytocin) among SBAs Effectiveness (Quant) PPH rate decreased by 90% from 2.2% to 0.2% of vaginal deliveries (Niger) Mean PPH rate decreased from 0.9% to <0.3% (Mali) Continuous shared learning across multiple facilities Engaging teams of healthcare providers, managers + staff in improving care Analysing how processes of care function within a system Changing processes + systems of care to implement best practices Continuous analysis of data to assess results During Provider buy-in to adherence of PPH best practices Sustaining intervention after cessation of study Adoption (Quant) % of births receiving AMTSL (of which oxytocin administration is one component) increased from 2% at baseline to >90% (Niger) + from 17% in the first month of reporting to 95–97% (Mali) AE Garcia-Elorrio et al. (2014) Nicaragua Prophylactic use of the oxytocin in Uniject (OiU) a timetemperature indicator among SBAs for AMTSL Effectiveness (Quant) 2% increase in the use of AMTSL Rate of correct oxytocin use was significantly increased, reaching 97.4% (P = 0.003) Intervention developed onsite and by staff Providers engaged and motivated to implement Prior During Use of opinion leaders, interactive workshops to develop guidelinesAcademic detailingUse of remindersFeedback on the rates of oxytocin use Adoption Acceptability (All quant) 86/104 (82.7%) reported that the use of oxytocin in a preloaded device was easier for administration of oxytocin than the traditional method (acceptability) 77 (74.0%) reported that OiU was preferred to standard oxytocin administration (acceptability) 84 (80.8%) reported that they would not stop using prophylactic oxytocin whether or not OiU was available Researchers noted 95% intervention uptake (adoption) AE Moucheraud et al. (2015) India Advance distribution of oxytocin Effectiveness Timeliness (All quant) Preparing injection prior to delivery + making it available at bedside before birth was associated with increased likelihood of its rapid postpartum use in a hospital During Adoption (Quant) Having the intervention on hand increased likelihood of uptake (adoption) Oxytocin was prepared + available at bedside for 38.8% of deliveries AE Article . LMIC . Intervention . Implementation . Evidence grade (FAME)a . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Uterotonics—oxytocin Boucar et al. (2014)b Niger, Mali AMTSL (including use of oxytocin) among SBAs Effectiveness (Quant) PPH rate decreased by 90% from 2.2% to 0.2% of vaginal deliveries (Niger) Mean PPH rate decreased from 0.9% to <0.3% (Mali) Continuous shared learning across multiple facilities Engaging teams of healthcare providers, managers + staff in improving care Analysing how processes of care function within a system Changing processes + systems of care to implement best practices Continuous analysis of data to assess results During Provider buy-in to adherence of PPH best practices Sustaining intervention after cessation of study Adoption (Quant) % of births receiving AMTSL (of which oxytocin administration is one component) increased from 2% at baseline to >90% (Niger) + from 17% in the first month of reporting to 95–97% (Mali) AE Garcia-Elorrio et al. (2014) Nicaragua Prophylactic use of the oxytocin in Uniject (OiU) a timetemperature indicator among SBAs for AMTSL Effectiveness (Quant) 2% increase in the use of AMTSL Rate of correct oxytocin use was significantly increased, reaching 97.4% (P = 0.003) Intervention developed onsite and by staff Providers engaged and motivated to implement Prior During Use of opinion leaders, interactive workshops to develop guidelinesAcademic detailingUse of remindersFeedback on the rates of oxytocin use Adoption Acceptability (All quant) 86/104 (82.7%) reported that the use of oxytocin in a preloaded device was easier for administration of oxytocin than the traditional method (acceptability) 77 (74.0%) reported that OiU was preferred to standard oxytocin administration (acceptability) 84 (80.8%) reported that they would not stop using prophylactic oxytocin whether or not OiU was available Researchers noted 95% intervention uptake (adoption) AE Moucheraud et al. (2015) India Advance distribution of oxytocin Effectiveness Timeliness (All quant) Preparing injection prior to delivery + making it available at bedside before birth was associated with increased likelihood of its rapid postpartum use in a hospital During Adoption (Quant) Having the intervention on hand increased likelihood of uptake (adoption) Oxytocin was prepared + available at bedside for 38.8% of deliveries AE AMTSL: active management of third stage of labour; SBA: skilled birth attendant. b Comprehensive intervention—intervening on multiple levels. Open in new tab Table 6 Service and implementation outcomes for interventions to prevent OH: provider skills training Article . LMIC . Intervention . Implementation . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Clinical provider skills training Ajeani et al. (2017)b Uganda Mentoring (including prevention of PPH) of HWs delivered through a cascade model Strengthening leadership for MNH at district + facility level Motivation of health workers Effectiveness Efficiency Timeliness (All qual) Mentorship competencies + responsiveness to emergencies + HW professionalism Improved supplies/medicine availability, teamwork + innovative local problem-solving approaches due to better district leadership for MNH HWs were empowered to perform better Implemented in two phases: Initial mentoring of local mentors by central mentors Actual mentoring process by local mentors Driven by a theory of change modelled around WHO health systems’ building blocks Pre During Cascade mentoring process needed strong external mentorship support at the start to ensure improved capacity among local mentors to provide mentorship among local district staff Adoption (Qual) Increased health worker productivity through skills improvement. Improved responsiveness from health professionals through new record keeping methods Better patient management as a result of improved treatment before referral Anastacia Nkwonta and Oyetunde (2017) Nigeria LSS for the prevention of OH among midwives in primary health centre Effectiveness (Quant) 52.5% of the midwives were highly knowledgeable about LSS after training A positive relationship was found between midwives' knowledge + practice, with knowledge varying by academic qualifications Continuous training, frequent monitoring, super-vision During After Lack of an assistant when on duty (23.7%) Shortage of oxytocin (20.3%) Adoption (Quant) Although 70.6% self-reported a high level of implementing LSS procedures, only 21.7% of the midwives observed actually demonstrated a high level of implementation (adoption) All administered oxytocin, most frequently when delivering the anterior shoulders Chang et al. (2019)b Malawi Alliance for Innovation on Maternal Health Malawi programme Bundle contains 13 ‘best practices’ including: prevention of OH by employing haemorrhage risk assessment tools Effectiveness Safety (All quant) Decreased rate of maternal mortality Increase in hospital survey of patient safety composite safety scores Delivered via: classroom didactics on OH, teamwork protocols, skills laboratory activities, simulation training Pre During Post Holding health care facilities accountable for errors, and creating a ‘Just Culture’ environment for a more concise way to address issues Adoption (Quant) The change in utilization rate of PPH procedural interventions Evans et al. (2018) Uganda Simulation-based, PAL to improve + maintain intrapartum skills of providers for PPH prevention Effectiveness (Quant) Observed care of mother + newborn improved in all groups Incidence of PPH declined 17% from during the intervention period compared with the 6- to 9-month period after the intervention Three approaches to facilitating training: some received simulation-based training alone some received training plus PAL some receive training plus PAL plus phone call reminders to facilitate practice During Post Clinical skills lacking in health care facility teams and no improvement in implementation even after months of drills Adoption Sustainability (All quant) Proportion of births with uterotonic given within 1 min of birth increased. Uterotonic coverage within 1 min increased from: full group: 8% (CI 4–12%) to 50% (CI 42–59%); partial group: 19% (CI 9–30%) to 42% (CI 31–53%); control group: 11% (5–7%) to 51% (40–61%) Simulated skills maintenance for PPH prophylaxis remained high across groups 7–8 months after the intervention Article . LMIC . Intervention . Implementation . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Clinical provider skills training Ajeani et al. (2017)b Uganda Mentoring (including prevention of PPH) of HWs delivered through a cascade model Strengthening leadership for MNH at district + facility level Motivation of health workers Effectiveness Efficiency Timeliness (All qual) Mentorship competencies + responsiveness to emergencies + HW professionalism Improved supplies/medicine availability, teamwork + innovative local problem-solving approaches due to better district leadership for MNH HWs were empowered to perform better Implemented in two phases: Initial mentoring of local mentors by central mentors Actual mentoring process by local mentors Driven by a theory of change modelled around WHO health systems’ building blocks Pre During Cascade mentoring process needed strong external mentorship support at the start to ensure improved capacity among local mentors to provide mentorship among local district staff Adoption (Qual) Increased health worker productivity through skills improvement. Improved responsiveness from health professionals through new record keeping methods Better patient management as a result of improved treatment before referral Anastacia Nkwonta and Oyetunde (2017) Nigeria LSS for the prevention of OH among midwives in primary health centre Effectiveness (Quant) 52.5% of the midwives were highly knowledgeable about LSS after training A positive relationship was found between midwives' knowledge + practice, with knowledge varying by academic qualifications Continuous training, frequent monitoring, super-vision During After Lack of an assistant when on duty (23.7%) Shortage of oxytocin (20.3%) Adoption (Quant) Although 70.6% self-reported a high level of implementing LSS procedures, only 21.7% of the midwives observed actually demonstrated a high level of implementation (adoption) All administered oxytocin, most frequently when delivering the anterior shoulders Chang et al. (2019)b Malawi Alliance for Innovation on Maternal Health Malawi programme Bundle contains 13 ‘best practices’ including: prevention of OH by employing haemorrhage risk assessment tools Effectiveness Safety (All quant) Decreased rate of maternal mortality Increase in hospital survey of patient safety composite safety scores Delivered via: classroom didactics on OH, teamwork protocols, skills laboratory activities, simulation training Pre During Post Holding health care facilities accountable for errors, and creating a ‘Just Culture’ environment for a more concise way to address issues Adoption (Quant) The change in utilization rate of PPH procedural interventions Evans et al. (2018) Uganda Simulation-based, PAL to improve + maintain intrapartum skills of providers for PPH prevention Effectiveness (Quant) Observed care of mother + newborn improved in all groups Incidence of PPH declined 17% from during the intervention period compared with the 6- to 9-month period after the intervention Three approaches to facilitating training: some received simulation-based training alone some received training plus PAL some receive training plus PAL plus phone call reminders to facilitate practice During Post Clinical skills lacking in health care facility teams and no improvement in implementation even after months of drills Adoption Sustainability (All quant) Proportion of births with uterotonic given within 1 min of birth increased. Uterotonic coverage within 1 min increased from: full group: 8% (CI 4–12%) to 50% (CI 42–59%); partial group: 19% (CI 9–30%) to 42% (CI 31–53%); control group: 11% (5–7%) to 51% (40–61%) Simulated skills maintenance for PPH prophylaxis remained high across groups 7–8 months after the intervention HWs: health workers; MNH: maternal and newborn health; LSS: life-saving skills. a •••. b Comprehensive intervention—intervening on multiple levels. Open in new tab Table 6 Service and implementation outcomes for interventions to prevent OH: provider skills training Article . LMIC . Intervention . Implementation . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Clinical provider skills training Ajeani et al. (2017)b Uganda Mentoring (including prevention of PPH) of HWs delivered through a cascade model Strengthening leadership for MNH at district + facility level Motivation of health workers Effectiveness Efficiency Timeliness (All qual) Mentorship competencies + responsiveness to emergencies + HW professionalism Improved supplies/medicine availability, teamwork + innovative local problem-solving approaches due to better district leadership for MNH HWs were empowered to perform better Implemented in two phases: Initial mentoring of local mentors by central mentors Actual mentoring process by local mentors Driven by a theory of change modelled around WHO health systems’ building blocks Pre During Cascade mentoring process needed strong external mentorship support at the start to ensure improved capacity among local mentors to provide mentorship among local district staff Adoption (Qual) Increased health worker productivity through skills improvement. Improved responsiveness from health professionals through new record keeping methods Better patient management as a result of improved treatment before referral Anastacia Nkwonta and Oyetunde (2017) Nigeria LSS for the prevention of OH among midwives in primary health centre Effectiveness (Quant) 52.5% of the midwives were highly knowledgeable about LSS after training A positive relationship was found between midwives' knowledge + practice, with knowledge varying by academic qualifications Continuous training, frequent monitoring, super-vision During After Lack of an assistant when on duty (23.7%) Shortage of oxytocin (20.3%) Adoption (Quant) Although 70.6% self-reported a high level of implementing LSS procedures, only 21.7% of the midwives observed actually demonstrated a high level of implementation (adoption) All administered oxytocin, most frequently when delivering the anterior shoulders Chang et al. (2019)b Malawi Alliance for Innovation on Maternal Health Malawi programme Bundle contains 13 ‘best practices’ including: prevention of OH by employing haemorrhage risk assessment tools Effectiveness Safety (All quant) Decreased rate of maternal mortality Increase in hospital survey of patient safety composite safety scores Delivered via: classroom didactics on OH, teamwork protocols, skills laboratory activities, simulation training Pre During Post Holding health care facilities accountable for errors, and creating a ‘Just Culture’ environment for a more concise way to address issues Adoption (Quant) The change in utilization rate of PPH procedural interventions Evans et al. (2018) Uganda Simulation-based, PAL to improve + maintain intrapartum skills of providers for PPH prevention Effectiveness (Quant) Observed care of mother + newborn improved in all groups Incidence of PPH declined 17% from during the intervention period compared with the 6- to 9-month period after the intervention Three approaches to facilitating training: some received simulation-based training alone some received training plus PAL some receive training plus PAL plus phone call reminders to facilitate practice During Post Clinical skills lacking in health care facility teams and no improvement in implementation even after months of drills Adoption Sustainability (All quant) Proportion of births with uterotonic given within 1 min of birth increased. Uterotonic coverage within 1 min increased from: full group: 8% (CI 4–12%) to 50% (CI 42–59%); partial group: 19% (CI 9–30%) to 42% (CI 31–53%); control group: 11% (5–7%) to 51% (40–61%) Simulated skills maintenance for PPH prophylaxis remained high across groups 7–8 months after the intervention Article . LMIC . Intervention . Implementation . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Clinical provider skills training Ajeani et al. (2017)b Uganda Mentoring (including prevention of PPH) of HWs delivered through a cascade model Strengthening leadership for MNH at district + facility level Motivation of health workers Effectiveness Efficiency Timeliness (All qual) Mentorship competencies + responsiveness to emergencies + HW professionalism Improved supplies/medicine availability, teamwork + innovative local problem-solving approaches due to better district leadership for MNH HWs were empowered to perform better Implemented in two phases: Initial mentoring of local mentors by central mentors Actual mentoring process by local mentors Driven by a theory of change modelled around WHO health systems’ building blocks Pre During Cascade mentoring process needed strong external mentorship support at the start to ensure improved capacity among local mentors to provide mentorship among local district staff Adoption (Qual) Increased health worker productivity through skills improvement. Improved responsiveness from health professionals through new record keeping methods Better patient management as a result of improved treatment before referral Anastacia Nkwonta and Oyetunde (2017) Nigeria LSS for the prevention of OH among midwives in primary health centre Effectiveness (Quant) 52.5% of the midwives were highly knowledgeable about LSS after training A positive relationship was found between midwives' knowledge + practice, with knowledge varying by academic qualifications Continuous training, frequent monitoring, super-vision During After Lack of an assistant when on duty (23.7%) Shortage of oxytocin (20.3%) Adoption (Quant) Although 70.6% self-reported a high level of implementing LSS procedures, only 21.7% of the midwives observed actually demonstrated a high level of implementation (adoption) All administered oxytocin, most frequently when delivering the anterior shoulders Chang et al. (2019)b Malawi Alliance for Innovation on Maternal Health Malawi programme Bundle contains 13 ‘best practices’ including: prevention of OH by employing haemorrhage risk assessment tools Effectiveness Safety (All quant) Decreased rate of maternal mortality Increase in hospital survey of patient safety composite safety scores Delivered via: classroom didactics on OH, teamwork protocols, skills laboratory activities, simulation training Pre During Post Holding health care facilities accountable for errors, and creating a ‘Just Culture’ environment for a more concise way to address issues Adoption (Quant) The change in utilization rate of PPH procedural interventions Evans et al. (2018) Uganda Simulation-based, PAL to improve + maintain intrapartum skills of providers for PPH prevention Effectiveness (Quant) Observed care of mother + newborn improved in all groups Incidence of PPH declined 17% from during the intervention period compared with the 6- to 9-month period after the intervention Three approaches to facilitating training: some received simulation-based training alone some received training plus PAL some receive training plus PAL plus phone call reminders to facilitate practice During Post Clinical skills lacking in health care facility teams and no improvement in implementation even after months of drills Adoption Sustainability (All quant) Proportion of births with uterotonic given within 1 min of birth increased. Uterotonic coverage within 1 min increased from: full group: 8% (CI 4–12%) to 50% (CI 42–59%); partial group: 19% (CI 9–30%) to 42% (CI 31–53%); control group: 11% (5–7%) to 51% (40–61%) Simulated skills maintenance for PPH prophylaxis remained high across groups 7–8 months after the intervention HWs: health workers; MNH: maternal and newborn health; LSS: life-saving skills. a •••. b Comprehensive intervention—intervening on multiple levels. Open in new tab Table 7 Service and implementation outcomes for interventions to prevent OH: provision of clinical guidelines Article . LMIC . Intervention . Implementation . Evidence grade (FAME)a . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Provision of clinical guidelines Braddick et al. (2016) Uganda PPH guidelines (including use of oxytocin) Effectiveness (Quant) Oxytocin was administered during AMTSL in 105 (68.2%) deliveries, but it was used in varying doses + by differing routes Midwives, obstetricians and medical officers worked to incorporate all 3 components of AMTSL as recommended by the WHO During Healthcare system and resource issues (understaffing; oxytocin accessible on the hospital ward but variable in community health centre; issues with correct storage) Current knowledge, awareness, + use of clinical guideline Healthcare practitioner attitudes to updating their clinical practice Acceptability (Quant) Adoption (Quant) Adoption (Qual) High compliance with provision of uterotonics, although not of the other components of AMTSL (adoption) AE Article . LMIC . Intervention . Implementation . Evidence grade (FAME)a . