Health Policy and Planning

doi: 10.1093/heapol/czz051pmid: N/A

This content is only available as a PDF. Author notes This article abstract was originally published in English. This translation has not been verified and should not be relied upon—it is provided for reference purposes only. The Publishers, Editors and the London School of Hygiene and Tropical Medicine have not checked this translation and accept no liability for completeness or accuracy of this translation or the use of this translation for whatever purpose. This translation may be incomplete and inaccurate in whole or in part. If you need to rely upon a translation of this abstract, a professional human translator should be engaged to supply an accurate translation of the original English. When referencing articles from this journal, please always refer to the original English version, rather than a translated equivalent. © The Author(s) 2019. 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)

doi: 10.1093/heapol/czz046pmid: 31135900

This content is only available as a PDF. Author notes This article abstract was originally published in English. This translation has not been verified and should not be relied upon—it is provided for reference purposes only. The Publishers, Editors and the London School of Hygiene and Tropical Medicine have not checked this translation and accept no liability for completeness or accuracy of this translation or the use of this translation for whatever purpose. This translation may be incomplete and inaccurate in whole or in part. If you need to rely upon a translation of this abstract, a professional human translator should be engaged to supply an accurate translation of the original English. When referencing articles from this journal, please always refer to the original English version, rather than a translated equivalent. © The Author(s) 2019. 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)

doi: 10.1093/heapol/czz052pmid: 31228236

This content is only available as a PDF. Author notes This article abstract was originally published in English. This translation has not been verified and should not be relied upon—it is provided for reference purposes only. The Publishers, Editors and the London School of Hygiene and Tropical Medicine have not checked this translation and accept no liability for completeness or accuracy of this translation or the use of this translation for whatever purpose. This translation may be incomplete and inaccurate in whole or in part. If you need to rely upon a translation of this abstract, a professional human translator should be engaged to supply an accurate translation of the original English. When referencing articles from this journal, please always refer to the original English version, rather than a translated equivalent. © The Author(s) 2019. 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)

Kavanagh, Matthew, M;Dubula-Majola,, Vuyiseka

doi: 10.1093/heapol/czy103pmid: 30629158

Abstract Efforts to improve the effectiveness of global health aid rarely take full account of the micro-politics of policy change and implementation. South Africa’s HIV/AIDS epidemic is a case in point, where the US President’s Emergency Plan for AIDS Relief (PEPFAR) has provided essential support to the national AIDS response. With changing political context, PEPFAR has shifted focus several times—most recently reversing the policy of ‘transition’ out of direct aid to a policy of re-investing in front-line services in priority districts to improve aid effectiveness. However, this policy shift has not led to the expected impact on health services. This paper reports the findings of a study on the implementation of the recent policy through interviews at randomly selected sites in high HIV-burden districts of South Africa that capture the experiences of public-sector health leaders. We find little evidence to support the explanation that the new aid policy displaced government staff and resources. Instead, our findings suggest that legacies of the previous policy remained as local aid managers did not shift funding and practice at sufficient scale to drive the planned service delivery expansion. Human resource support, the main PEPFAR contribution to service delivery at front-line facilities, was not adequate or distributed based on the size of the HIV programme, leaving notable gaps in outreach, defaulter tracing, and community service delivery. Instead, services that better fit the previous policy paradigm, like training and data-sharing, are common at site-level but provide diminishing returns. Together, our findings suggest opportunities for PEPFAR South Africa to revisit its model and increase service delivery intensity, in particular through community-based services. More broadly, this case illustrates the need for greater attention to the multiple actors with discretion in the policy system of health aid and the mechanisms through which political priority is translated into programming as policy shifts are made. HIV/AIDS, South Africa, aid, global health, policy implementation, human resources for health, PEPFAR Key Messages Translating political priority on aid effectiveness, achievement of disease response goals or the proper role of international assistance for middle-income countries faces policy implementation challenges. In South Africa, the limited impact of policy changes in the US President’s Emergency Plan for AIDS Relief stems from insufficient implementation through shifting allocations to front-line facilities rather than from a failure of the model of funding direct services and human resources. Introduction Financing for the HIV/AIDS response in South Africa has been characterized by a series of dramatic policy pivots driven by shifting politics. With the world’s highest burden of HIV and a middle-income economy with a fluctuating growth trajectory, the functioning of international aid for HIV in South Africa illuminates key challenges in policy change and implementation that are not often acknowledged in international debates about global health aid effectiveness and the achievement of national and international health goals. It is now credible to talk about ending the public health crisis of HIV through scaling up antiretroviral treatment (ART) for both health and prevention benefits alongside ‘combination prevention’ to halt HIV transmission (Fauci and Marston, 2015). In the face of a global infectious pandemic, whether this potential will be realized depends heavily on how quickly this scale-up takes place (Sidibé et al., 2016). The South African National Strategic Plan has thus adopted UN-backed goals of achieving ‘90–90–90’ targets by 2020—90% of people living with HIV will know their status, 90% of those will receive antiretroviral therapy, and 90% of those will achieve viral suppression (United Nations, 2016; SANAC, 2017). Despite significant increased investment and political commitment, however, scale-up and quality of South African HIV services is not yet on track to achieve these goals. The US President’s Emergency Plan for AIDS Relief (PEPFAR) has been a key part of financing the South African AIDS response for the last 15 years. While the South African government now finances the majority of its AIDS response, PEPFAR remains the largest external funding source for this largest AIDS response in the world (PEPFAR, 2017a). PEPFAR policy has shifted several times in that period—from initial focus on building treatment programmes to a major ‘transition’ out of funding front-line services. PEPFAR most recently reversed that transition with a policy shift toward font-line services in selected priority districts to improve effectiveness and impact. In this article, we explore how this most recent policy change has translated at the front-lines of the HIV/AIDS response in the country. We draw on policy change and implementation literature to explore this shift. Studies of policy implementation have long shown how interpretation, coordination, values clashes and bureaucratic autonomy all challenge simplistic translation of policy changes into practice (Pressman and Wildavsky, 1984; Walt, 1994; Barrett, 2004). Studies highlight the ways in which policy is made and implemented through sub-systems that function based on belief structures, with multiple actors and layers of authority (Sabatier and Jenkins-Smith, 1993). Workers at the ‘bottom’ are themselves policy-makers with significant discretion in operational decisions (Lipsky, 2010). Government agencies and organizations are far from monolithic and back-stage ‘micro-politics’ can re-shape what might appear to be clear policy direction (Burns, 1961). Local officials and managers play a critical role in aligning resources and organizational environments with policy goals to influence the discourses and incentives that can foster, or hinder, implementation at the front lines (Gilson et al., 2014). In this case, local aid officials include both ‘local’ US administrators within US aid agencies and officers with implementing NGOs holding longstanding contracts. Each year the Office of the Global AIDS Coordinator in the US State Department sets policy through the country operational plan (COP) and other mechanisms. However, the actual contracts and work plans that dictate how PEPFAR funds will be spent in practice are set by agency officials assigned to a given country from the US Agency for International Development (USAID), Centers for Disease Control and Prevention and other agencies. NGO administrators then translate policies into action as they decide exactly how to programme the funds they receive. These officials are not directly accountable to PEPFAR headquarters and have significant discretion in interpreting PEPFAR’s central policy directives. Models of the policy process also emphasize how high-profile policy decisions of the past influence implementation of subsequent policies. Especially when the new policy reflects a significant shift, policy legacies may hinder implementation if the process is not carefully organized to account for the ideas and incentives of those on the front lines (Sabatier and Jenkins-Smith, 1993; Pierson, 2004; Béland and Ridde, 2016). PEPFAR’s previous decision to ‘transition’ out of direct services is a good example of such a prior decision which set new ideas and incentives not likely to be easily discarded. Given these expectations, this paper addresses several questions. First, it examines the degree to which the most recent PEPFAR policy shift to emphasize financing ‘direct services’, such as salaries of health workers and supplies at clinics has been implemented. Literature and anecdotal evidence give reason to expect that legacies of the previous transition policy would undercut implementation of the new policy. Second, it explores whether the shift to direct service investments is needed, fills a clear gap and is prioritized by the public-sector facility managers who run much of the AIDS response. Since PEPFAR funds flow primarily to NGOs working to augment public-sector services, key decisions about how funds are used lie with in-country US government officials and NGOs. They could be failing to implement policy change or, alternatively, could be quickly attempting to implement the new policies but facing resistance from facility-level managers or finding that new policies prove duplicative or unnecessary. Overall our analysis finds that the high-level policy shifts announced by PEPFAR in South Africa have not translated at the front lines. Implementation of direct service interventions have been limited and slow to roll out. This, we argue, explains why programme goals have not been met in recent years, rather than the alternative explanation that facility-level investment in priorities like health worker salaries are not working. Evidence gathered from interviews with public-sector facility managers suggests that this is not for lack of need or priority within the broader AIDS response. PEPFAR has an opportunity for greater impact going forward through focusing on implementation of the stated policy and by more closely defining its investment priorities while engaging more deeply with public-sector health leaders. In the context of global policy debates over how to make foreign aid more effective (Beracochea, 2016) and about whether and how foreign aid for health should be deployed to middle-income countries (Markham et al., 2015; Resch and Hecht, 2018), this paper highlights the need to focus on how policy imperatives translate politically into front-line delivery to realize impact. Politics and PEPFAR policy in South Africa PEPFAR has played a critical role in supporting the AIDS response in South Africa. Hailed as one of the world’s most effective foreign aid programmes, it is credited with saving millions of lives in the country (Walensky and Kuritzkes, 2010). In 2004, PEPFAR began funding prevention and treatment programmes at a time of ambivalence about HIV treatment under Mbeki’s administration. In the early years, a significant portion of PEPFAR funding was focused on building new HIV treatment programmes—supporting efforts in public-sector facilities, NGO-run clinics, and within general practitioner networks. From 184 initial facilities in 2005, PEPFAR expanded its programmes into thousands of sites (Larson et al., 2012). Support included funding direct patient-serving staff, drugs, commodities and equipment as well as training, mentoring and information management. By 2010, South Africa had a new minister of health committed to closing the book on the era of HIV denialism and a rising government HIV budget. Importantly, the country was also seen by the USA as a rising power—a BRICS (Brazil, Russia, India, China, South Africa) member with a growing economy that no longer justified such significant PEPFAR investments—a stance increasingly taken by global health donors with respect to powerful middle-income countries. In a major policy shift, the US announced a ‘transition’ out of supporting front-line HIV treatment in the country and conducted a high-profile political process that culminated in Secretary of State Hilary Clinton signing a new partnership framework that set out an end to funding for ‘direct services’ (SAG & USG, 2010). PEPFAR funding was to decline by 48% to $250 million by 2017 and focus away from site-level direct services and toward supporting the health system. This process, which sparked controversy, included ending support for health worker salaries and moving a significant number of people on treatment from non-governmental sites to public-sector facilities (Kavanagh, 2014). Within a few years, however, it became evident that, even with dramatically increased government commitment and funding, South Africa’s burden of HIV and health systems challenges made reaching HIV goals nearly impossible through domestic financing alone. PEPFAR made a second major change—this one far less high profile—suspending the planned funding drawdown in the 2016 COP (PEPFAR, 2016; U.S. Mission South Africa, 2016). Direction from Washington shifted away from transition and towards re-investing in direct patient care where it could augment public-sector health services to speed achievement of 90–90–90 treatment goals (PEPFAR, 2017a). PEPFAR also named South Africa a priority for the DREAMS programme and injected $66 million in new funding for prevention programmes for adolescent girls and young women. In 2018, PEPFAR announced a new ‘surge’ in funding that will add several hundred million dollars on top of the current base funding of $483 million—the details of which are being negotiated at the time of this writing. In 2017 PEPFAR provided approximately one-quarter of all HIV funding in South Africa, sufficient funding to secure the policy shift to direct service and a figure that will increase to roughly 30% by 2019 (PEPFAR, 2017a).1 This new official policy was to shift funding into ‘direct services’ focused on 27 (out of 52) high HIV-burden districts. Four districts were prioritized as ‘scale-up saturation’ where significant PEPFAR investments were meant to achieve the 90–90–90 treatment goals by the end of fiscal year 2017. Those four districts—eThekwini, uMgungundlovu, Ekurhuleni and City of Johannesburg Metropolitan—experience some of the highest burdens of HIV in the country. While significant strides have been made, these districts fell significantly short of the goals—achieving only between 71 and 76% of their FY2017 goals for people currently receiving treatment (see Figure 1). Notably, only eThekwini achieved the testing targets for identifying people living with HIV, but programme quality was a significant problem. In these districts, PEPFAR reports 361 391 people were newly identified in 2017, but only 258 598 people were newly added to treatment; up to 30% of those already on treatment were ‘lost to follow-up’ (PEPFAR, 2018). These overall trends suggest there is more that is needed from PEPFAR programming to fill the key gaps toward reaching saturation and beyond. Figure 1 View largeDownload slide Saturation districts missed targets. FY17 treatment result vs target Figure 1 View largeDownload slide Saturation districts missed targets. FY17 treatment result vs target How has this global health aid policy shift—meant to improve aid effectiveness through focusing geographically and re-engaging in direct service delivery—been implemented at the front lines? Why have the shifts set by PEPFAR leadership not resulted in achievement of the stated goals? We focus in on these four districts to understand more deeply how the PEPFAR policy that reversed transition is being translated into interventions to improve HIV treatment coverage in the highest-burden districts. As PEPFAR looks to expand its impact by both shifting its strategies and increasing its investment through ‘surge’ funding in the coming years, a variety of choices present themselves about how funding can be invested to achieve the strategic objectives of the programme. Methods and facilities sample We visited a randomly selected sample of PEPFAR-supported health facilities in the four ‘saturation’ districts between November 2017 and January 2018 and conducted semi-structured interviews to understand how policy is realized at the clinic level (Mosley, 2013). These districts account for 40% of all those supported on treatment through direct services by PEPFAR in 2017 (in all 52 districts).2 Fifty-three facilities appear in the final random sample. From the 437 sites supported by PEPFAR, we excluded those sites in the bottom quartile of each district in the number of people on treatment, which we assume would be low on the list of priorities for increased direct service investment. The clinics sampled had a mean of 3335 people on ART. We also excluded mobile sites and those inside correctional facilities. Managers at all but five selected sites agreed to participate in interviews, with one excluded because no administrator had been at the facility longer than 6 months. We conducted interviews with lead staff at the remaining 53 sites—most often including the facility manager (usually a nurse) and/or nurse administering the HIV programme—and, wherever possible, we cross-checked answers with other staff and public records. All interviews were anonymous with the names of facility managers not linked to notes and recordings. Facility names are masked by codes below. Interviews were recorded, transcribed, and coded with checks for inter-coder reliability. PEPFAR funding flows largely to a set of NGO ‘implementing partners’ that work in public-sector clinics based on contracts with one of several US government agencies. To ensure our sample was representative, the facilities visited included those contracted through both the Centers for Disease Control and Prevention and the Agency for International Development and served by several lead implementing partners, including Right to Care, Anova, Wits RHI, MatCH, Health Systems Trust and Kheth'Impilo. We did not find significant, systematic differences between observations at clinics served by different implementing partners. The results below are based on these interviews. Our data therefore represent not what implementing partners say they are doing, but instead what public-sector nurses responsible for managing the throughput of the clinic report they have observed and experienced. These are, of course, imperfect data subject to limitations of recall and bias, much as any qualitative data of this nature. We found that these managers were highly reliable narrators when it came to the size and tasks of the staff at the clinic, and the information they provided about the ART programme was verified against outside reports whenever it was possible. An important benefit of this approach is that the subjective input of facility managers about the major barriers to increasing the quality and effectiveness of the ART programme adds an important, and often missing, perspective to conversations about how to improve the efficacy and impact of health aid. Results and discussion PEPFAR’s primary investment in ‘direct service’ at supported facilities since 2015 has been in human resources for health (HRH): staff paid through NGOs placed at facilities, rotating or roving teams visiting facilities to provide direct services, and training and mentoring of the existing government staff. PEPFAR in recent years has not procured significant antiretrovirals or equipment for front-line clinics (PEPFAR, 2017a). As such, we focussed data collection on HRH as the best indicator of the implementation of the policy change away from transition and into front-line services. Investments in HRH, however, run counter to the ‘transition’ paradigm of just a few years ago when local aid officials laid off many direct service workers and were encouraged to think of the role of PEPFAR-funded NGOs as mentors and technical experts not direct providers. Officials working for the US government and implementing NGOs in South Africa have significant discretion in how they translate the new high-level directives to re-invest in direct services into staffing and models service provision. Looking at HRH deployment therefore give us empirical insight about the degree to which aid officials are still acting under the previous paradigm rather than implementing path-departing change. Meanwhile, the views of front-line public-sector nurses managing the facilities largely align with PEPFAR’s official policy shift toward direct services—yet the continued gaps they experience underscore the limited implementation of that policy by aid officials as well as insights about how incentives could be better aligned. Characteristics of sample of public-sector facilities The facilities we visited had large numbers of people on treatment: 85% had at least 1500 people on treatment, and 17% had more than 5000 (Figure 2). Figure 2 View largeDownload slide People on HIV treatment at visited clinics Figure 2 View largeDownload slide People on HIV treatment at visited clinics The need for increases in human resources in South Africa to support rapid expansion of the AIDS response is well documented, with these public-sector clinics providing a wide range of services as well as initiating and maintaining people on ART (Van Damme et al., 2008; Mayosi and Benatar, 2014). The sampled clinics are serving very large numbers of patients and have significant staff complements. Figures 3 and 4 show the portion of clinics in our sample with different numbers of government-employed clinicians and lay staff, respectively. Over half of clinics have at least 10 clinicians, and almost half have 20 or more lay staff who are paid directly by government, most of whom have some level of engagement in the ART programme. While we do not have access to overall patient loads to reflect overall staffing ratios, these raw HRH levels are worth noting, because the additional impact of PEPFAR-supported staff is related to what is already in place as well as the gap in staffing needed to scale high-quality HIV services. Figure 3 View largeDownload slide Government's Nurses and doctors working (at least partly) on HIV per facility Figure 3 View largeDownload slide Government's Nurses and doctors working (at least partly) on HIV per facility Figure 4 View largeDownload slide Government’s lay staff working on HIV per facility Figure 4 View largeDownload slide Government’s lay staff working on HIV per facility Of particular note, despite having significant government staffing complements, only half of clinics have any staff focussed primarily on HIV adherence counselling and support; 89% report having no government staff focussed on outreach or tracing those lost to follow-up from the ART programme (Figures 5 and 6). Figure 5 View largeDownload slide Facilities with government HR focussed on default tracing Figure 5 View largeDownload slide Facilities with government HR focussed on default tracing Figure 6 View largeDownload slide Facilities with government HR focussed on adherence counselling Figure 6 View largeDownload slide Facilities with government HR focussed on adherence counselling PEPFAR support for HRH at front-line facility level Most facilities in our sample report that there are some staff paid for by PEPFAR through local implementing partners who are based at the facility on a full-time or near full-time basis. Our sample was home to a total of 305 such staff, which is notable because our sampled clinics serve 20% of the PEPFAR-supported ART patients in these four districts (Figure 7). Figure 7 View largeDownload slide Facilities with at least one direct service staff supported by PEPFAR Figure 7 View largeDownload slide Facilities with at least one direct service staff supported by PEPFAR In terms of clinical staff, facilities have a relatively small PEPFAR-supported complement of staff. A significant number of sites in these districts report they have no clinical staff based regularly at the facility. The modal configuration among those that do have clinical staff is a single nurse—30% of all clinics—with another 19% that have a second nurse, and a similar portion with a regular doctor funded by PEPFAR. A handful of clinics report more than two nurses, and a few facilities have regular pharmacy staff paid by PEPFAR (Figure 8). Figure 8 View largeDownload slide PEPFAR-supported clinical staff. Number of clinical HR by cadre and percent of clinics Figure 8 View largeDownload slide PEPFAR-supported clinical staff. Number of clinical HR by cadre and percent of clinics About half of clinics have each of several PEPFAR-supported lay cadre of community health workers including HIV testing counsellors, data capturers and adherence or linkage counsellors, most often one person. Very few of the PEPFAR-supported staff are devoted to tracing patients who are lost to follow-up or providing community outreach or services. Our understanding of the PEPFAR strategy based on COP16 and COP17 is that direct service delivery staff are focussed at public-sector facilities and integrated into service delivery. We asked, on this basis, about staff based at these clinics who focus primarily on this task—but this means we may not have captured staff who are primarily based in communities reporting to NGOs who may be engaged in outreach and default tracing. We also note that some of the lay staff primarily tasked with other roles do spend some time reaching out to those who are lost to follow-up—though upon closer questioning, it was evident that this is usually only a small part of the day-to-day work of those reported as adherence counsellors, nurses and others (Figure 9). Figure 9 View largeDownload slide PEPFAR-supported lay staff Figure 9 View largeDownload slide PEPFAR-supported lay staff Overall, public-sector managers report that these facility-based HRH additions are playing an important role in the service delivery structures of the clinics. Patients’ waiting time has been reduced because of the professional nurse that has been allocated for HIV and ART initiation (GP-JNB-04). With the addition of data capturers in our facility, we have improved in the way we collect and analyse statistics, and as a result, our service delivery to the patients has also improved (KZN-uM-03). Figure 10 represents how PEPFAR-supported HRH are distributed compared with the number of people on ART at each clinic. As reflects the frequencies shown in Table 1, most of the observations have relatively few nurses and overall HRH, clustering towards the chart bottom. Our observations cluster in the lower left where there are fewer people on ART (though all of the visited clinics have significant ART rolls) and few HRH. The upper right quadrant, meanwhile, is largely empty, reflecting the apparent lack of a systematic increase in the number of HRH or nurses as the size of the treatment rolls in a clinic increases. Figure 10 View largeDownload slide Distribution of PEPFAR-supported HRH Figure 10 View largeDownload slide Distribution of PEPFAR-supported HRH Table 1 Ratios of PEPFAR-supported HRH People on HIV treatment per staff person (all) Range 92–2767 Average no. on ART per staff 999 People on HIV treatment per nurse Range 415–15 034 Average no. on ART per nurse 2895 People on HIV treatment per staff person (all) Range 92–2767 Average no. on ART per staff 999 People on HIV treatment per nurse Range 415–15 034 Average no. on ART per nurse 2895 View Large Table 1 Ratios of PEPFAR-supported HRH People on HIV treatment per staff person (all) Range 92–2767 Average no. on ART per staff 999 People on HIV treatment per nurse Range 415–15 034 Average no. on ART per nurse 2895 People on HIV treatment per staff person (all) Range 92–2767 Average no. on ART per staff 999 People on HIV treatment per nurse Range 415–15 034 Average no. on ART per nurse 2895 View Large At visited facilities, the ratio of people on ART per staff person (including all cadre) varied significantly between clinics, with an average of 999 people on ART per staff person. The patient-to-nurse ratio was similarly large, with an average of 2895 people on ART per nurse. We did not have access to ‘total’ patient numbers per clinic, but we note that the ratio of people on ART to total ‘government’ staff was 127:1. These ratios are not directly comparable, because government staff are almost all doing more than ART, but it may help understand the limited impact of adding only small numbers of staff to a clinic. Overall, many clinic leaders interviewed identified both the benefits of PEPFAR-supported HRH and the continuing gaps in HRH needs. I would not say a lot has changed, because our clinic is very big. The NGOs are just doing initiation of some patients, and all the follow-ups are done by the clinic nurses, as well as the whole TB/HIV co-infection part, though ART initiation has improved (GP-JNB-16). One possible explanation for why additional PEPFAR HRH has not had a larger impact is that government staff are shifting out of HIV services as PEPFAR-supported staff are added to the clinic—moving a nurse out of ART initiation to focus on childhood vaccination, for example. The result would then be no net increase in HRH working on HIV. In seeking to address this, we began with an open-ended question about what had changed about the work of staff at the clinics since the addition of PEPFAR-supported staff and followed up with a specific question about whether staff had shifted to other areas, and if so, to what areas. Overall, we found little evidence of a significant shifting of government staff out of working on HIV as PEPFAR-supported staff were added. In most clinics, no such shifting was supported. As one respondent explained: No, it will never happen that government staff stop doing ART. We have so many thousands, if we just left HIV to them they would bleed through their nose and ears… (GP-JNB-15). Instead, at most facilities, staff shifted their work within HIV in ways meant to increase programme quality. Before the arrival of NGO staff, we could not do HCT [HIV counseling and testing] and initiation at the same time, on the same day. They have made our work much easier, as we no longer experience long queue, and waiting time for patients has been cut down dramatically. But no, there is no way to stop doing HIV, we are still very short-staffed (KZN-eTH-09). In a few clinics (13% of our sample), however, there was some report of government staff shifting away from HIV care. The professional nurse who was doing HIV is now concentrating on general follow-ups. She also assists in immunization when the clinic is too busy (GP-JNB-07). We also note that we do not have data to understand whether such shifts are occurring at a higher government level—shifts of staff into or out of clinics based on district or provincial decisions. As such, there is need to ensure that policy is clear and clearly communicated to leaders in this area. At clinic level, however, there does not seem to be widespread displacement (Figure 11). Figure 11 View largeDownload slide Clinics where government staff have/have not shifted from HIV to other pressing issues Figure 11 View largeDownload slide Clinics where government staff have/have not shifted from HIV to other pressing issues PEPFAR HRH support through visiting teams PEPFAR-funded NGOs also regularly come to the facilities as visitors to provide support. Nearly half of facilities receive such visits several times a week. These visits include both direct-service visitors—such as ‘roving teams’ of clinicians to boost service provision for difficult cases or at peak times—and visits for mentoring, technical assistance, and training. In our data and the experience of facility managers, these visits are not distinct. The same NGOs often provide both direct service and technical visits. When functioning well this interaction seems beneficial because it ties technical assistance closely to the life of the clinic. Clinic officials, when asked to subjectively rank the value of various types of NGO visiting support, identified providing care and seeing patients as the most valuable intervention, in their opinion. This was followed by the collection of data to review performance with them—something many officials valued highly. These two more structured interventions were far more often identified among the most valuable compared with training, mentoring, providing advice, and similar activities (Figure 12). Figure 12 View largeDownload slide How often NGOs visited. Most frequent visit by percent of clinics Figure 12 View largeDownload slide How often NGOs visited. Most frequent visit by percent of clinics PEPFAR-supported NGOs regularly share data with most facilities, which, as noted, is among the services most valued by facility leaders. Every Wednesday, we sit and look and discuss statistics and look at gaps and how to improve service (KZN-UM-02). This usually happens on a weekly or monthly basis, though a smaller number of clinics do not report such regular data sharing. Worryingly, however, fewer than half could identify specific ways that these data have resulted in a change in how they provide HIV services or run the clinic to improve performance. This, of course, does not mean that no changes were made—and indeed, management studies have long shown that simply identifying problems and showing staff how they are performing can improve performance. These data do, however, suggest a thinner relationship between data and mentoring than might be hoped (Figures 13 and 14). Figure 13 View largeDownload slide Do NGOs regularly share data with clinic managers? Figure 13 View largeDownload slide Do NGOs regularly share data with clinic managers? Figure 14 View largeDownload slide Can identify example of how data have changed practice at the facility Figure 14 View largeDownload slide Can identify example of how data have changed practice at the facility Clinic officials overall praised the trainings conducted by PEPFAR partners. In particular, the most valuable trainings identified by facility managers in the past 12 months were in nurse-initiated/managed ART (NIMART) and in the Tier.net data system, along with several mentions of centralized chronic medicines dispensing and distribution (CCMDD). We used to ask a nurse from the other clinics to come and initiate our patients on ART. [NGO] funded the NIMART training for our professional nurses. This NIMART training allows us as nurses to diagnose, make assessment, take bloods and offer treatment. Our numbers moved from 200 to 1000 clients taking treatment a month. It is a great achievement (KZN-uM-09). The training on the Tier.net, which helped us to understand what data needs to be captured and why. There is no guesswork anymore, and no running around when the district office is asking for certain numbers (KZN-eTH-12). While it is beyond the scope of this report to evaluate broader pre- and in-service training regimes, it is notable that these two subjects came out by far most often, reflecting a surprising unmet need. It is not clear how much is due to staff turnover or whether these facilities have not previously received such training, but centralizing and regularizing it might well increase efficacy and efficiency. These topics are also largely one-off trainings that should not require significant repeated and ongoing training (Figure 15). Figure 15 View largeDownload slide Can give example of changes, innovations, shifts in practice because of training in the last 12 months Figure 15 View largeDownload slide Can give example of changes, innovations, shifts in practice because of training in the last 12 months Again, worryingly, 58% could not identify any specific practice, ways of providing care, protocols, or innovations introduced at the facility because of the trainings. Front-line health leaders’ priorities We asked facility managers to identify the biggest barriers or challenges holding back increasing the number of people identified, initiated and retained in care. The biggest three identified were the following: Staff shortages resulting in lack of capacity to trace lost patients and build effective retention programmes, especially for mobile populations: We are dealing with a mobile community. They migrate from one place to the other after case finding and become lost to follow-up and come back when they are seriously sick, and sometimes they give us wrong addresses and wrong names, and we have nobody to follow up (GP-JNB-07). If we can trace more, we will need more nurses—otherwise the waiting period will increase, or else other nurses from chronic will have to do initiation and the general chronic will suffer. We will require more nurse initiators. But as of now, we do not have someone to trace, so we have a problem before that one (KZN-uM-02). Both physical space and lab infrastructure remained a significant problem: Infrastructure is the biggest problem. We cannot accommodate as many people as we would love to. Even [NGO] has a problem, because we cannot give them enough working space. They just manage to get a corner somewhere and do their work (KZN-uM-10). The lab results, the turn-around time, sometimes the errors, the queries on the results. So often the patients have no results to report and then they stop coming back. They say, ‘I’ll come next time’ but don’t (GP-EK-08). The nurses still lack working spaces (rooms) and this impacts negatively to the privacy of the patients, so more park homes are needed here (KZN-eTH-07). Particularly notable was the sense among respondents that clinics were not particularly well suited to supporting patient retention in many cases, and that some barriers might best be addressed by community-based service delivery. Patients do not want to wait for long at the clinic each and every time. They have to go to work, and sometimes the attitude of the nurses is not right for the patients and that makes them just stop coming in, which you can understand (GP-JNB-16). We do not work on weekends, and so we miss the working population. We need to rethink this, maybe make better use of these ideas going outside into community or workplaces for the chronics (KZN-uM-10). Many facility leaders mentioned the benefits of the CCMDD models but noted the gap between CCMDD and facility-based efforts to support adherence and retention. The overall sense from facility leaders is that PEPFAR-funded NGOs could be more focussed on addressing these challenges with additional capacity. Only 56% said they believed the work of these NGOs was focussed on these barriers. When asked how they would reprogramme existing funds or spend any increase in available funding, facility managers say they would prioritize increasing the number of paid staff regularly based at their facility and funding the direct-service work at the clinic above all else. While data and performance review received some support for increase, only a handful of clinic leaders said they need more training or mentoring as a priority (Figures 16 and 17). Figure 16 View largeDownload slide Are PEPFAR-supported NGOs addressing the biggest challenges identified by facility managers? Figure 16 View largeDownload slide Are PEPFAR-supported NGOs addressing the biggest challenges identified by facility managers? Figure 17 View largeDownload slide What would you rather NGOs focus on? Figure 17 View largeDownload slide What would you rather NGOs focus on? Limited incentives for scale-up and retention in the public sector Finally, we asked clinic managers about the incentive structure within the public sector. While PEPFAR implementing partners have targets they are required to hit and are incentivized to speed up testing and enrolment and increase retention rates, we were curious about whether the public-sector staff managing these facilities had similar incentives. We therefore asked if there was any benefit to scale up faster or improve retention rates—or any consequence if they did not. Overall, respondents repeatedly expressed that they understood the benefits to the community of reaching HIV treatment saturation and felt a professional obligation to move as quickly as possible. As one put it: The benefit will be that the clinic won’t be crowded with terminally ill people, the nurses will be able to concentrate on other chronics and follow-ups. That is the hope at least that keeps us going (GP-JNB-13). However, they also expressed a recognition that doing so would come with significantly increased workload and very few benefits to themselves, the clinic, or their staff. ‘We get more work is what it means’, noted one respondent, ‘and I don’t think my nurses want more work, they are very unhappy with me right now’ (GP-JNB-15). This suggests a significant area of work for PEPFAR and Government of South Africa—to seek to align incentive structures such that the NGOs and public-sector workers have clear and similar incentives to focus on scale-up (Figure 18). Figure 18 View largeDownload slide Is there any incentive (beyond personal sense of duty) for manager or clinic staff to add more people to treatment faster? Figure 18 View largeDownload slide Is there any incentive (beyond personal sense of duty) for manager or clinic staff to add more people to treatment faster? Conclusion External financing to fight HIV in South Africa has been a remarkable success among international aid programmes—scaling up HIV treatment and prevention programmes from a time before the government embraced a science-based response. In recent years, efforts by PEPFAR to adapt to a changing political and economic context have led to substantial policy shifts aimed at improving aid effectiveness and addressing with the proper role of aid in middle-income countries—issues that have occupied significant policy attention globally. PEPFAR had previously decided to transition out of aid for direct HIV services in South Africa, including pulling back from investments in human resources. The recent decision to halt this transition and re-invest in funding for direct front-line HIV services in high-priority districts faced policy legacies that undermined implementation. We conducted visits to a random sample of aid-supported public health facilities with large numbers of people on HIV treatment in high-priority districts. Data gathered from health facility managers and clinicians suggest an important role for PEPFAR-supported initiatives. However, they also reveal the degree to which local aid officials—in this case local PEPFAR and NGO officials—have not yet shifted financing to implement the policy change through expanding human resources and other front-line investments. This stands in contrast to concerns expressed that implementing the new policy resulted in duplication or resistance from public-sector workers. PEPFAR’s primary investment at this level is in a small cadre of health workers based at or visiting each clinic. With high ratios and low per-clinic numbers, these are not sufficient to provide the additional service intensity needed. Insufficient progress against the stated goals to reach 90–90–90 in these focus districts is likely due in part to an insufficient ‘dose’ of additional direct service support to achieve the desired ‘response’. We also found that the distribution of PEPFAR-supported HRH is not well aligned with the patient load of a facility. Mentoring and training, on the other hand, aligns far better with the ideas and incentives of the previous transition paradigm of ‘transition’ out of front-line services into a support and technical assistance role. While front-line facility leaders value training and mentoring, they are valued less than other inputs and we show evidence they are not resulting in significant changes in practice. South Africa has a mature AIDS response, in which many of these training and mentoring activities have been ongoing for years, so a level of diminishing returns could be expected. Moving some funding out of these areas and into new priorities is warranted, but doing so requires departing from the previous policy paradigm. We note that PEPFAR has taken on board some of these insights and is planning a 2019 ‘surge’ of front-line investments.3 As PEPFAR considers re-prioritizing, public-sector facility managers have insights about what is needed. They spotlight increased direct-service staff—especially for outreach, treatment literacy, and lost-to-follow-up tracing, which our data suggest is a gap in both government- and PEPFAR-funded capacities. Meanwhile, interviewees identified significant limitations in facility-based models, which mirrors the growing consensus that building community-based alternatives is necessary (Duncombe et al., 2015). In the words of one uMgungundlovu facility leader, ‘We need to rethink this’. Achieving ambitious goals will require differentiated service delivery, community-based drug pick-up and adherence support, and disruptive models that are better at reaching young people, men, key populations and others. Models including those piloted in South Africa by MSF and from the SEARCH study could be taken to scale to address this need (Bemelmans et al., 2014; Perriat et al., 2018). On a broader level, this case study reveals the need for a focus on the micro-politics of implementation in global health aid policy. The legacies of the previous policy of transition away from aid have not been easily shed. This is predictable in light of policy implementation research that suggests a shift like this, which directly contradicts the ideas, incentives, and underlying beliefs behind the prior policy, will face resistance. Multiple actors including locally-based aid officials and NGOs receiving funding have significant discretion that was not fully considered. While PEPFAR South Africa’s struggles to achieve its goals suggested a problem, the limited degree of policy implementation was not immediately apparent to senior leaders in PEPFAR, which suggests a need to set clearer benchmarks and indicators of implementation when aid policy shifts like these are undertaken. Front-line public-sector managers, meanwhile, may be untapped allies in implementation—with motivations and knowledge distinct from that of the aid agency and NGO officials most often involved in PEPFAR processes. This is likely especially true in the broader universe of middle-income countries where a significant, well-capacitated bureaucracy has a large role to play in fostering or hindering policy change. As current debates on achieving disease-fighting targets and making foreign assistance more effective are translated into specific policy change efforts, identifying ways to bring front-line health leaders into the process could help drive more rapid and comprehensive implementation. Footnotes 1 Currency fluctuations make exact figures difficult, but PEPFAR COP-level spending (excluding additional HQ top-ups) represents between 22 and 27% of total HIV expenditures in 2017. 2 PEPFAR differentiates between those served through direct service delivery versus technical assistance (PEPFAR, 2017b). 3 A version of this analysis was presented at the PEPFAR Regional Planning Meetings in Johannesburg in February 2018. Acknowledgements We thank the South African National Department of Health and US Office of the Global AIDS Coordinator for key information. Essential research assistance was provided by Nokhwezi Hoboyi, Duduzile Zwana, Gethwana Mahlase, Portia Serote, Themba Sokhela, Mukambilwa Mazambi and Sandile Khumalo, for which we send our appreciation. Funding This work was supported by amfAR: the foundation for AIDS Research (Public Policy Award 109727-62-PAGN). Conflict of interest statement. None declared. References Barrett SM. 2004 . Implementation studies: time for a revival? Personal reflections on 20 years of implementation studies . Public Administration 82 : 249 – 62 . Google Scholar Crossref Search ADS Béland D , Ridde V. 2016 . Ideas and policy implementation: understanding the resistance against free health care in Africa . Global Health Governance 10 : 9 – 23 . Bemelmans M , Baert S , Goemaere E et al. 2014 . Community‐supported models of care for people on HIV treatment in sub‐Saharan Africa . Tropical Medicine & International Health 19 : 968 – 77 . Google Scholar Crossref Search ADS Beracochea E. 2016 . Improving Aid Effectiveness in Global Health . Berlin : Springer . Burns T. 1961 . Micropolitics: mechanisms of institutional change . Administrative Science Quarterly 6 : 257 – 81 . Google Scholar Crossref Search ADS Duncombe C , Rosenblum S , Hellmann N et al. 2015 . Reframing HIV care: putting people at the centre of antiretroviral delivery . Tropical Medicine & International Health 20 : 430 – 47 . Google Scholar Crossref Search ADS Fauci AS , Marston HD. 2015 . Ending the HIV–AIDS pandemic—follow the science . The New England Journal of Medicine 373 : 2197 – 9 . Google Scholar Crossref Search ADS PubMed Gilson L , Schneider H , Orgill M. 2014 . Practice and power: a review and interpretive synthesis focused on the exercise of discretionary power in policy implementation by front-line providers and managers . Health Policy and Planning 29 : iii51 – 69 . Google Scholar Crossref Search ADS PubMed Kavanagh MM. 2014 . The Politics and Epidemiology of Transition: pEPFAR and AIDS in South Africa . Journal of Acquired Immune Deficiency Syndromes (1999) 65 : 247 – 50 . Google Scholar Crossref Search ADS PubMed Larson E , O’Bra H , Brown JW , Mbengashe T , Klausner JD. 2012 . Supporting the massive scale-up of antiretroviral therapy: the evolution of PEPFAR-supported treatment facilities in South Africa, 2005–2009 . BMC Public Health 12 : 173 . Google Scholar Crossref Search ADS PubMed Lipsky M. 2010 . Street-Level Bureaucracy, 30th ann. Ed.: Dilemmas of the Individual in Public Service . New York : Russell Sage Foundation . Markham A , Akerfeldt K , Philips M. 2015 . Political pressure from donors to reduce or end health aid on basis of country income classification risks damage to people’s health in middle income countries and undermining effectiveness of global health initiatives . Tropical Medicine & International Health 20 : 96 . Mayosi BM , Benatar SR. 2014 . Health and health care in South Africa—20 years after Mandela . The New England Journal of Medicine 371 : 1344 – 53 . Google Scholar Crossref Search ADS PubMed Mosley L. 2013 . Interview Research in Political Science . Ithaca : Cornell University Press . PEPFAR. 2016 . South Africa Country Operational Plan 2016, Strategic Direction Summary. PEPFAR. 2017a . South Africa Country Operational Plan 2017, Strategic Direction Summary. PEPFAR. 2017b . Monitoring, Evaluation, and Reporting (MER 2.0) Indicator Reference Guide. PEPFAR. 2018 . PANORAMA Database. U.S. Department of State. Perriat D , Balzer L , Hayes R et al. 2018 . Comparative assessment of five trials of universal HIV testing and treatment in sub-Saharan Africa . Journal of the International AIDS Society 21 : e25048. Google Scholar Crossref Search ADS Pierson P. 2004 . Politics in Time: History, Institutions, and Social Analysis . Princeton : Princeton University Press . Pressman JL , Wildavsky A. 1984 . Implementation: How Great Expectations in Washington Are Dashed in Oakland; Or, Why It’s Amazing that Federal Programs Work at All, This Being a Saga of the Economic Development Administration as Told by Two Sympathetic Observers Who Seek to Build Morals on a Foundation . Oakland, CA : University of California Press . Resch S , Hecht R. 2018 . Transitioning financial responsibility for health programs from external donors to developing countries: key issues and recommendations for policy and research . Journal of Global Health 8 : 01031-01036 . Google Scholar Crossref Search ADS Sabatier PA , Jenkins-Smith H. 1993 . Policy Change and Learning: An Advocacy Coalition Framework . Westview : Boulder CO . SAG & USG. 2010 . Partnership Framework in Support of South Africa’s National HIV & AIDS and TB Response 2012/13–2016/17. SANAC. 2017 . South African National Strategic Plan for HIV, TB and STIs 2017–2022. South African National AIDS Council. Sidibé M , Loures L , Samb B. 2016 . The UNAIDS 90–90–90 target: a clear choice for ending AIDS and for sustainable health and development . Journal of the International AIDS Society 19 : 2133 – 2134 . United Nations. 2016 . Political declaration on HIV and AIDS: on the fast-track to accelerate the fight against HIV and to end the AIDS epidemic by 2030, A/70/L.52. U.S. Mission South Africa. 2016 . Media Advisory: PEPFAR Investing More Than $410 Million Towards an AIDS-Free Generation in South Africa, Pretoria. Van Damme W , Kober K , Kegels G. 2008 . Scaling-up antiretroviral treatment in Southern African countries with human resource shortage: how will health systems adapt? Social Science & Medicine (1982) 66 : 2108 – 21 . Google Scholar Crossref Search ADS PubMed Walensky RP , Kuritzkes DR. 2010 . The impact of the President’s Emergency Plan for AIDS Relief (PEPfAR) beyond HIV and why it remains essential . Clinical Infectious Diseases 50 : 272 – 5 . Google Scholar Crossref Search ADS PubMed Walt G. 1994 . Health policy: an introduction to process and power. Johannesburg: Witwatersrand University Press. © The Author(s) 2019. 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)