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Provision of clinical guidelines Braddick et al. (2016) Uganda PPH guidelines (including use of oxytocin) Effectiveness (Quant) Oxytocin was administered during AMTSL in 105 (68.2%) deliveries, but it was used in varying doses + by differing routes Midwives, obstetricians and medical officers worked to incorporate all 3 components of AMTSL as recommended by the WHO During Healthcare system and resource issues (understaffing; oxytocin accessible on the hospital ward but variable in community health centre; issues with correct storage) Current knowledge, awareness, + use of clinical guideline Healthcare practitioner attitudes to updating their clinical practice Acceptability (Quant) Adoption (Quant) Adoption (Qual) High compliance with provision of uterotonics, although not of the other components of AMTSL (adoption) AE AMTSL: active management of the third stage of labour. Open in new tab Table 7 Service and implementation outcomes for interventions to prevent OH: provision of clinical guidelines Article . LMIC . Intervention . Implementation . Evidence grade (FAME)a . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Provision of clinical guidelines Braddick et al. (2016) Uganda PPH guidelines (including use of oxytocin) Effectiveness (Quant) Oxytocin was administered during AMTSL in 105 (68.2%) deliveries, but it was used in varying doses + by differing routes Midwives, obstetricians and medical officers worked to incorporate all 3 components of AMTSL as recommended by the WHO During Healthcare system and resource issues (understaffing; oxytocin accessible on the hospital ward but variable in community health centre; issues with correct storage) Current knowledge, awareness, + use of clinical guideline Healthcare practitioner attitudes to updating their clinical practice Acceptability (Quant) Adoption (Quant) Adoption (Qual) High compliance with provision of uterotonics, although not of the other components of AMTSL (adoption) AE Article . LMIC . Intervention . Implementation . Evidence grade (FAME)a . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Provision of clinical guidelines Braddick et al. (2016) Uganda PPH guidelines (including use of oxytocin) Effectiveness (Quant) Oxytocin was administered during AMTSL in 105 (68.2%) deliveries, but it was used in varying doses + by differing routes Midwives, obstetricians and medical officers worked to incorporate all 3 components of AMTSL as recommended by the WHO During Healthcare system and resource issues (understaffing; oxytocin accessible on the hospital ward but variable in community health centre; issues with correct storage) Current knowledge, awareness, + use of clinical guideline Healthcare practitioner attitudes to updating their clinical practice Acceptability (Quant) Adoption (Quant) Adoption (Qual) High compliance with provision of uterotonics, although not of the other components of AMTSL (adoption) AE AMTSL: active management of the third stage of labour. Open in new tab Table 8 Service and implementation outcomes for interventions to prevent OH: multi-component quality improvement bundles Article . LMIC . Intervention . Implementation . Evidence grade (FAME)a . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Provision of clinical guidelines Ajeani et al. (2017) Uganda Mentoring (including prevention of PPH) of HWs delivered through a cascade model Strengthening leadership for MNH at district + facility level Motivation of HWs Effectiveness Efficiency Timeliness (All qual) Mentorship competencies + responsiveness to emergencies + HW professionalism Improved supplies/medicine availability, teamwork + innovative local problem-solving approaches due to better district leadership for MNH HWs were empowered to perform better Implemented in two phases: Initial mentoring of local mentors by central mentors Actual mentoring process by local mentors Driven by a theory of change modelled around WHO health systems’ building blocks Cascade mentoring process needed strong external mentorship support at the start to ensure improved capacity among local mentors to provide mentorship among local district staff Adoption (Qual) AME Boucar et al. (2014) Niger, Mali AMTSL (including use of oxytocin) Essential newborn care Routine monitoring of mother + newborn after birth Effectiveness (Quant) PPH rate decreased by 90% from 2.2% to 0.2% of vaginal deliveries (Niger) Mean PPH rate decreased from 0.9% to <0.3% (Mali) Adherence to a set of postpartum minimum monitoring standards for mother + newborn for early detection of complications increased from 19% to 86% + from 19% to 89% in Mali, respectively (not measured as a discrete indicator in Niger) Continuous shared learning across multiple facilities Engaging teams of healthcare providers, managers, + staff in improving care Analysing how processes of care function within a system Changing processes + systems of care to implement best practices Continuous analysis of data to assess results Adoption (Quant) % of births receiving AMTSL (of which oxytocin administration is one component) increased from 2% at baseline to >90% (Niger) + from 17% in the first month of reporting to 95–97% (Mali) AE Chang et al. (2019) Malawi Alliance for Innovation on Maternal Health Malawi programme Bundle contains 13 ‘best practices’ including: prevention of OH by employing haemorrhage risk assessment tools Effectiveness Safety (All quant) Decreased rate of maternal mortality Increase in Hospital Survey of Patient Safety composite safety scores Delivered via: classroom didactics on OH, teamwork protocols, skills laboratory activities, simulation training Pre During Post Adoption (Quant) The change in utilization rate of PPH procedural interventions AME Article . LMIC . Intervention . Implementation . Evidence grade (FAME)a . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Provision of clinical guidelines Ajeani et al. (2017) Uganda Mentoring (including prevention of PPH) of HWs delivered through a cascade model Strengthening leadership for MNH at district + facility level Motivation of HWs Effectiveness Efficiency Timeliness (All qual) Mentorship competencies + responsiveness to emergencies + HW professionalism Improved supplies/medicine availability, teamwork + innovative local problem-solving approaches due to better district leadership for MNH HWs were empowered to perform better Implemented in two phases: Initial mentoring of local mentors by central mentors Actual mentoring process by local mentors Driven by a theory of change modelled around WHO health systems’ building blocks Cascade mentoring process needed strong external mentorship support at the start to ensure improved capacity among local mentors to provide mentorship among local district staff Adoption (Qual) AME Boucar et al. (2014) Niger, Mali AMTSL (including use of oxytocin) Essential newborn care Routine monitoring of mother + newborn after birth Effectiveness (Quant) PPH rate decreased by 90% from 2.2% to 0.2% of vaginal deliveries (Niger) Mean PPH rate decreased from 0.9% to <0.3% (Mali) Adherence to a set of postpartum minimum monitoring standards for mother + newborn for early detection of complications increased from 19% to 86% + from 19% to 89% in Mali, respectively (not measured as a discrete indicator in Niger) Continuous shared learning across multiple facilities Engaging teams of healthcare providers, managers, + staff in improving care Analysing how processes of care function within a system Changing processes + systems of care to implement best practices Continuous analysis of data to assess results Adoption (Quant) % of births receiving AMTSL (of which oxytocin administration is one component) increased from 2% at baseline to >90% (Niger) + from 17% in the first month of reporting to 95–97% (Mali) AE Chang et al. (2019) Malawi Alliance for Innovation on Maternal Health Malawi programme Bundle contains 13 ‘best practices’ including: prevention of OH by employing haemorrhage risk assessment tools Effectiveness Safety (All quant) Decreased rate of maternal mortality Increase in Hospital Survey of Patient Safety composite safety scores Delivered via: classroom didactics on OH, teamwork protocols, skills laboratory activities, simulation training Pre During Post Adoption (Quant) The change in utilization rate of PPH procedural interventions AME HW: health worker; MNH: maternal and newborn health. Open in new tab Table 8 Service and implementation outcomes for interventions to prevent OH: multi-component quality improvement bundles Article . LMIC . Intervention . Implementation . Evidence grade (FAME)a . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Provision of clinical guidelines Ajeani et al. (2017) Uganda Mentoring (including prevention of PPH) of HWs delivered through a cascade model Strengthening leadership for MNH at district + facility level Motivation of HWs Effectiveness Efficiency Timeliness (All qual) Mentorship competencies + responsiveness to emergencies + HW professionalism Improved supplies/medicine availability, teamwork + innovative local problem-solving approaches due to better district leadership for MNH HWs were empowered to perform better Implemented in two phases: Initial mentoring of local mentors by central mentors Actual mentoring process by local mentors Driven by a theory of change modelled around WHO health systems’ building blocks Cascade mentoring process needed strong external mentorship support at the start to ensure improved capacity among local mentors to provide mentorship among local district staff Adoption (Qual) AME Boucar et al. (2014) Niger, Mali AMTSL (including use of oxytocin) Essential newborn care Routine monitoring of mother + newborn after birth Effectiveness (Quant) PPH rate decreased by 90% from 2.2% to 0.2% of vaginal deliveries (Niger) Mean PPH rate decreased from 0.9% to <0.3% (Mali) Adherence to a set of postpartum minimum monitoring standards for mother + newborn for early detection of complications increased from 19% to 86% + from 19% to 89% in Mali, respectively (not measured as a discrete indicator in Niger) Continuous shared learning across multiple facilities Engaging teams of healthcare providers, managers, + staff in improving care Analysing how processes of care function within a system Changing processes + systems of care to implement best practices Continuous analysis of data to assess results Adoption (Quant) % of births receiving AMTSL (of which oxytocin administration is one component) increased from 2% at baseline to >90% (Niger) + from 17% in the first month of reporting to 95–97% (Mali) AE Chang et al. (2019) Malawi Alliance for Innovation on Maternal Health Malawi programme Bundle contains 13 ‘best practices’ including: prevention of OH by employing haemorrhage risk assessment tools Effectiveness Safety (All quant) Decreased rate of maternal mortality Increase in Hospital Survey of Patient Safety composite safety scores Delivered via: classroom didactics on OH, teamwork protocols, skills laboratory activities, simulation training Pre During Post Adoption (Quant) The change in utilization rate of PPH procedural interventions AME Article . LMIC . Intervention . Implementation . Evidence grade (FAME)a . Activities . Outcome type . Outcome description . Strategy . Phase . Challenges . Outcome type . Outcome description . Provision of clinical guidelines Ajeani et al. (2017) Uganda Mentoring (including prevention of PPH) of HWs delivered through a cascade model Strengthening leadership for MNH at district + facility level Motivation of HWs Effectiveness Efficiency Timeliness (All qual) Mentorship competencies + responsiveness to emergencies + HW professionalism Improved supplies/medicine availability, teamwork + innovative local problem-solving approaches due to better district leadership for MNH HWs were empowered to perform better Implemented in two phases: Initial mentoring of local mentors by central mentors Actual mentoring process by local mentors Driven by a theory of change modelled around WHO health systems’ building blocks Cascade mentoring process needed strong external mentorship support at the start to ensure improved capacity among local mentors to provide mentorship among local district staff Adoption (Qual) AME Boucar et al. (2014) Niger, Mali AMTSL (including use of oxytocin) Essential newborn care Routine monitoring of mother + newborn after birth Effectiveness (Quant) PPH rate decreased by 90% from 2.2% to 0.2% of vaginal deliveries (Niger) Mean PPH rate decreased from 0.9% to <0.3% (Mali) Adherence to a set of postpartum minimum monitoring standards for mother + newborn for early detection of complications increased from 19% to 86% + from 19% to 89% in Mali, respectively (not measured as a discrete indicator in Niger) Continuous shared learning across multiple facilities Engaging teams of healthcare providers, managers, + staff in improving care Analysing how processes of care function within a system Changing processes + systems of care to implement best practices Continuous analysis of data to assess results Adoption (Quant) % of births receiving AMTSL (of which oxytocin administration is one component) increased from 2% at baseline to >90% (Niger) + from 17% in the first month of reporting to 95–97% (Mali) AE Chang et al. (2019) Malawi Alliance for Innovation on Maternal Health Malawi programme Bundle contains 13 ‘best practices’ including: prevention of OH by employing haemorrhage risk assessment tools Effectiveness Safety (All quant) Decreased rate of maternal mortality Increase in Hospital Survey of Patient Safety composite safety scores Delivered via: classroom didactics on OH, teamwork protocols, skills laboratory activities, simulation training Pre During Post Adoption (Quant) The change in utilization rate of PPH procedural interventions AME HW: health worker; MNH: maternal and newborn health. Open in new tab The results of our review are presented under four sections: (1) service outcomes, (2) implementation strategies, (3) implementation challenges and successes and (4) implementation outcomes. Service outcomes Most studies (n = 11) were focused on primary prevention of OH. Only one study evaluated secondary prevention (i.e. OH risk assessment in the clinical setting) as part of a comprehensive programme that also included early recognition of OH and coordinated response among clinicians (Chang et al., 2019). Intervention outcomes that were most commonly reported were effectiveness (n = 11) and safety (n = 4). One study also included timeliness (n = 1) and efficiency (n = 1). Only one study did not examine effectiveness, due to the rarity of the primary outcome (incident drop in maternal haemoglobin over 20% with home birth) (Weeks et al., 2013). Notably patient-centredness and equity were not assessed in any of the studies. Prophylactic use of uterotonics Among the studies on prophylactic use of uterotonics, a few (n = 3) reported administration within clinical facilities in Nicaragua, Niger, Mali or India (Boucar et al., 2014; Garcia-Elorrio et al., 2014; Moucheraud et al., 2015), while others (n = 4) reported community-based distribution in Ethiopia, Nigeria or Uganda (Weeks et al., 2013; Spangler et al., 2014; Cannon et al., 2017; Orobaton et al., 2017). Studies of facility-based administration of uterotonics focused on the use of oxytocin and reported on effectiveness (i.e. reduction in the rates of PPH) with high rates of correct use as per WHO guidelines (Garcia-Elorrio et al., 2014). In a hospital in India, advanced distribution of oxytocin by skilled birth attendants via prepared injections made available at the bedside before birth was associated with increased likelihood of timely postpartum use (Moucheraud et al., 2015) and acceptability was high for a similar preloaded injector of oxytocin with a time temperature indicator at a health facility in Nicaragua (Garcia-Elorrio et al., 2014). Of the three studies of community-based distribution and use of uterotonics (Weeks et al., 2013; Cannon et al., 2017; Orobaton et al., 2017), all were focused on misoprostol. The drug was reported as more effective and safe for reducing bleeding during delivery than non-medical, traditional remedies (Cannon et al., 2017) with few or no side effects (Weeks et al., 2013; Orobaton et al., 2017). Community distribution was supported by evidence from Uganda, Ethiopia and Nigeria. While most studies (n = 11) focused on the implementation of programmes to prevent OH, only one study reported on policy implementation (Spangler et al., 2014). In a study of implementation of national policy on community-based distribution of uterotonics in Ethiopia, findings from stakeholder interviews suggested officials perceived high effectiveness in the community during implementation. However, there were safety concerns, including fear of mishandling by untrained persons, misuse of uterotonics as abortifacients for unsafe abortion, administration in the event of an undiagnosed multiple pregnancy (e.g. twin) leading to a fatal outcome and administration during labour leading to uterine rupture. Provision of clinical guidelines In the work of Braddick et al. (2016) on the provision of PPH clinical guidelines (Braddick et al., 2016), oxytocin was administered during AMTSL in 68.2% of deliveries as per WHO guidelines over a 2-month observation period in 2014 in an Ugandan healthcare facility, but in varying doses and by differing routes. Provider skills training Among the four studies of provider skills training (Ajeani et al., 2017; Anastacia Nkwonta and Oyetunde, 2017; Evans et al., 2018; Chang et al., 2019) in Malawi, Uganda and Nigeria, intervention outcomes focused on effectiveness, efficiency, timeliness and safety. One study reported on the cascade model, a model for the provision of mentorship in which external mentors (i.e. expert obstetricians and paediatricians) initially provided mentorship and training on OH prevention to local health workers in three districts in Uganda, who then conducted mentoring of others that was observed by the external mentors. Findings from key informant interviews with district-level managers and local mentors suggested that mentorship improved several aspects of leadership and healthcare delivery, including health worker empowerment and professionalism, improved availability of medicines and innovative approaches for local problem-solving (Ajeani et al., 2017). In a pragmatic trial from Uganda, simulation-based, peer-assisted learning (PAL) to support onsite practice after training was used to improve peripartum management with a view to OH prevention; this resulted in improved proportion of births where uterotonics were provided (Evans et al., 2018). Uterotonic coverage within 1 min increased from: 11% (5–7%) to 51% (40–61%) in the control group (who received simulation-based training alone); 19% [confidence interval (CI) 9–30%] to 42% (CI 31–53%) in the partial group (who received training plus PAL); and 8% (CI 4–12%) to 50% (CI 42–59%) in the full group (training plus PAL plus phone call reminders to facilitate practice). Incidence of PPH declined by 17% across all groups by 6–9 months after the intervention. Another study focused on the provision of training on life-saving skills (LSS) (for the provision of emergency obstetric care to address the major causes of maternal and neonatal morbidity and mortality, including for the prevention of OH) among midwives in a primary health centre in Nigeria. About half of those who were trained were highly knowledgeable on LSS after training, which was associated with good clinical practice (GCP) of LSS; however, GCP was not well defined in this context (Anastacia Nkwonta and Oyetunde, 2017). Multi-component quality improvement bundles The only study that touched on secondary prevention was conducted in Malawi (Chang et al., 2019) wherein a bundle of OH ‘best practices’ was deployed at a central hospital and district health centre. The interventions focused on the prevention of OH via use of uterotonics during AMTSL (i.e. primary prevention) and use of OH risk assessment tools (i.e. secondary prevention), in addition to the recognition of OH through accurate measurement of blood loss, which can guide timely management to mitigate the risk of serious outcomes. The independent effect of OH risk assessment was not assessed. Interventions that were evaluated spanned different levels: patient (e.g. self-administration of uterotonics); provider (e.g. administration of uterotonics, training on AMTSL, provision of OH risk assessment); community (e.g. distribution of uterotonics); and health system level (e.g. roll-out of regional and national policies for OH prevention). Except for three studies (Boucar et al., 2014; Ajeani et al., 2017; Chang et al., 2019), most were not multimodal in which comprehensive interventions (programmes or policies) were implemented with multiple components at various levels. Implementation strategies All studies reported on the development and testing of implementation strategies, even if not the objective of the study. Implementation strategies included early community engagement in the value of intervention (Cannon et al., 2017); stakeholder engagement in designing delivery (Boucar et al., 2014; Garcia-Elorrio et al., 2014; Cannon et al., 2017); use of a theory of change that clarifies and guides how intervention activities lead to outcomes (Ajeani et al., 2017); continuous shared learning for quality improvement (Boucar et al., 2014); training and monitoring (Anastacia Nkwonta and Oyetunde, 2017; Cannon et al., 2017); training through simulation (Evans et al., 2018; Chang et al., 2019); training through peer learning (Evans et al., 2018); utilization of reminders to use evidence-based interventions (Garcia-Elorrio et al., 2014; Evans et al., 2018); use of opinion leaders (Garcia-Elorrio et al., 2014); guidance by external mentors (Ajeani et al., 2017); and early advocacy efforts at the state level for national policy implementation (Cannon et al., 2017). Some studies linked successful delivery (i.e. adoption, compliance, uptake) to implementation strategies and outcomes. For example, Cannon et al. (2017) reported that early introduction of the intervention to community members and their engagement during programme preparation stages contributed to the intervention’s acceptance and adoption (Cannon et al., 2017). Notably, most studies did not report an explicit theoretical framework to guide strategies for intervention or implementation development or evaluation; however, Ajeani et al. stated that their work was guided by a theory of change modelled on the health systems’ building blocks described by the WHO (2000). In addition, Orobataon et al. draw on a framework for scale up in discussing implications of their work (Simmons and Shiffman, 2007). Implementation challenges and successes Studies reported various challenges to the implementation of interventions to prevent OH. These included: challenges among intervention beneficiaries, such as pervasive myths/fears of uterotonics and other sociocultural barriers (Cannon et al., 2017); challenges among providers (i.e. intervention implementers), including issues with knowledge, awareness and use of evidence-based interventions (Spangler et al., 2014; Braddick et al., 2016); and challenges from the health system, including supply chain bottlenecks (Anastacia Nkwonta and Oyetunde, 2017) and human resource limitations (Anastacia Nkwonta and Oyetunde, 2017; Cannon et al., 2017). Implementation outcomes The studies reported implementation research outcomes spanning the implementation process. Reporting included the pre-implementation (n = 1), during (n = 10) and post-implementation (n = 4) phases, with some reporting outcomes from more than one phase. Often, implementation research outcomes were reported in the paper but were not a primary outcome. Quantitative assessment was more commonly reported for acceptability, feasibility, adoption, penetration and sustainability, while qualitative assessment was more commonly reported for appropriateness. Fidelity and cost were not assessed within any articles included in our review. Adoption Tables report the strength of the evidence based on the implementation research outcomes reported in each study. Adoption, assessed within all articles, was the most commonly examined implementation research outcome and was referred to interchangeably as compliance, use, utilization, implementation, successful introduction and uptake of a given intervention, which was aligned with Proctor et al.’s (2011) definition of adoption. Adoption was mostly assessed quantitatively based on self-reporting during the implementation phase; some studies attempted to correlate self-reporting with the observation of uptake of the intervention. For example, Anastacia Nkwonta and Oyetunde (2017) reported that 70.6% of midwives in a primary health centre in Nigeria reported implementing the intervention for LSS. However, when observed, only 21.7% of midwives demonstrated a proper adoption, as noted by enacting GCPs. In the work of Braddick et al. (2016), there was high adoption of the individual components of AMTSL, including the critical component of oxytocin administration, but not the whole package (Braddick et al., 2016). Acceptability Acceptability was assessed in five studies. This was undertaken both quantitatively (n = 3) and qualitatively (n = 2) to determine whether the intervention was ‘well-received’ (Cannon et al., 2017), the beneficiaries were satisfied (Weeks et al., 2013) or the implementers (i.e. administrative officials, skilled birth attendants) had positive attitudes towards the intervention (Garcia-Elorrio et al., 2014; Spangler et al., 2014). There was geographic variation in the acceptability of a given intervention. For example, the acceptability of self-administered misoprostol in Uganda was high among mothers (97% reported satisfaction) (Weeks et al., 2013). By contrast, following community-based distribution of misoprostol in Ethiopia in response to a national policy, regional implementers (i.e. administrative officials) reported low acceptability due to perceived risk of misuse of the medication for abortion and its potential to encourage home births (Spangler et al., 2014). Other implementation outcomes Appropriateness, penetration and feasibility were examined to a lesser extent across the literature. Appropriateness was assessed qualitatively in two studies, focused on whether the intervention fit (Cannon et al., 2017) or did not fit (Spangler et al., 2014) within the local context. Penetration and feasibility were assessed quantitatively in one study. Specifically, Orobaton et al. (2017) reported that community-based distribution of misoprostol was successful at reaching mothers in all labour wards within the target area (penetration) in Sokoto State in Nigeria. The authors also reported that community-data collection systems to record maternal deaths were operational (feasibility) (Orobaton et al., 2017). Sustainability was quantitatively assessed in one article (Evans et al., 2018). The investigators reported the maintenance of intrapartum skills among providers 7–8 months after the intervention of simulation-based, PAL in Uganda; however, no studies reported longer-term follow-up. Discussion The findings from this systematic review suggest that studies of implementation of innovations to prevent OH in LMICs focus on the following key interventions: use of uterotonics, clinical guidelines and clinical provider skills training, with some interventions integrated into multi-component bundles of OH prevention and management. The focus has been overwhelmingly in favour of primary prevention; only one group evaluated secondary prevention efforts (Chang et al., 2019) as part of an OH bundle that also included early recognition of OH and coordinated response among clinicians. Chang et al. remind us how prevention efforts within OH bundles developed in high-income settings need context-specific adaptations to be successful. Tools for risk assessment developed in high-income settings (secondary prevention) will need to integrate context-specific risk factors for LMICs. For example, anaemia is much more common among women of reproductive age in LMICs and anaemic women are more vulnerable to even moderate amounts of blood loss. Whereas recognition and response programme components do not prevent OH, they do prevent and reduce the severity of consequences through improved management. Although clinical and surgical management of OH is critical to preventing severe maternal morbidity and mortality, there is a noticeable lack of focus on OH risk assessment within efforts to prevent haemorrhage and mitigate risk of haemorrhage-related morbidity when bleeding occurs. By summarizing service outcomes in addition, this review allows us to discuss safety and effectiveness across various OH preventative efforts while examining implementation. As other reviews have previously established, strong evidence exists to support the clinical efficacy and effectiveness of prophylactic use of uterotonics in LMICs (Tanna, 2013; Salati et al., 2019). This includes administration of oxytocin in clinical facilities (including preloaded injections) and of misoprostol in communities (including self-administered). Each uterotonic has demonstrated efficacy in the management of OH, although management of bleeding once it occurs depends on the cause. Although uterotonics target causes of OH due to tone, they will not resolve OH related to tears, tissue or thrombin; however, an estimated 80% of OH is due to atony (Likis Fe et al., 2015), which further supports the rationale to prioritize prevention. Patient-centredness and equity were not assessed in any of the studies, presenting a gap that future research could explore. The need to improve the implementation of evidence-based interventions for the primary and secondary prevention of OH has been a growing concern in the USA, also. Such attention could provide a means to address disparities in maternal health and health care, as reflected by the disproportionately higher rates of maternal mortality observed among minority women (Petersen et al., 2019). Implementation of evidence-based OH bundles in the USA has had mixed results, in part due to lower incidence of OH and associated difficulties attaining power for studies of rare events. However, there remains an opportunity to study these strategies in LMICs where the burden of OH is higher and well-powered studies are more feasible. In addition, lessons for OH prevention can be gleaned from efforts focused on the secondary prevention of other contributors to maternal morbidity and mortality. For example, studies on developing, validating and implementing tools for the risk assessment of pre-eclampsia in LMICs (von Dadelszen et al., 2012; Payne et al., 2014; Lim et al., 2015) suggest the importance of ensuring implementation feasibility, appropriateness and user acceptability in local contexts. This study highlights challenges in the implementation of OH prevention policies at various levels, including limited acceptability of some uterotonics among patients due to side effects or among health officials due to fears of misuse if implemented in community settings, difficulties with adoption by clinical providers, supply chain bottlenecks and human resource limitations. This is in line with previously reported considerations for implementing evidence-based practices for intrapartum care in LMICs (WHO, 2018b). In line with the WHO quality of care framework (WHO, 2016), healthcare systems and facilities in LMICs will need implementation research to guide the roll-out of appropriate OH prevention interventions. This may include providing clinical protocols for OH prevention and management, ensuring that facilities have uterotonics and related supplies (i.e. syringes, needles) and requiring provider training and annual refresher sessions. Presently, the know–do gap persists between what recommendations state is best practice and how policymakers and providers implement those recommendations. For example, the Federal Ministry of Health in Nigeria in its national guideline for the community use of misoprostol for the prevention and treatment of PPH identified roles for community health extension workers, community health officers and community-oriented resource persons including patent propriety medicine vendors, traditional birth attendants and teachers (Federal Ministry of Health, 2012). However, these roles are yet to be fully explored in implementation research for OH prevention. Implementation challenges at the policy level also remain, as policymakers may view community-based distribution of uterotonics as being in conflict with existing maternal health policies. Current Tanzanian guidance on PPH included in the BEmOC training manual developed in 2010 includes all three components of AMTSL, while the International Federation of Obstetrics and Gynaecology (Otolorin et al., 2015) and emerging WHO evidence (WHO, 2018a) recommends oxytocin administration for OH prevention and selective use of umbilical cord clamping, controlled traction of the cord and uterine massage. As some uterotonics, such as ergometrine, are contraindicated for anaemia and hypertension, common clinical conditions in Tanzania and many other LMICs, front-line healthcare providers must restrict the use until exclusion of these contraindications; however, they may receive conflicting messages from implementing partners, visiting obstetricians, volunteer doctors and research projects who are less knowledgeable on the context in a low-resource clinical setting where shortage of staff, frequent staff transition and maintenance of the cold chain for preservation of oxytocin efficacy may be a challenge. This demonstrates that the unique circumstances of context must be considered when developing and implementing guidelines and protocols of care. For this reason, future implementation research should examine OH prevention, including through collaborations between clinicians, policymakers and implementation scientists. As most of the existing studies occurred during the implementation period and were solely focused on adoption, rigourous evaluation of implementation is needed, including application of standardized measures of the various implementation outcomes. Indeed, recent efforts are being made to standardize reporting of implementation research outcomes, but many of these studies were carried out before these reporting guidelines were developed (Pinnock et al., 2017). This study has limitations. First, this systematic review is novel in its focus on prevention but is limited in that it is not targeting all OH prevention efforts, rather it includes interventions with a focus on implementation research outcomes. Our review only captures articles in which the authors themselves categorize the intervention as a prevention efforts; this limitation is likely to cause under-reporting of primary prevention, in particular. As we limited our search to articles written in English, there is potential for reporting bias. In addition, some studies were not included as they focused on the management of OH or did not explicitly identify the intervention as focused on prevention. For example, the innovative non-pneumatic anti-shock garment may prevent the severity of outcomes but does not prevent OH. Lastly, this review was restricted to databases of peer-reviewed journals and thus did not include a web search or grey literature. Nonetheless, the strengths of this study include its methodological approach. It was methodologically rigourous and controlled for selection bias through duplication of screening at each stage of the review. Additional strategies for rigour included the use of multidisciplinary expert debriefing. Conclusion The findings from this systematic review suggest that studies of evidence-based interventions to prevent OH, such as use of uterotonics, clinical guidelines and clinical provider skills training, highlight various challenges to implementation and scale up in LMICs. We highlight gaps in the literature, such as few studies on bundles that encourage best practices for OH management and prevention. A greater focus on the prevention of OH in LMICs is needed, particularly through the implementation of innovative programmes and policies guided by theory. A significant gap exists for secondary prevention efforts and for opportunities to integrate clinicians and policymakers into the programme and policy development and evaluation process. Addressing barriers to the implementation of interventions to prevent and treat OH is critical to achieving the targets of the Sustainable Development Goals Acknowledgements This research is supported in part by an NYU CTSA grant TL1 TR001447 from the National Center for Advancing Translational Sciences, National Institutes of Health. Supplementary data Supplementary data are available at Health Policy and Planning online. Conflict of interest statement. None declared. Ethical approval. No ethical approval was required for this study. References Ajeani J , Mangwi Ayiasi R , Tetui M et al. 2017 . A cascade model of mentorship for frontline health workers in rural health facilities in Eastern Uganda: processes, achievements and lessons . Global Health Action 10 : 1345497 . Google Scholar Crossref Search ADS PubMed WorldCat Anastacia Nkwonta C , Oyetunde MO. 2017 . Midwives utilization of life saving skills for prevention and management of hemorrhage in primary health centers in Anambra State, Nigeria . International Journal of Childbirth 7 : 87 – 100 . Google Scholar Crossref Search ADS WorldCat Bates I , Chapotera GK , Mckew S , van den Broek N. 2008 . Maternal mortality in sub-Saharan Africa: the contribution of ineffective blood transfusion services . BJOG 115 : 1331 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat Begley CMGG , Devane D , Mcguire W , Weeks A , Biesty LM. 2019 . Active versus expectant management for women in the third stage of labor . Cochrane Database Syst Rev 2 : CD007412 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Boucar M , Hill K , Coly A et al. ; USAID Applying Science to Strengthen Improve Systems University Research Co., LLC (USAID ASSIST-URC). 2014 . Improving postpartum care for mothers and newborns in Niger and Mali: a case study of an integrated maternal and newborn improvement programme . BJOG: An International Journal of Obstetrics & Gynaecology 121 : 127 – 33 . Google Scholar Crossref Search ADS WorldCat Braddick L , Tuckey V , Abbas Z et al. 2016 . A mixed-methods study of barriers and facilitators to the implementation of postpartum hemorrhage guidelines in Uganda . International Journal of Gynecology & Obstetrics 132 : 89 – 93 . Google Scholar Crossref Search ADS WorldCat Cannon M , Charyeva Z , Oguntunde O et al. 2017 . A case study of community-based distribution and use of misoprostol and chlorhexidine in Sokoto State, Nigeria . Global Public Health 12 : 1553 – 67 . Google Scholar Crossref Search ADS PubMed WorldCat Chang OH , Levy B , Lytle H et al. 2019 . Implementation of the alliance for innovation on maternal health program to reduce maternal mortality in Malawi . Obstetrics and Gynecology 133 : 507 – 14 . Google Scholar Crossref Search ADS PubMed WorldCat Chaudoir SR , Dugan AG , Barr CH. 2013 . Measuring factors affecting implementation of health innovations: a systematic review of structural, organizational, provider, patient, and innovation level measures . Implementation Science 8 : 22 . Google Scholar Crossref Search ADS PubMed WorldCat Critical Appraisal Skills Programme. 2020 a. CASP Cohort Study Checklist [Online]. Oxford, UK : CASP UK . https://casp-uk.net/casp-tools-checklists/, accessed October 2019. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Critical Appraisal Skills Programme. 2020 b. CASP Qualitative Checklist [Online]. Oxford, UK : CASP UK . https://casp-uk.net/casp-tools-checklists/, accessed October 2019. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Critical Appraisal Skills Programme. 2020 c. CASP Randomised Control Trial Checklist . Oxford, UK : CASP UK . https://casp-uk.net/casp-tools-checklists/, accessed October 2019. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Evans CL , Bazant E , Atukunda I et al. 2018 . Peer-assisted learning after onsite, low-dose, high-frequency training and practice on simulators to prevent and treat postpartum hemorrhage and neonatal asphyxia: a pragmatic trial in 12 districts in Uganda . PLoS One 13 : e0207909 . Google Scholar Crossref Search ADS PubMed WorldCat Evensen A , Anderson J , Fontaine P. 2017 . Postpartum hemorrhage: prevention and treatment . American Family Physician 7 : 442 – 9 . Google Scholar OpenURL Placeholder Text WorldCat Federal Ministry of Health. 2012 . National Guidelines for the Community Use of Misoprostol for the Prevention and Treatment of Postpartum Haemorrhage in Nigeria. Nigeria: Nigeria Federal Ministry of Health. https://www.health.gov.ng/. Gallos I , Williams H , Price M et al. 2019 . Uterotonic drugs to prevent postpartum haemorrhage: a network meta-analysis . Health Technology Assessment 23 : 1 – 356 . Google Scholar Crossref Search ADS PubMed WorldCat Garcia-Elorrio E , Aleman A , Cafferata ML et al. 2014 . A multifaceted intervention to increase prophylactic oxytocin use during the third stage of labor and to reduce routine episiotomies in Nicaragua . International Journal of Gynecology & Obstetrics 127 : 31 – 4 . Google Scholar Crossref Search ADS WorldCat Gaucher L , Occelli P , Deneux-Tharaux C et al. 2019 . Non-clinical interventions to prevent postpartum haemorrhage and improve its management: a systematic review . European Journal of Obstetrics & Gynecology and Reproductive Biology 240 : 300 – 9 . Google Scholar Crossref Search ADS WorldCat Goffman D , Ananth CV , Fleischer A et al. ; for the Safe Motherhood Initiative Obstetric Hemorrhage Work Group. 2019 . The New York state safe motherhood initiative: early impact of obstetric hemorrhage bundle implementation . American Journal of Perinatology 36 : 1344 – 50 . Google Scholar Crossref Search ADS PubMed WorldCat Institute of Medicine (US) Committee on Quality of Health Care in America. 2001 . Crossing the Quality Chasm: A New Health System for the 21st Century . Washington, DC : National Academies Press (US ). Google Scholar PubMed OpenURL Placeholder Text Google Preview WorldCat COPAC Kemp CG , Jarrett BA , Kwon C-S et al. 2019 . Implementation science and stigma reduction interventions in low- and middle-income countries: a systematic review . BMC Medicine 17 : 6 . Google Scholar Crossref Search ADS PubMed WorldCat Knight HE , Self A , Kennedy SH. 2013 . Why are women dying when they reach hospital on time? A systematic review of the ‘third delay’ . PLoS One 8 : e63846 . Google Scholar Crossref Search ADS PubMed WorldCat Lefebvre C , Glanville J , Briscoe S et al. 2019 . Chapter 4: Searching for and selecting studies. In: Higgins JPT , Thomas J , Chandler J , Cumpston M , Li T , Page MJ , Welch VA (eds). Cochrane Handbook for Systematic Reviews of Interventions Version 6.0 (Updated July 2019) . Cochrane . www.training.cochrane.org/handbook, accessed 4 March 2020. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Likis Fe SN , Morgans AK , Hartmann KE et al. 2015 . Management of postpartum hemorrhage. Agency for Healthcare Research and Quality (US), Rockville, MD. Report No.: 15-EHC013-EF. Lim J , Cloete G , Dunsmuir DT et al. 2015 . Usability and feasibility of PIERS on the move: an mHealth app for pre-eclampsia triage . JMIR Mhealth and Uhealth 3 : e37 . Google Scholar Crossref Search ADS PubMed WorldCat Mclintock C , James A. 2011 . Obstetric hemorrhage . Journal of Thrombosis and Haemostasis 9 : 1441 – 51 . Google Scholar Crossref Search ADS PubMed WorldCat Moher D , Liberati A , Tetzlaff J , Altman DG ; The PRISMA Group. 