Renggli, Sabine; Mayumana, Iddy; Mshana, Christopher; Mboya, Dominick; Kessy, Flora; Tediosi, Fabrizio; Pfeiffer, Constanze; Aerts, Ann; Lengeler, Christian

doi: 10.1093/heapol/czy091pmid: 30689879

Abstract In Tanzania, the health financing system is extremely fragmented with strategies in place to supplement funds provided from the central level. One of these strategies is the Community Health Fund (CHF), a voluntary health insurance scheme for the informal rural sector. As its implementation has been challenging, we investigated different CHF implementation practices and how these practices and the wider health financing context affect CHF implementation and potentially enrolment. Two councils were purposively selected for this study. Routine data relevant for understanding CHF implementation in the wider health financing context were collected at council and public health facility level. Additionally, an economic costing approach was used to estimate CHF administration cost and analyse its financing sources. Our results showed the importance of considering different CHF implementation practices and the wider health financing context when looking at CHF performance. Exemption policies and healthcare-seeking behaviour influenced negatively the maximum potential enrolment rate of the voluntary CHF scheme. Higher revenues from user fees, user fee policies and fund pooling mechanisms might have furthermore set incentives for care providers to prioritize user fees over CHF revenues. Costing results clearly pointed out the lack of financial sustainability of the CHF. The financial analysis however also showed that thanks to significant contributions from other health financing mechanisms to CHF administration, the CHF could be left with more than 70% of its revenues for financing services. To make the CHF work, major improvements in CHF implementation practices would be needed, but given the wider health financing context and healthcare-seeking behaviours, it is questionable whether such improvements are feasible, scalable and value for money. Thus, our results call for a reconsideration of approaches taken to address the challenges in health financing and demonstrate that the CHF cannot be looked at as a stand-alone system. Tanzania, community-based health insurance, health financing, health system research, operations research Introduction Key Messages When looking at CHF performance, it is important to consider council-specific CHF implementation practices and the wider health financing context. Exemption policies and healthcare-seeking behaviour influences negatively the maximum potential enrolment rate of the voluntary CHF scheme Higher revenues from user fees, user fee policies and fund pooling mechanisms can set incentives for care providers to prioritize user fees over CHF revenues Major improvements in CHF implementation practices would be needed to make the CHF work, but given the wider health financing context and healthcare-seeking behaviours, it is questionable whether such improvements are feasible, scalable and value for money Following the publication of the World Health Report 2010 and the formulation of the health-related Sustainable Development Goal 3, Universal Health Coverage (UHC) has gained high priority globally (World Health Organization, 2010; Sustainable Development Solution Network, 2015). UHC implies that everyone has access to needed health services of sufficient quality to be effective without incurring financial hardship (World Health Organization, 2010). However, many low- and middle-income countries have been struggling to implement sustainable health financing strategies. A major problem is the informal nature of their economies, which makes revenue collection to fund health systems difficult. Underlying mechanisms of health financing systems also pose challenges (World Health Organization, 2013). The basis to address these challenges lies in the in-depth understanding of the context-specific and often complex designs and implementation practices of existing health financing systems (World Health Organization, 2010, 2013). In Tanzania, the healthcare system primarily depends on central level funding coming from tax revenues or external donors (Dutta, 2015). There are also several insurance schemes and out-of-pocket payments account for around 23% of total health expenditure (World Health Organization, 2014). Overall, the health financing system is extremely fragmented, both in terms of insurance schemes and within the central level funding system (McIntyre et al., 2008; Haazen, 2012; Borghi et al., 2013; Dutta, 2015). User fees paid out of pocket are levied at the point of access, whereby the councils define the amount to be paid in their user fee policies. National exemption policies stipulate that the poor and other priority groups (children under five, pregnant women, elderly above 60 and people with certain disease conditions, including chronic illnesses, HIV/AIDS, TB and leprosy) are supposed to receive free services at public health facilities (Mubyazi, 2004). All public servants are compulsorily enrolled in the National Health Insurance Fund (NHIF) (McIntyre et al., 2008). Voluntary insurance schemes include the Community Health Funds (CHFs) for the informal rural population (Haazen, 2012). Each council is responsible for administrating its own CHF and defining the benefit package and flat rate premium per year. The CHF scheme covers a whole household. CHF funds raised are doubled through matching grants from the central government via the NHIF (Joseph and Maluka, 2016). Resources collected through CHF revenues, matching grants, user fees and NHIF reimbursements are referred to as ‘Cost Sharing and Insurance Funds (CSIFs)’ (Ifakara Health Institute, 2013). The pooling mechanism of these funds is defined by the councils. Key CSIFs stakeholders within a council are described in Box 1 and Figure 1. Box 1. Key stakeholders of Cost Sharing and Insurance Funds within a council (Figure 1) Council level The Council Health Service Board (CHSB), consisting of community and private health sector representatives, is the governance body overseeing the Council Health Management Team (CHMT) (Kessy et al., 2008; Kessy, 2014). The CHSB is responsible for the management and administration of the CSIFs (Mtei and Mulligan, 2007; Kessy et al., 2008). This includes mobilizing and allocating funds, issuing CHF membership cards to exempted households and verifying the collection and expenditure of funds (United Republic of Tanzania, 2001). The CHSB receives technical input from the CHMT through the Council Medical Officer. The CHMT is in charge of monitoring and assuring the quality of services provided (United Republic of Tanzania, 2001). The CHF and NHIF coordinators are typically members or co-opted members of the CHMT (Borghi et al., 2015). The CHF coordinator, who is supported by a council health accountant, oversees the operation of the CHF and tracks membership, fund generation and use (Borghi et al., 2015). It is the duty of the council (often the CHF coordinator) to claim the matching funds from the NHIF. The NHIF coordinator compiles the NHIF claim forms and forwards them to the NHIF office. NHIF reimburses the council or directly the health facility for expenses based on the submitted claim forms. Ward and village level The Ward Development Committee (WDC) at ward level and the Village Council (VC) at village level are in charge of sensitizing and mobilizing community members (e.g. during the Village Assembly) and identifying poor households eligible for exemptions (United Republic of Tanzania, 2001). Health facility level At facility level the Health Facility Governing Committees (HFGCs), composed of community representatives, oversee the facility operations. They are responsible for the mobilization of financial resources to run the health facility and liaising with the CHSB (Kessy et al., 2008; Kessy, 2014). The Health Facility Management Team (HFMT) enrols community members into the CHF, collects contributions (CHF revenues, user fees) and completes NHIF claim forms (Kessy, 2014; Borghi et al., 2015). Figure 1. Open in new tabDownload slide Key stakeholders of cost sharing and insurance funds within a council. Solid lines indicate official reporting hierarchies, dashed lines indicate further relevant interactions and stakeholders within the dotted box belong to the health facility level Figure 1. Open in new tabDownload slide Key stakeholders of cost sharing and insurance funds within a council. Solid lines indicate official reporting hierarchies, dashed lines indicate further relevant interactions and stakeholders within the dotted box belong to the health facility level National CHF enrolment rate in 2015 was around 4.5% (Ministry of Health Community Development Gender Elderly and Children et al., 2016), indicating that the target of 30% enrolment by 2015 had not been reached (Ministry of Health and Social Welfare, 2009, 2015). Numerous studies have investigated reasons for low enrolment. Among them are low quality of care, high premium rates, limited benefit packages, lack of trust in the scheme or healthcare provider and failure to see the rationale of an insurance scheme (Kamuzora and Gilson, 2007; Mtei and Mulligan, 2007; Kessy et al., 2008; Stoermer et al., 2011, 2012; Ministry of Health and Social Welfare, 2012; Borghi et al., 2013; Macha et al., 2014; Maluka and Bukagile, 2014; Kalolo et al., 2015, 2018; Kapologwe et al., 2017). Additionally, issues in governance were observed in terms of insufficiently capacitated or functioning CHSBs, HFGCs and WDCs and regarding the role of the NHIF in managing the CHF (Kamuzora and Gilson, 2007; Mtei and Mulligan, 2007; Kessy et al., 2008; Stoermer et al., 2011, 2012; Borghi et al., 2013, 2015; Ministry of Health and Social Welfare, 2013; Kessy, 2014; Maluka and Bukagile, 2014; Mkumbo and Masbayi, 2014; Kalolo et al., 2015, 2018; Joseph and Maluka, 2016). Some studies also described problems of insufficient council management commitment, high administration cost, inadequate supportive supervision, a weak medical supply chain and missing mechanisms for service purchasing, claim processing and risk equalization or cross-subsidization (Kamuzora and Gilson, 2007; Mtei and Mulligan, 2007; Kessy et al., 2008; Stoermer et al., 2011, 2012; Borghi et al., 2013, 2015; Macha et al., 2014; Maluka and Bukagile, 2014; Joseph and Maluka, 2016). Furthermore, inadequate fund pooling, insufficient transparency and accountability, as well as poor data quality and management were mentioned in connection with low CHF enrolment (Kamuzora and Gilson, 2007; Kessy et al., 2008; Stoermer et al., 2011, 2012; Ministry of Health and Social Welfare, 2012; Borghi et al., 2013, 2015; Frumence et al., 2014; Macha et al., 2014; Maluka and Bukagile, 2014; Mkumbo and Masbayi, 2014; Kalolo et al., 2015, 2018; Joseph and Maluka, 2016). Lastly, exemption policies were reported to potentially discourage people from joining the CHF (Kamuzora and Gilson, 2007; Mtei and Mulligan, 2007; Kessy et al., 2008; Nangawe, 2012; Idd et al., 2013; Maluka, 2013; Ministry of Health and Social Welfare, 2013; Macha et al., 2014). However, little detailed evidence has been provided about how CHF implementation is affected by council-specific CHF implementation decisions. These council-specific implementation practices, which differ from one council to the other, include the overall CHF administration and the definition of the premium and benefit package. Neither is there much information about how the success of these council-specific CHF implementation practices is influenced by the wider health financing context, meaning council defined user fee policies and fund pooling mechanisms as well as exemption policies and other health financing mechanisms. Hence, this article aims to investigate council-specific CHF implementation practices and how these practices and the wider health financing context within a council affect CHF implementation and therewith potentially enrolment. Methods Description of study councils Two rural councils ‘A’ and ‘B’ from the same region were selected. Both benefited from the ‘Initiative to Strengthen Affordability and Quality of Healthcare (ISAQH)’, with which the authors were associated and which aimed to expand CHF coverage through: (1) CHF implementation training for all relevant stakeholders (2012), (2) CHF forum (2013), (3) CHF radio spots (2012–14), (4) supportive supervision on CHF data management (2012–14) and [5] village sensitization meetings (2012 for both councils and 2013 for council A only). Councils were chosen because of their difference in perceived CHF implementation capacity as judged by ISAQH staff. Council A was perceived as better performing than council B. Relevant council characteristics and specific health financing decisions (CHF premium, CHF benefit package, user fee policies and fund pooling mechanisms) are described in Table 1. Supplementary Figure S1 summarizes CHF administration activities reported to be conducted by each council. Table 1. Description of study councils (status 2014) Characteristics . Council A . Council B . Population sizea ∼250 000 ∼400 000 Average household sizea 4.9 4.3 Number of health facilitiesb 38 59 Number of public health facilities (hospitals/health centres/ dispensaries)b 27 (23/3/1) 25 (20/5/0g) Perceived CHF implementation capacity Medium Low Year of CHF introductionc 2003 2008/9 CHF premiumc 3.01/6.02 USDe,f 6.02 USDf CHF benefit packagec,d Maximum of six beneficiaries from one household per CHF card and unlimited access to all services offered at any public health facility within the council, including the council hospital Maximum of five beneficiaries from one household per CHF card with access limited to all services offered at the health facility, where CHF registration took place User fee policyd ‘Fixed’ (independent of treatment): 0.90 USD at public dispensaries or health centres including all services; 1.20 USD at the public hospital for registration/consultation and various prices for medical supplies, diagnostics or any other additional services ‘Flexible’ (depending on treatment): 0.12–1.08 USD for registration/consultation and various prices for medical supplies, diagnostics or any other additional services at all public health facilities Fund poolingd Cost Sharing and Insurance Funds pooled at council level Cost Sharing and Insurance Funds pooled at health facility level Role of CHF coordinator Dental Medical Officer at council hospital Health facility in-charge (medical officer) at main council health centre Characteristics . Council A . Council B . Population sizea ∼250 000 ∼400 000 Average household sizea 4.9 4.3 Number of health facilitiesb 38 59 Number of public health facilities (hospitals/health centres/ dispensaries)b 27 (23/3/1) 25 (20/5/0g) Perceived CHF implementation capacity Medium Low Year of CHF introductionc 2003 2008/9 CHF premiumc 3.01/6.02 USDe,f 6.02 USDf CHF benefit packagec,d Maximum of six beneficiaries from one household per CHF card and unlimited access to all services offered at any public health facility within the council, including the council hospital Maximum of five beneficiaries from one household per CHF card with access limited to all services offered at the health facility, where CHF registration took place User fee policyd ‘Fixed’ (independent of treatment): 0.90 USD at public dispensaries or health centres including all services; 1.20 USD at the public hospital for registration/consultation and various prices for medical supplies, diagnostics or any other additional services ‘Flexible’ (depending on treatment): 0.12–1.08 USD for registration/consultation and various prices for medical supplies, diagnostics or any other additional services at all public health facilities Fund poolingd Cost Sharing and Insurance Funds pooled at council level Cost Sharing and Insurance Funds pooled at health facility level Role of CHF coordinator Dental Medical Officer at council hospital Health facility in-charge (medical officer) at main council health centre aNational Bureau of Statistics (2013). bSource: Comprehensive Council Health Plans of selected councils collected by SR and IM. cSource: CHF reports of selected councils collected by SR and IM. dSource: Informal personal communication and observational data from selected councils collected by SR and IM. eCHF premium changed from 3.01 USD to 6.02 USD mid-October 2014. fAnnual average exchange rate for 2014 (1662 TSh = 1 USD) (Bank of Tanzania, 2017). gThere is a designated non-public referral hospital in council B. Open in new tab Table 1. Description of study councils (status 2014) Characteristics . Council A . Council B . Population sizea ∼250 000 ∼400 000 Average household sizea 4.9 4.3 Number of health facilitiesb 38 59 Number of public health facilities (hospitals/health centres/ dispensaries)b 27 (23/3/1) 25 (20/5/0g) Perceived CHF implementation capacity Medium Low Year of CHF introductionc 2003 2008/9 CHF premiumc 3.01/6.02 USDe,f 6.02 USDf CHF benefit packagec,d Maximum of six beneficiaries from one household per CHF card and unlimited access to all services offered at any public health facility within the council, including the council hospital Maximum of five beneficiaries from one household per CHF card with access limited to all services offered at the health facility, where CHF registration took place User fee policyd ‘Fixed’ (independent of treatment): 0.90 USD at public dispensaries or health centres including all services; 1.20 USD at the public hospital for registration/consultation and various prices for medical supplies, diagnostics or any other additional services ‘Flexible’ (depending on treatment): 0.12–1.08 USD for registration/consultation and various prices for medical supplies, diagnostics or any other additional services at all public health facilities Fund poolingd Cost Sharing and Insurance Funds pooled at council level Cost Sharing and Insurance Funds pooled at health facility level Role of CHF coordinator Dental Medical Officer at council hospital Health facility in-charge (medical officer) at main council health centre Characteristics . Council A . Council B . Population sizea ∼250 000 ∼400 000 Average household sizea 4.9 4.3 Number of health facilitiesb 38 59 Number of public health facilities (hospitals/health centres/ dispensaries)b 27 (23/3/1) 25 (20/5/0g) Perceived CHF implementation capacity Medium Low Year of CHF introductionc 2003 2008/9 CHF premiumc 3.01/6.02 USDe,f 6.02 USDf CHF benefit packagec,d Maximum of six beneficiaries from one household per CHF card and unlimited access to all services offered at any public health facility within the council, including the council hospital Maximum of five beneficiaries from one household per CHF card with access limited to all services offered at the health facility, where CHF registration took place User fee policyd ‘Fixed’ (independent of treatment): 0.90 USD at public dispensaries or health centres including all services; 1.20 USD at the public hospital for registration/consultation and various prices for medical supplies, diagnostics or any other additional services ‘Flexible’ (depending on treatment): 0.12–1.08 USD for registration/consultation and various prices for medical supplies, diagnostics or any other additional services at all public health facilities Fund poolingd Cost Sharing and Insurance Funds pooled at council level Cost Sharing and Insurance Funds pooled at health facility level Role of CHF coordinator Dental Medical Officer at council hospital Health facility in-charge (medical officer) at main council health centre aNational Bureau of Statistics (2013). bSource: Comprehensive Council Health Plans of selected councils collected by SR and IM. cSource: CHF reports of selected councils collected by SR and IM. dSource: Informal personal communication and observational data from selected councils collected by SR and IM. eCHF premium changed from 3.01 USD to 6.02 USD mid-October 2014. fAnnual average exchange rate for 2014 (1662 TSh = 1 USD) (Bank of Tanzania, 2017). gThere is a designated non-public referral hospital in council B. Open in new tab Routine data collection Routine data relevant for the understanding of council-specific CHF implementation practices (overall CHF administration and definition of premium and benefit package) and the wider health financing context (user fee policies, fund pooling mechanisms, exemption policies and other health financing mechanisms) were collected at public health facility and council level for the financial year (FY) 2013/14 or the calendar year 2014 between February and March 2015. Data collected at public health facilities We collected data on the number of households enrolled in the CHF, the number of out-patient visits by financing source (CHF, NHIF, exempted, user fee), as well as the amount of revenues by financing source (CHF, user fee, other) and expenditures from all public health facility for each month in 2014. In council B, one dispensary could not be reached due to its remote location. Yearly averages for CHF enrolment, the number of out-patient visits, revenues and expenditures by health facility level (dispensary, health centre, hospital) were calculated for 2014 (if not specified otherwise). Total council figures were based on health facility level averages and the total number of public health facilities per council, except where indicated otherwise. Revenues and expenditure were converted from Tanzanian Shillings (TSh) to USD using the annual average exchange rate for 2014 (1662 TSh = 1 USD) (Bank of Tanzania, 2017). The required routine data were often available owing to a data collection sheet distributed to all public health facilities by ISAQH. To cross verify the data and fill gaps, other available documentation was used. This included CHF counter books, CHF register books designed by NHIF, CHF membership cards, CHF receipt books, out-patient registers, monthly or yearly out-patient or financial health facility reports and cash books. In rare cases in council A where no other data source was available reports from the CHF coordinator or ISAQH were used to obtain CHF enrolment data. If data for a particular month could not be found in any of the sources, the average of available months was taken to compute the missing data. In case this could not reliably be estimated, the health facility was excluded from average calculations for that particular value, leading to different numbers of units considered (N) in Table 2. Table 2 Routine data collected at public health facilities for the year 2014 by level of care and for the total council . Council A . Council B . . Dispensary (N = 23) . Health centre (N = 3) . Hospital (N = 1) . Total council . Dispensary (N = 20) . Health centre (N = 5) . Total council . . . N . . N . . N . . . N . . N . . Yearly CHF enrolment Households 146 23 328 3 975 1 5327 19 19 97 5 866 Yearly number of out-patient visits at public health facilities by financing source Total 5946 16 19 458a 1 12 821a 1 207 951 4127 19 15 115 4 158 108  CHF (% of total) 3202 (54%) 16 6908 (36%)a 1 3398 (27%)a 1 97 760 (47%) 347 (8%) 2 NA 0 NA  NHIF (% of total) 87 (1%) 16 272 (1%)a 1 1018 (8%)a 1 3829 (2%) 64 (2%) 2 NA 0 NA  User fee (% of total) 151 (3%) 16 1630 (8%)a 1 7831 (61%)a 1 16 203 (8%) 1325 (32%) 19 6522 (43%) 4 59 103 (37%)  Exempted (% of total) 2506 (42%) 16 10 648 (55%)a 1 574 (4%)a 1 90 158 (43%) 2390 (58%) 2 NA 0 NA Yearly revenues and expenditure at public health facilities in USD by financing source Total revenue 694 18 2303 2 NA 0 22 881b 3008 19 22 125 5 170 781  CHF (% of total) 546 (79%) 18 845 (37%) 2 NA 0 15 094 (66%)b 114 (4%) 19 589 (3%) 5 5225 (3%)  User fee (% of total) 142 (20%) 18 1458 (63%) 2 NA 0 7633 (33%)b 2865 (95%) 19 19 337 (87%)c 5 153 982 (90%)  Other (% of total) 7 (1%) 18 0 (0%) 1 NA 0 154 (1%)b 29 (1%) 19 2199 (10%) 5 11 575 (7%) Total expenditure 11 18 193 2 NA 0 834b 2619 19 14 167 4 123 222 % spent 2% 18 8% 2 NA 0 4%b 87% 19 87% 4 87% . Council A . Council B . . Dispensary (N = 23) . Health centre (N = 3) . Hospital (N = 1) . Total council . Dispensary (N = 20) . Health centre (N = 5) . Total council . . . N . . N . . N . . . N . . N . . Yearly CHF enrolment Households 146 23 328 3 975 1 5327 19 19 97 5 866 Yearly number of out-patient visits at public health facilities by financing source Total 5946 16 19 458a 1 12 821a 1 207 951 4127 19 15 115 4 158 108  CHF (% of total) 3202 (54%) 16 6908 (36%)a 1 3398 (27%)a 1 97 760 (47%) 347 (8%) 2 NA 0 NA  NHIF (% of total) 87 (1%) 16 272 (1%)a 1 1018 (8%)a 1 3829 (2%) 64 (2%) 2 NA 0 NA  User fee (% of total) 151 (3%) 16 1630 (8%)a 1 7831 (61%)a 1 16 203 (8%) 1325 (32%) 19 6522 (43%) 4 59 103 (37%)  Exempted (% of total) 2506 (42%) 16 10 648 (55%)a 1 574 (4%)a 1 90 158 (43%) 2390 (58%) 2 NA 0 NA Yearly revenues and expenditure at public health facilities in USD by financing source Total revenue 694 18 2303 2 NA 0 22 881b 3008 19 22 125 5 170 781  CHF (% of total) 546 (79%) 18 845 (37%) 2 NA 0 15 094 (66%)b 114 (4%) 19 589 (3%) 5 5225 (3%)  User fee (% of total) 142 (20%) 18 1458 (63%) 2 NA 0 7633 (33%)b 2865 (95%) 19 19 337 (87%)c 5 153 982 (90%)  Other (% of total) 7 (1%) 18 0 (0%) 1 NA 0 154 (1%)b 29 (1%) 19 2199 (10%) 5 11 575 (7%) Total expenditure 11 18 193 2 NA 0 834b 2619 19 14 167 4 123 222 % spent 2% 18 8% 2 NA 0 4%b 87% 19 87% 4 87% aEstimations were based on average data from 2013 as no data for 2014 was available, but this was considered as realistic because CHF enrolment rate at the particular health centre only changed by 0.3% and at the hospital by 6%. bTotal council figures do not include the hospital due to unavailability of data. cIncludes also user fees collected for in-patient services as this amount could not clearly be separated from the total revenues documented in the health facility. Open in new tab Table 2 Routine data collected at public health facilities for the year 2014 by level of care and for the total council . Council A . Council B . . Dispensary (N = 23) . Health centre (N = 3) . Hospital (N = 1) . Total council . Dispensary (N = 20) . Health centre (N = 5) . Total council . . . N . . N . . N . . . N . . N . . Yearly CHF enrolment Households 146 23 328 3 975 1 5327 19 19 97 5 866 Yearly number of out-patient visits at public health facilities by financing source Total 5946 16 19 458a 1 12 821a 1 207 951 4127 19 15 115 4 158 108  CHF (% of total) 3202 (54%) 16 6908 (36%)a 1 3398 (27%)a 1 97 760 (47%) 347 (8%) 2 NA 0 NA  NHIF (% of total) 87 (1%) 16 272 (1%)a 1 1018 (8%)a 1 3829 (2%) 64 (2%) 2 NA 0 NA  User fee (% of total) 151 (3%) 16 1630 (8%)a 1 7831 (61%)a 1 16 203 (8%) 1325 (32%) 19 6522 (43%) 4 59 103 (37%)  Exempted (% of total) 2506 (42%) 16 10 648 (55%)a 1 574 (4%)a 1 90 158 (43%) 2390 (58%) 2 NA 0 NA Yearly revenues and expenditure at public health facilities in USD by financing source Total revenue 694 18 2303 2 NA 0 22 881b 3008 19 22 125 5 170 781  CHF (% of total) 546 (79%) 18 845 (37%) 2 NA 0 15 094 (66%)b 114 (4%) 19 589 (3%) 5 5225 (3%)  User fee (% of total) 142 (20%) 18 1458 (63%) 2 NA 0 7633 (33%)b 2865 (95%) 19 19 337 (87%)c 5 153 982 (90%)  Other (% of total) 7 (1%) 18 0 (0%) 1 NA 0 154 (1%)b 29 (1%) 19 2199 (10%) 5 11 575 (7%) Total expenditure 11 18 193 2 NA 0 834b 2619 19 14 167 4 123 222 % spent 2% 18 8% 2 NA 0 4%b 87% 19 87% 4 87% . Council A . Council B . . Dispensary (N = 23) . Health centre (N = 3) . Hospital (N = 1) . Total council . Dispensary (N = 20) . Health centre (N = 5) . Total council . . . N . . N . . N . . . N . . N . . Yearly CHF enrolment Households 146 23 328 3 975 1 5327 19 19 97 5 866 Yearly number of out-patient visits at public health facilities by financing source Total 5946 16 19 458a 1 12 821a 1 207 951 4127 19 15 115 4 158 108  CHF (% of total) 3202 (54%) 16 6908 (36%)a 1 3398 (27%)a 1 97 760 (47%) 347 (8%) 2 NA 0 NA  NHIF (% of total) 87 (1%) 16 272 (1%)a 1 1018 (8%)a 1 3829 (2%) 64 (2%) 2 NA 0 NA  User fee (% of total) 151 (3%) 16 1630 (8%)a 1 7831 (61%)a 1 16 203 (8%) 1325 (32%) 19 6522 (43%) 4 59 103 (37%)  Exempted (% of total) 2506 (42%) 16 10 648 (55%)a 1 574 (4%)a 1 90 158 (43%) 2390 (58%) 2 NA 0 NA Yearly revenues and expenditure at public health facilities in USD by financing source Total revenue 694 18 2303 2 NA 0 22 881b 3008 19 22 125 5 170 781  CHF (% of total) 546 (79%) 18 845 (37%) 2 NA 0 15 094 (66%)b 114 (4%) 19 589 (3%) 5 5225 (3%)  User fee (% of total) 142 (20%) 18 1458 (63%) 2 NA 0 7633 (33%)b 2865 (95%) 19 19 337 (87%)c 5 153 982 (90%)  Other (% of total) 7 (1%) 18 0 (0%) 1 NA 0 154 (1%)b 29 (1%) 19 2199 (10%) 5 11 575 (7%) Total expenditure 11 18 193 2 NA 0 834b 2619 19 14 167 4 123 222 % spent 2% 18 8% 2 NA 0 4%b 87% 19 87% 4 87% aEstimations were based on average data from 2013 as no data for 2014 was available, but this was considered as realistic because CHF enrolment rate at the particular health centre only changed by 0.3% and at the hospital by 6%. bTotal council figures do not include the hospital due to unavailability of data. cIncludes also user fees collected for in-patient services as this amount could not clearly be separated from the total revenues documented in the health facility. Open in new tab Data collected at council level At council level, Comprehensive Council Health Plans (CCHPs) and annual combined Technical and Financial Performance Implementation Reports (TFPIRs) were used to analyse the contribution of various funding sources to overall health financing in the FY2013/14. Except for the central government’s in-kind contributions through the Medical Store Department (MSD), funds