2009 . Preferred Reporting Items for Systematic Reviews and Meta-Analyses: the PRISMA statement . PLoS Medicine 6 : e1000097 . Google Scholar Crossref Search ADS PubMed WorldCat Moucheraud C , Gass J , Lipsitz S et al. 2015 . Bedside availability of prepared oxytocin and rapid administration after delivery to prevent postpartum hemorrhage: an observational study in Karnataka, India . Global Health: Science and Practice 3 : 300 – 4 . Google Scholar Crossref Search ADS WorldCat Orobaton N , Abdulazeez J , Abegunde D et al. 2017 . Implementing at-scale, community-based distribution of misoprostol tablets to mothers in the third stage of labor for the prevention of postpartum haemorrhage in Sokoto State, Nigeria: early results and lessons learned . PLos One 12 : e0170739 . Google Scholar Crossref Search ADS PubMed WorldCat Otolorin E , Gomez P , Currie S , Thapa K , Dao B. 2015 . Essential basic and emergency obstetric and newborn care: from education and training to service delivery and quality of care . International Journal of Gynecology & Obstetrics 130 : S46 – 53 . Google Scholar Crossref Search ADS WorldCat Payne BA , Hutcheon JA , Ansermino JM et al. ; MINIPIERS STUDY WORKING GROUP. 2014 . A risk prediction model for the assessment and triage of women with hypertensive disorders of pregnancy in low-resourced settings: the miniPIERS (Pre-eclampsia Integrated Estimate of RiSk) multi-country prospective cohort study . PLoS Medicine 11 : e1001589 . Google Scholar Crossref Search ADS PubMed WorldCat Petersen E , Davis N , Goodman D et al. 2019 . Racial/ethnic disparities in pregnancy-related deaths—United States, 2007–2016 . MMWR. Morbidity and Mortality Weekly Report 68 : 762 – 5 . Google Scholar Crossref Search ADS PubMed WorldCat Pinnock H , Barwick M , Carpenter CR et al. ; StaRI Group. 2017 . Standards for Reporting Implementation Studies (StaRI) statement . BMJ (Clinical Research ed.) 356 : i6795 . Google Scholar Crossref Search ADS PubMed WorldCat Prata N , Bell S , Weidert K. 2013 . Prevention of postpartum hemorrhage in low-resource settings: current perspectives . International Journal of Women's Health 5 : 737 – 52 . Google Scholar Crossref Search ADS PubMed WorldCat Proctor E , Silmere H , Raghavan R et al. 2011 . Outcomes for implementation research: conceptual distinctions, measurement challenges, and research agenda . Administration and Policy in Mental Health and Mental Health Services Research 38 : 65 – 76 . Google Scholar Crossref Search ADS PubMed WorldCat Proctor EK , Landsverk J , Aarons G et al. 2009 . Implementation research in mental health services: an emerging science with conceptual, methodological, and training challenges . Administration and Policy in Mental Health 36 : 24 – 34 . Google Scholar Crossref Search ADS PubMed WorldCat Salati JA , Leathersich SJ , Williams MJ , Cuthbert A , Tolosa JE. 2019 . Prophylactic oxytocin for the third stage of labour to prevent postpartum haemorrhage . Cochrane Database of Systematic Reviews 29 : CD001808 . Google Scholar OpenURL Placeholder Text WorldCat Say L , Chou D , Gemmill A et al. 2014 . Global causes of maternal death: a WHO systematic analysis . The Lancet Global Health 2 : e323 – 33 . Google Scholar Crossref Search ADS PubMed WorldCat Sheldon WR , Blum J , Vogel JP et al. 2014 . Postpartum haemorrhage management, risks, and maternal outcomes: findings from the World Health Organization Multicountry Survey on Maternal and Newborn Health . BJOG 121 : 5 – 13 . Google Scholar Crossref Search ADS PubMed WorldCat Simmons R , Shiffman J. 2007 . Scaling up health service innovations: a framework for action. In: Simmons R , Fajans P , Ghiron L (eds). Scaling Up Health Service Delivery: From Pilot Innovations to Policies Program . Geneva : WHO, 1–30 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Smith HJ , Colvin CJ , Richards E et al. 2016 . Programmes for advance distribution of misoprostol to prevent post-partum haemorrhage: a rapid literature review of factors affecting implementation . Health Policy and Planning 31 : 102 – 13 . Google Scholar Crossref Search ADS PubMed WorldCat Spangler SA , Gobezayehu AG , Getachew T , Sibley LM. 2014 . Interpretation of national policy regarding community-based use of misoprostol for postpartum hemorrhage prevention in Ethiopia: a tale of two regions . Journal of Midwifery & Women's Health 59 : S83 – 90 . Google Scholar Crossref Search ADS PubMed WorldCat Stanton C , Armbruster D , Knight R et al. 2009 . Use of active management of the third stage of labour in seven developing countries . Bulletin of the World Health Organization 87 : 207 – 15 . Google Scholar Crossref Search ADS PubMed WorldCat Tanna S. 2013 . Priority Medicines for Europe and the World “A Public Health Approach to Innovation” . Essential Medicines and Health Products Information Portal . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tiruneh GT , Yakob B , Ayele WM et al. 2019 . Effect of community-based distribution of misoprostol on facility delivery: a scoping review . BMC Pregnancy and Childbirth 19 : 404 . Google Scholar Crossref Search ADS PubMed WorldCat Verhey IJ , Ryan GK , Scherer N , Magidson JF. 2020 . Implementation outcomes of cognitive behavioural therapy delivered by non-specialists for common mental disorders and substance-use disorders in low- and middle-income countries: a systematic review . International Journal of Mental Health Systems 14 : 40 . Google Scholar Crossref Search ADS PubMed WorldCat von Dadelszen P , Firoz T , Donnay F et al. 2012 . Preeclampsia in low and middle income countries-health services lessons learned from the PRE-EMPT (PRE-Eclampsia-Eclampsia Monitoring, Prevention and Treatment) project . Journal of Obstetrics and Gynaecology Canada 34 : 1701 – 2163 . Google Scholar Crossref Search ADS WorldCat Weeks A , Ditai J , Ononge S et al. 2013 . Self-administered misoprostol to prevent bleeding after homebirths in Uganda: a placebo-controlled randomised trial . BJOG 120 : 76 . Google Scholar OpenURL Placeholder Text WorldCat WHO. 2000 . World Health Report 2000—Health Systems: Improving Performance . Geneva, Switzerland : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC WHO. 2016 . Standards for Improving Quality of Maternal and Newborn Care in Health Facilities . Geneva, Switzerland : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC WHO. 2018 a. WHO Recommendations for the Prevention and Treatment of Postpartum Haemorrhage . Geneva, Switzerland : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC WHO. 2018 b. WHO Recommendations: Intrapartum Care for a Positive Childbirth Experience . Geneva, Switzerland : World Health Organization . Google Scholar PubMed OpenURL Placeholder Text Google Preview WorldCat COPAC © The Author(s) 2020. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. All rights reserved. For permissions, please e-mail: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Young,, Ruth;Ayiasi, Richard, Mangwi;Shung-King,, Maylene;Morgan,, Rosemary

doi: 10.1093/heapol/czaa111pmid: 33179072

Intersectionality is a useful lens to understand how inequities in health develop and are experienced. Intersectionality is both a theory and methodological approach that demonstrates how overlaying social stratifiers (e.g. gender, ableness, sexual orientation and identity) can result in mutually enforced vulnerabilities that render some groups at an advantage and others at a disadvantage (Bowleg, 2008; Nixon, 2019). For example, as described in Crenshaw’s (1989) work, the experience of Black women is not wholly captured by being Black or a woman alone. Rather, the intersection of being both Black and a woman interacts in ways that result in a unique experience unable to be explained by either single identity (Crenshaw, 1989). This process is dynamic and advantage/disadvantage is not dichotomous but on a continuum (Larson et al., 2016). Even though equity is one of the driving aims for a health system, indicators for performance, such as infant mortality rate (World Health Organization, 2010), are often measured at the population level, masking inequities. Such aggregate level analyses do not tell us where inequities in health outcomes might exist, what the root causes are, or how to address root causes of inequity (Larson et al., 2016). We use an intersectional approach to explore how health systems can perpetuate systems and structures of oppression and describe how an intersectional-informed approach can be used to understand and highlight inequities in health and health systems. This topic is particularly pertinent, as the ongoing COVID-19 pandemic highlights how health systems can entrench and perpetuate broader societal power hierarchies, resulting in differential and inequitable health outcomes (Ahmed et al., 2020). This commentary serves to start a conversation on how intersectionality and health systems are linked through illustrative examples. Health systems are based within existing societal systems Decisions on how health systems are governed, financed and delivered are driven by established governance structures, which are rooted in historical and contemporary systems of oppression and power. For example, South Africa’s history of oppression of Black people during the apartheid era,1 coupled with a patriarchal society, resulted in restricting access to education and work opportunities (Shefer, 2010; Shung-King et al., 2018). The repercussions of these historical legacies of oppression continue into present day South Africa, where the confluence of gender, race and professional hierarchy still influence the nature of leadership and management in today’s health system (Shung-King et al., 2018). During apartheid, the career advancement and training for Black men, and even more so for Black women, were limited, contributing to poor representation in management positions, especially in senior eschelons (Digby, 2005; Coovadia et al., 2009). Primary research conducted by one of our authors in South Africa found that Black women had fewer senior management positions and opportunities for advancement (Shung-King et al., 2018). Women took on additional family demands as part of dual-career households, which led to vocational choices that better accommodated ‘family-friendly’ hours, and resulted in women taking on roles that were deemed lower in the professional hierarchy (Shung-King et al., 2018). In South African society at the time, women who wished to pursue a professional career were generally steered towards nursing or teaching, roles deemed befitting of women (Nkabinde, 1997). Men were more likely to train as doctors and, consequently, were more likely to progress faster and further into management positions (Shung-King et al., 2018). Women who entered nursing were appointed to management positions only more recently, as these positions were traditionally occupied by doctors, the majority of whom were men (Shung-King et al., 2018). These historical barriers to advancement, coupled with gender norms, result in fewer Black women holding health management (and even fewer senior) positions (Shung-King et al., 2018). This example demonstrates how understanding the intersectional drivers of leadership and representation in health systems can show how health systems are based within existing societal structures and as a result reflect and perpetuate existing systems of oppression and power. Health information systems reflect and perpetuate existing systems of oppression What is measured reflects what is implicitly valued or perceived to be the biggest priorities in a health system. Increasingly, datasets (e.g. Demographic Health Surveys) are being disaggregated by specific variables such as sex, age, location or wealth quintile. While this allows us to see variation between groups, it does not show us how groups may be disadvantaged as a result of the intersection of stratifiers. In addition, when data are disaggregated it is often presented per variable, looking at differences between men and women or between different age groups, as opposed to differences among women or among men by different age groups. By not exploring within group differences, such datasets risk treating everyone’s lived experiences as if they were the same. Often, if something is not measured, it is presumed to not exist. Such data gaps can mask health inequities and needs experienced by specific groups, often characterized by multiple, intersecting social stratifiers, rendering them invisible (Bauer, 2014). ‘Invisible populations’ may be less likely to access services they may need due to a range of reasons, such as inappropriate service design, little awareness of services provided and feeling excluded when trying access services. In Uganda, for example, the sexual and reproductive health needs of women with physical disabilities were left unaddressed (Apolot et al., 2019). Women with physical disabilities were perceived as not having the same reproductive and sexual needs as women without walking disabilities as they were often perceived to be asexual and/or unable to look after offspring (Apolot et al., 2019). Without data describing the sexual and reproductive health needs of women with physical disabilities, preconceptions of their health needs prevail and services are not designed to meet their needs. Explicit recognition of potentially marginalized groups through data can directly inform funding, design and implementation; these should include a way to measure if specific needs are being met and if outcomes improve over time (Gruskin et al., 2017). As data inform service design and policy, if health systems are going to recognize and meet the needs of the most vulnerable and marginalized—rendering the invisible visible—the experiences of key groups must be incorporated within data collection tools and activities, data needs to be disaggregated by sex and other biological and social stratifiers (and remain disaggregated within reporting), and intersections between these variables must be explored and reported. Health service delivery excludes marginalized groups Different groups experience different barriers to accessing and utilizing health services. In addition, the root causes of these barriers may differ. For example, recent research by one of our authors in Masindi, Uganda, found that immigrant women perceived or experienced additional barriers to accessing maternal and newborn care compared to nonimmigrant women (Ayiasi et al., 2020, unpublished data). Unpublished findings from a precursor study (Ayiasi et al., 2016) of pregnant immigrant women in Masindi and Kiryandongo districts indicated that, compared to local resident populations, immigrant women were less likely to seek newborn vaccination services [adjusted odds ratio (aOR) = 4.80(3.31–6.93)], or institutionalized delivery [aOR = 2.43(1.87–3.14)] (Ayiasi, personal communication). Follow-up qualitative studies on pregnant immigrants in Masindi found that women perceived discrimination at health facilities, which they attributed to their inability to speak the local dialect, or their names that signalled they were migrants (Ayiasi et al., 2020, unpublished data). Immigrant women were likely to delay maternal care during delivery due to potential discrimination by health workers, cultural norms that perceive women as ‘strong’ if they did not complain during labour pains, and belief that delivery is a woman-only issue (Ayiasi et al., 2020, unpublished data). In this case, the confluence of being seen as an outsider when seeking health care, and gender norms that prioritize stoicism in women even during childbirth reduced access to care and changed care-seeking behaviour (Ayiasi et al., 2020, unpublished data). These findings demonstrate the value of using an intersectional lens to understand how overlaying social stratifiers can elevate or oppress individual health (Nixon, 2019). By considering the unique needs of migrant women within maternal health care, for example, services can be redesigned to promote health among this at-risk population, making health delivery more equitable (Kapilashrami and Hankivsky, 2018). What does an intersectional lens add to health systems? Applying an intersectional lens to health systems can provide a more nuanced and robust way to examine health inequities. Failing to use an intersectional lens can result in a superficial analysis of inequity in health outcomes, focused on the outcomes rather than the entrenched and root causes of the inequity. Acknowledging that health systems reflect broader multifaceted oppression can help policymakers and planners take appropriate steps to help mitigate the impact of entrenched inequity. Researchers should recognize that health systems are products of their societies and acknowledge the broader sociocultural and historical contexts that they operate within. Intentionally examining power structures that the system operates within may highlight appropriate disaggregation of data and point to reasons why certain health inequities may persist (Gruskin et al., 2017). An intersectional approach can be adopted throughout the research process, from actively engaging communities in research design to designing data collection systems that measure multiple social stratifiers, and to communicating differences in health outcomes or experiences within groups. The aforementioned ways to incorporate intersectionality into health systems research are not complete but rather the start of a discussion on ways to understand how health systems can perpetuate and challenge systems of oppression. Applying an intersectional lens is not just theory or methodology but a fundamental step to dismantling systems of oppression both in the health sector and beyond. Footnotes 1 Apartheid was a ‘legalized’ system of extreme oppression based on race and ethnicity, which resulted in the majority of the population, Black South Africans, experiencing harsh social, economic and political discrimination and disadvantage, the repercussions of which continue today (Kon and Lackan, 2008). Conflict of interest statement. None declared. Ethical approval. None required. References Ahmed F , Ahmed N , Pissarides C , Stiglitz J. 2020 . Why inequality could spread COVID-19 . The Lancet Public Health 5 : e240 . Google Scholar Crossref Search ADS PubMed WorldCat Apolot RR , Ekirapa E , Waldman L et al. 2019 . Maternal and newborn health needs for women with walking disabilities; “the twists and turns”: a case study in Kibuku District Uganda . International Journal for Equity in Health 18 : 43 . Google Scholar Crossref Search ADS PubMed WorldCat Ayiasi RM , Kolsteren P , Batwala V , Criel B , Orach CG. 2016 . Effect of village health team home visits and mobile phone consultations on maternal and newborn care practices in Masindi and Kiryandongo, Uganda: a community-intervention trial . PLoS One 11 : e0153051 . Google Scholar Crossref Search ADS PubMed WorldCat Ayiasi RM , Ochola Mangwi AJ , Garimoi Orach C. 2020 . Ethnicity, gender and migration status: applying intersectionality methodology to explore barriers to equitable health systems for maternal and newborn health among immigrant populations in Masindi, Uganda, unpublished. Bauer GR. 2014 . Incorporating intersectionality theory into population health research methodology: challenges and the potential to advance health equity . Social Science & Medicine 110 : 10 – 7 . Google Scholar Crossref Search ADS WorldCat Bowleg L. 2008 . When black + lesbian + woman ≠ black lesbian woman: the methodological challenges of qualitative and quantitative intersectionality research . Sex Roles 59 : 312 – 25 . Google Scholar Crossref Search ADS WorldCat Coovadia H , Jewkes R , Barron P , Sanders D , McIntyre D. 2009 . The health and health system of South Africa: historical roots of current public health challenges . The Lancet 374 : 817 – 34 . Google Scholar Crossref Search ADS WorldCat Crenshaw K. 1989 . Demarginalizing the intersection of race and sex: a black feminist critique of antidiscrimination doctrine, feminist theory and antiracist politics . University of Chicago Legal Forum 1989 : 139 – 67 . Google Scholar OpenURL Placeholder Text WorldCat Digby A. 2005 . Early black doctors in South Africa . The Journal of African History 46 : 427 – 54 . Google Scholar Crossref Search ADS WorldCat Gruskin S , Ferguson L , Kumar S et al. 2017 . A novel methodology for strengthening human rights based monitoring in public health: family planning indicators as an illustrative example . PLoS One 12 : e0186330 . Google Scholar Crossref Search ADS PubMed WorldCat Kapilashrami A , Hankivsky O. 2018 . Intersectionality and why it matters to global health . The Lancet 391 : 2589 – 91 . Google Scholar Crossref Search ADS WorldCat Kon Z R , Lackan N. 2008 . Ethnic Disparities in Access to Care in Post-Apartheid South Africa . American Journal of Public Health 98 : 2272 – 7 . 10.2105/AJPH.2007.127829 Google Scholar Crossref Search ADS PubMed WorldCat Larson E , George A , Morgan R , Poteat T. 2016 . 10 Best resources on… intersectionality with an emphasis on low- and middle-income countries . Health Policy and Planning 31 : 964 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat Nixon SA. 2019 . The coin model of privilege and critical allyship: implications for health . BMC Public Health 19 : 1637 . Google Scholar Crossref Search ADS PubMed WorldCat Nkabinde ZP. 1997 . An Analysis of Educational Challenges in the New South Africa . Lanham: University Press of America . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Shefer T. 2010 . Narrating gender and sex in and through apartheid divides . South African Journal of Psychology 40 : 382 – 95 . Google Scholar Crossref Search ADS WorldCat Shung-King M , Gilson L , Mbachu C et al. 2018 . Leadership experiences and practices of South African health managers: what is the influence of gender?—a qualitative, exploratory study . International Journal for Equity in Health 17 : 148 . Google Scholar Crossref Search ADS PubMed WorldCat World Health Organization. 2010 . Monitoring the building blocks of health systems: a handbook of indicators and their measurement strategies. Geneva, Switzerland: World Health Organization. http://www.who.int/healthinfo/systems/WHO_MBHSS_2010_full_web.pdf, accessed 26 August 2020. © The Author(s) 2020. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. All rights reserved. For permissions, please e-mail: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Gilson,, Lucy;Marchal,, Bruno;Ayepong,, Irene;Barasa,, Edwine;Dossou,, Jean-Paul;George,, Asha;Guinaran,, Ryan;Maceira,, Daniel;Molyneux,, Sassy;Prashanth, N, S;Schneider,, Helen;Shawar,, Yusra;Shiffman, Jeremy, R;Sheikh,, Kabir;Spicer,, Neil;, Van Belle, Sara;Whyle,, Eleanor