outside council accounts (contributions from multi- and bilateral partners) were excluded as they could not reliably be tracked within the council system (Ministry of Health and Social Welfare and Prime Minister’s Office Regional Administration and Local Government, 2011). Yet, for reference the contributions from multi- and bilateral partners in council A and B were budgeted to be 1 741 395 USD and 2 338 951 USD in the FY2013/14. In council A, receipts of money submitted by health facility in charges and monthly revenue reports from cash books were obtained from the health accountant. In council B, no such detailed documentation could be obtained. TSh were converted to USD using the annual average exchange rate for the FY2013/2014 (1626 TSh = 1 USD) (Bank of Tanzania, 2017). Cost of CHF administration and its financing sources To explore CHF administration, which is handled independently by each council, we investigated the cost of CHF administration and how the wider health financing context, in particular other financing sources, contributes to this cost. Therefore, an approach similar to the methodology used previously for the CHF in Tanzania was adopted (Borghi et al., 2015). Yearly recurrent costs required for administrating the CHF at health facility and council level were estimated for 2014. For this an ingredient approach was used, whereby quantities of each resource were identified, and valued with the appropriate unit cost (Drummond et al., 2005). Costs were classified by resource (personnel, per diem, transport, other expenses), financing sources (CHF, NHIF, user fee, other public health financing sources, other public or non-public sources), cost type (variable, fixed) and activity (mobilization, fund pooling, stewardship, purchasing). For categorizing activities, the framework of Mathauer and Nicolle (2011) was used. Personnel cost was defined as the cost of staff time and estimated based on their salary and time spent. When estimating the time spent on activities that were not solely conducted to administer the CHF (e.g. HFGC meetings), costs were apportioned accordingly based on information given by respondents (e.g. proportion of time spent on CHF-related issues) (Supplementary Table S1). To identify activities, time spent, resources required and financing sources, a pre-defined data collection template was used to interview 22 informants: CHF coordinator, health accountant, Council Medical Officer and one responsible person for CHF administration at six public dispensaries and two public health centres per council. However, in council B only at one of the three visited health centres an informant was available and willing to provide the required information. This resulted in 11 informants in council A and 10 in council B. Details on cost calculations can be found in Supplementary Annex S1. Overall, council cost was computed by multiplying the average health facility cost with the number of public health facilities per council and adding the council level cost. All costs were calculated in TSh and converted to USD using the annual exchange rate for 2014. Results Routine data collected at public health facilities Table 2 displays routine data collected at public health facilities relevant for understanding CHF implementation in the wider health financing context. CHF population coverage in 2014 was 11.0% in council A and 1% in council B. Strikingly, in council A most out-patients were either exempted or CHF members and only few paid user fees. This was different in council B, where patients were either exempted or paid user fees. Consequently, a big share of revenues collected at public health facilities in council A came from CHF contributions, while in council B the main source of revenues was user fees. Council B had more than seven times higher total revenues. This was primarily due to the greater number of patients paying user fees and the flexible user fee amount, but also because of a smaller CHF benefit package and bigger CHF premium (Table 1). Council A is therefore losing out financially as a result of higher CHF coverage, a smaller CHF premium, a bigger CHF benefit package and fixed user fees (Table 1). The percentage of revenues spent at public health facilities in council A reflected the fund pooling mechanisms in place (Table 1), with a single council level fund pool (account), where only little cash was transferred back to the health facilities for rehabilitation and renovation (Figure 2). In contrast, the proportion of collected money spent was much higher in council B, with individual health facility level fund pools (accounts). Generally, observations across health facilities revealed that reporting formats were inconsistent, patient registers did not capture the financing source of out-patients (CHF, NHIF, exempted, user fee) and in places with more than one person consulting patients CSIFs data was not consolidated. Figure 2. Open in new tabDownload slide Spending pattern of CHF revenues in council A for the FY2013/14. In council B, no such detailed documentation could be obtained Figure 2. Open in new tabDownload slide Spending pattern of CHF revenues in council A for the FY2013/14. In council B, no such detailed documentation could be obtained Routine data collected at council level To further understand CHF implementation in the wider health financing context, Table 3 shows the contribution of various funding sources to overall health financing in the two study councils for the FY2013/14 based on routine data collected at council level. Funds are divided into funds approved, brought forward, received and spent. Funds brought forward are unspent funds from the previous year (FY2012/13). CHF revenues only made up around 2% of total funds available for health (sum of brought forward and received). The proportion of CHF money brought forward was high compared with its share in the funds approved, received and spent. This reflected the greater difficulty to spend this money relative to funds from other sources. Council A had less problems receiving (81% of approved budget) and spending (41% of brought forward and received) CHF money in comparison to council B (0.3% received of approved budget and 0% spent of brought forward and received). In contrast to the CHF revenues, revenues coming from other CSIFs were spent easier in both councils. Table 3. Contribution of various funding sources to overall health financing by resources approved, brought forward, received and spent for each council in the FY2013/14 [USD] (% of total) . Council A . Council B . Approved budget . Brought forward . Received . Spent . Approved budget . Brought forward . Received . Spent . Personal emolument (LGBGa) 1 421 846 (61%) 0 892 258 (57%) 892 258 (49%) 1 593 944 (49%) 0 1 571 962 (57%) 1 571 962 (59%) Other charges (LGBGa) 119 741 (5%) 18 132 (3%) 130 044 (8%) 103 012 (6%) 221 997 (7%) 0 188 932 (7%) 148 005 (6%) Health Sector Basket Fund 318 478 (14%) 137 892 (26%) 318 478 (20%) 369 029 (20%) 492 600 (15%) 263 348 (51%) 492 600 (18%) 474 540 (18%) Health Sector Development Grant 74 124 (3%) 105 677 (20%) 23 067 (1%) 90 023 (5%) 113 809 (4%) 173 893 (34%) 0 164 399 (6%) Local Government Development Grant 116 875 (5%) 241 604 (45%) 12 303 (1%) 203 936 (11%) 0 14 749 (3%) 0 0 Central government other source 0 0 0 0 246 052 (8%) 0 246 052 (9%) 37 587 (1%) Council own source 12 303 (1%) 0 0 0 123 026 (4%) 0 0 0 Receipt in kind (Medical Store Department) 167 780 (7%) 0 113 628 (7%) 113 628 (6%) 223 538 (7%) 0 223 538 (8%) 223 538 (8%) Cost sharing and insurance funds National Health Insurance Fund 19 721 (1%) 0 9421 (1%) 9421 (1%) 24 605 (1%) 10 102 (2%) 0 10 102 (0%) Community Health Fund 44 412 (2%) 21 986 (4%) 36 131c (2%) 23 795 (1%) 169 530 (5%) 55 060 (11%) 554(0%) 0 User fee 23 873 (1%) 0 19 242 (1%) 13 274 (1%) 14 563 (0%) 0 14 563 (1%) 14 563 (1%) Drug Revolving Fundb 7382 (0%) 12 841 (2%) 12 215 (1%) 19 944 (1%) 0 0 0 0 Total 2 326 535 538 131 1 566 786 1 838 319 3 223 664 517 152 2 738 200 2 644 696 [USD] (% of total) . Council A . Council B . Approved budget . Brought forward . Received . Spent . Approved budget . Brought forward . Received . Spent . Personal emolument (LGBGa) 1 421 846 (61%) 0 892 258 (57%) 892 258 (49%) 1 593 944 (49%) 0 1 571 962 (57%) 1 571 962 (59%) Other charges (LGBGa) 119 741 (5%) 18 132 (3%) 130 044 (8%) 103 012 (6%) 221 997 (7%) 0 188 932 (7%) 148 005 (6%) Health Sector Basket Fund 318 478 (14%) 137 892 (26%) 318 478 (20%) 369 029 (20%) 492 600 (15%) 263 348 (51%) 492 600 (18%) 474 540 (18%) Health Sector Development Grant 74 124 (3%) 105 677 (20%) 23 067 (1%) 90 023 (5%) 113 809 (4%) 173 893 (34%) 0 164 399 (6%) Local Government Development Grant 116 875 (5%) 241 604 (45%) 12 303 (1%) 203 936 (11%) 0 14 749 (3%) 0 0 Central government other source 0 0 0 0 246 052 (8%) 0 246 052 (9%) 37 587 (1%) Council own source 12 303 (1%) 0 0 0 123 026 (4%) 0 0 0 Receipt in kind (Medical Store Department) 167 780 (7%) 0 113 628 (7%) 113 628 (6%) 223 538 (7%) 0 223 538 (8%) 223 538 (8%) Cost sharing and insurance funds National Health Insurance Fund 19 721 (1%) 0 9421 (1%) 9421 (1%) 24 605 (1%) 10 102 (2%) 0 10 102 (0%) Community Health Fund 44 412 (2%) 21 986 (4%) 36 131c (2%) 23 795 (1%) 169 530 (5%) 55 060 (11%) 554(0%) 0 User fee 23 873 (1%) 0 19 242 (1%) 13 274 (1%) 14 563 (0%) 0 14 563 (1%) 14 563 (1%) Drug Revolving Fundb 7382 (0%) 12 841 (2%) 12 215 (1%) 19 944 (1%) 0 0 0 0 Total 2 326 535 538 131 1 566 786 1 838 319 3 223 664 517 152 2 738 200 2 644 696 aLocal Government Block Grants (LGBGs) are divided into ‘Personal emolument’ (salaries) and ‘Other charges’ (statutory employment benefits). bMoney obtained from selling medicines at hospital level (only in councils with a public hospital) (McIntyre et al., 2008). cComposition of CHF (45%) and matching fund (34%) contributions from all levels of care as well as NHIF (14%) and user fees (6%) from health centres and dispensaries. A total of 2% are of unknown source. Open in new tab Table 3. Contribution of various funding sources to overall health financing by resources approved, brought forward, received and spent for each council in the FY2013/14 [USD] (% of total) . Council A . Council B . Approved budget . Brought forward . Received . Spent . Approved budget . Brought forward . Received . Spent . Personal emolument (LGBGa) 1 421 846 (61%) 0 892 258 (57%) 892 258 (49%) 1 593 944 (49%) 0 1 571 962 (57%) 1 571 962 (59%) Other charges (LGBGa) 119 741 (5%) 18 132 (3%) 130 044 (8%) 103 012 (6%) 221 997 (7%) 0 188 932 (7%) 148 005 (6%) Health Sector Basket Fund 318 478 (14%) 137 892 (26%) 318 478 (20%) 369 029 (20%) 492 600 (15%) 263 348 (51%) 492 600 (18%) 474 540 (18%) Health Sector Development Grant 74 124 (3%) 105 677 (20%) 23 067 (1%) 90 023 (5%) 113 809 (4%) 173 893 (34%) 0 164 399 (6%) Local Government Development Grant 116 875 (5%) 241 604 (45%) 12 303 (1%) 203 936 (11%) 0 14 749 (3%) 0 0 Central government other source 0 0 0 0 246 052 (8%) 0 246 052 (9%) 37 587 (1%) Council own source 12 303 (1%) 0 0 0 123 026 (4%) 0 0 0 Receipt in kind (Medical Store Department) 167 780 (7%) 0 113 628 (7%) 113 628 (6%) 223 538 (7%) 0 223 538 (8%) 223 538 (8%) Cost sharing and insurance funds National Health Insurance Fund 19 721 (1%) 0 9421 (1%) 9421 (1%) 24 605 (1%) 10 102 (2%) 0 10 102 (0%) Community Health Fund 44 412 (2%) 21 986 (4%) 36 131c (2%) 23 795 (1%) 169 530 (5%) 55 060 (11%) 554(0%) 0 User fee 23 873 (1%) 0 19 242 (1%) 13 274 (1%) 14 563 (0%) 0 14 563 (1%) 14 563 (1%) Drug Revolving Fundb 7382 (0%) 12 841 (2%) 12 215 (1%) 19 944 (1%) 0 0 0 0 Total 2 326 535 538 131 1 566 786 1 838 319 3 223 664 517 152 2 738 200 2 644 696 [USD] (% of total) . Council A . Council B . Approved budget . Brought forward . Received . Spent . Approved budget . Brought forward . Received . Spent . Personal emolument (LGBGa) 1 421 846 (61%) 0 892 258 (57%) 892 258 (49%) 1 593 944 (49%) 0 1 571 962 (57%) 1 571 962 (59%) Other charges (LGBGa) 119 741 (5%) 18 132 (3%) 130 044 (8%) 103 012 (6%) 221 997 (7%) 0 188 932 (7%) 148 005 (6%) Health Sector Basket Fund 318 478 (14%) 137 892 (26%) 318 478 (20%) 369 029 (20%) 492 600 (15%) 263 348 (51%) 492 600 (18%) 474 540 (18%) Health Sector Development Grant 74 124 (3%) 105 677 (20%) 23 067 (1%) 90 023 (5%) 113 809 (4%) 173 893 (34%) 0 164 399 (6%) Local Government Development Grant 116 875 (5%) 241 604 (45%) 12 303 (1%) 203 936 (11%) 0 14 749 (3%) 0 0 Central government other source 0 0 0 0 246 052 (8%) 0 246 052 (9%) 37 587 (1%) Council own source 12 303 (1%) 0 0 0 123 026 (4%) 0 0 0 Receipt in kind (Medical Store Department) 167 780 (7%) 0 113 628 (7%) 113 628 (6%) 223 538 (7%) 0 223 538 (8%) 223 538 (8%) Cost sharing and insurance funds National Health Insurance Fund 19 721 (1%) 0 9421 (1%) 9421 (1%) 24 605 (1%) 10 102 (2%) 0 10 102 (0%) Community Health Fund 44 412 (2%) 21 986 (4%) 36 131c (2%) 23 795 (1%) 169 530 (5%) 55 060 (11%) 554(0%) 0 User fee 23 873 (1%) 0 19 242 (1%) 13 274 (1%) 14 563 (0%) 0 14 563 (1%) 14 563 (1%) Drug Revolving Fundb 7382 (0%) 12 841 (2%) 12 215 (1%) 19 944 (1%) 0 0 0 0 Total 2 326 535 538 131 1 566 786 1 838 319 3 223 664 517 152 2 738 200 2 644 696 aLocal Government Block Grants (LGBGs) are divided into ‘Personal emolument’ (salaries) and ‘Other charges’ (statutory employment benefits). bMoney obtained from selling medicines at hospital level (only in councils with a public hospital) (McIntyre et al., 2008). cComposition of CHF (45%) and matching fund (34%) contributions from all levels of care as well as NHIF (14%) and user fees (6%) from health centres and dispensaries. A total of 2% are of unknown source. Open in new tab Finally, the spending pattern of CHF revenues from council A (23 795 USD) revealed that the revenues were spent as stipulated in the guidelines with at least 70% of expenditure on medicines and supplies (Figure 2). Cost of CHF administration and its financing sources Table 4 shows personnel costs (based on salary and time spent) and financial costs (per diem, transport and other expenses) for CHF administration in the councils A and B. In both councils financial costs only made up about 15% of total cost. Mobilizing people to join the CHF (including enrolment) was the most resource-intense activity at health facility level, both in terms of financial and overall cost. At council level, stewardship of the CHF scheme caused the biggest overall cost, but mobilization activities remained with the largest share of financial cost. Fund pooling and purchasing only marginally contributed to the total cost because little time was spent on these activities (Figure 3). In both councils, important drivers for financial cost were CHF supplies (cards, receipt books), transport cost for fund pooling and per diem cost for mobilization, fund pooling and stewardship. Financial as well as overall cost for administrating the CHF was about double in council A compared with council B. Table 4. Average annual health facility level, council level and council overall cost in USD by input, council, type of resource and activitya for 2014 . Council A . Council B . . Personnel . Per diem . Transport . Other expensesb . Total financialc . Total overalld . Personnel . Per diem . Transport . Other expensesb . Total financialc . Total overalld . Dispensary level Mobilization 2735 0 0 127 127 (4%) 2861 (87%) 753 68 0 18 86 (10%) 839 (63%) Fund pooling 103 0 68 0 68 (40%) 171 (5%) 197 0 68 0 68 (26%) 265 (20%) Stewardship 134 86 30 0 116 (46%) 250 (8%) 160 11 65 0 75 (32%) 235 (18%) Total 2971 86 98 127 310 (9%) 3282 1110 79 133 18 229 (17%) 1340 Health Centre level Mobilization 1296 337 0 282 619 (32%) 1915 (76%) 1776 159 0 85 244 (12%) 2019 (79%) Fund pooling 107 0 68 0 68 (39%) 175 (7%) 6 0 0 0 0.4 (5%) 7 (0%) Stewardship 301 55 60 0 115 (28%) 416 (17%) 399 12 108 0 120 (23%) 519 (20%) Total 1703 392 128 282 802 (32%) 2505 2181 171 108 85 364 (14%) 2545 Hospital level Mobilization 3613 193 0 837 1029 (22%) 4642 (81%) Fund pooling 154 0 68 0 68 (31%) 222 (4%) Stewardship 496 245 67 39 351 (41%) 847 (15%) Total 4263 438 135 875 1448 (25%) 5712 Council level Mobilization 4288 2396 752 0 3148 (42%) 7435 (28%) 1823 1745 376 0 2121 (54%) 3944 (17%) Fund pooling 1100 1092 215 7 1314 (54%) 2414 (9%) 2215 0 0 2 2 (0%) 2217 (9%) Stewardship 10 238 2396 44 581 3022 (23%) 13 260 (49%) 9913 892 78 52 1021 (9%) 10 935 (47%) Purchasing 3723 0 0 2 2 (0%) 3725 (14%) 6367 0 0 2 2 (0%) 6369 (27%) Total 19 350 5884 1011 590 7485 (28%) 26 835 20 318 2637 454 56 3147 (13%) 23 465 Overall council Mobilization 74 687 3599 752 4597 8949 (11%) 83 635 (72%) 25 758 3904 376 781 5061 (16%) 30 819 (49%) Fund pooling 3945 1092 2043 7 3142 (44%) 7087 (6%) 6192 0 1364 4 1368 (18%) 7560 (12%) Stewardship 14 710 4783 984 620 6387 (30%) 21 097 (18%) 15 107 1163 1911 52 3126 (17%) 18 233 (29%) Purchasing 3723 0 0 2 2 (0%) 3725 (3%) 6367 0 0 2 2 (0%) 6369 (10%) Total 97 065 9474 3779 5226 18 479 (16%) 115 545 53 424 5067 3651 839 9557 (15%) 62 981 . Council A . Council B . . Personnel . Per diem . Transport . Other expensesb . Total financialc . Total overalld . Personnel . Per diem . Transport . Other expensesb . Total financialc . Total overalld . Dispensary level Mobilization 2735 0 0 127 127 (4%) 2861 (87%) 753 68 0 18 86 (10%) 839 (63%) Fund pooling 103 0 68 0 68 (40%) 171 (5%) 197 0 68 0 68 (26%) 265 (20%) Stewardship 134 86 30 0 116 (46%) 250 (8%) 160 11 65 0 75 (32%) 235 (18%) Total 2971 86 98 127 310 (9%) 3282 1110 79 133 18 229 (17%) 1340 Health Centre level Mobilization 1296 337 0 282 619 (32%) 1915 (76%) 1776 159 0 85 244 (12%) 2019 (79%) Fund pooling 107 0 68 0 68 (39%) 175 (7%) 6 0 0 0 0.4 (5%) 7 (0%) Stewardship 301 55 60 0 115 (28%) 416 (17%) 399 12 108 0 120 (23%) 519 (20%) Total 1703 392 128 282 802 (32%) 2505 2181 171 108 85 364 (14%) 2545 Hospital level Mobilization 3613 193 0 837 1029 (22%) 4642 (81%) Fund pooling 154 0 68 0 68 (31%) 222 (4%) Stewardship 496 245 67 39 351 (41%) 847 (15%) Total 4263 438 135 875 1448 (25%) 5712 Council level Mobilization 4288 2396 752 0 3148 (42%) 7435 (28%) 1823 1745 376 0 2121 (54%) 3944 (17%) Fund pooling 1100 1092 215 7 1314 (54%) 2414 (9%) 2215 0 0 2 2 (0%) 2217 (9%) Stewardship 10 238 2396 44 581 3022 (23%) 13 260 (49%) 9913 892 78 52 1021 (9%) 10 935 (47%) Purchasing 3723 0 0 2 2 (0%) 3725 (14%) 6367 0 0 2 2 (0%) 6369 (27%) Total 19 350 5884 1011 590 7485 (28%) 26 835 20 318 2637 454 56 3147 (13%) 23 465 Overall council Mobilization 74 687 3599 752 4597 8949 (11%) 83 635 (72%) 25 758 3904 376 781 5061 (16%) 30 819 (49%) Fund pooling 3945 1092 2043 7 3142 (44%) 7087 (6%) 6192 0 1364 4 1368 (18%) 7560 (12%) Stewardship 14 710 4783 984 620 6387 (30%) 21 097 (18%) 15 107 1163 1911 52 3126 (17%) 18 233 (29%) Purchasing 3723 0 0 2 2 (0%) 3725 (3%) 6367 0 0 2 2 (0%) 6369 (10%) Total 97 065 9474 3779 5226 18 479 (16%) 115 545 53 424 5067 3651 839 9557 (15%) 62 981 aActivities were categorized according to Mathauer and Nicolle (2011). bOthers included supplies (e.g. CHF cards and receipts, registration books, printouts) as well as rent, food and refreshment during meetings if applicable. cValues in brackets indicate the percentage of total overall cost for the specific activity. dValues in brackets indicate the percentage of total overall cost for the specific health system level (dispensary, health centre, council or overall council). Open in new tab Table 4. Average annual health facility level, council level and council overall cost in USD by input, council, type of resource and activitya for 2014 . Council A . Council B . . Personnel . Per diem . Transport . Other expensesb . Total financialc . Total overalld . Personnel . Per diem . Transport . Other expensesb . Total financialc . Total overalld . Dispensary level Mobilization 2735 0 0 127 127 (4%) 2861 (87%) 753 68 0 18 86 (10%) 839 (63%) Fund pooling 103 0 68 0 68 (40%) 171 (5%) 197 0 68 0 68 (26%) 265 (20%) Stewardship 134 86 30 0 116 (46%) 250 (8%) 160 11 65 0 75 (32%) 235 (18%) Total 2971 86 98 127 310 (9%) 3282 1110 79 133 18 229 (17%) 1340 Health Centre level Mobilization 1296 337 0 282 619 (32%) 1915 (76%) 1776 159 0 85 244 (12%) 2019 (79%) Fund pooling 107 0 68 0 68 (39%) 175 (7%) 6 0 0 0 0.4 (5%) 7 (0%) Stewardship 301 55 60 0 115 (28%) 416 (17%) 399 12 108 0 120 (23%) 519 (20%) Total 1703 392 128 282 802 (32%) 2505 2181 171 108 85 364 (14%) 2545 Hospital level Mobilization 3613 193 0 837 1029 (22%) 4642 (81%) Fund pooling 154 0 68 0 68 (31%) 222 (4%) Stewardship 496 245 67 39 351 (41%) 847 (15%) Total 4263 438 135 875 1448 (25%) 5712 Council level Mobilization 4288 2396 752 0 3148 (42%) 7435 (28%) 1823 1745 376 0 2121 (54%) 3944 (17%) Fund pooling 1100 1092 215 7 1314 (54%) 2414 (9%) 2215 0 0 2 2 (0%) 2217 (9%) Stewardship 10 238 2396 44 581 3022 (23%) 13 260 (49%) 9913 892 78 52 1021 (9%) 10 935 (47%) Purchasing 3723 0 0 2 2 (0%) 3725 (14%) 6367 0 0 2 2 (0%) 6369 (27%) Total 19 350 5884 1011 590 7485 (28%) 26 835 20 318 2637 454 56 3147 (13%) 23 465 Overall council Mobilization 74 687 3599 752 4597 8949 (11%) 83 635 (72%) 25 758 3904 376 781 5061 (16%) 30 819 (49%) Fund pooling 3945 1092 2043 7 3142 (44%) 7087 (6%) 6192 0 1364 4 1368 (18%) 7560 (12%) Stewardship 14 710 4783 984 620 6387 (30%) 21 097 (18%) 15 107 1163 1911 52 3126 (17%) 18 233 (29%) Purchasing 3723 0 0 2 2 (0%) 3725 (3%) 6367 0 0 2 2 (0%) 6369 (10%) Total 97 065 9474 3779 5226 18 479 (16%) 115 545 53 424 5067 3651 839 9557 (15%) 62 981 . Council A . Council B . . Personnel . Per diem . Transport . Other expensesb . Total financialc . Total overalld . Personnel . Per diem . Transport . Other expensesb . Total financialc . Total overalld . Dispensary level Mobilization 2735 0 0 127 127 (4%) 2861 (87%) 753 68 0 18 86 (10%) 839 (63%) Fund pooling 103 0 68 0 68 (40%) 171 (5%) 197 0 68 0 68 (26%) 265 (20%) Stewardship 134 86 30 0 116 (46%) 250 (8%) 160 11 65 0 75 (32%) 235 (18%) Total 2971 86 98 127 310 (9%) 3282 1110 79 133 18 229 (17%) 1340 Health Centre level Mobilization 1296 337 0 282 619 (32%) 1915 (76%) 1776 159 0 85 244 (12%) 2019 (79%) Fund pooling 107 0 68 0 68 (39%) 175 (7%) 6 0 0 0 0.4 (5%) 7 (0%) Stewardship 301 55 60 0 115 (28%) 416 (17%) 399 12 108 0 120 (23%) 519 (20%) Total 1703 392 128 282 802 (32%) 2505 2181 171 108 85 364 (14%) 2545 Hospital level Mobilization 3613 193 0 837 1029 (22%) 4642 (81%) Fund pooling 154 0 68 0 68 (31%) 222 (4%) Stewardship 496 245 67 39 351 (41%) 847 (15%) Total 4263 438 135 875 1448 (25%) 5712 Council level Mobilization 4288 2396 752 0 3148 (42%) 7435 (28%) 1823 1745 376 0 2121 (54%) 3944 (17%) Fund pooling 1100 1092 215 7 1314 (54%) 2414 (9%) 2215 0 0 2 2 (0%) 2217 (9%) Stewardship 10 238 2396 44 581 3022 (23%) 13 260 (49%) 9913 892 78 52 1021 (9%) 10 935 (47%) Purchasing 3723 0 0 2 2 (0%) 3725 (14%) 6367 0 0 2 2 (0%) 6369 (27%) Total 19 350 5884 1011 590 7485 (28%) 26 835 20 318 2637 454 56 3147 (13%) 23 465 Overall council Mobilization 74 687 3599 752 4597 8949 (11%) 83 635 (72%) 25 758 3904 376 781 5061 (16%) 30 819 (49%) Fund pooling 3945 1092 2043 7 3142 (44%) 7087 (6%) 6192 0 1364 4 1368 (18%) 7560 (12%) Stewardship 14 710 4783 984 620 6387 (30%) 21 097 (18%) 15 107 1163 1911 52 3126 (17%) 18 233 (29%) Purchasing 3723 0 0 2 2 (0%) 3725 (3%) 6367 0 0 2 2 (0%) 6369 (10%) Total 97 065 9474 3779 5226 18 479 (16%) 115 545 53 424 5067 3651 839 9557 (15%) 62 981 aActivities were categorized according to Mathauer and Nicolle (2011). bOthers included supplies (e.g. CHF cards and receipts, registration books, printouts) as well as rent, food and refreshment during meetings if applicable. cValues in brackets indicate the percentage of total overall cost for the specific activity. dValues in brackets indicate the percentage of total overall cost for the specific health system level (dispensary, health centre, council or overall council). Open in new tab Figure 3. Open in new tabDownload slide Estimated annual number of hours spent on CHF administration within a council by type of personnel and activity in 2014 Figure 3. Open in new tabDownload slide Estimated annual number of hours spent on CHF administration within a council by type of personnel and activity in 2014 Similar to the overall cost, time spent administrating the CHF in council A was more than double the amount of council B (Figure 3). It was however interesting that the number of hours spent by public health personnel in council A was less than in council B. This was mainly because in council A front-line workers at health facility level spent less time on CHF administration (particularly mobilization) than in council B (7% and 25% of a single full-time person at dispensary and health centre level in council A vs 12% and 33% in council B; data not shown) and a large share of this work was taken over by HFGC members. As a consequence of responsibilities being more equally shared amongst stakeholders in council A (especially with those outside the public sector), personnel costs in council A were financed to a large extent by non-public money (Figure 4). In contrary, in council B personnel costs were mainly carried by the public sector as most of the activities were implemented by public employees. Personnel costs in both councils were exclusively financed through non-CHF money. Figure 4. Open in new tabDownload slide Contribution of different financing sources to personnel and financial cost incurred for CHF administration by council in 2014. Percentage figures indicate the proportion financed by CHF revenues Figure 4. Open in new tabDownload slide Contribution of different financing sources to personnel and financial cost incurred for CHF administration by council in 2014. Percentage figures indicate the proportion financed by CHF revenues Remarkably, only 25% and 8% of the total financial cost for CHF administration were directly financed by CHF revenues in council A and B, respectively. The percentage in council A was higher because these financial costs (CHF cards and receipt books) were pure variable cost and depended on the number of CHF member households. All additional financial costs for CHF administration were borne by other financing sources, including contributions from NHIF and user fees. In both councils, overall costs mainly consisted out of fixed cost (data not shown). As a result, the administration cost per CHF member household was lower in council A than in council B (Table 5), although overall administration cost was bigger (Table 4). The cost–revenue ratio was 0.50 and 0.92 in councils A and B when only the financial costs were considered. This means the financial administration cost was below the premium paid by a CHF household. When the cost of personnel time was included, the ratio increased to around 3 in council A and 6 in council B, meaning administration cost was more than three or six times above the premium paid by a CHF household. If only considering the administrative cost directly financed through CHF revenues, the cost revenue ratio decreased to 0.12 in council A and 0.07 in council B. This ratio was smaller in council B because administration cost directly financed through CHF money was the same for each household in either council, but premiums were higher in council B. Most importantly, this meant that there was >70% of the CHF revenue left to purchase medicines and supplies and do minor facility renovations (cost paid by CHF revenues/total revenues < 0.3). Table 5. Summary of cost revenue ratios and cost per CHF member household for the year 2014 . Council A . Council B . Enrolment Total number of individuals enrolled (%) 29 048 (11%) 4186 (1%) Total number of households enrolled 5327 866 Premium paid by each household [USD] 3.46 6.02 Total revenues (including matching fund) [USD] 18 408 (36 816) 5212 (10 423) Administration cost [USD] Cost paid by CHF revenues 4565 742 Financial cost 18 479 9557 Total overall cost (including personnel) 115 545 62 981 Cost revenue ratio (including matching fund) Cost paid by CHF revenues/total revenues 0.25 (0.12) 0.14 (0.07) Financial cost/total revenues 1.00 (0.50) 1.83 (0.92) Total overall cost/total revenues 6.28 (3.14) 12.08 (6.04) Cost per CHF member household [USD] Cost paid by CHF revenues/household 0.86 0.86 Financial cost/household 3.47 11.03 Total overall cost/household 21.69 72.72 . Council A . Council B . Enrolment Total number of individuals enrolled (%) 29 048 (11%) 4186 (1%) Total number of households enrolled 5327 866 Premium paid by each household [USD] 3.46 6.02 Total revenues (including matching fund) [USD] 18 408 (36 816) 5212 (10 423) Administration cost [USD] Cost paid by CHF revenues 4565 742 Financial cost 18 479 9557 Total overall cost (including personnel) 115 545 62 981 Cost revenue ratio (including matching fund) Cost paid by CHF revenues/total revenues 0.25 (0.12) 0.14 (0.07) Financial cost/total revenues 1.00 (0.50) 1.83 (0.92) Total overall cost/total revenues 6.28 (3.14) 12.08 (6.04) Cost per CHF member household [USD] Cost paid by CHF revenues/household 0.86 0.86 Financial cost/household 3.47 11.03 Total overall cost/household 21.69 72.72 Open in new tab Table 5. Summary of cost revenue ratios and cost per CHF member household for the year 2014 . Council A . Council B . Enrolment Total number of individuals enrolled (%) 29 048 (11%) 4186 (1%) Total number of households enrolled 5327 866 Premium paid by each household [USD] 3.46 6.02 Total revenues (including matching fund) [USD] 18 408 (36 816) 5212 (10 423) Administration cost [USD] Cost paid by CHF revenues 4565 742 Financial cost 18 479 9557 Total overall cost (including personnel) 115 545 62 981 Cost revenue ratio (including matching fund) Cost paid by CHF revenues/total revenues 0.25 (0.12) 0.14 (0.07) Financial cost/total