doi: 10.1093/heapol/czaa112pmid: 33020815

Introduction To say that we live in turbulent times is a massive understatement. COVID-19 ruthlessly exposes the fault lines of health services and systems, and the responses put in place to prevent its spread or mitigate its effects may affect people more than the actual infection. The outbreak in Wuhan quickly grew to a pandemic that has affected countries and regions all over the world in many, and as of yet, little understood ways. This is a global infectious disease outbreak of a scale not seen since the Spanish Flu. For many countries, it is an extreme stress test of the health system and of society at large. All over the world, people, patients, providers, health service managers, health and other sectoral policymakers and politicians, are dealing with high levels of uncertainty and severe challenges to the resilience of their systems. The governance not only of health, at national and global levels, but also of trade, communication and globalization itself is under scrutiny. The virus exposes, yet again, the structural determinants that lead to health inequalities (Shadmi et al., 2020), including racism and colonial legacies. Many see this as a key moment of reckoning, nationally and globally: the pandemic and its responses have precipitated unprecedented economic, social and health crises that may shape the decades ahead. At the same time, the role of health systems in responding to COVID-19 and the need to (re-)invest in these systems through the state offers transformative opportunities. In the light of this, we outline how health policy and systems research (HPSR) can both address current short-term challenges, and support the system transformations needed to strengthen people-centred and equitable health systems over the long term. The HPSR community has responded to the COVID-19 pandemic quickly, following the wave of publications on epidemiological and clinical aspects of the disease. Initial studies have included those describing the capacity of hospitals, intensive care units and first line health services required to respond to the disease, and those reporting specific experiences at community and local levels, including the denial of care and the inequitable effects of disease control measures. Many commentaries and calls for action have been published (COVID-19 Clinical Research Coalition, 2020; English et al., 2020; Shamasunder et al., 2020). Inevitably, however, due to the acute nature of the crisis, few papers have yet focused on how health systems are coping with or adapting to the pandemic, or how health policy-making and decision-making has (or has not) changed in this time of crisis. Yet, there is an urgent need to develop a structured research agenda to inform health policy and system responses to COVID-19 that can move us beyond the current crisis, and into the future. This commentary makes proposals towards such an agenda. In line with the audience of Health Policy and Planning, we specifically focus on low- and middle-income country (LMIC) HPSR needs, drawing on our collective experience as a group of HPS researchers based around the world. The Health System Research and Health Policy Processes section editors initiated the process and purposefully sought inputs from HPS researchers in a range of LMICs . The process was also supported by both the Alliance for Health Policy and Systems Research and Health System Global. We organized an online consultation process, whereby the first authors invited the co-authors to identify research priorities, questions and themes. In a second round, these were compiled and categorized in themes and sent out for further comment. In a final round, the issues and questions within each theme were examined and gaps and overlaps eliminated. We consider, first, key dimensions of the overall approach of HPSR to frame the further work needed, and second, in an annex, we suggest an initial categorization and listing of possible research topics. We present these ideas to prompt wider reflection—and we conclude by proposing ways of engaging further with these ideas, acknowledging the fast-changing nature of the pandemic and the need to review research priorities regularly. HPSR lenses and areas of focus We start by reviewing how the defining features of HPSR, including the systems approach, multi-disciplinarity and the emphasis on policy and power may be applied to the pandemic. One defining feature of HPSR is its systems approach. This frames COVID-19 and the responses to it in a critical–analytical perspective, zooming out from specific experiences to seek the root causes of the differential impact of the pandemic across individuals and population groups as a function of society’s power structures and dominant cultures—as reflected in their social, political and economic position, and their race, gender, caste, class and more. Groups that become vulnerable due to systemic and structural inequities include those living in informal settlements in cities or in geographically isolated areas, informal workers, migrant and refugee communities, people without citizenship rights, sex workers, single-women households, LGBTQI+ communities and indigenous peoples. The systems lens equips HPS researchers to understand how, for such groups, health system fault lines interact with the histories of discrimination and disenfranchisement that underpin other determinants of vulnerability, risks and infection. Some people have already faced worse effects from COVID-19 responses than from (the possibility of) infection itself, due to social exclusion, racism and human rights abuse. Meanwhile, privilege and social networks have largely allowed wealthy elites to avoid the negative social and economic effects of lockdowns, deepening inequalities. Health systems have, in turn, generally been poorly equipped to respond fast enough either to the medical and psychological demands of the pandemic or to the wider public and social action needed to address multiple disadvantage and vulnerability. HPSR is well placed to contribute in understanding the layered causes and effects of the pandemic on people and systems. Identifying how health system gaps and weaknesses interact with the root causes of vulnerability to COVID-19 is a vital responsibility that HPSR needs to shoulder—not only as an obligation of itself (referring to HPSR’s social justice agenda), but also as an opportunity to strengthen health systems in more just and inclusive ways than before. The multi-disciplinary approach of HPSR will be key to this task because of the complex nature of health and its determinants, as well as of health policies and health systems. This multi-disciplinary approach supports consideration of how agents and systems interact and of the inter-connections among the systems relevant to health. These include community, workplace and government systems; health and other government sectors; local, national and global systems and markets, corporate actors and the private sector in its full heterogeneity. HPSR can assist in distinguishing COVID-19-related challenges that are simple problems, from those that are better considered as complex problems and that demand appropriate context-sensitive response strategies. It can also nurture and sustain the systemic responses to COVID-19 and its impacts that are vital for the long-term. It supports a system-based approach in anticipating the collateral effects of policies aimed at responding to the pandemic, generating ideas about how mitigate the damages and optimize the gains. The pandemic specifically demands recognition of the interactions of human, animal and ecological systems. The spill-over of the virus from animal to human reservoirs inevitably calls for a critical exploration of how humans continue to interfere with fragile equilibria in the natural world through urbanization, deforestation and more. Beyond ‘one health’ research, attention should be paid to the dominant economic growth paradigm on health, development, trade, society and the natural world and the role of populist and isolationist ideologies in framing the response. Another defining feature of HPSR is its focus on policy, policy-making and how health systems are nested in and influenced by power dynamics and political forces, histories and cultures: the ‘p’ in HPSR. This is another vital lens for research in the time of COVID-19. It supports inquiry into governance, decision-making and health policy responses in times of crisis, as well as how both to inform and strengthen system change. It encourages consideration of the values driving decision-making and the ethical demands of leadership. In addition, this lens underpins investigation of the political economy of the pandemic response and whether and how health systems and political action align to address the structural determinants of ill health and inequity which COVID-19 exposes. It can expose the lack of accountability towards some marginalized groups and the focus on politically important constituencies, the lack of stewardship/leadership at multiple levels and how efforts towards decentralizing and commercializing health system responsibilities lead to fragmented health systems. It could focus on new forms of sub-national and national governance arrangements and investigate how that deepens community-level and inter-sectoral action for health and societal development. The pandemic forces us to consider governance not only at national level, but also at global level: Should we redefine global health? How can we promote better global leadership to coordinate and enforce efforts across countries, including the need for consideration of global public goods and global health ethics? Drawing on well-established bodies of knowledge and evidence from other disciplines (e.g. political science, policy studies, public administration, sociology, complexity theory, critical studies), HPSR can make major contributions to learning how better to deal with pandemics. But HPSR must also do more than help systems absorb future shocks. It must help establish the foundations of more just, equitable and better health systems—health systems that demonstrate resilience through their capacity to be transformative as they respond to shock and stress. Crucial in this effort will be research around how health systems can be transformed for the better during or in the wake of the pandemic. This includes work around adaptive governance as well as on other fundamental system drivers—such as the health workforce, and information and financing systems. Prioritizing research topics Prioritizing among research topics is not straightforward. Such prioritization must, first, be informed by the views of vulnerable groups, community groups, health system decision-makers and health policymakers in the broadest sense and be situated in specific settings. Below, we discuss a few waypoints that may help in thinking about HPSR priorities in relation to COVID-19. As Marquette (2020) has argued for social science more generally, more immediate needs for research are likely to include understanding the root causes of vulnerability to support response activities and limit their negative effects. However, she notes that even mid- to longer-term research to understand secondary effects and long-term impacts and recovery must start now, accompanied by political analysis, and this will be needed to sustain recovery and support the emergence of new and better systems. Critical to this work will be revealing opportunity, agency and resilience, even in the midst of multi-layered challenges. For HPSR, we can also be guided by our understanding of people-centred health systems. This points, e.g. to the importance of understanding community-level COVID-19 experiences, including the experience of marginalized groups, and considering how system software (including power, trust and values) interacts with other system changes to influence the impacts of COVID-19; examining how health systems and COVID-19 responses may exclude people on the basis of their gender, race, income and other characteristics, and how histories of colonialism and racism underpin such exclusion; analysing the power and influence of ideas and framing, and the role of communication in decision-making at every level; and finally, purposefully considering the political economy influences driving COVID-19 and responses to it . Our research must offer new ideas for future health systems—building evidence around new ways of organizing, new ways of caring, new strategies of health development. Ways of doing and being In responding to COVID-19 and offering new ideas for future health systems, the HPSR community must also consider how to go about doing research. Issues to be reflected upon include the distribution of power within HPSR communities, the balance between global concerns and priorities on one hand and context-sensitivity on the other hand, the challenge of researching what is essentially a fast moving target, the practical problems induced by the control measures (such as social distancing) when collecting data, and finally the issue of research governance. Power-balancing strategies must be reflected in how the research is done—considering, e.g. with whom and how we collaborate, and what forms of knowledge are valued and enhanced through this work. Research responding to COVID-19 must also be relevant to the contexts in which it is located, and acknowledge the imperatives of this moment. In the short term, then, it must be conducted quickly to address immediate needs, and be fed-back into decision-making rapidly. Innovative knowledge translation efforts and new models of collaborations between research, policymaking and stakeholder organizations are of special interest. Being systematic and rigorous will always be important, but we must capture current experience even as we also develop longer-term research activities. To support such work, COVID-related HPSR can build on and deepen new ways of doing research. These include embedded research approaches (Olivier et al., 2017), participatory action research (Loewenson et al., 2014), action learning processes (The RESYST/DIAHLS Learning Site Team, 2020), insider research and prospective policy analysis and deliberative processes (Buse, 2008). All value multiple forms of knowledge and provide opportunities for researchers to learn through partnership with others in supporting the COVID response in real time. They represent ways of working that open up opportunities for HPSR to help shape the world after COVID-19. At the same time, in assessing the response to COVID-19 it always remains important to consider the ethical dimensions and relevance of ethical principles and methods. In the short-term, we will also need to think about research methods and data collection approaches that can be adapted to physical distancing and still engage vulnerable groups (Samuels, 2019), building—among others—on opportunities offered by online survey tools and online interviews. Learning from people’s own experiences, as they report them, their own stories, is essential, and social media could be a relevant source. Arts-based research forms offer other new ways of gathering and sharing such experience (Brady et al., 2019). Reflecting other initiatives,1 an online repository of methods, tools and exemplar studies could be developed to support the HPSR community. Analysis of secondary data and simple survey work can also bring important, immediate insights about social lockdown and contact tracing among other experiences. Wider evidence synthesis, drawing on both qualitative and quantitative data and evidence, is needed. Stretching into the longer-term, longitudinal and process tracing research (Mahoney, 2015) will be important to tease out the patterns of experience. Historical analysis is also needed both to understand the current experience and draw lessons from similar, past experience (Olivier, 2019). Ethnographies, autoethnographies and detailed case study work can offer rich insights into the experience of particular groups and into particular responses to inform future action (Mathew, 2020). The application of multi-level methodologies will support understanding of system complexity, whilst purposeful sub-national or cross-national analyses can throw light, e.g. on critical governance influences over responses. Intersectional analyses of experience will also always be important to understand the layered vulnerability of people and systems, and the privilege, agency and ability to act, and must become a hallmark of HPSR (Larson et al., 2016). Finally, it is critical now to set up and transform research governance processes at all levels to ensure important, ethical, timely research that supports social justice is possible and is supported, and to identify and limit potentially harmful research. We must also track if and how research governance processes, and the research that is undertaken, are themselves impacted by COVID-19 and responses to it. Who produces the research of today and is credited for it is part of this impact: will journals ensure not only high-quality work but also papers led by women and led by LMIC researchers? At another level, how can we contribute to decolonize global health (research)? Who speaks for whom in the current research on COVID-19? Conclusion COVID-19 challenges HPS researchers to think and act differently in producing research that contributes to creating new health systems fit for the future. HPSR that addresses current needs and experiences will offer important insights and understandings both about how COVID-19 impacts on our societies, and about the responses to the pandemic. It can help interrogate how these responses exacerbate existing vulnerabilities and fragilities or open up new opportunities for strengthening socially just health systems. The HPSR done now in LMICs simply must support long-overdue shifts in the place and capacity for people- and nationally driven, responsive research, with appropriate international engagement. Health and research institutions in LMICs must play leading roles, and, as needed, be strengthened to do so. Research alliances for change must also include communities, patients, providers, health service managers and health and other policy makers, in addition to civil society and media. De-colonizing HPSR must be an ongoing imperative, as it is for global health more broadly, and these are steps in that process. Recognizing the current moment, a new COVID-19 stream has just been announced for the sixth Global Symposium on Health Systems Research , the bi-annual gathering of the HPSR community. Abstracts are welcomed within its overarching theme of ‘re-imagining health systems for better health and social justice’ and for any of the three sub-themes: Engaging political forces; Engaging social, economic and environmental forces and Engaging technological, data and social innovation. In the annex, we also present a set of emerging HPSR themes and topics addressing the areas of importance we have already flagged. These range from studies investigating the differential impact on people, disadvantage and agency, to those focused on the impacts on health systems, including studies on governance and system level levers of health system change towards better and socially just health systems. We welcome responses to our ideas and proposals for a COVID-provoked HPSR agenda, in the form of blogs to be published in Health Policy and Planning Debated. We also envisage a future call for papers for a 2021 special issue of this journal that may include: Papers outlining and assessing the COVID-19 pandemic and the responses developed at global, national, sub-national and ‘district’-level, as well as emergent responses by people, communities organizations, health and social providers, decision-makers including community and religious leaders, NGOs, etc.—preferably from a forgotten/neglected perspective; Papers analysing COVID-19 responses in terms of health and social impact, including equity and intersectional analyses; Papers providing insights on the innerworkings of decision-making in a time of crisis; Papers presenting methodologies to document the emerging impact of the pandemic and the responses developed at global, national, sub-national and ‘district’-level through e.g. multi-level methodologies, ethnography, comparative studies, policy and decision-making analysis, etc. Footnotes 1 Doing Fieldwork in a Pandemic—crowdsourced document of resources: https://docs.google.com/document/d/1clGjGABB2h2qbduTgfqribHmog9B6P0NvMgVuiHZCl8/edit?ts=5e88ae0a; No Panic in the #Phdpandemic: curated links to helpful resources and strategies—https://www.virtualnotviral.com; Overseas Development Institute live repository—https://www.odi.org/publications/16977-primary-data-collection-covid-19-era Acknowledgement Conflict of interest statement. Kabir Sheikh is a staff member of the World Health Organization and is himself alone responsible for the views expressed in the article, which do not necessarily represent the views, decisions or policies of the World Health Organization. Ethical approval. No ethical approval was required for this study. References Brady L , De Vries S , Gallow R et al. 2019 . Paramedics, poetry, and film: health policy and systems research at the intersection of theory, art, and practice . Human Resources for Health 17 : 64 . Google Scholar Crossref Search ADS PubMed WorldCat Buse K. 2008 . Addressing the theoretical, practical and ethical challenges inherent in prospective health policy analysis . Health Policy and Planning 23 : 351 – 60 . Google Scholar Crossref Search ADS PubMed WorldCat COVID-19 Clinical Research Coalition. 2020 . Global coalition to accelerate COVID-19 clinical research in resource-limited settings' . The Lancet 395 : 1322 – 5 . Crossref Search ADS WorldCat English M , Moshabela M , Nzinga J et al. 2020 . Systems and implementation science should be part of the COVID-19 response in low resource settings' . BMC Medicine 18 . Google Scholar OpenURL Placeholder Text WorldCat Larson E , George A , Morgan R , Poteat T. 2016 . 10 Best resources on… intersectionality with an emphasis on low-and middle-income countries . Health Policy and Planning 31 : 964 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat Loewenson R , Laurell AC , Hogstedt C , D'Ambruoso L , Shroff Z. 2014 . Participatory Action Research in Health Systems: A Methods Reader . Harare: TARSC, AHPSR, WHO, IDRC Canada, Equinet . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mahoney J. 2015 . Process tracing and historical explanation . Security Studies 24 : 200 – 18 . Google Scholar Crossref Search ADS WorldCat Marquette H. 2020 . On COVID-19 Social Science Can Save Lives: Where do We Start? Blog post. https://oxfamblogs.org/fp2p/on-covid-19-social-science-can-save-lives-where-do-we-start/ Mathew B. 2020 . Engagement and Responsibility: Ethical Challenges of Doing an Ethnography on the Streets of Delhi. Blog post. https://www.internationalhealthpolicies.org/featured-article/engagement-and-responsibility-ethical-challenges-of-doing-an-ethnography-on-the-streets-of-delhi/ Olivier J. 2019 . LMIC Health Systems in Danger of Losing Their Own History. https://www.internationalhealthpolicies.org/featured-article/lmic-health-systems-in-danger-of-losing-their-own-history/l Olivier J , Scott V , Molosiwa D , Gilson L. 2017 . Embedded systems approaches to health policy and systems research. In: De Savigny D , Blanchet K , Adam T (eds). Applied Systems Thinking for Health Systems Research: A Methodological Handbook, Chapter 2 . London : Open University Press , pp. 14 – 52 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Samuels F. 2019 . Tips for Collecting Primary Data in a COVID-19 Era . London : Overseas Development Institute . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Shadmi E , Chen Y , Dourado I et al. 2020 . Health equity and COVID-19: global perspectives . International Journal for Equity in Health 19 : 104 . Google Scholar Crossref Search ADS PubMed WorldCat Shamasunder S , Holmes SM , Goronga T et al. 2020 . COVID-19 reveals weak health systems by design: why we must re-make global health in this historic moment . Global Public Health 15 : 1083 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat The RESYST/DIAHLS Learning Site Team. 2020 . Learning sites for health system governance in Kenya and South Africa: reflecting on our experience . Health Research Policy and Systems 18 : 1 – 12 . Crossref Search ADS PubMed WorldCat Annex—Towards an HPSR agenda for COVID-19 and beyond: themes, topics and questions Studies assessing the differential impact of COVID-19 and the response on disadvantaged groups and identifying the root causes of vulnerability What is the effect of the pandemic and the response on the availability and utilization of first line services for and by vulnerable groups (incl. people with chronic diseases and the elderly), reproductive and sexual health services, response services for gender-based violence and LGBTQI+ service provision? What is the differential impact of COVID-19 on livelihoods and how does it affect the health of poor people, and low-income wage earners, people who cannot work from home, etc. What are the gendered impacts of COVID-19 containment measures? How to advance sex disaggregation in surveillance and monitoring systems to understand the gender implications of COVID-19? How to adapt COVID-19 containment measures to reduce undesired impacts on vulnerable groups, such as those living in informal settlements, refugee camps, the elderly and the disabled? What are the drivers of stigmatization and how can stigmatization of those who test positive for COVID-19 and those who have recovered from COVID-19 (including health workers) be offset? How (if in any way) is stigmatization linked to trust and mis-trust of government/the official response to a disease outbreak? Studies exploring power dynamics and political forces, histories and cultures, the p in HPSR How does power play out in the governance of COVID-19 responses and within new forms of governance and organization that emerge? How do pre-existing power relationships and public-private identities condition the COVID-19 response? Which and whose evidence and knowledge is being used in decision-making around COVID-19? What is the balance of political and technical/bureaucratic leadership, and how do governance histories and structures shape this balance? Whether and how bureaucratic and organizational cultures resist or limit necessary change over the longer-term, and how to address this constraint? Is COVID-19 changing global health agendas (and donor funding in LMICs), e.g. more emphasis on global health security and away from universal health coverage? What will COVID-19 change in the global political economy of health? Is COVID-19 changing attitudes to multilateralism and collective approaches to addressing global health problems? Is it leading to entrenchment of bilateral approaches and coloniality? Which knowledge and whose voices are influencing COVID-19 responses, how does this differ between countries, and what is influencing this? Health system governance and decision-making at national and global levels How is cross-sectoral coordination within government and with private agents impacted and enabled, and what are longer-term gains/losses for action on the social determinants of health? Which experts are invited to support the response, using what evidence and how is that changing and contested over time? What histories and structures shape these experiences? Who is held responsible for what and who is invited as an expert and who is not? Is there policy coordination or confusion through rapid policy change in COVID-19 responses? What is the role and influence of bottom-up flows of experience for decision-making, including personal narratives and stories from communities and frontline health workers? What is the role of mis-information and fake news, and the impact on communication of the politicization of information? What is the practice of policy communication in a time of crisis, including the framing of messages, the role of scientific leaders/advisors (‘chief scientists’) and scientific differences and controversy, and the consequences for trust in COVID-19 responses and in government? What innovative policies or programmes that support or protect vulnerable populations in COVID-19 responses have been implemented, and what decision-making dynamics led to the prioritization of these groups? What is the impact on the role of WHO in global health governance? How is geopolitics shaping the global response? What new forms of global leadership are emerging that can assist in coordinating and enforcing efforts across countries, including the need for consideration of global public goods and global health ethics? Studies exploring how (local) health systems are impacted by and support COVID-19 responses System-wide effects What is the impact of national/global COVID-19 responses on attention to and implementation of solutions to other pressing health issues, including onehealth approaches? What crowding-out effects are occurring and how can these be mitigated? How can critical non-COVID-19 services be maintained alongside the COVID-19 response in resource-poor settings? What is the impact of COVID-19 on the physical and mental health of the health workforce, and of rapid change in ways of working? What is the impact of COVID-19 on the position of (frontline) health workers in the political/social arena in countries (power, gender-related issues, work conditions, etc.)? What is the role of community-based services, including Community Health Workers, in the COVID-19 response? System responses How effective is e-learning/online learning for health human resources? How do COVID-19 responses impact on the interaction of system hardware and software, and with what consequences for system functionality? How is the national–sub-national coordination happening in decentralized systems (pooling, division of tasks, oversight, coordination)? How prepared are local and national governments for a pandemic as opposed to natural disasters? Is the COVID-19 pandemic challenging our views on resilience? What is the impact on changes in global financing patterns and flows in national health systems? What are the changes in funding and global investment in health, in health security, in health system strengthening? What is the role of public finance management in the pandemic response? What are the successes and failures of global pandemic financing mechanisms? To note: whilst HPSR particularly emphasizes the system level issues, we include in this category some issues related to the service-level, emergency preparedness and logistics that are currently important. Service-level responses What are the interfaces and gaps between levels of care (community/primary care level/hospital) during the response? Are preventive services recentralized during COVID? What are the role private sector actors in service responses? How can an optimal balance of care be maintained within the health system under pressure? How to strengthen workplace safety/occupational health for the health workforce? How to support/care for the carers, acknowledging anxiety, stress, fear? Public health surveillance and emergency preparedness What is the place of public health surveillance within the wider health system, and how can it be strengthened? What is the role of national health institutes in managing the COVID-19 response within LMICs? What is the place of laboratories in the health system and in the COVID-19 response? The issues of biosafety, infection prevention and control mechanisms in the rush to build and accredit laboratories What is the effectiveness of health information systems in contributing to COVID-19 responses (e.g. Trackers, GIS mapping)? Can mathematical models and infectious disease models be improved with HPSR insights, such as by integrating health systems factors and questing the dominant assumptions underlying models? What are the ethical pros and cons of community based or bottom up surveillance processes? Technology, equipment and therapy needs for the COVID response The equipment, infrastructure, etc. required to implement an effective COVID-19 response, and the challenges of inadequate preparation and supply lines The challenges of procurement and distribution of medical equipment in the light of global and local supply chains of equipment and supplies The spike in demand for telemedicine and information technology use in health in LMIC Intellectual property regimes (patent laws, regulatory bodies etc.) and access to COVID-19 therapies and vaccines How to address gender biases in Personal Protection Equipment (PPE), for whom is PPE designed and who gets access? The ethics of digital tracking: Whose privacy and data is trespassed and with what consequences? Health technology assessment during a pandemic Studies exploring how the pandemic can inform the foundations of more just, equitable and better health systems Whether, how and which new forms of enabling governance and leadership emerge at local or other levels, supporting work with community organizations and engaging mid-level and front-line actors, and if and how they are sustained or undermined over time? New strategic purchasing arrangements—what is the potential for system-wide gains post COVID-19 from, e.g. price-capping for tests or beds secured from the private sector Are regional health actors and agendas strengthened and how, with what consequences? What is the role and place of community-level action for health—including approaches to engaging with and responding to community/public priorities and concerns, and beyond health? How can initiatives that have been transformative for vulnerable populations be embedded and sustained through health policy and system change into the longer-term? What forms of local level intersectoral collaboration develop that address food, housing, transport, school, social safety nets, i.e. the social determinants of health: what are the lessons learned for more people-centred systems? How does the pandemic inform and energize wider action on onehealth and climate change? Research governance What is the role of national and international academic and research institutions in sudden and rapidly evolving global health crises like COVID-19? How can research governance processes during crises such as COVID-19 support important, ethical, timely research that supports social justice? How is COVID-19 impacting on who is submitting and publishing articles? What is the role of multilateral organizations, international banks, global donors in supporting CPVI-19 research and setting research priorities? What innovative knowledge translation efforts and models of collaborations between research, policymaking and stakeholder organizations emerge during COVID? Author notes †Lucy Gilson and Bruno Marchal contributed equally to this study © The Author(s) 2020. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. All rights reserved. For permissions, please e-mail: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Haacker,, Markus;Hallett, Timothy, B;Atun,, Rifat

doi: 10.1093/heapol/czaa073pmid: 33450767

Abstract The issue of time horizons has received scant attention in discussions pertaining to health economic evaluations unlike discounting or translation of health outcomes into life-cycle measures (e.g. quality-adjusted life years or disability-adjusted life years). The available guidelines do not offer clear and consistent guidance for many problems addressed in health economic evaluations. In practice, variation of time horizons between studies for the same diseases is a matter of concern, as results on cost-effectiveness depend on the time horizon. Our paper contributes to establishing a consistent approach to setting time horizons across common types of health economic evaluations and mitigating potential bias where the choice of a time horizon may affect results of the evaluation. We find that available guidance is clear only for patient-focused interventions, but not in the presence of population-level effects owing to transmission of infections or other linkages. We distinguish between a policy period—over which an intervention is delivered or initiated—and an evaluation period over which the effects are measured. One important challenge in establishing a time horizon for evaluation is that, at least for infectious diseases, the state of the epidemic at the end of the policy period cannot be evaluated precisely and incorporated in the results of an economic evaluation. While longer policy periods partly mitigate this challenge, they are subject to greater uncertainty, and outcomes may not adequately reflect the cost-effectiveness of current policies because outcomes reflect an average over the policy period. Incremental analysis on interventions implemented in sub-periods of the policy period (especially at the beginning) potentially improves accuracy and helps to identify potential for improving cost-effectiveness by varying the path of implementation or the mix of interventions offered over time. Time horizon, cost-effectiveness analysis, benefit–cost analysis, health economics, economic evaluation, public health, global health, methods, policy evaluation KEY MESSAGES Available guidance on time horizons is not clear or consistent with regard to population-level effects owing to infectious disease transmission or other linkages, and the practice on setting time horizons is inconsistent. At least for infectious diseases, the state of the epidemic at the end of the policy period cannot be evaluated precisely. Evaluations over longer policy periods mitigate this challenge, but are subject to greater uncertainty and return averages of possibly time-varying effects. Incremental analysis of the effects of interventions implemented at a point in time within the policy period improves accuracy and helps identify opportunities in realigning the timing and mix of interventions under consideration. Introduction The measurement of health outcomes and costs over time in health economic evaluations poses several challenges—capturing health effects which often extend well beyond the study period (notably owing to longer survival), the scope of future costs, weighing health outcomes and costs over time through discounting, and setting time horizons for policy implementation and evaluation. Some of these issues have been well explored. The immediate effects of health interventions are typically transformed into quality-adjusted life years (QALYs, endorsed by the Panel on Cost-Effectiveness in Health and Medicine, see Feeny et al., 2016) or in global health to disability-adjusted life years (DALYs) (Jamison et al., 2017; Kyu et al., 2018). Claxton et al., (2019) and Haacker et al. (2019) provide recent access points to a longstanding discussions on discounting in global health going back some 40 years (Weinstein and Stason, 1977). The issues around future costs, and in particular their scope, included in health economic evaluations are well understood (van Baal et al., 2011), but not considered consistently in practice (Sanders et al., 2016; van Baal et al., 2019). In contrast, the issue of time horizons in health economic evaluations has not been addressed comprehensively, even though the choice of a ‘time period over which to evaluate an intervention is a vexing question’ (Tan-Torres Edejer et al., 2003). Influential guidance on cost-effectiveness analysis (Table 1) provides general recommendations, e.g. advising that the time horizon ‘should extend far enough into the future to capture important differences in consequences’ (Owens et al., 2016). Table 1 Perspectives on time horizons in influential guidelines Torrance et al. (1996) (1st Panel on cost-effectiveness) ‘The time horizon of the analysis for a cost-effectiveness study should extend far enough into the future to capture the major health and economic outcomes-both intended effects and unintended side effects. As a result, some analyses follow patients for the duration of their lives. For certain interventions, such as the removal of environmental toxins, the effects of a program may run even longer, requiring a time horizon that extends for generations. Frequently, the appropriate time horizon extends beyond the availability of primary data, and modeled data must be used in the analysis’. ‘When a positive discount rate is used, as is recommended for the Reference Case Analysis, the time horizon of the study will in many cases be effectively limited by the discount rate’. Drummond and Jefferson (1996) (Guidelines for authors and peer reviewers of economic submissions to the BMJ) ‘The time horizon should be long enough to capture all the differential effects of the options. It should often extend to the whole life of the treated individuals and even to future generations. If the time horizon is shortened for practical reasons this decision should be justified and an estimate made of any possible bias introduced’. Tan-Torres Edejer et al. (2003) (WHO guide to cost-effectiveness analysis) ‘All interventions should be evaluated under the assumption that they are implemented over a period of 10 years. However, costs and health effects related to the intervention should be followed for the duration of the lifetime of the beneficiaries’. Wilkinson et al. (2014) (‘iDSI’ Reference Case) ‘The time horizon used in an economic evaluation should be of sufficient length to capture all costs and effects relevant to the decision problem’. ‘Lifetime time horizon should be used in first instance. Shorter time horizon can be used where shown that all costs and effects that are relevant to the decision problem have been captured’. Owens et al. (2016) (2nd panel on cost-effectiveness) ‘The time horizon of the CEA should extend far enough into the future to capture important differences in consequences. […] Using a time horizon that does not encompass all consequences can provide a misleading estimate of the cost-effectiveness of an intervention. […] At a minimum, modeling should be used to estimate gains in life expectancy due to differential survival. As noted, the time horizon should also be sufficient to capture relevant costs. […] When a positive discount rate is used, as is recommended for Reference