revenues 1.00 (0.50) 1.83 (0.92) Total overall cost/total revenues 6.28 (3.14) 12.08 (6.04) Cost per CHF member household [USD] Cost paid by CHF revenues/household 0.86 0.86 Financial cost/household 3.47 11.03 Total overall cost/household 21.69 72.72 . Council A . Council B . Enrolment Total number of individuals enrolled (%) 29 048 (11%) 4186 (1%) Total number of households enrolled 5327 866 Premium paid by each household [USD] 3.46 6.02 Total revenues (including matching fund) [USD] 18 408 (36 816) 5212 (10 423) Administration cost [USD] Cost paid by CHF revenues 4565 742 Financial cost 18 479 9557 Total overall cost (including personnel) 115 545 62 981 Cost revenue ratio (including matching fund) Cost paid by CHF revenues/total revenues 0.25 (0.12) 0.14 (0.07) Financial cost/total revenues 1.00 (0.50) 1.83 (0.92) Total overall cost/total revenues 6.28 (3.14) 12.08 (6.04) Cost per CHF member household [USD] Cost paid by CHF revenues/household 0.86 0.86 Financial cost/household 3.47 11.03 Total overall cost/household 21.69 72.72 Open in new tab Discussion Strikingly, although population coverage in council A was just above 10%, only few patients at dispensary and health centre level paid user fees. This clearly indicated that the people seeking public care the most were the exempted and insured. The others were either seeking care in the non-public sector, not at all or only at very late stages, when they had to attend hospital level services (as indicated by a high proportion of user fee patients for the hospital). This suggested and confirmed previous findings that CHF enrolment was likely to be affected by healthcare-seeking behaviour and exemption policies, which stipulate free service provision to groups with a higher likelihood to be in need of care (Mtei and Mulligan, 2007; Macha et al., 2014). These factors also undoubtedly influence negatively the maximum potential enrolment rate which could possibly be reached with a voluntary scheme. On the other hand, the number of patients paying user fees and the council-specific user fee policy, CHF premium and benefit package seemed to impact the total revenues collected. Compared with council B, council A was losing out financially as a result of high CHF coverage (low number of patients paying user fees), fixed user fees independent of the treatment received, a small CHF premium and a bigger CHF benefit package. In contrast, council B had substantial higher revenues due to lower CHF coverage (greater number of patients paying user fees), flexible user fee amounts depending on the treatment received, a smaller CHF benefit package and a bigger CHF premium. Furthermore, the fund pooling mechanism in place had an influence on the availability of money and the subsequent spending pattern at health facility level. This meant that higher revenues from user fees, a flexible user fee policy and fund pooling at health facility level might have set incentives for the supply side to prioritize user fees over CHF revenues, which also poses a problem for equity. Thus, the situation in council B, where revenues from flexible user fees were high and funds were pooled at health facilities, might have provided little incentives for healthcare workers and HFGC members to conduct CHF mobilization activities. At the same time, the higher CHF premium and a smaller benefit package in council B might neither have provided incentives for the demand side to join the CHF despite the high user fees, which is in contrary to expectations (Kessy et al., 2008). This altogether would contribute to explain why enrolment rate was so low in council B. Additionally, the decision in council B to pool and use the CSIFs at health facility level led to insufficient documentation at council level. This made it impossible for the council to know what CSIFs were received at health facility level and how they were spent. Neither did it allow applying for matching funds. Fund pooling at health facility level also made it more difficult to put a mechanism in place for balancing the risk across the many smaller pools, which emerged as a consequence. Documentation was about to be improved at the time when the study was conducted, but without addressing the problem of matching fund application or risk pooling. The latter problems were also reported from other councils elsewhere in the country, whereby the fragmented risk pools were seen as a challenge to equity (Borghi et al., 2013). In contrast, pooling of CSIFs at council level in council A facilitated planning and budgeting as well as risk pooling and other CHF administration processes. This was observed based on the bigger percentage of budgeted CHF revenue received and available revenues spent as well as due to the possibility to request for matching funds, track how available revenues were used and allow for risk sharing through need-based reallocation of funds. Both councils were facing difficulties to spend CHF revenue, because of lengthy and cumbersome overall CHF administration processes attached to it. For example, in council B CHF money collected at council level (prior to the implementation of individual health facility fund pooling) was stuck in the council account and could not be spent because of not clearly defined processes. In council A, use of funds was impeded by the closure of the CHF account and its consolidation with other council accounts, which changed fund access rights. Similar problems with fund usage have been reported by others (Mubyazi et al., 2006; Mtei and Mulligan, 2007; Kessy et al., 2008; Ministry of Health and Social Welfare, 2012; Borghi et al., 2013; Ministry of Health and Social Welfare, 2013; Macha et al., 2014). Also, not knowing the number of CHF patients treated at each health facility from routine data impeded in either of the two councils the possibility of risk-adjusted reallocation of the CHF money. Overall, these administrative hurdles had an impact on the quality of data available for planning and budgeting and made activities planned to be implemented through CHF revenue more unlikely to happen. The problems of CHF administration additionally led to a financial loss as matching funds could not be requested due to the lack of household registration details and/or proof of money submission, similar to what had been noted earlier (Borghi et al., 2013; Kalolo et al., 2015). Consequently, all these bottlenecks in administration led to CHF implementation failures and therewith diminished potential positive effects of a council level health insurance scheme. This may ultimately also have contributed to CHF member dissatisfaction and low enrolment. The selection of the same study approach as used previously by Borghi et al. (2015) for assessing the cost of CHF administration allowed for comparison across studies. Importantly, several key findings could be confirmed: (1) lack of financial sustainability of the CHF as such, (2) substantial personnel cost with a share of around 85% of total cost, (3) workload of front-line health workers in a very similar percentage range of a single full-time person, (4) mobilization as the most significant task at health facility level and CHF stewardship at council level, (5) similar relative cost of different administration activities at health facility, (6) comparable average annual health facility level cost for an average dispensary in council B and (7) higher cost per CHF member household in area where enrolment was lower due to considerable fixed costs. However, in our study we found the total annual council-wide cost to be higher than what was published by Borghi et al. (2015). Yet, detailed comparison with Borghi et al. was difficult because council level cost only included stewardship activities and it was unclear how dispensary and health centre costs were calculated given the number of health facilities in a council and the average annual health facility level cost. Consequently, cost to revenue ratios and cost per CHF member households were also higher than reported previously (Borghi et al., 2015). In contrast to Borghi et al., we found in council A strong engagement of HFGC members in CHF mobilization activities, which reduced the burden of public health workers (Borghi et al., 2015). This showed the importance of considering council-specific CHF implementation practices and suggested that contrary to other places in Tanzania, HFGCs in council A were well informed about their roles and responsibilities (Kessy et al., 2008; Kessy, 2014). Additionally, it was argued before that cost resulting from mobilization activities could be reduced if all or most out-patients in public health facilities were covered by insurance (Borghi et al., 2015). However, we found that substantial mobilization activities would still be needed even if most out-patients had insurance coverage as seen in council A, where population coverage was just 11%, even though only 8% of out-patients were paying user fees. This demonstrated the relevance of taking into account the wider health financing context when looking at CHF implementation. Lastly, although our results undoubtedly confirm the lack of financial sustainability of the CHF observed by Borghi et al. (2015), they additionally showed that because the CHF was built into existing structures, there was considerable cross-subsidization in terms of financing sources paying for CHF administration (e.g. national tax-financed salaries, NHIF and user fee funds). This also meant that the CHF would be left with >70% of its revenues to purchase medicines and supplies and implement quality improvement activities at the health facility. It therefor again highlighted the importance of other health financing mechanisms in the analysis of CHF implementation. Way forward In-line with what has been suggested by others, the results made clear that in order to make the CHF work, major improvements in CHF implementation practices would be indispensable (Stoermer et al., 2011, 2012; Mtei and Enemark, 2013; Kalolo et al., 2015, 2018; Ministry of Health and Social Welfare, 2015). Most importantly, our findings showed the importance of considering council-specific CHF implementation practices and the wider health financing context when looking at CHF performance. Changes in CHF implementation practices would need to go hand in hand with adaptions in other health financing policies (e.g. exemption, user fee, fund pooling policies) as the CHF cannot be looked at as a stand-alone system. It is highly questionable whether improvements in CHF implementation practices alone were feasible and scalable given the council-specific CHF premiums, CHF benefit packages, user fee policies and fund pooling mechanisms as well as when taking into account the exemption policies, other health financing mechanisms and healthcare-seeking behaviours. The question also remains whether such efforts to improve CHF implementation were value for money taking into account the already considerable contributions of other health financing mechanisms to CHF administration and the small contribution of the CHF to overall health financing. Limited resources might potentially be better invested if in a first place the focus was on improving processes of major health financing sources coming from central level (Block Grants, Health Sector Basket Fund, Development Grants and MSD supply chain) in order to increase resource utilization and predictability of funding flows. This would lead more likely to a noticeable change in quality of care, because even little improvements in these processes could free up a substantial amount of money and human capacity. Improved quality might then in turn increase willingness of the community to contribute to health services as suggested by others (Bonu et al., 2003; World Health Organization, 2013; Adebayo et al., 2015). However, this would imply that for protecting the informal sector from financial hardship, they would need to be at least temporarily exempted from user fees until certain level of healthcare quality could be guaranteed. This could obviously not be done without increasing the level of funding for healthcare from central level through existing or new innovative financing solutions (Gilson and McIntyre, 2005; Dutta, 2015). Such changes may also have implications on several other parts of the system, including a potential increase in service utilization followed by a possible drop of quality of care (Gilson and McIntyre, 2005; Borghi et al., 2012; McIntyre et al., 2013; World Health Organization, 2013). However, given the problems with CHF implementation or CSIFs more generally, it could be worth considering conducting further research in this direction and advocate for the most pro-poor and cost-effective approach. In particular, a comprehensive study ought to be done, which compares the cost and other implications of abolishing user fees with the efforts required for effectively improving CHF implementation. Limitations of the study Some data presented were collected from routine data and its documentation might have been erroneous. Yet, by verifying the numbers with additional sources available, it was assured to obtain data of reliable quality. Part of the analysis could only be done in council A, where detailed enough data were available. The lack of sufficient data in council B further supported the findings discussed above. For the cost calculations, also the cost of activities that would need to be done in the absence of the CHF was included. Though, these costs were apportioned according to the share of time spent on CHF administration. Additionally, it could be argued that the sample of informants providing costing information was too small to be representative for the council. However, most findings overlap well with what has been shown previously (Borghi et al., 2015). Finally, activities done by HFGCs were indirectly reported through the person responsible for CHF administration at the health facility. These estimates could thus be overestimated. Yet, even if the reported values were halved, apart from the absolute values for cost and time spent no statement reported in this study would change. Conclusion Our results showed the importance of considering council-specific CHF implementation practices (overall CHF administration and the definition of the premium and benefit package) and the wider health financing context (council defined user fee policies and fund pooling mechanisms as well as exemption policies and other health financing mechanisms) when looking at CHF performance. Findings demonstrated that exemption policies and healthcare-seeking behaviour influenced negatively the maximum potential enrolment rate. Higher revenues from user fees, user fee policies and fund pooling mechanisms might have furthermore set incentives for care providers to prioritize user fees over CHF revenues. Bottlenecks in overall CHF administration diminished potential positive effects of a council level health insurance scheme and may ultimately have affected CHF enrolment. Costing results clearly pointed out the lack of financial sustainability of the CHF. The financial analysis however also showed that due to significant contributions from other financing mechanisms to CHF administration, the CHF could be left with >70% of its revenues for financing services. Given the wider health financing context and healthcare-seeking behaviours, it is highly questionable whether improvements in CHF implementation practices alone were feasible and scalable. The question also certainly remains whether such efforts were value for money, and if limited resources were not better invested through primarily focusing on improving utilization and predictability of major health financing sources coming from central level. Therefore, this article calls for a realistic reconsideration of approaches taken to address the challenges in health financing and demonstrated that the CHF cannot be looked at as a stand-alone system. Acknowledgements We thank the councils that participated in our study for their trust, especially the study participants. We acknowledge the ISAQH team for supportive supervision on CHF data management. Further, we highly appreciate logistic support provided by Dominik Shamba during the study implementation. Also, we thank Paola Salari and Nikhil Mandalia for fruitful inputs during manuscript development. Funding The research presented here was fully funded by Novartis Foundation. Ethical considerations Permission to publish the findings was obtained from the National Institute for Medical Research (NIMR) in Tanzania. Ethical clearance was granted by the same institution (original: NIMR/HQ/R.8a/Vol. IX/1839, extension: NIMR/HQ/R.8c/Vol. II/521), the Institutional Review Board of the Ifakara Health Institute (IHI/IRB/No: 37-2014) and the Ethic Commission of Northeast and Central Switzerland (EKNZ 2014-347). Oral informed consent was obtained from all respondents. Conflict of interest statement. None declared. References Adebayo EF , Uthman OA, Wiysonge CSet al. . ( 2015 ). A systematic review of factors that affect uptake of community-based health insurance in low-income and middle-income countries . BMC Health Services Research 15 : 543 . Google Scholar Crossref Search ADS PubMed WorldCat Bank of Tanzania. ( 2017 ). Interbank Foreign Exchange Market (IFEM) Summaries [Online] http://www.bot.go.tz/FinancialMarkets/IFEMsummaries/IFEMsummaries.asp, (accessed 9 January 2017). Bonu S , Rani M, Bishai D. ( 2003 ). Using willingness to pay to investigate regressiveness of user fees in health facilities in Tanzania . Health Policy and Planning 18 : 370 – 82 . Google Scholar Crossref Search ADS PubMed WorldCat Borghi J , Makawia S, Kuwawenaruwa A. ( 2015 ). The administrative costs of community-based health insurance: a case study of the Community Health Fund in Tanzania . Health Policy and Planning 30 : 19 – 27 . Google Scholar Crossref Search ADS PubMed WorldCat Borghi J , Maluka S, Kuwawenaruwa Aet al. . ( 2013 ). Promoting universal financial protection: a case study of new management of community health insurance in Tanzania . Health Research Policy and Systems 11 : 21 . Google Scholar Crossref Search ADS PubMed WorldCat Borghi J , Mtei G, Ally M. ( 2012 ). Modelling the implications of moving towards universal coverage in Tanzania . Health Policy and Planning 27 ( Suppl 1 ): i88 – 100 . Google Scholar Crossref Search ADS PubMed WorldCat Drummond F , Sculpher M, Torrance G, O’Brien B, Stoddart G. ( 2005 ). Methods for the Economic Evaluation of Health Care Programmes . Oxford : Oxford University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Dutta A . ( 2015 ). Prospects for Sustainable Health Financing in Tanzania - Baseline Report . Washington, DC : Health Policy Project, Futures Group . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Frumence G , Nyamhanga T, Mwangu M, Hurtig AK. ( 2014 ). Participation in health planning in a decentralised health system: experiences from facility governing committees in the Kongwa district of Tanzania . Global Public Health 9 : 1125 – 38 . Google Scholar Crossref Search ADS PubMed WorldCat Gilson L , McIntyre D. ( 2005 ). Removing user fees for primary care in Africa: the need for careful action . BMJ 331 : 762 – 5 . Google Scholar Crossref Search ADS PubMed WorldCat Haazen D . ( 2012 ). Making Health Financing Work for the Poor People in Tanzania: A Health Financing Policy Note . Washington, DC : Human Development Department, African Region, World Bank . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Idd A , Yohana O, Maluka SO. ( 2013 ). Implementation of pro-poor exemption policy in Tanzania: policy versus reality . The International Journal of Health Planning and Management 28 : e298 – 309 . Google Scholar Crossref Search ADS PubMed WorldCat Ifakara Health Institute. ( 2013 ). Spotlight: A Closer Look at Cost Sharing Revenue Availability and Use in the Public Sector . Dar es Salaam : Ifakara Health Institute . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Joseph C , Maluka SO. ( 2016 ). Do management and leadership practices in the context of decentralisation influence performance of Community Health Fund? Evidence from Iramba and Iringa Districts in Tanzania . International Journal of Health Policy and Management 6 : 257 – 65 . Google Scholar Crossref Search ADS PubMed WorldCat Kalolo A , Gautier L, Radermacher Ret al. . ( 2018 ). Implementation of the redesigned Community Health Fund in the Dodoma region of Tanzania: A qualitative study of views from rural communities . The International Journal of Health Planning and Management 33 : 121 – 35 . Google Scholar Crossref Search ADS PubMed WorldCat Kalolo A , Radermacher R, Stoermer M, Meshack M, Allegri DE, M. ( 2015 ). Factors affecting adoption, implementation fidelity, and sustainability of the Redesigned Community Health Fund in Tanzania: a mixed methods protocol for process evaluation in the Dodoma region . Global Health Action 8 : 29648 . Google Scholar Crossref Search ADS PubMed WorldCat Kamuzora P , Gilson L. ( 2007 ). Factors influencing implementation of the Community Health Fund in Tanzania . Health Policy and Planning 22 : 95 – 102 . Google Scholar Crossref Search ADS PubMed WorldCat Kapologwe NA , Kagaruki GB, Kalolo Aet al. . ( 2017 ). Barriers and facilitators to enrollment and re-enrollment into the Community Health Funds/Tiba Kwa Kadi (CHF/TIKA) in Tanzania: a cross-sectional inquiry on the effects of socio-demographic factors and social marketing strategies . BMC Health Services Research 17 : 308 . Google Scholar Crossref Search ADS PubMed WorldCat Kessy F . ( 2014 ). Improving health services through community participation in health governance structures in Tanzania . Journal of Rural and Community Development 9 : 14 – 3 . Google Scholar OpenURL Placeholder Text WorldCat Kessy F , Mtei G, Mbwambo J, Manumbu K, Mbuya D. ( 2008 ). Technical Review of Council Health Service Boards and Health Facility Governing Committees in Tanzania . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Macha J , Kuwawenaruwa A, Makawia S, Mtei G, Borghi J. ( 2014 ). Determinants of Community Health Fund membership in Tanzania: a mixed methods analysis . BMC Health Services Research 14 : 538 . Google Scholar Crossref Search ADS PubMed WorldCat Maluka SO . ( 2013 ). Why are pro-poor exemption policies in Tanzania better implemented in some districts than in others? International Journal for Equity in Health 12 : 80 . Google Scholar Crossref Search ADS PubMed WorldCat Maluka SO , Bukagile G. ( 2014 ). Implementation of Community Health Fund in Tanzania: why do some districts perform better than others? The International Journal of Health Planning and Management 29 : e368 – 82 . Google Scholar Crossref Search ADS PubMed WorldCat Mathauer I , Nicolle E. ( 2011 ). A global overview of health insurance administrative costs: what are the reasons for variations found? Health Policy 102 : 235 – 46 . Google Scholar Crossref Search ADS PubMed WorldCat McIntyre D , Garshong B, Mtei Get al. . ( 2008 ). Beyond fragmentation and towards universal coverage: insights from Ghana, South Africa and the United Republic of Tanzania . Bulletin of the World Health Organization 86 : 871 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat McIntyre D , Ranson MK, Aulakh BK, Honda A. ( 2013 ). Promoting universal financial protection: evidence from seven low- and middle-income countries on factors facilitating or hindering progress . Health Research Policy and Systems 11 : 36 . Google Scholar Crossref Search ADS PubMed WorldCat Ministry of Health and Social Welfare ( 2009 ). Health Sector Strategic Plan III: July 2009–June (2015) . Dar es Salaam : Ministry of Health and Social Welfare, United Republic of Tanzania . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Ministry of Health and Social Welfare ( 2012 ). Health Sector Public Expenditure Review (2010)/11 . Dar es Salaam : Directorate of Policy and Planning, Ministry of Health and Social Welfare, United Republic of Tanzania . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Ministry of Health and Social Welfare ( 2013 ). Mid Term Review of the Health Sector Strategic Plan III 2009-–2015: Health Care Financing . Dar es Salaam : Ministry of Health and Social Welfare, United Republic of Tanzania . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Ministry of Health and Social Welfare ( 2015 ). Health Sector Strategic Plan (HSSP IV): July 2015-June 2020 . Dar es Salaam : Ministry of Health and Social Welfare, United Republic of Tanzania . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Ministry of Health and Social Welfare & Prime Minister’s Office Regional Administration and Local Government ( 2011 ). Comprehensive Council Health Planning Guidelines . Dar es Salaam : Ministry of Health and Social Welfare and Prime Minister’s Office Regional Administration and Local Government, United Republic of Tanzania . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Ministry of Health Community Development Gender Elderly and Children, Ministry of Health, National Bureau of Statistics, Office of the Chief Government Statistician & ICF ( 2016 ). Demographic and Health Survey and Malaria Indicator Survey 2015–16 . Dar es Salaam, Tanzania, and Rockville, MD : Ministry of Health, Community Development, Gender, Elderly and Children [Tanzania Mainland], Ministry of Health [Zanzibar], National Bureau of Statistics [Tanzania Mainland], Office of the Chief Government Statistician [Zanzibar], and ICF, Rockville, MD . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mkumbo E , Masbayi V. ( 2014 ). Assessment findings of the functionality of Community Health Funds in Misenyi, Musoma Rural, and Sengerema districts . In: Tibu Homa Project for the United States Agency for International Development (ed.) Dissemination Workshop Report . Mwanza, Tanzania : University Research Co., LLC . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mtei G , Enemark U. ( 2013 ). Development of the Tanzania Health Financing Strategy . Options Paper No. 6: Reform Options for the CHF System - Final Report. Dar es Salaam . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mtei G , Mulligan JA. ( 2007 ). Community Health Funds in Tanzania: A Literature Review . Dar es Salaam : Consortium for Research on Equitable Health Systems, Ifakara Health Institute . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mubyazi G , Massaga J, Kamugisha Met al. . ( 2006 ). User charges in public health facilities in Tanzania: effect on revenues, quality of services and people’s health-seeking behaviour for malaria illnesses in Korogwe district . Health Services Management Research 19 : 23 – 35 . Google Scholar Crossref Search ADS PubMed WorldCat Mubyazi GM . ( 2004 ). The Tanzanian policy on health-care fee waivers and exemptions in practice as compared with other developing countries: evidence from recent local studies and international literature . East African Journal of Public Health 1 : 11 – 7 . Google Scholar OpenURL Placeholder Text WorldCat Nangawe E . ( 2012 ). Situation Analysis of Quality Improvement in Health Care, Tanzania . Dar es Salaam : Ministry of Health and Social Welfare, United Republic of Tanzania . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC National Bureau of Statistics. ( 2013 ). (2012) population and housing census . In: Population Distribution by Administrative Area . Dar es Salaam : National Bureau of Statistics, Ministry of Finance, United Republic of Tanzania . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Prime Minister’s Office Regional Administration and Local Government. ( 2013 ). Salary Structure and Scale of (2013)/(2014) and (2014)/(2015) . Dar es Salaam, Tanzania : Prime Minister’s Office Regional Administration and Local Government, United Republic of Tanzania . Renggli S , Mayumana I, Mboya Det al. . ( 2018 ). Towards improved health service quality in Tanzania: an approach to increase efficiency and effectiveness of routine supportive supervision . PLoS One 13 : e0202735 . Google Scholar Crossref Search ADS PubMed WorldCat Stoermer M , Hanlon P, Tawa M, Macha J, Mosha D. ( 2012 ). Community Health Funds (CHFs) in Tanzania: innovations study - final report . Swiss Tropical and Public Health Institute (SwissTPH), Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Stoermer M , Radermacher R, Vanderhyden M. ( 2011 ). Transforming Community Health Funds in Tanzania into viable social health insurance schemes: the challenges ahead . MMS Bulletin 120 : 22 – 8 . Google Scholar OpenURL Placeholder Text WorldCat Sustainable Development Solution Network. ( 2015 ). Indicators and a monitoring framework for the sustainable development goals . Launching a Data Revolution . Sustainable Development Solution Network, United Nations . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC United Republic of Tanzania. ( 2001 ). The Community Health Fund Act . Dar es Salaam : United Republic of Tanzania . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. ( 2010 ). The World Health Report: health system financing . In: The Path to Universal Coverage . Geneva : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. ( 2013 ). State of Health Financing in the African Region . Republic of Congo : Regional Office for Africa, World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. ( 2014 ). Global Health Expenditure Database, NHA Indicators [Online], http://apps.who.int/nha/database/Select/Indicators/en (accessed 22 April 2017). © The Author(s) 2019. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © The Author(s) 2019. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine.