Case analyses, the time horizon of the study will in many cases be effectively limited by the discount rate’. Torrance et al. (1996) (1st Panel on cost-effectiveness) ‘The time horizon of the analysis for a cost-effectiveness study should extend far enough into the future to capture the major health and economic outcomes-both intended effects and unintended side effects. As a result, some analyses follow patients for the duration of their lives. For certain interventions, such as the removal of environmental toxins, the effects of a program may run even longer, requiring a time horizon that extends for generations. Frequently, the appropriate time horizon extends beyond the availability of primary data, and modeled data must be used in the analysis’. ‘When a positive discount rate is used, as is recommended for the Reference Case Analysis, the time horizon of the study will in many cases be effectively limited by the discount rate’. Drummond and Jefferson (1996) (Guidelines for authors and peer reviewers of economic submissions to the BMJ) ‘The time horizon should be long enough to capture all the differential effects of the options. It should often extend to the whole life of the treated individuals and even to future generations. If the time horizon is shortened for practical reasons this decision should be justified and an estimate made of any possible bias introduced’. Tan-Torres Edejer et al. (2003) (WHO guide to cost-effectiveness analysis) ‘All interventions should be evaluated under the assumption that they are implemented over a period of 10 years. However, costs and health effects related to the intervention should be followed for the duration of the lifetime of the beneficiaries’. Wilkinson et al. (2014) (‘iDSI’ Reference Case) ‘The time horizon used in an economic evaluation should be of sufficient length to capture all costs and effects relevant to the decision problem’. ‘Lifetime time horizon should be used in first instance. Shorter time horizon can be used where shown that all costs and effects that are relevant to the decision problem have been captured’. Owens et al. (2016) (2nd panel on cost-effectiveness) ‘The time horizon of the CEA should extend far enough into the future to capture important differences in consequences. […] Using a time horizon that does not encompass all consequences can provide a misleading estimate of the cost-effectiveness of an intervention. […] At a minimum, modeling should be used to estimate gains in life expectancy due to differential survival. As noted, the time horizon should also be sufficient to capture relevant costs. […] When a positive discount rate is used, as is recommended for Reference Case analyses, the time horizon of the study will in many cases be effectively limited by the discount rate’. Open in new tab Table 1 Perspectives on time horizons in influential guidelines Torrance et al. (1996) (1st Panel on cost-effectiveness) ‘The time horizon of the analysis for a cost-effectiveness study should extend far enough into the future to capture the major health and economic outcomes-both intended effects and unintended side effects. As a result, some analyses follow patients for the duration of their lives. For certain interventions, such as the removal of environmental toxins, the effects of a program may run even longer, requiring a time horizon that extends for generations. Frequently, the appropriate time horizon extends beyond the availability of primary data, and modeled data must be used in the analysis’. ‘When a positive discount rate is used, as is recommended for the Reference Case Analysis, the time horizon of the study will in many cases be effectively limited by the discount rate’. Drummond and Jefferson (1996) (Guidelines for authors and peer reviewers of economic submissions to the BMJ) ‘The time horizon should be long enough to capture all the differential effects of the options. It should often extend to the whole life of the treated individuals and even to future generations. If the time horizon is shortened for practical reasons this decision should be justified and an estimate made of any possible bias introduced’. Tan-Torres Edejer et al. (2003) (WHO guide to cost-effectiveness analysis) ‘All interventions should be evaluated under the assumption that they are implemented over a period of 10 years. However, costs and health effects related to the intervention should be followed for the duration of the lifetime of the beneficiaries’. Wilkinson et al. (2014) (‘iDSI’ Reference Case) ‘The time horizon used in an economic evaluation should be of sufficient length to capture all costs and effects relevant to the decision problem’. ‘Lifetime time horizon should be used in first instance. Shorter time horizon can be used where shown that all costs and effects that are relevant to the decision problem have been captured’. Owens et al. (2016) (2nd panel on cost-effectiveness) ‘The time horizon of the CEA should extend far enough into the future to capture important differences in consequences. […] Using a time horizon that does not encompass all consequences can provide a misleading estimate of the cost-effectiveness of an intervention. […] At a minimum, modeling should be used to estimate gains in life expectancy due to differential survival. As noted, the time horizon should also be sufficient to capture relevant costs. […] When a positive discount rate is used, as is recommended for Reference Case analyses, the time horizon of the study will in many cases be effectively limited by the discount rate’. Torrance et al. (1996) (1st Panel on cost-effectiveness) ‘The time horizon of the analysis for a cost-effectiveness study should extend far enough into the future to capture the major health and economic outcomes-both intended effects and unintended side effects. As a result, some analyses follow patients for the duration of their lives. For certain interventions, such as the removal of environmental toxins, the effects of a program may run even longer, requiring a time horizon that extends for generations. Frequently, the appropriate time horizon extends beyond the availability of primary data, and modeled data must be used in the analysis’. ‘When a positive discount rate is used, as is recommended for the Reference Case Analysis, the time horizon of the study will in many cases be effectively limited by the discount rate’. Drummond and Jefferson (1996) (Guidelines for authors and peer reviewers of economic submissions to the BMJ) ‘The time horizon should be long enough to capture all the differential effects of the options. It should often extend to the whole life of the treated individuals and even to future generations. If the time horizon is shortened for practical reasons this decision should be justified and an estimate made of any possible bias introduced’. Tan-Torres Edejer et al. (2003) (WHO guide to cost-effectiveness analysis) ‘All interventions should be evaluated under the assumption that they are implemented over a period of 10 years. However, costs and health effects related to the intervention should be followed for the duration of the lifetime of the beneficiaries’. Wilkinson et al. (2014) (‘iDSI’ Reference Case) ‘The time horizon used in an economic evaluation should be of sufficient length to capture all costs and effects relevant to the decision problem’. ‘Lifetime time horizon should be used in first instance. Shorter time horizon can be used where shown that all costs and effects that are relevant to the decision problem have been captured’. Owens et al. (2016) (2nd panel on cost-effectiveness) ‘The time horizon of the CEA should extend far enough into the future to capture important differences in consequences. […] Using a time horizon that does not encompass all consequences can provide a misleading estimate of the cost-effectiveness of an intervention. […] At a minimum, modeling should be used to estimate gains in life expectancy due to differential survival. As noted, the time horizon should also be sufficient to capture relevant costs. […] When a positive discount rate is used, as is recommended for Reference Case analyses, the time horizon of the study will in many cases be effectively limited by the discount rate’. Open in new tab Applying such general recommendations across common types of health economic evaluations is a challenge. The most clear-cut case is a patient-focused evaluation of an intervention for which costs and effects can be measured or projected through the life course of the patient(s), or a shorter period if there are no lasting health and cost effects. This approach works because we can be certain that all the effects of the intervention have been realized by that time. Setting an appropriate time horizon is less straightforward, if the disease targeted by an intervention is communicable or when behavioural externalities are present, as the effects of an intervention spread across the population and are not tied to the lifetime of the immediate beneficiaries of an intervention. The design of an evaluation with respect to time horizons becomes more complex when an evaluation considers the impact of a policy, by which we refer to an intervention delivered or initiated (e.g. by placing patients with HIV on antiretroviral therapy, who then receive treatment for life) successively across a targeted population over a period of time. A common motive for such an evaluation is the desire to assess progress towards policy targets, for example reducing incidence or the number of deaths from a disease below a threshold. In such an evaluation, it is necessary to make decisions on the time horizon for the assumed implementation of the policy of which the effects are studied (although the policy might continue after this period, outside the scope of the evaluation), and the time horizon over which the effects are evaluated. While some considerations call for adopting longer time horizons (e.g. sufficient ones to assess progress towards disease elimination), others go against it (e.g. increasing uncertainty for more distant periods). Our objectives are informing choices on setting time horizons across common types of health economic evaluations, and improving accuracy where results may be sensitive to the choice of a time horizon. We first provide an overview of available guidelines and of the practice on setting time horizons. We then discuss options on setting time horizons for the evaluation of policies. Finally, we describe approaches that can help refine the practice on setting time horizons for health economic evaluations. Background The important role of the time horizon in health economic evaluations is reflected in influential guidelines. These guidelines broadly recommend the time horizon should ‘be of sufficient length to capture all costs and effects relevant to the decision problem’, ‘extend far enough into the future to capture important differences in consequences’ or ‘be long enough to capture all the differential effects of the options’ (Table 1). These guidelines do not make explicit reference to the role of transmission effects for interventions related to communicable diseases, however. The focus on lifetime effects of ‘treated individuals’ or the beneficiaries of an intervention, and the absence of references to disease transmission suggests a focus only on direct patient-level effects of an intervention. Only one guideline differentiates between a policy period—‘all interventions should be evaluated under the assumption that they are implemented over a period of 10 years’ (Tan-Torres Edejer et al., 2003)—and the period over which the effects are evaluated. Some more specific guidance has emerged for interventions like vaccines where disease transmission is an important consideration. Ultsch et al. (2016), e.g. suggest the ‘time horizon of a dynamic model should last until the steady state is achieved in order to deliver valuable results’ in an economic evaluation of immunization. The practical importance of time horizons is underscored by a number of studies. McCreesh et al. (2018) demonstrate that results on cost-effectiveness of expanding treatment eligibility for people living with HIV strongly depend on the time horizon chosen. Kim et al. (2017), in a review focusing on the USA, report similar findings for treatment of Hepatitis C or various cardiovascular diseases. These studies suggest that the choice of the time horizon is not rendered moot by discounting, for time horizons commonly used. Focusing on the USA, Kim et al. (2017) also report that a majority of a sample of 800 cost-effectiveness studies adopted a time horizon of at least 50 years, but many of the studies adopting a shorter time horizon did so even though a long-term outlook seemed appropriate, considering the type of disease and intervention studied. Only a small minority of studies provided a sensitivity analysis with respect to the time horizon. A similar picture emerges in relation to cost-effectiveness analyses in global health (Figure 1), as measured by the studies captured in the Global Health Cost Effectiveness Analysis Registry (CEVR, 2019). The time horizon adopted in these studies reflects properties of the disease (e.g. shorter for rotaviral enteritis, which is an acute and time-limited viral infection), but there is high dispersion of time horizons used in studies within disease categories. Less than one-half of studies adopt a lifetime horizon, and only ∼5% provide an analysis for multiple time horizons. The studies covered translate direct health outcomes into DALYs (a condition for inclusion in this database), which is a common method of accounting for life-time health consequences of an intervention, but it is not clear how the corresponding future costs have been accommodated (problematic in particular for HIV/AIDS, where survival depends on continuing treatment). As a matter of concern, three of the four diseases covered most frequently in the studies analysed (namely, HIV/AIDS, malaria and tuberculosis) are both chronic and communicable: i.e. diseases for which the consequences of an intervention spread across populations and therefore beyond the lifetime of patients or populations targeted directly. Figure 1 Open in new tabDownload slide Time horizons across cost-effectiveness studies for selected diseases in the global cost-effectiveness analysis registry Figure 1 Open in new tabDownload slide Time horizons across cost-effectiveness studies for selected diseases in the global cost-effectiveness analysis registry Discussion Our discussion is guided by three considerations. First, as pointed out by O’Mahony et al. (2015), it is helpful to distinguish different dimensions of time horizons. Specifically, we consider a policy period over which intervention is assumed to be implemented (although it could be expected to continue, but outside the scope of the evaluation), and an evaluation period over which the effects of an intervention or of a policy are realized and measured, which can be equal to or extend beyond the policy period. We set aside a third dimension discussed by O’Mahony et al. (2015)—intervention duration, i.e. ‘length of time over which the intervention is applied per person or cohort’—which, for our purposes, can be captured under the evaluation period. Second, we distinguish three stylized settings which raise different methodological or practical issues—(1) analysis of an intervention which is delivered or initiated at one point in time; (2) analysis of an intervention which is delivered over some policy period and evaluated over the same period and (3) an analysis in which the evaluation period extends beyond the policy period. We note that there are shades between these stylized cases regarding the scope of outcomes and costs covered. Third, we recognize that an evaluation may adopt a shorter time horizon than theoretically warranted, for practical or political reasons—e.g. measuring progress towards global targets like the Sustainable Development Goals through 2030 (United Nations General Assembly, 2015). The most straightforward case is the ‘analysis of an intervention delivered or initiated at a point in time’. For non-communicable diseases, the effects can be tracked by following the beneficiaries of the intervention over time, or applying some modelling, to establish the relevant changes in health outcomes and costs, up to a lifetime horizon if required. For communicable diseases (or in the presence of linkages or externalities), establishing an appropriate evaluation period is less straightforward, as the health effects extend beyond the targeted population, and the relevant time horizon becomes dependent on the disease transmission and progression dynamics. In this setting, it is desirable to project the consequences of an intervention across the population until the effects peter out (Ultsch et al., 2016). Adopting a shorter horizon results in truncation of health gains or costs, and leads to biased estimates on cost-effectiveness. The relevance of these points is illustrated by Haacker et al. (2016) for male circumcision for HIV prevention. For a male individual circumcised at age 20, projected HIV infections averted among the population circumcised are spread over ∼10 years. HIV infections averted among sexual partners and through mother-to-child transmission exceed the direct effects by a factor of 1.5, and are spread over a longer period of ∼20 years. HIV-related treatment costs peak only after 30 years and are spread over ∼70 years. These findings—indirect effects of a similar order of magnitude as the direct effects of a prevention intervention and the realization of indirect benefits over a longer time horizon—apply to other communicable diseases, such as tuberculosis and malaria, although the time horizons are unusually long for HIV. The second case is an economic evaluation involving a policy period (perhaps of some considerable length) and an evaluation period equal to the policy period. This approach is typical for policy-oriented studies designed to analyse both the effectiveness and cost-effectiveness of an intervention/policy and its contributions to controlling a disease. For example, this approach has been adopted in global policy targets and related studies, including on ‘ending AIDS’ (UNAIDS, 2014), ‘eliminating Hepatitis C’ (WHO, 2016), the investment case for the 6th Global Fund Replenishment (Global Fund, 2019), Lancet Commissions on ‘Global Health 2035’ (Jamison et al., 2013), diabetes in sub-Saharan Africa (Atun et al., 2017), tuberculosis (Reid et al., 2019) and cancer (Atun et al., 2015), and in numerous academic papers covering similar ground [e.g. Eaton et al. (2014) and Stover et al. (2016) on HIV; Bertram et al. (2018) on NCDs; Ozawa et al. (2016) on immunization]. The long policy period averaging 17 years in these 11 reports and studies is motivated by the objective of assessing the effectiveness of a policy, or of its components, in terms of reducing incidence and mortality and meeting ‘elimination’ targets. Adopting a long time horizon, though, comes at a cost in terms of the accuracy and practical relevance of results on cost-effectiveness. One reason is uncertainty—the larger the weight of developments further out, the more uncertain the results on cost-effectiveness become. Moreover, if cost-effectiveness is calculated as the ratio of health effects and costs, each summed over the policy period (possibly weighed by some discount rate), the result reflects an average cost-effectiveness over the policy period. This average figure can mask changes in the cost-effectiveness of a policy and its components over the policy period. Hence, adopting a long time horizon can result in weak guidance on ‘current’ policies. This shortcoming is particularly relevant for the design of policies to control infectious diseases, where the effectiveness of interventions depends on the prevalence of the disease and other endogenous factors. For example, in case of HIV, average cost-effectiveness ratios over the policy period have been shown to mask changes in cost-effectiveness across sub-periods [such that that front-loading of investments improves cost-effectiveness overall; Anderson et al. (2018)] and changes in the relative effectiveness of prevention interventions [so that the optimal policy mix changes over the policy period; Shattock et al. (2016)]. Hence, setting a long policy period for evaluation may be appropriate to study the potential effectiveness of a policy in controlling a disease, and mitigates bias owing to truncation of health gains or costs in the presence of long time lags. However, the issue of averaging can be addressed by estimating the ‘incremental contributions’ of interventions implemented at some point in time within the policy period—in particular at the beginning, which is most relevant for contemporaneous policy analysis. For example, the coverage rate of some intervention(s) could be varied for the first period, keeping policies for the remainder of the policy period unchanged. If this incremental analysis produces results (and—for several interventions—rankings) different from the average cost-effectiveness, this could result in a reassessment of sequencing or pace of scale-up of the policy. The second challenge regarding the design of health economic evaluations with respect to time horizons is the valuation of the consequences beyond the end of the policy period. In addition to outcomes over the policy period, there are costs and health outcomes beyond this period which can be attributed to actions within the policy period. In practice, this problem is frequently addressed by ‘extending the time horizon of the evaluation beyond the policy period’—the third type noted above. There are several