Haley, Connie A; Brault, Marie A; Mwinga, Kasonde; Desta, Teshome; Ngure, Kenneth; Kennedy, Stephen B; Maimbolwa, Margaret; Moyo, Precious; Vermund, Sten H; Kipp, Aaron M

doi: 10.1093/heapol/czy105pmid: 30698696

Abstract Despite numerous international and national efforts, only 12 countries in the World Health Organization’s African Region met the Millennium Development Goal #4 (MDG#4) to reduce under-five mortality by two-thirds by 2015. Given the variability across sub-Saharan Africa, a four-country study was undertaken to examine barriers and facilitators of child survival prior to 2015. Liberia and Zambia were chosen to represent countries making substantial progress towards MDG#4, while Kenya and Zimbabwe represented countries making less progress. Our individual case studies suggested that strong health governance and leadership (HGL) was a significant driver of the greater success in Liberia and Zambia compared with Kenya and Zimbabwe. To elucidate specific components of national HGL that may have substantially influenced the pace of reductions in child mortality, we conducted a cross-country analysis of national policies and strategies pertaining to maternal, neonatal and child health (MNCH) and qualitative interviews with individuals working in MNCH in each of the four study countries. The three aspects of HGL identified in this study which most consistently contributed to the different progress towards MDG#4 among the four study countries were (1) establishing child survival as a top national priority backed by a comprehensive policy and strategy framework and sufficient human, financial and material resources; (2) bringing together donors, strategic partners, health and non-health stakeholders and beneficiaries to collaborate in strategic planning, decision-making, resource-allocation and coordination of services; and (3) maintaining accountability through a ‘monitor-review-act’ approach to improve MNCH. Although child mortality in sub-Saharan Africa remains high, this comparative study suggests key health leadership and governance factors that can facilitate reduction of child mortality and may prove useful in tackling current Sustainable Development Goals. Child health, governance, Millennium Development Goals, accountability, health services, qualitative research Key Messages Stable and consistent health governance and leadership was a key factor contributing to the variable progress towards the Millennium Development Goal Four (MDG#4) target of reducing under-five mortality by two-thirds by 2015. Three main aspects of successful health governance and leadership effecting improved child survival identified in this study were (1) establishing child survival as a top national priority backed by a comprehensive policy and strategy framework and sufficient human, financial and material resources; (2) bringing together donors, strategic partners, health and non-health stakeholders and beneficiaries to collaborate in strategic planning, decision-making, resource-allocation and coordination of services; and (3) maintaining accountability through a ‘monitor-review-act’ approach to improve MNCH. Countries that made inadequate progress towards MDG#4, struggled to fully support MNCH care, implement policies and strategies, maintain a functional health system, coordinate stakeholders to integrate programmes and services, or ensure effective monitoring and use of health data to identify and overcome gaps in health services. Introduction Substantial progress in child survival led to an estimated decline in under-five mortality (U5M) worldwide from 12.7 million in 1990 to 5.9 million in 2015 (UNICEF et al., 2015). However, progress was limited in many regions, such that Millennium Development Goal #4 (MDG#4) to reduce global U5M by two-thirds between 1990 and 2015 was not met (United Nations, 2000). Although the rate of U5M in sub-Saharan Africa (SSA) remains the highest in the world, estimated at 83 deaths per 1000 live births in 2015 (UNICEF et al., 2015), 12 SSA countries met their MDG#4 target: Eritrea, Ethiopia, Liberia, Madagascar, Malawi, Mozambique, Niger, Rwanda, Senegal, Tanzania, Uganda and Zambia (UNICEF et al., 2015; You et al., 2015). These successes demonstrate that substantial reduction in childhood deaths is possible in low- and middle-income countries (LMICs). Most childhood morbidity and mortality can be prevented or cured with known, affordable technologies and treatments (Friberg et al., 2010; Berman et al., 2016; Moucheraud et al., 2016). Yet, inadequate health systems in many LMICs hinder progress such that essential drugs and interventions are not distributed reliably, in sufficient quantity, equitably or at reasonable cost. Published case studies highlight how some SSA countries accelerated progress to reduce U5M, providing valuable insights regarding implementation and scale-up of child survival strategies (Bellagio Group, 2003; Amouzou et al., 2012; Mbonye et al., 2012; Zimba et al., 2012; Kuruvilla et al., 2014; Afnan-Holmes et al., 2015; Requejo et al., 2015; United Nations, 2015; Kanyuka et al., 2016; Moucheraud et al., 2016; Ruducha et al., 2017), but few comprehensively evaluate countries making insufficient progress towards MDG#4. We previously conducted country-specific case studies of four SSA nations with different annual rates of reduction (ARR) in U5M to identify specific barriers and facilitators that influenced their progress towards MDG#4 (Figure 1) (Kipp et al., 2016; Brault et al., 2017, 2018; Haley et al., 2017). Liberia and Zambia were on track for MDG#4 when the study began (and have now met MDG#4) while Kenya and Zimbabwe were not on track (and did not meet MDG#4) (Figure 1). Country trends for infant mortality mirrored those of U5M. Neonatal mortality declined by ∼50% for Liberia and Zambia, yet remained stagnant for Kenya and Zimbabwe. Each country has unique historical, social and political experiences, while sharing characteristics such as high poverty levels, developing economies and large rural populations (see Boxes 1–4). Notably, we identified health governance and leadership (HGL) as a factor influencing progress in reducing U5M in these four countries. Figure 1 Open in new tabDownload slide Trends in under-five mortality and progress towards Millennium Development Goal #4 for Liberia, Zambia, Kenya and Zimbabwe, 1990–2015 Figure 1 Open in new tabDownload slide Trends in under-five mortality and progress towards Millennium Development Goal #4 for Liberia, Zambia, Kenya and Zimbabwe, 1990–2015 Box 1 Historical and contextual factors impacting Liberia during the study period Located on the western coast of Africa with a small population of about 3.5 million, half of whom reside in urban areas Fourteen years of civil war ending in 2003 destroyed most of the national infrastructure, internally displaced many people and cost at least 200 000 lives First democratic election held in 2005 began a new phase of national reforms and rebuilding Prudent macroeconomic management, social stability and substantial foreign investments have facilitated efforts to overcome the civil crisis and re-establish its economy Economic growth affected by infrastructure constraints, unemployment, a narrow base of the economy and the country’s dependency on food and fuel imports Compounding challenges include flooding and drought in some areas, outbreaks of communicable diseases, influx of more than 150 000 refugees from neighbouring Cote D’Ivoire and increasing dependence on international aid Box 2 Historical and contextual factors impacting Zambia during the study period Gained independence in 1964, has since enjoyed decades of political stability and freedom from conflict enabling a consistent focus on development and reforms Experienced consistent economic growth and strong macroeconomic indicators over several decades Growing population, high level of urbanization and increasing life expectancy Young and increasing population also intensifies the burden of health needs on the economy Economic growth has not translated into significant poverty reduction at household level; more than half of the population lives below the poverty line, most considered to be in extreme poverty Unemployment has been high and income inequality is significant Box 3 Historical and contextual factors impacting Kenya during the study period Largest and most diversified economy in East Africa Strategically located to serve as an important transport hub for much of Eastern Africa Large and growing population consisting of most major ethno-racial and linguistic groups found in Africa High absolute poverty; more than two-thirds of urban population living in slums Violence following the 2007 presidential elections worsened mistrust between different political and ethnic groups Decades of globalization, political instability, regional and national macroeconomic challenges and climate change have contributed to high inequities Box 4 Historical and contextual factors impacting Zimbabwe during the study period Overcame a decade of civil war to gain independence in 1980, and successfully established one of the strongest economies and health systems in southern Africa Long period of relative stability and progress following independence until it experienced a drastic economic decline and hyperinflation beginning in the late-1990s Prior to the economic crisis, Zimbabwe had a highly performing health delivery system supporting a long track record of delivering comprehensive health services across the country. Nearly a quarter of the population left the country, including a large proportion of the workforce High poverty rates, unemployment and food insecurity persisted during the study period Despite challenges, education and literacy rates remained high among both men and women HGL has been defined in different ways in the literature (Barbazza and Tello, 2014; Mishra et al., 2015; Moucheraud et al., 2016). We used the World Health Organization’s (WHO) definition which provides a practical country-level framework for HGL: ‘ensuring strategic policy frameworks exist and are combined with effective oversight, coalition-building, regulation, attention to system-design and accountability’ (WHO, 2007). Widely considered the most critical of WHO’s health system building blocks, HGL links all health system components together, providing strategic direction for ensuring availability of high quality health services, managing the health workforce, providing medicines, financing health services and generating information needed for effective decision-making (WHO, 2007; Cavagnero et al., 2008). In this study, we re-analysed data from all four country case-studies, including a review of national policies and strategies pertaining to the larger scope of maternal, neonatal and child health (MNCH) under which U5M falls, and qualitative interviews with individuals working in MNCH to elucidate specific components of HGL that influenced achieving (Liberia and Zambia) or not achieving (Kenya and Zimbabwe) MDG#4. Methods We reviewed national policies and strategies issued between 2000 and 2013 and conducted key informant (KI) interviews in 2013 to explore eight content areas influencing child survival (WHO, 2006, 2007, 2010, 2012; Ban, 2010; WHO and PMNCH, 2011): (1) health care system (including HGL, structure, human resources for health, access & utilization, monitoring & evaluation and accountability), (2) national health strategies and policies, (3) MNCH interventions, (4) clinical standards and guidelines, (5) commodities and essential medicines, (6) health financing, (7) partnerships and (8) contextual factors (e.g. conflict, political environment, hygiene and sanitation, nutrition and food security, education and human rights). Four SSA countries (Liberia, Zambia, Kenya and Zimbabwe) were chosen based on their U5M ARR between 1990 and 2011 (data available when the study was designed, Figure 1) and their national governments’ willingness to participate. Detailed study methods for each country case study have been published (Kipp et al., 2016; Brault et al., 2017, 2018; Haley et al., 2017). Review of MNCH policies and strategies A national document review was conducted for each country to evaluate the MNCH policy framework affecting progress towards MDG#4. Policies and strategies pertaining to overall national health, MNCH and other related determinants were obtained from the WHO African Region office, WHO country focal points and Ministry of Health (MOH) for Liberia, Zambia, Kenya and Zimbabwe. Additional MNCH-related documents referenced in initial sources were subsequently obtained and reviewed (see individual case study supplementary tables in Kipp et al., 2016; Brault et al., 2017, 2018; Haley et al., 2017). An abstraction guide was developed based on the eight study content areas and several cross-cutting questions (Table 1). Each document was reviewed by one author (CAH), who consulted with a second reviewer (MAB) as needed. Information from original documents was recorded verbatim in the abstraction guide to avoid observer bias. Table 1 Key questions and deductive themes explored during the review of national health policies and strategies and key informant interviews that cut across child survival content areas Specific questions for review of national policies and strategies . Specific themes explored across content areas with key informants . What policies and strategies related to MNCH were in place between 2000 and 2013 (including changes during this period)? What challenges were stated as hindering progress towards MDG#4? What facilitators were stated as enabling progress towards MDG#4? What changes or improvements to MNCH policies and strategies were proposed or newly implemented towards the end of the study period but were not yet measurable? Issues related to programme evaluation, access and utilization, coverage, impact and sustainability, as appropriate Knowledge and experiences related to MNCH across the health care continuum (prenatal care through age 5 years) Knowledge and experiences related to MNCH across the health system continuum (community to tertiary hospitals) Specific questions for review of national policies and strategies . Specific themes explored across content areas with key informants . What policies and strategies related to MNCH were in place between 2000 and 2013 (including changes during this period)? What challenges were stated as hindering progress towards MDG#4? What facilitators were stated as enabling progress towards MDG#4? What changes or improvements to MNCH policies and strategies were proposed or newly implemented towards the end of the study period but were not yet measurable? Issues related to programme evaluation, access and utilization, coverage, impact and sustainability, as appropriate Knowledge and experiences related to MNCH across the health care continuum (prenatal care through age 5 years) Knowledge and experiences related to MNCH across the health system continuum (community to tertiary hospitals) Open in new tab Table 1 Key questions and deductive themes explored during the review of national health policies and strategies and key informant interviews that cut across child survival content areas Specific questions for review of national policies and strategies . Specific themes explored across content areas with key informants . What policies and strategies related to MNCH were in place between 2000 and 2013 (including changes during this period)? What challenges were stated as hindering progress towards MDG#4? What facilitators were stated as enabling progress towards MDG#4? What changes or improvements to MNCH policies and strategies were proposed or newly implemented towards the end of the study period but were not yet measurable? Issues related to programme evaluation, access and utilization, coverage, impact and sustainability, as appropriate Knowledge and experiences related to MNCH across the health care continuum (prenatal care through age 5 years) Knowledge and experiences related to MNCH across the health system continuum (community to tertiary hospitals) Specific questions for review of national policies and strategies . Specific themes explored across content areas with key informants . What policies and strategies related to MNCH were in place between 2000 and 2013 (including changes during this period)? What challenges were stated as hindering progress towards MDG#4? What facilitators were stated as enabling progress towards MDG#4? What changes or improvements to MNCH policies and strategies were proposed or newly implemented towards the end of the study period but were not yet measurable? Issues related to programme evaluation, access and utilization, coverage, impact and sustainability, as appropriate Knowledge and experiences related to MNCH across the health care continuum (prenatal care through age 5 years) Knowledge and experiences related to MNCH across the health system continuum (community to tertiary hospitals) Open in new tab Qualitative methods Study location and participants Utilizing country Demographic and Health Surveys (DHS) closest to 1990 and 2011, one or two provinces were selected from each country that had U5M ARRs comparable with the national ARR and were logistically accessible. Specific rural and urban sites were selected to evaluate differences in MNCH that can exist between urban and rural areas (Table 2). Table 2 Selected study sites within Kenya, Liberia, Zambia and Zimbabwe Country . Capital . Urban . Rural . Kenyaa Nairobi (Nairobi Province) Embu (Eastern Province) Liberia Monrovia (Montserrado County) Gbarnga (Bong county) Zambia Lusaka Livingstone (Southern Province) Kazungula (Southern Province) Zimbabwe Harare Chinhoyi (Mashonaland West Province) Banket (Mashonaland West Province) Country . Capital . Urban . Rural . Kenyaa Nairobi (Nairobi Province) Embu (Eastern Province) Liberia Monrovia (Montserrado County) Gbarnga (Bong county) Zambia Lusaka Livingstone (Southern Province) Kazungula (Southern Province) Zimbabwe Harare Chinhoyi (Mashonaland West Province) Banket (Mashonaland West Province) a Nairobi Province is now Nairobi County; Eastern Province now consists of eight counties (established in 2013), including Embu County as the rural study site. Open in new tab Table 2 Selected study sites within Kenya, Liberia, Zambia and Zimbabwe Country . Capital . Urban . Rural . Kenyaa Nairobi (Nairobi Province) Embu (Eastern Province) Liberia Monrovia (Montserrado County) Gbarnga (Bong county) Zambia Lusaka Livingstone (Southern Province) Kazungula (Southern Province) Zimbabwe Harare Chinhoyi (Mashonaland West Province) Banket (Mashonaland West Province) Country . Capital . Urban . Rural . Kenyaa Nairobi (Nairobi Province) Embu (Eastern Province) Liberia Monrovia (Montserrado County) Gbarnga (Bong county) Zambia Lusaka Livingstone (Southern Province) Kazungula (Southern Province) Zimbabwe Harare Chinhoyi (Mashonaland West Province) Banket (Mashonaland West Province) a Nairobi Province is now Nairobi County; Eastern Province now consists of eight counties (established in 2013), including Embu County as the rural study site. Open in new tab Study participants Semi-structured interviews were conducted with KIs involved in MNCH from the MOH, donor organizations, community-based organizations (CBO) and health care providers (HCP) (Tables 3 and 4). CBO participants and HCPs were selected from both urban and rural sites. National level KIs (see below) were recruited from the capital and each local site. In-country research teams collaborated with the MOH and WHO to identify potential KIs representing a range of ages, work experiences and positions/roles balanced between urban and rural sites. Table 3 Additional inclusion criteria for each key informant group Key informant type . Description . All participants Age 18 years or older Have adequate knowledge or experiences related to childhood survival specified for each participant group below Speak English or the most common local language, Able to provide written or verbal informed consent. Ministry of Health National or provincial-level officials working in government-level health care system administration, policy-making, programme development or leadership. All officials working in areas related to MNCH were eligible. Donor partners Individuals working as directors, managers or other leaders of entities providing financial or other aid for MNCH services, or serving as the implementing partner. International or national organizations focusing entirely on MNCH or with MNCH as one component of their mission. Organizations had to be officially registered in the country. Members of community-based organizations Directors, leaders, managers working for a CBO involved in or providing referrals to MNCH services within the study site. Organizations had to be officially registered in the country. Health care providers Professionally trained physicians, nurses, clinical officers or other health-related staff such as environmental health technicians, pharmacists or community health workers. Working in a health facility providing MNCH care. Key informant type . Description . All participants Age 18 years or older Have adequate knowledge or experiences related to childhood survival specified for each participant group below Speak English or the most common local language, Able to provide written or verbal informed consent. Ministry of Health National or provincial-level officials working in government-level health care system administration, policy-making, programme development or leadership. All officials working in areas related to MNCH were eligible. Donor partners Individuals working as directors, managers or other leaders of entities providing financial or other aid for MNCH services, or serving as the implementing partner. International or national organizations focusing entirely on MNCH or with MNCH as one component of their mission. Organizations had to be officially registered in the country. Members of community-based organizations Directors, leaders, managers working for a CBO involved in or providing referrals to MNCH services within the study site. Organizations had to be officially registered in the country. Health care providers Professionally trained physicians, nurses, clinical officers or other health-related staff such as environmental health technicians, pharmacists or community health workers. Working in a health facility providing MNCH care. Open in new tab Table 3 Additional inclusion criteria for each key informant group Key informant type . Description . All participants Age 18 years or older Have adequate knowledge or experiences related to childhood survival specified for each participant group below Speak English or the most common local language, Able to provide written or verbal informed consent. Ministry of Health National or provincial-level officials working in government-level health care system administration, policy-making, programme development or leadership. All officials working in areas related to MNCH were eligible. Donor partners Individuals working as directors, managers or other leaders of entities providing financial or other aid for MNCH services, or serving as the implementing partner. International or national organizations focusing entirely on MNCH or with MNCH as one component of their mission. Organizations had to be officially registered in the country. Members of community-based organizations Directors, leaders, managers working for a CBO involved in or providing referrals to MNCH services within the study site. Organizations had to be officially registered in the country. Health care providers Professionally trained physicians, nurses, clinical officers or other health-related staff such as environmental health technicians, pharmacists or community health workers. Working in a health facility providing MNCH care. Key informant type . Description . All participants Age 18 years or older Have adequate knowledge or experiences related to childhood survival specified for each participant group below Speak English or the most common local language, Able to provide written or verbal informed consent. Ministry of Health National or provincial-level officials working in government-level health care system administration, policy-making, programme development or leadership. All officials working in areas related to MNCH were eligible. Donor partners Individuals working as directors, managers or other leaders of entities providing financial or other aid for MNCH services, or serving as the implementing partner. International or national organizations focusing entirely on MNCH or with MNCH as one component of their mission. Organizations had to be officially registered in the country. Members of community-based organizations Directors, leaders, managers working for a CBO involved in or providing referrals to MNCH services within the study site. Organizations had to be officially registered in the country. Health care providers Professionally trained physicians, nurses, clinical officers or other health-related staff such as environmental health technicians, pharmacists or community health workers. Working in a health facility providing MNCH care. Open in new tab Table 4 Numbers of key informants interviewed for each country . Ministry of Health . Donor organization . Community-based organization . Health care worker . Total . Kenya 9 8 13 13 43 Liberia 11 8 14 14 47 Zambia 6 6 10 9 31 Zimbabwe 6 6 6 12 30 Total 32 28 43 48 151 . Ministry of Health . Donor organization . Community-based organization . Health care worker . Total . Kenya 9 8 13 13 43 Liberia 11 8 14 14 47 Zambia 6 6 10 9 31 Zimbabwe 6 6 6 12 30 Total 32 28 43 48 151 Open in new tab Table 4 Numbers of key informants interviewed for each country . Ministry of Health . Donor organization . Community-based organization . Health care worker . Total . Kenya 9 8 13 13 43 Liberia 11 8 14 14 47 Zambia 6 6 10 9 31 Zimbabwe 6 6 6 12 30 Total 32 28 43 48 151 . Ministry of Health . Donor organization . Community-based organization . Health care worker . Total . Kenya 9 8 13 13 43 Liberia 11 8 14 14 47 Zambia 6 6 10 9 31 Zimbabwe 6 6 6 12 30 Total 32 28 43 48 151 Open in new tab Data collection and analysis Guides for KI interviews were developed and piloted, mirroring the eight content areas and cross-cutting questions explored in the national document review (Table 1). Interviews were audio recorded, transcribed and translated into English (as needed) by trained research assistants. Transcripts were coded using deductive themes based on study content areas plus additional themes identified upon transcript review. Analyses were conducted using the qualitative software Atlas.ti (Murh, 2004), grouping the on-track countries (Liberia and Zambia) and not on-track countries (Kenya and Zimbabwe) for comparison. Analyses focused on codes related to HGL based on the WHO definition (WHO, 2007). The Institutional Review Boards at the authors’ institutes and both the national and local ethics and research committees for each country approved the qualitative component of the study as follows (see Supplementary file S1 for copies of approval letters): Vanderbilt University Medical Center (Coordinating Center), Kenyatta National Hospital Ethics & Research Committee (Kenya), University of Liberia Office of the Institutional Review Board (Liberia), ERES Converge Institutional Review Board (Zambia), Joint Parirenyatwa Hospital and University of Zimbabwe College of Health Sciences Research Ethics Committee and the Medical Research Council of Zimbabwe. Results Liberia Prioritization and support of child survival National documents and KIs described Liberia’s focused efforts to rebuild the healthcare system and establish essential services following a prolonged civil crisis. A strong policy framework was devised and implemented, including a triple planning approach using immediate, short- and long-term plans concurrently focusing on health, social welfare and development [Liberia Ministry of Health and Social Welfare (MoHSW), 2008; MoHSW, 2011d; Liberia Ministry of Planning and Economic Affairs (MPEA), 2012]. Liberia’s first post-conflict national health policy and strategic plan (MoHSW, 2007) prioritized MNCH through primary health care, community empowerment and cross-sectoral partnerships. Within 5 years, Liberia updated its national policies, integrating health and social determinants to increase equitable access to comprehensive packages of MNCH services delivered closer to communities (MoHSW, 2011b). Nearly all KIs felt these policies spearheaded by Liberia’s president enabled rapid recovery of the health system and increased utilization of MNCH services. … the President had launched the revised road map for accelerating the reduction of maternal mortality, maternal and newborn mortality and morbidity in Liberia … initiatives that we believe … [have] shown government own commitment … (49-year-old male donor partner). With significant donor support for overall development, Liberia increased total government expenditure on health (TGEH) to exceed the Abuja Declaration target of at least 15% of a country’s annual budget [African Union (AU), 2006]. National documents and KIs reported that resources supporting MNCH were generally allocated appropriately and directed towards high priority areas, but that additional government funding was needed to fully implement MNCH interventions. Collaboration, coordination and inclusion National documents and KIs asserted that Liberia’s government developed collaborative multi-sectoral partnerships at all levels of the health system, aligning local MNCH activities with national priorities (MoHSW, 2011d). A 2009 decentralization policy shifted health services funding and allocation decisions to sub-national leaders more knowledgeable about local needs. In addition, a 2011 community health services policy established services closer to the populations in need (MoHSW, 2011c). Moreover, KIs felt the government effectively coordinated international donors, national and local programme leaders and community health providers and beneficiaries, to integrate delivery of MNCH services at each point of care. Although Liberia maintained programme-specific policies and strategies (e.g. for HIV/AIDS, malaria, immunization and food security), the Ministries of Child Health and Social Welfare were merged to enable a holistic approach to MNCH, which was viewed favourably by KIs. … it [effective external partnerships] was develop[ed] through coordination meetings …. When they came, some of them started doing their own thing …. But when the Ministry of Health said we have to meet and coordinate and know exactly what each partner is doing … people started coordinating and started working together and looking at best practices and start planning to have one focus … (50-year-old male urban CBO partner). [W]henever there is a new policy in place, the ministry will inform the county health team, they will do trainers of trainers, from county level, district level and then they will train facility level staff to implement these policies and then down to the community level (49-year-old female donor partner). Accountability Liberia set specific health targets, timeframes, roles and responsibilities within its child health policy framework aimed specifically at reaching MDG#4. An effective national and district-level health management information system (HMIS) enabled reporting of surveillance data, vital statistics and health services data from local facilities and providers up to county and national levels. National documents (MoHSW, 2011a) and KIs described timely collection and review of data as facilitating ongoing monitoring, evaluation and data-driven decision-making. County health and social welfare boards and community health committees further encouraged stakeholder and community involvement in HGL and ensured accountability for MNCH resource allocation. … the Ministry of Health and Social Welfare bases its plans on evidence; every activity, every move to improve [MNCH] is based on data, based on situations analysis that was conducted and high impact interventions identified to affect situations …. They are constantly monitored, evaluated and discussed and reviewed … (45-year-old male MOH representative). One other thing that promoted the effective partnership was accountability. Because we started understanding that if somebody gives you money, to give back you need to give account …. Nobody wants to give somebody something who doesn’t know his left hand from his right hand … (48-year-old male urban CBO partner). The Republic of Zambia Prioritization and support of child survival Zambia’s achievements in MNCH and health sector reforms steadily evolved over decades of political stability with a commitment to reducing U5M by focusing on immediate, medium- and long-term goals. Health system restructuring was intentionally aligned with development and poverty reduction efforts through five consecutive National Health Strategic Plans, six corresponding National Development Plans and a long-term National Development Strategy [Zambia Ministry of Health (MOH), 2006, 2012]. Zambia prioritized reduction of U5M through a comprehensive health policy framework that reflected international recommendations and resolutions related to MNCH [Zambia MOH, 2012; Zambia Ministry of Community Development and Mother and Child Health (MCDMCH) and MOH, 2013]. Expanded access to MNCH care was facilitated through a policy to remove user fees, adoption of a ‘Primary Health Care Approach’ (WHO, 2008) and delivery of integrated packages of basic health services from pregnancy thorough adolescence and across health system levels (Zambia MOH, 2012). In 2011, MNCH services were moved into an expanded Ministry of Community Development, Mother and Child Health to holistically address poverty, health and other social welfare issues. In addition, Zambia’s Constitution was amended to guarantee children’s right to health, and the government strengthened its policy framework to improve newborn health and provide a roadmap for achieving MDG#4 (MCDMCH, 2013b; MCDMCH and MOH, 2013). KIs described a well-structured national system for identifying and funding local MNCH priorities and needs and expanding community-level services. Though TGEH was increased to meet the Abuja Declaration (Countdown to 2015, 2012; USAID and UNAIDS, 2013), some KIs felt that additional government funding was needed to avoid reliance on donors. … more and more efforts are being made towards maternal and child health in terms of trying to increase funding and trying to make those facilities available and accessible and … now [the] creation of a new ministry which entirely looks at the mother and child health so that … it is prioritized … (51-year-old male MOH representative). … even if we have very few resources, we prioritize it, that finances at least should go to maternal and child health (43-year-old male urban healthcare provider). Collaboration, coordination and inclusion Zambia’s well-structured health system and MNCH policy framework promoted strong partnerships with external donors willing to align their support with domestic priorities. According to national documents and KIs, the government’s collaborative approach and decentralized HGL facilitated partnerships among health sector departments, between health and non-health ministries, and with a diversity of stakeholders at national and local levels (Zambia MOH and WHO, 2011; Zambia MOH, 2012). Local stakeholders were engaged in the coordination and integration of MNCH services, through an Interagency Coordinating Committee and technical working groups used to identify gaps, remove bottlenecks, mobilize resources and improve efficiency. … we have a sectorial advice group meeting and these are platforms that we use to try and persuade partners to buy into the health sector strategic plan … instead of them dreaming up something that they want to do, we actually present the activities that we have included in the strategic plan … [with] some partner input in them (43-year-old male MOH representative). They’ve known … that they need to have a community led strategy of people mobilizing fellow community members to go and have vaccinations so they … have what they call reaching every child … where they try to promote community efforts in supporting the program … (41-year-old female donor partner). Accountability Per national documents and KIs, Zambia fostered accountability throughout the health system by conducting ongoing and effective monitoring and evaluation efforts while encouraging feedback from stakeholders and beneficiaries. This process was facilitated by a highly functioning HMIS (Zambia MOH, 2013) and effective oversight of national electronic reporting for vital statistics, disease surveillance and response, human resources, pharmaceutical supply and distribution and finance and administration. An innovative electronic health records system was established to feed directly from the point of care into the HMIS, allowing detailed and timely reporting of MNCH service utilization, health expenditure and clinical outcomes (MCDMCH, 2013a). The data informed strategic planning, resource allocation and quality improvement, which along with a Zambian-led Countdown to 2015 initiative, accelerated achievement of MDG#4 (Zambia MOH, 2008). First and foremost, it’s identifying and having the right mix of priorities so in the development of the national health strategic plan … we use available data, mortality data, service data to look at where the need is greatest … (43-year-old male MOH representative). … Zambia is among very few countries who have done impact studies for a number of good years. To see how we are progressing, how those interventions we are employing whether they are working or not … (51-year-old male MOH representative). Kenya Prioritization and support of child survival During most of the study period, inadequate investment in the national health system led to stagnating public health sector performance, worsening health inequities and reversals of previous gains in child health outcomes [MOPHS, 2008; Kenya Ministry of Medical Services (MOMS) and Ministry of Public Health and Sanitation (MOPHS), 2012]. The government of Kenya also underwent several transitions, including a period of marked instability following the 2007 elections. Corresponding changes occurred in national HGL, with the MOH dividing into separate Ministry of Public Health and Sanitation (MOPHS, responsible for primary care at the community, dispensary and health centres levels) and Ministry of Medical Services (MOMS, responsible for the highest system levels) in 2008 before being re-unified in 2013. We thank God that now the MOMS and the MOPHS have come together, that is also what was causing a lot of division … [MOPHS] had a lot of resources than the MOMS, but now it is integrated … (57-year-old female urban healthcare provider). Kenya’s comprehensive national MNCH policy framework was described by both national documents and KIs as largely ineffective during most of the study period. One document described ‘years of erratic application of policy’ and ‘inadequate financial and human resources, inefficient support systems, and poorly coordinated responses to public health problems’ leading to poor health system performance (MOPHS, 2008). Later in the study period Kenya renewed its focus on health system strengthening and the right to health through a long-term national development plan (Government of Kenya, 2007) and a new Constitution (Government of Kenya, 2010), but progress was hindered by unresolved short-term challenges. Devolution of HGL to sub-national levels aimed to improve service delivery, accountability, citizen participation and equitable resource distribution, but this was not achieved during the study period. An updated National Health Sector Strategic Plan was issued to expand equitable access to care and strengthen community-level interventions through the Kenya Essential Package for Health (KEPH) and a Community Health Strategy (CHS) [Kenya Ministry of Health (MOH), 2007; MOMS et al., 2009; MOMS and MOPHS, 2013]. However, implementation was described as ‘slow’, and limited by inadequate human resources in many areas [National Coordinating Agency for Population and Development (NCAPD) [Kenya] et al., 2011]. Comprehensive strategies targeting newborn survival and U5M were also developed (MOPHS, 2008; MOPHS and MOMS, 2010), as were policies supporting adequate housing, nutrition, clean water, social security and education (MOMS and MOPHS, 2013). Unfortunately, as one KI stated, ‘[Kenya has] many strategic plans…the problem has been the strategies are there but the implementation is not there’ (40-year-old female urban healthcare provider). In 2012, Kenya’s National Health Policy was revised, promoting a ‘health in all policies’ approach to concurrently address all determinants of health. This revision’s effect could not be determined by the end of the study period (MOMS and MOPHS, 2012, 2013). National documents and KIs reported chronic government underfunding of Kenya’s health system and MNCH specifically, with nearly all KIs describing limited financial, material and human resources, particularly for primary care. Moreover, donor support was largely project-oriented and not necessarily aligned with Kenya’s priorities (NCAPD Kenya et al., 2011). Some KIs reported that the most successful MNCH programmes during the study period were those with steady funding from both the government and external partners. … And the government signed the Abuja Declaration to be able to fund health with at least 15% of the national budget. We’ve never gone beyond a 1/3rd of that budget that’s why we’re still struggling … (53-year-old male MOH representative). … [Priorities] seem to change unfortunately depending on where the funds have come from … [W]here the funds are from for HIV services, the HIV gets precedence. If you have a donor who says they want to look at TB, they’ll concentrate on TB, when Malaria, it’s that.… (37-year-old male donor partner). Collaboration, coordination and inclusion National documents indicated that persistently centralized HGL led to poor coordination between health system levels and inequitable distribution and financing of health services. KIs, however, expressed optimism that the recent devolution might alleviate this problem. An inter-ministerial National Council for Maternal and Child Health was created to harmonize national policy formulation, planning and coordination, resource mobilization, intervention delivery and monitoring and evaluation but was given no regulatory authority (MOMS and MOPHS, 2013). According to KIs, the lack of coordination, oversight or inclusion of beneficiaries in planning contributed to service gaps, duplication and poor quality of care. … there’s been very poor connection or cross sharing of skills, of resources to ensure continuum of care at a service delivery level … the HIV program came in and set up … a vertical PMTCT service in a health system where we had an MCH service and we would have easily integrated that within the MCH. There [is] lots of verticalization including of reporting and of monitoring … (41-year-old male urban healthcare provider). … an unfortunate thing is [in] this country people have been operating in silo[s] … so everybody operating independently …. Probably even one thing in improving child survival is making sure that all of you have the same goal, seeing … what can you complement each other to achieve the same goal or even at a lower cost (40-year-old female urban CBO partner). Accountability National documents described health sector ‘accountability deficits’ as contributing to inadequate MNCH service delivery, considerable inequities and poor health outcomes (MOPHS, 2008). Moreover, the country’s weak HMIS limited capacity for compiling, analysing and applying data to improve MNCH programmes or inform health policy (MOMS and MOPHS, 2009; NCAPD Kenya et al., 2011). Once we implement we need to have a way of having continuous monitoring and evaluation to see where we are at, what impact have we had, so that once an intervention is in place, we are able then to keep upgrading it … (40-year-old female urban healthcare provider). … in Kenya, a bulk of patients are seen in the private sector … we have to strengthen the M and E [monitoring and evaluation] system for all the sectors, whether public or private. We must get them somewhere they are analysed so that we can get the true picture [of the burden of disease] (50-year-old male urban healthcare provider). Later national health policies and strategies (MOMS and MOPHS, 2012, 2013) began to strengthen Kenya’s capacity to collect and apply local health data to improve availability and quality of MNCH services. Health management teams and local stakeholders (MOMS and MOPHS, 2009) were tasked with regular performance reviews, and mechanisms were implemented to improve public transparency and accountability. KIs did not discuss these reforms, making it difficult to determine their impact. Zimbabwe Prioritization and support of child survival Following independence in 1980 and a decade of civil war, Zimbabwe developed one of the strongest health systems in southern Africa, achieving lower U5M rates and higher coverage of MNCH interventions compared with other SSA countries. However, national documents and KIs described how Zimbabwe’s health system collapsed following the national socioeconomic crisis that began in the 1990s and peaked in 2009–2010 [Zimbabwe Ministry of Health and Child Welfare (MOHCW), 2010a,b]. Provision of MNCH services at that time was undermined by debilitated health infrastructure, a poorly functioning patient referral system, drug shortages and unaffordable out-of-pocket health care costs. Nearly all KIs and national documents stated that Zimbabwe’s critical shortage of health workers affected quality and availability of MNCH services [MOHCW, 2010b; Osika et al., 2010; Zimbabwe Ministry of Economic Planning and Investment Promotion (MEPIP) and United Nations Development Program (UNDP), 2012]. Health management was severely weakened by high attrition rates of experienced leaders, supervisors and programme managers. National health and re-development strategies addressing these limitations were not adequately implemented or funded (MOHCW, 2010b; Osika et al., 2010). … quality of maternal child born services … at all levels was highly compromised, it was very much substandard. It had something to do with shortage of human resources, had to do with WHO’s shortage of supplies and of course it had something to do with poo[r] supportive supervision and monitoring … (58-year-old male donor partner). … [W]here a nurse knows that I should manage … a sick child using their IMSI protocol but because there is a queue there … and there is just one nurse, they just do a shortcut … (52-year-old female donor partner). In the late 2000s, the government of Zimbabwe renewed its commitment to ‘kick-start’ the national health care system and re-focus on national development (MOHCW, 2007, 2010a,b). Zimbabwe’s 2009 National Health Strategy reinstituted measures to improve child survival such as the Primary Health Care Approach (WHO, 2008), delivery of MNCH intervention packages for all life stages at all health system levels, and community health services and outreach activities, but the overarching health policy framework remained outdated. To increase availability and utilization of MNCH services, Zimbabwe established a user fees exemption policy for the poor and vulnerable (including children) and a 5-year (2011–2015) multi-donor pooled Health Transition Trust Fund to enable health system improvements and increase access to care for mothers and young children. However, TGEH remained far below the Abuja recommendation, and many KIs felt that donor support was unsustainable. There is a challenge [that the money] allocated in health ministries [is] very low …. The strongest that has been funding the MNCH is the …. Health Transition Fund, but it has also a limit of … five years and then it goes (52-year-old female donor partner). Even at the end of the study period, KIs at various system levels felt that national strategies and policies related to MNCH were generally ‘good on paper’ but were not implemented, coordinated or enforced. l think we have the … RH [reproductive health] road map, the RH policy, the child survival strategy … l don’t think there is a serious problem with the policy and strategy, the major problem is translating these strategies and policies into action (58-year-old male donor partner). Collaboration, coordination and inclusion Although once decentralized, Zimbabwe’s HGL shifted towards national control over decision-making and resource allocation. This resulted in poor communication with local levels and ‘non-involvement of communities in health planning and management’ (MOHCW, 2010b; Osika et al., 2010). Health was considered a sectoral issue instead of a national priority integrated across policies (MOHCW, 2010b). Child Health and Maternal/Reproductive Health were separate departments within the Ministry of Health and Child Welfare (MOHCW), each coordinated by different officers with different reporting hierarchies (MOHCW, 2010a). Poorly synchronized health strategies, limited collaboration and ill-defined roles and responsibilities among stakeholders led to fragmented MNCH programmes and services (MOHCW, 2010b). Development of the National MNCH Steering Committee, National Child Survival Technical Working Group and National Child Welfare Council were intended to promote a participatory leadership structure, but these entities were described as ‘weak’, with limited stakeholder participation (MOHCW, 2010a,b). KIs also expressed concern that nearly every aspect of the MNCH system required the support of external partners, whose priorities were inconsistently aligned with the MOHCW. Vertical approaches intensified uneven distribution of aid and magnified inequities among programmes, populations and geographic areas. Heavy reliance on programme- or condition-specific donor aid also hindered the ‘supermarket approach’ intended to provide multiple MNCH services at one visit (MOHCW, 2010b; MEPIP and UNDP, 2012). I think the Ministry needs to continue discussing with lower levels of the health care system so that they understand what is it that is happening at [the] clinic level, and that the national level goes and procure things which cannot be used at clinic level that is a waste of resources … (60-year-old female donor partner). … if you go to a district you find there are a number of donors but if you go to the other, there is not even a single donor. I think the coordination, if possible at national level, should be improved so that there is an equitable distribution of services … (46-year-old female MOH representative). Accountability National documents frankly described Zimbabwe’s insufficient progress towards MDG#4 and other health goals, acknowledging limited public availability of health financing and service information and a failure of health committees to involve stakeholders (MOHCW, 2010b). Zimbabwe’s National HIMS was described as ineffective with inadequate oversight resulting in poorly harmonized monitoring and evaluation. More recent national documents noted Zimbabwe’s commitment to accountability, and KIs recognized efforts to improve health data to more effectively track indicators associated with MNCH. … you find that there are so many strategic documents, there is HIV/AIDS, MNCH, RH, so they are there but they are not integrated so you find each one will come up with their own M and E systems and they are donor driven programmes … (60-year-old female donor partner). … we are also trying to support the monitoring evaluation system including the … national health management of information system …. Now the provinces have restarted conducting their own planning review meetings every six months …[and] now the quality has started improving … (58-year-old male donor partner). Cross-country summary Table 5 summarizes the similarities and differences in the HGL themes described above for each country. Overwhelmingly, Liberia and Zambia successfully engaged with or implemented these elements during the study period. In contrast, Kenya and Zimbabwe struggled to do so, despite sometimes having the appropriate frameworks or approaches. Table 5 Comparison of health governance and leadership elements between progressing and non-progressing countries . Progressing . Non-progressing . . Liberia . Zambia . Kenya . Zimbabwe . Prioritization and support of child survival  Political support + + +/− +/−  Current policy framework + + + −  Policies and strategies implemented + + − −  Concurrent national policy focus on health, social welfare, development + + − −  Triple planning approach + + +/− −  Abuja Declaration target met during study + + − −  Non-financial health system resources (human, material, facility, etc.) + + − − Collaboration, coordination and inclusion  Donors aligned with national priorities + + − −  Collaborative strategic planning with partners/stakeholders + + − −  Coordination/collaboration between health and other sectors + + − −  Coordination and sharing resources among different health programmes + + − −  Coordination of MNCH services across health system levels + + − −  Integrate packages of health services at point of care + + − −  Decentralization of decision-making and resource allocation + + − −  Beneficiaries included in strategic planning (community input) + + − − Accountability  Clear roles, responsibilities and expectations + + +/− −  Updated, effective HMIS + + − −  Consistent data collection and reporting at all health system levels + + − −  Ongoing monitoring and evaluation of health programmes and interventions + + − −  Specifically monitoring of progress towards MDG#4 + + +/−a +/−  Data-driven planning and decision-making responsive to population needs + + − −  Local involvement (community planning boards and committees) + + − − . Progressing . Non-progressing . . Liberia . Zambia . Kenya . Zimbabwe . Prioritization and support of child survival  Political support + + +/− +/−  Current policy framework + + + −  Policies and strategies implemented + + − −  Concurrent national policy focus on health, social welfare, development + + − −  Triple planning approach + + +/− −  Abuja Declaration target met during study + + − −  Non-financial health system resources (human, material, facility, etc.) + + − − Collaboration, coordination and inclusion  Donors aligned with national priorities + + − −  Collaborative strategic planning with partners/stakeholders + + − −  Coordination/collaboration between health and other sectors + + − −  Coordination and sharing resources among different health programmes + + − −  Coordination of MNCH services across health system levels + + − −  Integrate packages of health services at point of care + + − −  Decentralization of decision-making and resource allocation + + − −  Beneficiaries included in strategic planning (community input) + + − − Accountability  Clear roles, responsibilities and expectations + + +/− −  Updated, effective HMIS + + − −  Consistent data collection and reporting at all health system levels + + − −  Ongoing monitoring and evaluation of health programmes and interventions + + − −  Specifically monitoring of progress towards MDG#4 + + +/−a +/−  Data-driven planning and decision-making responsive to population needs + + − −  Local involvement (community planning boards and committees) + + − − + Indicates clear activity, policy, participation and/or implementation of an element in the defined area during the study period; − indicates a lack of engagement of this element or merely planning, but not implementing policy/action during the study period; +/− Indicates ambiguous activity, policy, participation and/or implementation of an element in the defined area. a We found information indicating that a Kenya Country Countdown was conducted in 2013 (end of the study period), though this was not reported to our study team by Kenya’s MOH. Open in new tab Table 5 Comparison of health governance and leadership elements between progressing and non-progressing countries . Progressing . Non-progressing . . Liberia . Zambia . Kenya . Zimbabwe . Prioritization and support of child survival  Political support + + +/− +/−  Current policy framework + + + −  Policies and strategies implemented + + − −  Concurrent national policy focus on health, social welfare, development + + − −  Triple planning approach + + +/− −  Abuja Declaration target met during study + + − −  Non-financial health system resources (human, material, facility, etc.) + + − − Collaboration, coordination and inclusion  Donors aligned with national priorities + + − −  Collaborative strategic planning with partners/stakeholders + + − −  Coordination/collaboration between health and other sectors + + − −  Coordination and sharing resources among different health programmes + + − −  Coordination of MNCH services across health system levels + + − −  Integrate packages of health services at point of care + + − −  Decentralization of decision-making and resource allocation + + − −  Beneficiaries included in strategic planning (community input) + + − − Accountability  Clear roles, responsibilities and expectations + + +/− −  Updated, effective HMIS + + − −  Consistent data collection and reporting at all health system levels + + − −  Ongoing monitoring and evaluation of health programmes and interventions + + − −  Specifically monitoring of progress towards MDG#4 + + +/−a +/−  Data-driven planning and decision-making responsive to population needs + + − −  Local involvement (community planning boards and committees) + + − − . Progressing . Non-progressing . . Liberia . Zambia . Kenya . Zimbabwe . Prioritization and support of child survival  Political support + + +/− +/−  Current policy framework + + + −  Policies and strategies implemented + + − −  Concurrent national policy focus on health, social welfare, development + + − −  Triple planning approach + + +/− −  Abuja Declaration target met during study + + − −  Non-financial health system resources (human, material, facility, etc.) + + − − Collaboration, coordination and inclusion  Donors aligned with national priorities + + − −  Collaborative strategic planning with partners/stakeholders + + − −  Coordination/collaboration between health and other sectors + + − −  Coordination and sharing resources among different health programmes + + − −  Coordination of MNCH services across health system levels + + − −  Integrate packages of health services at point of care + + − −  Decentralization of decision-making and resource allocation + + − −  Beneficiaries included in strategic planning (community input) + + − − Accountability  Clear roles, responsibilities and expectations + + +/− −  Updated, effective HMIS + + − −  Consistent data collection and reporting at all health system levels + + − −  Ongoing monitoring and evaluation of health programmes and interventions + + − −  Specifically monitoring of progress towards MDG#4 + + +/−a +/−  Data-driven planning and decision-making responsive to population needs + + − −  Local involvement (community planning boards and committees) + + − − + Indicates clear activity, policy, participation and/or implementation of an element in the defined area during the study period; − indicates a lack of engagement of this element or merely planning, but not implementing policy/action during the study period; +/− Indicates ambiguous activity, policy, participation and/or implementation of an element in the defined area. a We found information indicating that a Kenya Country Countdown was conducted in 2013 (end of the study period), though this was not reported to our study team by Kenya’s MOH. Open in new tab Discussion Among the four study countries, Liberia and Zambia reduced U5M by two-thirds between 1990 and 2015, but both had almost double the U5M rates of Kenya and Zimbabwe in 1990. While slower progress in Kenya and Zimbabwe could have been influenced by the complexities of reducing preventable child deaths when starting at a lower baseline, this cross-study analysis identified HGL as a notable factor contributing to the differences in progress among study countries. Other published case studies from LMICs have also identified strong country HGL as a success factor for reducing U5M (Amouzou et al., 2012; Kuruvilla et al., 2014; Ahmed et al., 2016; Huicho et al., 2016; Kanyuka et al., 2016; Moucheraud et al., 2016; Ruducha et al., 2017). Effective HGL enables a solid health system foundation of national management capacity, comprehensive legislation, well-equipped workforce, functioning infrastructure, sufficient funding and robust data for decision-making, transparency and accountability. Our study expanded on these prior findings by identifying three overarching components of HGL that influenced progress in reducing U5M: (1) establishing child survival as a top national priority backed by a comprehensive policy and strategy framework and sufficient human, financial and material resources; (2) bringing together donors, strategic partners, health and non-health stakeholders and beneficiaries for strategic planning, decision-making, resource-allocation and coordination of services; and (3) maintaining accountability through a ‘monitor-review-act’ approach to improve MNCH. Liberia and Zambia clearly established child survival as a top priority supported by updated policy frameworks aligned with international recommendations and financed at the globally recommended level (African Union, 2006). Both countries integrated the health sector’s strategic direction with social welfare and development rather than having disconnected plans competing for attention and resources (Cavagnero et al., 2008; United Nations, 2010). Moreover, both Liberia and Zambia highlight the benefit of a ‘triple-planning approach’ in MNCH policy development, synchronously addressing urgent needs, adapting mid-term strategies to accelerate progress while also implementing sustainable long-term approaches, as shown in other countries achieving MDG#4 (Kuruvilla et al., 2014). High coverage of MNCH services has consistently been linked to cross-sector efforts addressing poverty, nutrition, education, gender equity, disease and sanitation (WHO and UNICEF, 2013; Mishra et al., 2015; Rasanathan et al., 2015), and approximately half of the reduction in maternal and child mortality in LMICs since 1990 is attributable to non-health sector investments (Kuruvilla et al., 2014; Bishai et al., 2016). While U5M can be reduced by leveraging limited resources across health programmes and other sectors (Ban, 2010; Jamison et al., 2013; Stenberg et al., 2014; Lie et al., 2015; Mishra et al., 2015), strong health systems require sustained investment (Every Woman Every Child, 2015). In contrast, persistently low