established methods of extending aspects of the evaluation beyond the policy period, such as translating health outcomes into QALYs (Feeny et al., 2016) or DALYs (Kyu et al., 2018), or assigning monetary values to improved health outcomes (Jamison et al., 2013; Robinson et al., 2019). While QALYs and DALYs are widely used in academic studies in public and global health, policy-oriented reports and studies often specify ‘economic returns’ in form of monetary valuations of health outcomes (including 8 of the 11 reports and studies quoted earlier). The issues around the accounting for future costs (‘related’ or ‘unrelated’) are well understood, and academic guidelines generally recommend a symmetric approach to health outcomes and costs with respect to the time horizon (Tan-Torres Edejer et al., 2003; Wilkinson et al., 2014; Owens et al., 2016; see Table 1). However, the academic practice on including future costs is inconsistent (Sanders et al., 2016; de Vries et al., 2019). Extending this approach to interventions implemented over some policy period is also possible, e.g. by including savings in treatment costs owing to reduced incidence that occur after the end of the policy period. Examples include assessments of comprehensive HIV control policies (Schwartländer et al., 2011) or of an HIV prevention programme (Njeuhmeli et al., 2011). Where the implementation of an intervention is studied over a period of time, though, the accounting for the future costs poses some additional challenges. Costs within the policy period include some of the lifetime costs of disease (or reflects corresponding savings) caused by the health events within the period, and it is important to account for lifetime costs and current spending in a way that avoids double-counting. This point is illustrated in Figure 2, which shows lifetime costs of treatment associated with two policies. Scenario 2 envisages improved prevention effort (the direct costs of which are not included in this illustration), but the same treatment eligibility as in Scenario 1. Because there are fewer new infections in Scenario 2, the costs caused by new cases, and the costs overall, are lower in Scenario 2 than in Scenario 1. Irrespective of any new cases, the costs of care and treatment are at least A + B, for the population already affected by the diseases at the beginning of the policy period, of which cost B occurs after the end of the policy period. The improved prevention effort, which results in lower number of new cases, reduces the lifetime costs of care and treatment for new cases by E + F, or from C + D+E + F to C + D. E + F thus represents financial savings against the additional costs of implementing the prevention policy described by Scenario 2. However, if spending during the policy period are already included in the policy analysis, only the remaining savings in lifetime costs of people affected by the disease at the end of the policy period should be counted (represented by F in comparing Scenario 2 with Scenario 1). Figure 2 Open in new tabDownload slide Accounting for lifetime costs of disease beyond policy period Figure 2 Open in new tabDownload slide Accounting for lifetime costs of disease beyond policy period Generally, implementing these approaches is conceptually straightforward. In many policy analyses, however, the lifetime costs of disease are also policy-dependent. For example, if the eligibility for accessing some treatment for some diseases expands, the projected lifetime costs for people affected by the disease, and for all new cases, changes. This is an example of an ‘unfunded liability’ (Premchand, 1995) which is created by a legislative or regulatory change and commits resources in future budget periods. Such unfunded liabilities can be captured with the framework described in Figure 2, but this might blur the link between expenditures over time and the underlying and preceding policy decisions. Accrual accounting methods to assess such future costs are commonly used in public finance (OECD and IFAC, 2017; IFAC and CIPFA, 2018)—e.g. to estimate the liabilities posed by insurance programmes (Congressional Budget Office, 2018)—but rarely applied in public and global health. The approaches on accounting for consequences beyond a policy period described above may not fully account for the implications of the policy beyond the endpoint of the policy period even with a very long evaluation period. For example, the state of an infectious diseases epidemic at the end of the policy period may have implications for the future trajectory of the epidemic. Assessing the implications (and even the direction) of such dynamic effects is not straightforward. Does a disease control strategy result in a permanently lower trajectory of incidence (and hence improved health outcomes, and lower costs of care and treatment) following the end of the policy period? Or will the epidemic rebound in the absence of continued investments in prevention, and incidence could be higher than otherwise following the policy period? Or will incidence remain low, but only subject to some continuity policy which comes at a cost? In health economic evaluations, there is no straightforward and fully satisfactory way to get around this endpoint problem. One approach would be to prolong the time horizon of the policy period to a point from which disease dynamics and costs no longer matter practically, e.g. because of discounting. However, longer time horizons result in less reliable estimates on cost-effectiveness of current policies (as discussed). In some cases, it is possible to evaluate the state of a disease at the end of the policy period (subject to assumptions on continuation policies) without explicitly running projections until the end of time. Such methods are used in engineering or macroeconomics (Bellman, 1957; Ljungqvist and Sargent, 2018), but they require that the state of a dynamic system can be summarized by one or very few state variables. This requirement is generally not met in health economics, as the population affected would typically be differentiated by sex, age, disease progression, treatment status, transmission risk (for infectious diseases), etc., resulting in a large number of state variables. In one special case, it is more feasible to evaluate the state of a disease at the end of the policy period. This is when the effects of a policy dissipate over time, or the epidemic attains a ‘steady state’ (Ultsch et al., 2016; Mauskopf et al., 2018), in which the relevant outcomes and costs under alternative scenarios are constant or grow at a constant rate. From this point, the present discounted value of the changes in long-term outcomes (following the point at which steady-state is reached) can be calculated as a multiple of the current value of the relevant variables. Conclusions Available guidelines on health economic evaluations (summarized in Table 1) broadly agree that the evaluation should extend over the lifetime of the beneficiaries, or a shorter period if that is sufficient (i.e. if the intervention regards a disease which does not affect survival and does not have other lasting consequences in terms of either effects or costs). We find this guidance incomplete and inadequate in informing choices in settings commonly encountered in health economic evaluations—especially concerning infectious diseases—for two reasons. First, the ‘lifetime of the beneficiaries’ is well-defined only for patient-focused interventions, but not when population-level effects occur because of disease transmission or behavioural and other externalities. Second, health economic evaluations frequently regard interventions offered over some policy period, and different considerations apply regarding the choice of the policy period and of the evaluation period over which the effects are measured, and for the scope of effects captured over the evaluation period. We find that there is no fully satisfactory general method to account for the consequences of interventions beyond the end of the policy period, except for special cases: when the effects of a policy dissipate eventually—over the lifetime horizon for patient-focused evaluations, if the population-level effects of an intervention taper off (as in the example on male circumcision), or if an epidemic attains a steady state. However, our discussion points to several methods which can help improve the accuracy of health economic evaluations. For example, it has been observed that cost-effectiveness ratios tend to improve with longer time horizons (Kim et al., 2017; McCreesh et al., 2018). ‘Sensitivity analysis’ with regard to the time horizon is rarely conducted (Kim et al., 2017), yet understanding how costs and effects evolve over time is a valuable insight in its own right. Moreover, if results depend on the time horizon adopted, this finding could point to misspecification of the evaluation, e.g. when important results are truncated and results on cost-effectiveness are therefore biased. One common cause of such misspecification is inconsistent accounting for costs and effects. While health economic evaluation commonly accounts for the health consequences beyond the policy period (by translating immediate health impacts to deaths or cases averted in QALYs or DALYs), it is less common to account for the future costs. Where a long policy period is adopted it is nevertheless important to bear in mind drawbacks of such an approach. Especially in the context of a disease control strategy, the cost-effectiveness of interventions, and their ranking, may change over the policy period. Results on cost-effectiveness ratios pertaining to the policy period overall may therefore be misleading in terms of identifying the most effective (possibly time-variant) policies. Incremental analysis of variations on the policy (especially on interventions delivered at the beginning of the policy period) can serve as a validity check on policy analyses and help to refine optimal policy design over time. Relatedly, accrual accounting offers a useful framework for linking costs and health outcomes to underlying decisions and events. In summary—unlike other aspects of health economic evaluations over time like discounting or translating immediate health outcomes into life-cycle measures like QALYs or DALYs, the issue of time horizons has received relatively little attention, and available guidelines do not offer clear guidance for commonly used policies in different settings. This is the case even though results of health economic evaluations have been shown to be sensitive to the choice of the time horizon. Perhaps one reason for this neglect is that the issue of time horizons does not lend itself to simple general solutions. Our contribution is to describe common settings in health economic evaluations with regard to time horizons, to propose a more differentiated approach to time horizons differentiating between policy periods and evaluation periods, while describing methods to improve accuracy and relevance of evaluations covering long policy periods. Funding T.B.H. acknowledges joint Centre funding from the UK Medical Research Council and Department for International Development (MR/R015600/1). No funding was received specifically for this work. Conflict of interest statement. None declared. Ethical approval. No ethical approval was required for this study. References Anderson SJ , Ghys PD , Ombam R , Hallett TB. 2018 . Frontloading HIV financing maximizes the achievable impact of HIV prevention . Journal of the International AIDS Society 21 : e25087 . Google Scholar Crossref Search ADS PubMed WorldCat Atun R , Davies JI , Gal EA et al. 2017 . Diabetes in sub-Saharan Africa: from clinical care to health policy . The Lancet Diabetes & Endocrinology 5 : 622 – 67 . Google Scholar Crossref Search ADS PubMed WorldCat Atun R , Jaffray DA , Barton MB et al. 2015 . Expanding global access to radiotherapy . The Lancet Oncology 16 : 1153 – 86 . Google Scholar Crossref Search ADS PubMed WorldCat Bellman R. 1957 . Dynamic Programming . Princeton, NJ : Princeton University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Bertram MY , Sweeny K , Lauer JA et al. 2018 . Investing in non-communicable diseases: an estimation of the return on investment for prevention and treatment services . The Lancet 391 : 2071 – 8 . Google Scholar Crossref Search ADS WorldCat CEVR (Center for Evaluation of Value and Risk in Health). 2019 . Global Health Cost Effectiveness Analysis Registry. Boston, MA: CEVR, Tufts Medical Center. http://healtheconomics.tuftsmedicalcenter.org/ ghcearegistry/, accessed 18 January 2019. Claxton K , Asaria M , Chansa C et al. 2019 . Accounting for Timing when Assessing Health-Related Policies . Journal of Benefit-Cost Analysis 10 : 73 – 105 . 10.1017/bca.2018.29 Google Scholar Crossref Search ADS PubMed WorldCat Congressional Budget Office. 2018 . Measuring the Costs of Federal Insurance Programs: Cash or Accrual ? Washington, DC : Congressional Budget Office . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC de Vries LM , van Baal PHM , Brouwer WBF. 2019 . Future costs in cost-effectiveness analyses: past, present, future . Pharmacoeconomics 37 : 119 – 30 . Google Scholar Crossref Search ADS PubMed WorldCat Drummond M , Jefferson T. 1996 . Guidelines for authors and peer reviewers of economic submissions to the BMJ . BMJ 313 : 275 – 83 . Google Scholar Crossref Search ADS PubMed WorldCat Eaton JW , Menzies NA , Stover J et al. 2014 . Health benefits, costs, and cost-effectiveness of earlier eligibility for adult antiretroviral therapy and expanded treatment coverage: a combined analysis of 12 mathematical models . The Lancet Global Health 2 : e23 – 34 . Google Scholar Crossref Search ADS PubMed WorldCat Feeny D , Krahn M , Prosser LA , Salomon JA , 2016 . Valuing health outcomes. In: Neumann PJ , Sanders GD , Russell LB et al. (eds). Cost-effectiveness in Health and Medicine . Oxford and New York : Oxford University Press , Chapter 7 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Global Fund to Fight AIDS, Tuberculosis and Malaria (Global Fund). 2019 . Step up the Fight — Investment Case, Sixth Replenishment 2019 . Geneva : Global Fund . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Haacker M , Fraser-Hurt N , Gorgens M. 2016 . Effectiveness of and financial returns to voluntary medical male circumcision for HIV prevention in South Africa: an incremental cost-effectiveness analysis . PLoS Medicine 13 : e1002012 . Google Scholar Crossref Search ADS PubMed WorldCat Haacker M , Hallett TB , Atun R. 2019 . On discount rates for economic evaluations in global health. Health Policy and Planning 35: 107 – 114 . International Federation Accountants (IFAC) and Chartered Institute of Public Finance and Accounting (CIPFA). 2018 . International Public Sector Financial Accountability Index . New York : IFAC . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Jamison DT , Gelb H , Horton S et al. 2017 . Disease Control Priorities: Improving Health and Reducing Poverty. Disease Control Priorities , Vol. 9, 3rd edn. Washington, DC : World Bank . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Jamison DT , Summers LH , Alleyne G et al. 2013 . Global Health 2035: a world converging within a generation . The Lancet 382 : 1898 – 955 . Google Scholar Crossref Search ADS WorldCat Kim DD , Wilkinson CL , Pope EF et al. 2017 . The influence of time horizon on results of cost-effectiveness analyses . Expert Review of Pharmacoeconomics & Outcomes Research 17 : 615 – 23 . Google Scholar Crossref Search ADS PubMed WorldCat Kyu HH , Abate D , Abate KH et al. 2018 . Global, regional, and national disability-adjusted life years (DALYs) for 359 diseases and injuries and healthy life expectancy (HALE) for 195 countries and territories, 1990–2017: a systematic analysis for the global burden of disease study 2017 . The Lancet 392 : 1859 – 922 . Google Scholar Crossref Search ADS WorldCat Ljungqvist L , Sargent T. 2018 . Recursive Macroeconomic Theory , 4th edn. Cambridge, MA : MIT Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mauskopf J , Standaert B , Connolly MP et al. 2018 . Economic analysis of vaccination programs . Value in Health 21 : 1133 – 49 . Google Scholar Crossref Search ADS PubMed WorldCat McCreesh N , Andrianakis I , Nsubuga RN et al. 2018 . Choice of time horizon critical in estimating costs and effects of changes to HIV programmes . PLoS One 13 : e0196480 . Google Scholar Crossref Search ADS PubMed WorldCat Njeuhmeli E , Forsythe S , Reed J et al. 2011 . Voluntary medical male circumcision: modeling the impact and cost of expanding male circumcision for HIV prevention in Eastern and Southern Africa . PLoS Medicine 8 : e1001132 . Google Scholar Crossref Search ADS PubMed WorldCat O’Mahony JF , Newall AT , van Rosmalen J. 2015 . Dealing with time in health economic evaluation: methodological issues and recommendations for practice . PharmacoEconomics 33 : 1255 – 68 . Google Scholar Crossref Search ADS PubMed WorldCat Organisation for Economic Co-operation and Development (OECD), and International Federation Accountants (IFAC). 2017 . Accrual Practices and Reform Experiences in OECD Countries . Paris : OECD . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Owens DK , Siegel JE , Sculpher MJ , Salomon JA , 2016 . Designing a cost-effectiveness analysis. In: Neumann PJ , Sanders GD , Russell LB et al. (eds). Cost-effectiveness in Health and Medicine . Oxford and New York : Oxford University Press , 75 – 104 . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Ozawa S , Clark S , Portnoy A et al. 2016 . Return on Investment from childhood immunization in low- and middle-income countries, 2011–20 . Health Affairs 35 : 199 – 207 . Google Scholar Crossref Search ADS PubMed WorldCat Premchand A. 1995 . Effective Government Accounting . Washington, DC : International Monetary Fund . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Reid MJ , Arinaminpathy N , Bloom A et al. 2019 . Building a tuberculosis-free world: the lancet commission on tuberculosis . Lancet (London, England) 393 : 1331 – 84 . Google Scholar Crossref Search ADS PubMed WorldCat Robinson LA , Hammitt JK , O’Keeffe L. 2019 . Valuing mortality risk reductions in global benefit–cost analysis . Journal of Benefit–Cost Analysis 10 : 15 – 50 . Google Scholar Crossref Search ADS PubMed WorldCat Sanders GD , Neumann PJ , Basu A et al. 2016 . Recommendations for conduct, methodological practices, and reporting of cost-effectiveness analyses: second panel on cost-effectiveness in health and medicine . JAMA 316 : 1093 – 103 . Google Scholar Crossref Search ADS PubMed WorldCat Schwartländer B , Stover J , Hallett T et al. 2011 . Towards an improved investment approach for an effective response to HIV/AIDS . The Lancet 377 : 2031 – 41 . Google Scholar Crossref Search ADS WorldCat Shattock AJ , Kerr CC , Stuart RM et al. 2016 . In the interests of time: improving HIV allocative efficiency modelling via optimal time‐varying allocations . Journal of the International AIDS Society 19 : 20627 . Google Scholar Crossref Search ADS PubMed WorldCat Stover J , Bollinger L , Izazola JA , et al. 2016 . What is required to end the AIDS epidemic as a public health threat by 2030? The cost and impact of the fast-track approach . PLoS One 11 : e0154893 . Google Scholar Crossref Search ADS PubMed WorldCat Tan-Torres Edejer T , Baltussen R , Adam T et al. (eds). 2003 . Making Choices in Health: WHO Guide to Cost-effectiveness Analysis . Geneva : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Torrance GW , Siegel JE , Luce BR. 1996 . Framing and designing the cost-effectiveness analysis. In: Gold MR , Siegel JE , Russell LB , Weinstein MC (eds). Cost-effectiveness in Health and Medicine . Oxford and New York : Oxford University Press , 54 – 81 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Ultsch B , Damm O , Beutels P et al. 2016 . Methods for health economic evaluation of vaccines and immunization decision frameworks: a consensus framework from a European vaccine economics community . PharmacoEconomics 34 : 227 – 44 . Google Scholar Crossref Search ADS PubMed WorldCat UNAIDS. 2014 . Fast-track — Ending the AIDS Epidemic by 2030 . Geneva : UNAIDS . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC United Nations General Assembly. 2015 . Transforming Our World: The 2030 Agenda for Sustainable Development. Resolution Adopted by the General Assembly on 25 September 2015. New York: United Nations. van Baal PH , Wong A , Slobbe LC et al. 2011 . Standardizing the inclusion of indirect medical costs in economic evaluations . Pharmacoeconomics 29 : 175 – 87 . Google Scholar Crossref Search ADS PubMed WorldCat van Baal P , Morton A , Meltzer D , Brouwer W. 2019 . Future unrelated medical costs need to be considered in cost effectiveness analysis . The European Journal of Health Economics 20 : 1 – 5 . Google Scholar Crossref Search ADS PubMed WorldCat Weinstein MC , Stason WB. 1977 . Foundations of cost-effectiveness analysis for health and medical practices . New England Journal of Medicine 296 : 716 – 21 . Google Scholar Crossref Search ADS WorldCat Wilkinson T , Chalkidou K , Walker D et al. 2014 . Bill and Melinda Gates Foundation Methods for Economic Evaluation Project (MEEP): Final Report . London : National Institute for Health and Clinical Excellence . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization (WHO). 2016 . Global Health Sector Strategy on Viral Hepatitis 2016–2021 . Geneva : WHO . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC © The Author(s) 2020. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. All rights reserved. For permissions, please e-mail: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)