health financing in Kenya and Zimbabwe hindered implementation of MNCH-related policies and strategies (Mishra et al., 2015). HGL in these countries remained focused on more immediate obligations and challenges rather than longer-term health system reforms. Collaborative partnerships offer LMICs a vehicle for aligning interests and obtaining additional resources to implement MNCH initiatives. However, strong HGL is required to effectively coordinate partners across the health system and to align donor assistance with national priorities (Organisation for Economic Co-operation Development, 2005; Atun et al., 2011; Mishra et al., 2015). HGL in both Liberia and Zambia collaborated with partners for strategic planning and persuaded them to support government-established MNCH initiatives. This balancing of donor investment in specific health interventions with more general health system strengthening can increase availability of health services (Kinney et al., 2010; Bryce et al., 2013; WHO and PMNCH; 2013; Stenberg et al., 2014; Mishra et al., 2015). In addition, a ‘health in all policies’ approach with integration of health and non-health programmes enabled HGL in Liberia and Zambia to synergize the efficient and effective provision of MNCH services (Kerber et al., 2007; Friberg et al., 2010; Were et al., 2015). Decentralization from national to sub-national levels can also improve responsiveness to local needs and priorities, further strengthening health systems (WHO, 1978, 2007; Kuruvilla et al., 2014; Mkoka et al., 2014; Maluka and Bukagile, 2016; Tsofa et al., 2017). Moreover, giving the community a voice in HGL promotes ownership, utilization of services and better health outcomes (WHO, 1978, 2008; Cornwall et al., 2000; Tsofa et al., 2017). In contrast, over-centralization of HGL, vertical programming and misalignment between partners, national priorities and local needs resulted in inefficient service delivery in Kenya and Zimbabwe. Accountability is a critical responsibility of national health leaders who must establish and implement mechanisms to monitor, review and act on results to improve child survival (WHO, 2011, 2015; Mishra et al., 2015; Schweitzer, 2015). In line with the global Countdown to 2015 expectations that countries monitor coverage of recommended MNCH interventions, identify gaps and propose new actions to improve survival (Bellagio Group, 2003; Bryce et al., 2006; Victora et al., 2016), both Liberia and Zambia were sharply focused on progress towards MDG#4 and quickly responded to deficiencies by implementing appropriate policies, strategies and initiatives. These strategic reforms were facilitated through a well-functioning HMIS (WHO, 2007) and a robust data-driven M&E approach, as has been shown in other LMICs that have met MDG#4 (Rowe, 2009; Kuruvilla et al., 2014). Although Countdown has raised the visibility and accountability for MNCH worldwide, many LMICs including Kenya and Zimbabwe lack sufficient data on vital statistics, disease surveillance, resource utilization or service availability to inform appropriate responses (Grove et al., 2015; Mikkelsen et al., 2015). Further investments are needed to ensure that MNCH data are collected at the point of care, transferred between health system levels, and compiled and reported at both national and local levels (AbouZahr et al., 2010; Agyepong et al., 2018). A major strength of this study is the comparison of two SSA nations that achieved MDG#4 with two that did not, highlighting successful strategies and persistent challenges influencing U5M. We conducted an extensive document review and obtained qualitative data from diverse participants. Limitations of our methods for the individual case studies have been published (Kipp et al., 2016; Brault et al., 2017, 2018; Haley et al., 2017). Because we were evaluating progress towards MDG#4 which measures U5M, we focused on pregnancy, the newborn period and early childhood, though we recognize that the continuum now also includes reproductive and adolescent periods (Countdown to 2030 Collaboration, 2018). Use of only four countries limits the study’s generalizability across SSA; however, our findings corroborate and extend findings from other countries that have successfully reduced U5M. Strong HGL can drive a significant reduction in U5M despite considerable financial, social and political challenges (Kuruvilla et al., 2014; Mishra et al., 2015). Political and health leaders must prioritize child survival on their development agendas, engage and align partners with national activities and commit adequate resources for universal availability of MNCH services (Bryce et al., 2013; Every Woman Every Child, 2015; United Nations, 2015). Cross-sector policies and strategies should concurrently address all determinants of MNCH, tackle inequities in access and quality of care, and encourage accountability (Agyepong et al., 2018). The experiences from our study countries can contribute to attaining the Sustainable Development Goal target of reducing U5M rates to <25 per 1000 live births in each country by 2030 (United Nations, 2015). Ethics The Institutional Review Board at Vanderbilt University Medical Center approved the qualitative component of the study, with Vanderbilt serving as the Coordinating Center (IRB# 130567). Local ethics approval was obtained from the following committees prior to data collection: Kenyatta National Hospital Ethics & Research Committee (KNH-ERC/A/A259; Kenya), University of Liberia Office of the Institutional Review Board (Liberia), ERES Converge Institutional Review Board (IRB# 00005948; Zambia), Joint Parirenyatwa Hospital and University of Zimbabwe College of Health Sciences Research Ethics Committee (JREC/193/13; Zimbabwe) and the Medical Research Council of Zimbabwe (MRCZ/A/1772; Zimbabwe). Acknowledgements We thank the research participants for sharing their time, experiences and opinions, and the research assistants who conducted and transcribed the interviews and focus group discussions (Kenya: Isabella Maina and Sophie Ngugi; Liberia: Wede M. Nagbe and Curtis H. Taylor; Zambia: Bisalomo Mwanza and Lawrence Mwenge; Zimbabwe: Abigail Mutsinze and Marigold Mupunga). We would also like to thank Dr Tigest Ketsela and Dr Charles Sagoe-Moses from the World Health Organization Regional Office for Africa who contributed to the initial conceptualization of this work. WHO AFRO Child Survival Study Team: Kenya: Stewart Kabaka, MBChB, MPH, Kenya Ministry of Health, Nairobi, Kenya; Kibet Sergon, MBChB, MSc WHO/Kenya Country Office, Nairobi, Kenya; Liberia: Adolphus T Clarke, BPharm, MPH, Liberia Ministry of Health, Monrovia, Liberia; Musu C Duworko, MD, MPH, WHO/Country Office, Monrovia, Liberia; Zambia: Penny Kalesha-Masumbu, MBChB, MPH, Zambia Ministry of Health, Lusaka, Zambia; Mary Katepa-Bwalya, MBChB, MMed, MPH, WHO/Zambia Country Office, Lusaka, Zambia; Zimbabwe: Bernard Madzima, MBChB, MPH, Zimbabwe Ministry of Health, Harare, Zimbabwe; Trevor Kanyowa, MBChB, MSc, WHO/Zimbabwe Country Office, Harare, Zimbabwe; WHO: Phanuel Habimana, MD, MPH, WHO/Regional Office for Africa, Brazzaville, Congo. Funding Funding for this project was provided by the World Health Organization Regional Office for Africa. Support for data management came from the Vanderbilt Institute for Clinical and Translational Research (grant UL1 TR000445 from the National Center for Advancing Translational Sciences at the National Institutes of Health). At the time of the study, Dr Kipp was a Scholar with the HIV/AIDS, Substance Abuse and Trauma Training Program (HA-STTP), at the University of California, Los Angeles; supported through an award from the National Institute on Drug Abuse of the National Institutes of Health (R25 DA035692). Dr Vermund’s participation was supported, in part, by NIH grant P30AI110527, the Tennessee Center for AIDS Research. Dr. Brault is currently supported by grant number K12HS023000 from the Agency for Healthcare Research and Quality. The content is solely the responsibility of the authors and does not represent the official views of the Agency for Healthcare Research and Quality. Conflict of interest statement. K.M., T.D., K.S., M.C.D., M.K.B., T.K. and P.H. are employees of the World Health Organization. All other authors have no competing interests. References Abouzahr C , Gollogly L, Stevens G. 2010 . Better data needed: everyone agrees, but no one wants to pay . The Lancet (London, England) 375 : 619 – 21 . Google Scholar Crossref Search ADS WorldCat Afnan-Holmes H , Magoma M, John T et al. 2015 . Tanzania's Countdown to 2015: an analysis of two decades of progress and gaps for reproductive, maternal, newborn, and child health, to inform priorities for post-2015 . The Lancet Global Health 3 : e396 – e409 . Google Scholar Crossref Search ADS PubMed WorldCat African Union (AU). 2006 . Abuja Call for Accelerated Action Towards Universal Access to HIV/AIDS, Tuberculosis and Malaria Services in Africa . Abuja, Nigeria: Abuja African Union . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Agyepong IA , Sewankambo N, Binagwaho A et al. 2018 . The path to longer and healthier lives for all Africans by 2030: the Lancet Commission on the future of health in sub-Saharan Africa . The Lancet (London, England) 390 : 2803 – 59 . Google Scholar Crossref Search ADS WorldCat Ahmed SM , Rawal LB, Chowdhury SA et al. 2016 . Cross-country analysis of strategies for achieving progress towards global goals for women’s and children health . Bulletin of the World Health Organization 94 : 351 – 61 . Google Scholar Crossref Search ADS PubMed WorldCat Amouzou A , Habi O, Bensaïd K, The Niger Countdown Case Study Working Group. 2012 . Reduction in child mortality in Niger: a Countdown to 2015 country case study . The Lancet (London, England) 380 : 1169 – 78 . Google Scholar Crossref Search ADS WorldCat Atun R , Pothapregada SK, Kwansah J, Degbotse D, Lazarus JV. 2011 . Critical interactions between the global fund-supported HIV programs and the health system in Ghana . JAIDS Journal of Acquired Immune Deficiency Syndromes 57(Suppl 2) : S72 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat Ban K. 2010 . Global Strategy for Women's and Children's Health . New York : United Nations . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Barbazza E , Tello JE. 2014 . A review of health governance: definitions, dimensions and tools to govern . Health Policy 116 : 1 – 11 . Google Scholar Crossref Search ADS PubMed WorldCat Bellagio Study Group on Child Survival. 2003 . Knowledge into action for child survival . The Lancet 362 : 323 – 7 . Crossref Search ADS WorldCat Berman P , Requejo J, Bhutta ZA et al. 2016 . Countries’ progress for women’s and children’s health in the Millennium Development Goal era: the Countdown to 2015 experience . BioMed Central 16(Suppl 2) : 791 . Google Scholar Crossref Search ADS WorldCat Bishai DM , Cohen R, Alfonso YN et al. 2016 . Factors contributing to maternal and child mortality reductions in 16 low- and middle-income countries between 1990 and 2010 . PLoS One 11 : e0144908 . Google Scholar Crossref Search ADS PubMed WorldCat Brault MA , Kennedy SB, Haley CA et al. 2018 . Factors influencing rapid progress in child health in post-conflict Liberia: a country case study on progress in child survival, 2000–2013 . BMJ Open 8 : e021879 . Google Scholar Crossref Search ADS PubMed WorldCat Brault MA , Ngure K, Haley CA et al. 2017 . The introduction of new policies and strategies to reduce inequities and improve child health in Kenya: a country case study on progress in child survival, 2000–2013 . PLoS One 12 : e0181777. Google Scholar Crossref Search ADS PubMed WorldCat Bryce J , Terreri N, Victora CG et al. 2006 . Countdown to 2015: tracking intervention coverage for child survival . The Lancet 368 : 1067 – 76 . Google Scholar Crossref Search ADS WorldCat Bryce J , Victora CG, Black RE. 2013 . The unfinished agenda in child survival . The Lancet (London, England) 382 : 1049 – 59 . Google Scholar Crossref Search ADS WorldCat Cavagnero E , Daelmans B, Gupta N, Scherpbier R, Shankar A. 2008 . Assessment of the health system and policy environment as a critical complement to tracking intervention coverage for maternal, newborn, and child health . The Lancet 371 : 1284 – 93 . Google Scholar Crossref Search ADS WorldCat Cornwall A , Lucas H, Pasteur K. 2000 . Introduction: accountability through participation: developing workable partnership models in the health sector . IDS Bulletin 31 : 1 – 13 . Google Scholar OpenURL Placeholder Text WorldCat Countdown to 2015. 2012 . Building a Future for Women and Children: The 2012 Report . Geneva : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Countdown to 2030 Collaboration. 2018 . Countdown to 2030: tracking progress towards universal coverage for reproductive, maternal, newborn, and child health . The Lancet 391 : 1538 . Crossref Search ADS WorldCat Every Woman Every Child. 2015 . The Global Strategy for Women's, Children's, and Adolescents' Health (2016–2030) . New York : United Nations . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Friberg IK , Kinney MV, Lawn JE et al. 2010 . Sub-Saharan Africa's mothers, newborns, and children: how many lives could be saved with targeted health interventions? PLoS Medicine 7 : e1000295. Google Scholar Crossref Search ADS PubMed WorldCat Government of Kenya. 2007 . Kenya Vision 2030 . Nairobi, Kenya : National Economic and Social Council of Kenya . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Government of Kenya. 2010 . The Constitution of Kenya. Nairobi, Kenya : National Council for Law Reporting . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Grove J , Claeson M, Bryce J et al. 2015 . Maternal, newborn, and child health and the Sustainable Development Goals—a call for sustained and improved measurement . The Lancet 386 : 1511 – 4 . Google Scholar Crossref Search ADS WorldCat Haley CA , Vermund SH, Moyo P et al. 2017 . Impact of a critical health workforce shortage on child health in Zimbabwe: a country case study on progress in child survival, 2000–2013 . Health Policy and Planning 32 : 613 – 24 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Huicho L , Segura ER, Huayanay-Espinoza CA et al. 2016 . Child health and nutrition in Peru within an antipoverty political agenda: a Countdown to 2015 country case study . The Lancet Global Health 4 : e414 – e26 . Google Scholar Crossref Search ADS PubMed WorldCat Jamison DT , Summers LH, Alleyne G et al. 2013 . Global health 2035: a world converging within a generation . The Lancet , 382 : 1898 – 1955 . Google Scholar Crossref Search ADS WorldCat Kanyuka M , Ndawala J, Mleme T et al. 2016 . Malawi and Millennium Development Goal 4: a Countdown to 2015 country case study . The Lancet Global Health 4 : e201 – e14 . Google Scholar Crossref Search ADS PubMed WorldCat Kenya Ministry of Health (MOH). 2007 . The Second National Health Sector Strategic Plan Mid Term Review . Nairobi, Kenya : Ministry of Health . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kenya Ministry of Medical Services (MOMS), Ministry of Public Health and Sanitation (MOPHS). 2009 . Health Sector Strategic Plan for Health Information System. Nairobi, Kenya : Ministry of Medical Services and Ministry of Public Health and Sanitation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kenya Ministry of Medical Services (MOMS), Ministry of Public Health and Sanitation (MOPHS). 2012 . Kenya Health Policy 2012–2030 . Nairobi, Kenya : Ministry of Medical Services and Ministry of Public Health & Sanitation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kenya Ministry of Medical Services (MOMS), Ministry of Public Health and Sanitation (MOPHS). 2013 . Health Sector Strategic and Investment Plan 2013–2017: The Second Medium Term Plan for Health . Nairobi, Kenya : Ministry of Medical Services and Ministry of Public Health & Sanitation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kenya Ministry of Medical Services (MOMS), Ministry of Public Health and Sanitation (MOPHS), World Health Organization (WHO). 2009 . Reversing the Trends, The Second National Health Sector Strategic Plan: Clinical Management and Referral Guidelines , Vols I–III . Nairobi, Kenya : Ministry of Medical Services and Ministry of Public Health and Sanitation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kenya Ministry of Public Health and Sanitation (MOPHS). 2008a . Reversing the Trends: The Second National Health Sector Strategic Plan of Kenya, 2008–2012 . Nairobi, Kenya : Republic of Kenya, Ministry of Public Health and Sanitation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kenya Ministry of Public Health and Sanitation (MOPHS). 2008b . Child Survival and Development Strategy 2008–2015 . Nairobi, Kenya : Ministry of Public Health and Sanitation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kenya Ministry of Public Health and Sanitation (MOPHS), Ministry of Medical Services (MOMS). 2010 . National Roadmap for Accelerating the Attainment of the MDGs Related to Maternal and Newborn Health in Kenya. Nairobi, Kenya : Ministry of Public Health and Sanitation and Ministry of Medical Services . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kerber KJ , DE Graft-Johnson JE, Bhutta ZA et al. 2007 . Continuum of care for maternal, newborn, and child health: from slogan to service delivery . The Lancet 370 : 1358 – 69 . Google Scholar Crossref Search ADS WorldCat Kinney MV , Kerber KJ, Black RE et al. 2010 . Sub-Saharan Africa's mothers, newborns, and children: where and why do they die? PLoS Medicine 7 : e1000294. Google Scholar Crossref Search ADS PubMed WorldCat Kipp AM , Maimbolwa M, Brault MA et al. 2017 . Improving access to child health services at the community level in Zambia: a country case study on progress in child survival, 2000–2013 . Health Policy and Planning 32 : 603 – 12 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Kuruvilla S , Schweitzer J, Bishai D et al. 2014 . Success factors for reducing maternal and child mortality . Bulletin of the World Health Organization 92 : 533B – 44B . Google Scholar Crossref Search ADS WorldCat Liberia Ministry of Health and Social Welfare (MOHSW). 2007 . The National Health Policy and Plan, 2007–2011 . Monrovia, Liberia : Ministry of Health and Social Welfare . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Liberia Ministry of Health and Social Welfare (MOHSW). 2008 . National Strategy for Child Survival in Liberia, 2008–2011 . Monrovia, Liberia : Ministry of Health and Social Welfare . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Liberia Ministry of Health and Social Welfare (MOHSW). 2011a . Country Situational Analysis Report. Monrovia, Liberia : Ministry of Health and Social Welfare . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Liberia Ministry of Health and Social Welfare (MOHSW). 2011b . National Health and Social Welfare Policy and Plan 2011–2021 . Monrovia, Liberia : Ministry of Health and Social Welfare . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Liberia Ministry of Health and Social Welfare (MOHSW). 2011c . Revised National Community Health Services Policy. Monrovia, Liberia : Ministry of Health and Social Welfare . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Liberia Ministry of Health and Social Welfare (MOHSW). 2011d . National Health and Social Welfare Policy and Plan, 2011–2021 (NHWPP). Monrovia, Liberia : Ministry of Health and Social Welfare . Liberia Ministry of Planning and Economic Affairs (MPEA). 2012 . Agenda for Transformation: Steps towards Liberia Rising 2030 . Monrovia, Liberia : Ministry of Planning and Economic Affairs . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Lie GS , Soucat AL, Basu S. 2015 . Financing women’s, children’s, and adolescent's health . BMJ 351 : h4267 . Maluka SO , Bukagile G. 2016 . Community participation in the decentralised district health systems in Tanzania: why do some health committees perform better than others? The International Journal of Health Planning and Management 31 : E86 – E104 . Google Scholar Crossref Search ADS PubMed WorldCat Mbonye AK , Sentongo M, Mukasa GK et al. 2012 . Newborn survival in Uganda: a decade of change and future implications . Health Policy and Planning 27(Suppl 3) : iii104 – 17 . Google Scholar Crossref Search ADS PubMed WorldCat Mikkelsen L , Phillips DE, Abouzahr C et al. 2015 . A global assessment of civil registration and vital statistics systems: monitoring data quality and progress . The Lancet 386 : 1395 – 406 . Google Scholar Crossref Search ADS WorldCat Mishra CK , Thomas J, Kumar R et al. 2015 . National leadership: driving forward the updated Global Strategy for Women's, Children's and Adolescents' Health . British Medical Journal 351 : h4282 . Google Scholar Crossref Search ADS PubMed WorldCat Mkoka DA , Kiwara A, Goicolea I, Hurtig A-K. 2014 . Governing the implementation of emergency obstetric care: experiences of rural district health managers, Tanzania . BMC Health Services Research 14 : 1 . Google Scholar Crossref Search ADS PubMed WorldCat Moucheraud C , Owen H, Singh NS, et al. 2016 . Countdown to 2015 country case studies: what have we learned about processes and progress towards MDGs 4 and 5? BMC Public Health 16 : 794 . Google Scholar Crossref Search ADS PubMed WorldCat Murh T. 2004 . User's Manual for Atlas.ti 6.0 . Berlin : Atlas.ti . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC National Coordinating Agency for Population and Development (NCAPD) [Kenya], Ministry of Medical Services (MOMS) [Kenya], Ministry of Public Health and Sanitation (MOPHS) [Kenya], Kenya National Bureau of Statistics (KNBS) [Kenya], ICF MACRO. 2011 . Kenya Service Provision Assessment Survey 2010. Nairobi, Kenya: National Coordinating Agency for Population and Development, Ministry of Medical Services, Ministry of Public Health and Sanitation, Kenya National Bureau of Statistics, and ICF Macro. Organisation for Economic Co-operation Development. 2005 . The Paris Declaration on Aid Effectiveness and the Accra Agenda for Action . Paris : Organisation for Economic Co-operation and Development Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Osika J , Altman D, Ekbladh L et al. 2010 . Zimbabwe Health System Assessment 2010 . Bethesda, MD : Health Systems 20/20 Project, Abt Associates, Inc . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Rasanathan K , Damji N, Atsbeha T et al. 2015 . Ensuring multisectoral action on the determinants of reproductive, maternal, newborn, child, and adolescent health in the post-2015 era . BMJ 351 : h4213 . Requejo JH , Bryce J, Barros AJ et al. 2015 . Countdown to 2015 and beyond: fulfilling the health agenda for women and children . The Lancet (London, England) 385 : 466 – 76 . Google Scholar Crossref Search ADS WorldCat Rowe A. 2009 . Potential of integrated continuous surveys and quality management to support monitoring, evaluation and the scale-up . The American Journal of Tropical Medicine and Hygiene 80 : 971 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat Ruducha J , Mann C, Singh NS et al. 2017 . How Ethiopia achieved Millennium Development Goal 4 through multisectoral interventions: a Countdown to 2015 case study . The Lancet Global Health 5 : e1142 – 51 . Google Scholar Crossref Search ADS PubMed WorldCat Schweitzer J on behalf of the Expert Consultative Group for Every Woman Every Child on Accountability. 2015 . Accountability in the 2015 Global Strategy for Women's, Children's, and Adolescent's Health . BMJ 351 : h4248 . Stenberg K , Axelson H, Sheehan P et al. 2014 . Advancing social and economic development by investing in women's and children's health: a new Global Investment Framework . The Lancet 383 : 1333 – 54 . Google Scholar Crossref Search ADS WorldCat Tsofa B , Molyneux S, Gilson L, Goodman C. 2017 . How does decentralisation affect health sector planning and financial management? A case study of early effects of devolution in Kilifi County, Kenya . International Journal for Equity in Health 16 : 151 . Google Scholar Crossref Search ADS PubMed WorldCat UNICEF, WHO, World Bank, UNDESA. 2015 . Levels and Trends in Child Mortality: Report 2015—Estimates Developed by the United Nations Inter-agency Group for Child Mortality Estimation . New York : United Nations Children’s Fund (UNICEF ). www.childmortality.org, accessed October 2015. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC United Nations. 2000 . United Nations Millennium Declaration . New York : United Nations . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC United Nations. 2010 . Global Strategy for Women's and Children's Health . New York : United Nations . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC United Nations. 2015 . Transforming Our World: The 2030 Agenda for Sustainable Development. New York, NY, USA : United Nations . USAID, UNAIDS. 2013 . Abuja +12: Shaping the Future of Health in Africa . Addis Ababa, Ethiopia : USAID . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Victora CG , Requejo JH, Barros AJ et al. 2016 . Countdown to 2015: a decade of tracking progress for maternal, newborn, and child survival . The Lancet (London, England) 387 : 2049 – 59 . Google Scholar Crossref Search ADS WorldCat Were WM , Daelmans B, Bhutta Z et al. 2015 . Children’s Health Priorities and Interventions. BMJ 351 : h4300 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC WHO, PMNCH. 2013 . A Policy Guide for Implementing Essential Interventions for Reproductive, Maternal, Newborn and Child Health (RMNCH): A Multisectoral Policy Compendium for RMNCH . Geneva : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC WHO, UNICEF. 2013 . End Preventable Deaths: Global Action Plan for Prevention and Control of Pneumonia and Diarrhoea . Geneva, Switzerland : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization (WHO). 1978 . Declaration of Alma-Ata: International Conference on Primary Health Care, Alma-Ata. USSR, 6–12 September 1978. Geneva, Switzerland : World Health Organization . World Health Organization (WHO). 2006 . Child Survival: A Strategy for the African Region . Brazzaville : World Health Organization, Regional Office for Africa . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization (WHO). 2007 . Everybody's Business: Strengthening Health Systems to Improve Health Outcomes: WHO's Framework for Action . Geneva : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization (WHO). 2008 . Ouagadougou Declaration on Primary Health Care and Health Systems in Africa: Achieving Better Health for Africa in the New Millennium . Brazzaville, Congo : World Health Organization Regional Office for Africa . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization (WHO). 2010 . Countdown to 2015 Decade Report (2000–2010): Taking Stock of Maternal, Newborn, and Child Survival . Geneva, Switzerland : WHO and UNICEF . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization (WHO). 2011 . Keeping Promises, Measuring Results: Commission on Information and Accountability for Women's and Children's Health . Geneva : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization (WHO). 2012 . Keeping Promises, Measuring Results: Commission on Information and Accountability for Women's and … Children's Health . World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization (WHO). 2015 . Every Woman, Every Child, Every Adolescent: Achievements and Prospects: The Final Report of the Independent Expert Review Group on Information and Accountability for Women's and Children's Health . Geneva : World Health Organization . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC WHO, PMNCH. 2011 . A Global Review of the Key Interventions Related to Reproductive, Maternal, Newborn and Child Health (RMNCH). Geneva, Switzerland : WHO and the Partnership for Maternal Newborn & Child Health (PMNCH ). Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC You D , Hug L, Ejdemyr S et al. 2015 . Global, regional, and national levels and trends in under-5 mortality between 1990 and 2015, with scenario-based projections to 2030: a systematic analysis by the UN Inter-agency Group for Child Mortality Estimation . The Lancet 386 : 2275 – 86 . Google Scholar Crossref Search ADS WorldCat Zambia Ministry of Community Development & Mother and Child Health (MCDMCH). 2013a . Strategic Plan for the Ministry of Community Development, Mother and Child Health 2013–2016 Lusaka . Zambia : Ministry of Community Development, Mother and Child Health . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zambia Ministry of Community Development & Mother and Child Health (MCDMCH). 2013b . Zambia Newborn Health Framework 2013 Lusaka . Zambia : Ministry of Community Development, Mother and Child Health . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zambia Ministry of Community Development & Mother and Child Health (MCDMCH), Ministry of Health (MOH). 2013 . Roadmap for Accelerating Reduction of Maternal, Newborn and Child Mortality 2013–2016 Lusaka . Zambia : Zambia Ministry of Community Development, Mother and Child Health . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zambia Ministry of Health (MOH). 2006 National Long Term Vision 2030. Lusaka, Zambia : Ministry of Health . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zambia Ministry of Health (MOH). 2008 Zambia 2008 Countdown to 2015 Conference Report. Lusaka, Zambia : Ministry of Health . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zambia Ministry of Health (MOH). 2012 National Health Strategic Plan 2011–2015. Lusaka, Zambia : Ministry of Health . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zambia Ministry of Health (MOH). 2013 . National Health Policy. Lusaka, Zambia : Ministry of Health . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zambia Ministry of Health (MOH), World Health Organization (WHO). 2011 . Services Availability and Resource Assessment: 2010 Summary Report . Lusaka, Zambia : Ministry of Health . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zimba E , Kinney MV, Kachale F et al. 2012 . Newborn survival in Malawi: a decade of change and future implications . Health Policy and Planning 27(Suppl 3) : iii88 – 103 . Google Scholar Crossref Search ADS PubMed WorldCat Zimbabwe Ministry of Economic Planning and Investment Promotion (MEPIP), United Nations Development Program (UNDP). 2012 . Zimbabwe Millennium Development Goals Progress Report . Harare, Zimbabwe : Ministry of Economic Planning and Investment Promotion . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zimbabwe Ministry of Health and Child Welfare (MOHCW). 2007 . The Zimbabwe National Maternal and Neonatal Health Road Map 2007–2015 . Harare, Zimbabwe : Ministry of Health and Child Welfare . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zimbabwe Ministry of Health and Child Welfare (MOHCW). 2010a . National Child Survival Strategy for Zimbabwe 2010–2015 . Harare, Zimbabwe : Ministry of Health and Child Welfare . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zimbabwe Ministry of Health and Child Welfare (MOHCW). 2010b . National Health Strategy 2009–2013: Equity and Quality in Health-A People’s Right . Harare, Zimbabwe : Ministry of Health and Child Welfare . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Author notes Connie A Haley and Marie A Brault authors contributed equally to this work. The members of the team have been provided in the Acknowledgments section. © The Author(s) 2019. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © The Author(s) 2019. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine.

Qian,, Jiwei;Jingwei He,, Alex;Dean-Chen Yin,, Jason

doi: 10.1093/heapol/czz001pmid: 30715314

Abstract The rapid diffusion of medical technologies is widely recognized as a key driver of healthcare cost escalation. The excessive duplication of technologies gives rise to the so-called medical arms race. Conventional wisdom tends to explain this phenomenon by external reimbursement mechanisms and hospitals’ competitive strategies, but has largely neglected the role played by health regulations that may also affect hospitals’ technology adoption decisions. This study sheds new light on the medical arms race with evidence from China, which has witnessed an unprecedented expansion of big tertiary hospitals and a keen pursuit of expensive medical technologies. Chinese hospitals aggressively pursue high-tech medical equipment as an opportunistic reaction to the peculiar health regulatory environment. By analysing a panel dataset collected from Shenzhen City, this study reveals a series of important impacts of the medical arms race in Chinese public hospitals. High-tech medical equipment is found to lead to an increase in hospital revenues and patient volumes, but no significant impact is noted on unit costs. While high-tech medical equipment is associated with a discernible improvement in clinical outcomes, no contribution to hospitals’ operational efficiency is noted. These findings are interpreted in the context of the broader health regulatory framework and China’s public hospital reforms. Health planning, regulation, technology, costs, hospitals Key Messages China has witnessed an unprecedented expansion of big tertiary hospitals and aggressive pursuit of expensive medical technologies. Driven by hospitals’ profit incentives, this fierce medical arms race is also an opportunistic reaction to China’s peculiar health regulatory environment. The empirical evidence derived from Shenzhen City suggests mixed outcomes of the race, but raises grave implications for health planning and regulation. Introduction The adoption and diffusion of medical technologies have been widely recognized as a major driver of the global escalation of healthcare costs (Okunade and Murthy, 2002). While technological innovations have undoubtedly improved the quality of care, they have also heightened the financial strains on health systems (Smith et al., 2009). Newhouse (1992) suggests that in the five decades preceding 1990, half of health spending in the USA was attributable to new technologies. Barros (1998) estimates that technological change may explain at least 30% of health expenditure growth in the Organization for Economic Co-operation and Development countries. The so-called medical arms race has been observed in many systems where excessive duplication of expensive medical technologies and services creates enormous waste, supplier-induced demands and rapid cost inflation (Kessler and McClellan, 2000; Cooper et al., 2011). When coupled with perverse economic incentives, the medical arms race may get even more intractable, not only fuelling a steep escalation in costs but also raising serious ethical issues (Hillman and Goldsmith, 2010). Most previous studies tend to attribute hospitals’ keen adoption of new technologies to their competitive strategy in the healthcare market (Devers et al., 2003; Berenson et al., 2006) as well as to economic incentives created by the external funding environment, such as health insurance (Romeo et al., 1984; Chou et al., 2004). In the USA, e.g., the introduction and penetration of managed care were found to have slowed down the diffusion of magnetic resonance imaging (MRI) equipment (Baker and Wheeler, 1998), mammography facilities (Baker and Brown, 1999) and an even broader collection of advanced medical technologies (Mas and Seinfeld, 2008), arguably in response to the cost-saving incentives embedded in managed care. Despite the received wisdom on the determinants and effects of the medical arms race, two research gaps are notable. First, while market structure and reimbursement mechanisms are certainly crucial factors leading to the aggressive acquisition of technologies, adequate attention has not been paid to the role of regulation (Selder, 2005). Given the sheer budget required, the purchase of expensive equipment is subject to government regulation in many health systems. Therefore, the regulatory environment may drastically alter hospitals’ economic incentives, which, in turn, will affect their technology acquisition and utilization decisions (Bech et al., 2009). Second, most previous studies were predominantly focused on the West, especially the USA, with very little knowledge gained elsewhere. The structural differences in healthcare delivery and regulatory environment may complicate the causal dynamics. Therefore, the medical arms race may be the result of many distinct contextual factors. This study aims to advance the existing knowledge on the medical arms race by examining the case of China, which has witnessed an unprecedented expansion of big tertiary hospitals and a zealous pursuit of state-of-the-art medical technologies in the past decade (Yip and Hsiao, 2014). What is the role played by health regulation in the medical arms race in China? Is the adoption of medical technologies a good strategy for hospitals to maximize their interests? What is its actual impact of such a strategy on various aspects of hospital operation? This study attempts to answer these research questions with secondary data collected from a city-level investigation. It illustrates that Chinese hospitals’ competition for high-tech medical equipment partially results from a peculiar regulatory environment in which essential resources are subject to varying degrees of government regulation. The loopholes within the regulatory framework enable profit-minded hospitals to purchase high-tech equipment to maximize revenues. This medical arms race impedes the efforts to build a reasonably structured delivery system and creates duplication and enormous waste in large tertiary hospitals. In China’s ongoing healthcare reform, policy attention should be paid to tweaking the health regulatory regime that is able to curb excess capacity of large hospitals. Background Under the old communist planned economy, public hospitals, the key providers in the Chinese health system, were run and financed by the state through regular subsidies and government-organized insurance programmes (Ma et al., 2008). In China’s marketization reforms that commenced in the 1980s, financial support from the government drastically dwindled, while most insurance programmes were either dismantled or significantly weakened (Hsiao, 1995). On average, government subsidies account for merely 10% of hospitals’ revenue, leaving the vast shortfall to be filled by user fees. Hospitals are allowed to earn revenues from drug sales and service charges. The government-regulated fee schedule sets prices for basic drugs and procedures below marginal costs but overprices high-tech diagnostic tests such as computerized tomography (CT) and MRI scans (Liu et al., 2000). This distorted fee schedule, coupled with the domination of fee-for-service in paying providers, powerfully motivated Chinese hospitals and physicians to overprescribe pharmaceuticals and high-tech diagnostic tests (Eggleston and Yip, 2004; Yip et al., 2010). Staff incomes were linked to revenue generation indicators, compounding the overprescription and administration of unnecessary procedures (Qian and He, 2018). This skewed incentive regime has paralleled the dramatic expansion of public hospitals in China. In contrast, primary care facilities are significantly underutilized, in part owing to patients’ distrust about their medical standards (Bhattacharyya et al., 2011). China’s health delivery system has thus been characterized by a big tertiary sector and a weak primary care foundation, despite laudable progress in strengthening the latter in recent years (Liu et al., 2015). In the absence of a functioning gate-keeping system, many Chinese patients are inclined to visit public hospitals instead of private hospitals or clinics, even for minor conditions (Eggleston et al., 2008; Wu and Lam, 2016). Since virtually all hospitals are self-financing facilities, their financial interest lies in keeping rather than referring patients elsewhere, even when a referral is clinically necessary (Yip and Hsiao, 2014). As such, patients continue to flood into big hospitals, which have developed a strong motivation to expand their capacity because that enables them to earn even more revenues. Between 1980 and 2010, the annual growth rate of the number of hospitals in China was 2.5%. There was also a 3.7-fold increase in the number of ‘very large hospitals’ (>800 beds) from 149 to 588 between 2000 and 2009 (Barber et al., 2014). Sixty-six giant hospitals are each equipped with at least 3000 beds (Kanyijie, 2018). Three elements are central to the expansion of hospitals: personnel (especially physicians), infrastructure (especially hospital beds) and medical equipment. Bech et al. (2009) highlight an important but often neglected trade-off between the regulation of hospitals’ physical capacity and their technology use. A similar regulatory trade-off also exists between the control of pharmaceutical use and technology adoption, suggesting the well-known cost-shifting strategies of hospitals. The aforementioned study inspired us to examine how the regulatory arrangements affect the medical arms race in the Chinese context. The Chinese hospital system manifests a peculiar mix of both tight regulation and laissez-faire policy. On the one hand, the government’s health bureaucracy has weak financial leverage to control the behaviours of public hospitals given its small contribution to their incomes (Hsiao, 2007). With limited policy tools, the health bureaucracy often has to resort to ad hoc administrative mandates or moral exhortations that generate a temporary effect at the most (He and Qian, 2013). In reality, hospital managers have considerable operational autonomy but limited checks and balances on them from other stakeholders (Yang, 2016). On the other hand, regulation is tightly exercised on some crucial aspects of hospital operation (Allen et al., 2014). The three elements aforementioned are subject to varying degrees of regulation in China. First, the employment of full-time medical professionals in Chinese public hospitals is subject to rigid civil service rules. The headcount is tightly controlled by the Institutional Organization Office (bianzhi bangongshi) of each level of government, over which even the health administration has little say (Eggleston et al., 2008). The actual recruitment of medical staff is administered by local personnel authorities. Hospital managers have very limited autonomy over hiring or firing decisions, even when some staff members do not perform (Eggleston et al., 2008; Yip and Hsiao, 2014). A severe shortage of nurses also plagues Chinese hospitals (Anand et al., 2008). In reality, hospital managers may overcome this manpower constraint by hiring part-time staff or full-time staff without permanent status. In other words, the regulatory constraint on personnel may not be as rigid as it appears on document. Second, infrastructural expansion of public hospitals is subject to regional health planning enacted by local governments. This administrative binding plan specifies key parameters, including the population-to-hospital-bed ratio in a locality in a given period (typically 5–10 years). Major infrastructure projects must obtain approval from the local planning commission (fa gai wei). However, in reality, considerable leeway exists where ambitious hospital managers may push the envelope. Many local governments also tend to turn a blind eye towards large projects as long as the hospital is able to finance them (World Bank, 2016; Yang, 2016). Unfortunately, many similar aggressive projects result in excessive borrowing and a high debt–asset ratio paralysing hospitals’ finance (Hu and Cai, 2014). In recent years, the Chinese government has significantly tightened up the regulation on the physical expansion of public hospitals, making it increasingly difficult to expand the number of hospital beds (National Commission of Health and Family Planning, 2015; World Bank, 2016). Third, constrained by the above regulations, investing in medical equipment could be a substitute expansionary strategy to increase a hospital’s capacity. Public hospitals’ acquisition of major medical equipment has been governed by a central government regulation enacted in 2005. This Chinese version of the certificate of needs system classifies medical equipment into Category A (unit price ≥5 million RMB) and Category B (unit price <5 million RMB). The acquisition of Category A equipment must be considered by provincial health bureaus in accordance with their respective regional plans and quota given by the National Health Commission, while city-level health bureaus may grant approval to acquisition applications for Category B equipment. In reality, its decentralized way of enforcement has created considerable loopholes, within which hospitals may manoeuvre (World Bank, 2016; Yang, 2016). Most important, given the regulation of physical and personnel expansion, purchasing high-tech medical equipment has become a possible and convenient way of expansion. In recent years, the Chinese government has introduced a series of measures to curb the notorious drug price inflation (such as the termination of the 15% drug profit margin and the introduction of the National Essential Medicines System), which has had a marked effect in drug cost reduction (Song et al., 2014; Zhou et al., 2015). Ironically, this has in fact made diagnostic equipment of even greater strategic value for the financial interest of hospitals. New evidence suggests increased diagnostic expenditures as hospitals have further speeded up the utilization of their CT and MRI scanners (Fu et al., 2018). In 2005 alone, China imported US$60 billion worth of medical equipment, accounting for >6% of large hospitals’ fixed assets. From 2002 to 2005, the number of MRI machines and CT scanners increased by 90.2% and 55.4%, respectively (Ma et al., 2008). A study in 2013 indicated that in four Chinese provinces, the number of CT and MRI scanners increased by 50% from 2006 to 2009 (He et al., 2013). According to the Chinese government, from 2010 to 2015, the total medical equipment value of all hospitals rose from 320 to 629 billion RMB, representing a doubling in expenditure (Liu et al., 2017). However, enormous wastes are created by overprescribing diagnostic tests. A study has reported that 17% of CT requisitions and 27% of MRI requisitions were found to be inappropriate and could not be justified (Li et al., 2005).The rapid acquisition of new technologies has been further accelerated by the aggressive move of foreign medical device companies into the lucrative Chinese market. Possibilities for rent seeking in the acquisition of high-tech equipment have further fuelled the medical arms race. Corruption scandals are frequently reported in the mass media (The Economic Observer, 2015). Conceptual framework and research hypotheses As the famous Roemer’s law states, hospital beds that are built tend to be used (Shain and Roemer, 1959). The same logic applies to medical technologies: new equipment that is purchased tends to be used. This is especially so when the prevalent incentives encourage over-utilization. In the Chinese hospital system, the low salary of physicians, the overpriced high-tech diagnostic tests and the fee-for-service payment mechanisms combine to create powerful incentives for physicians to overprescribe these services (Liu et al., 2017). The empirical analysis of this study is guided by a framework that incorporates both the hospital’s perspective and the patient’s perspective. Hospitals adopt new technologies to serve their own interests: increasing efficiency, attracting patients and generating revenues (Devers et al., 2003). The scramble to adopt medical technologies could be considered as a ‘non-price’ strategy for hospitals to remain competitive in the healthcare market (Kessler and McClellan, 2000; Yip et al., 2010). High-tech medical equipment is often understood as a signal for the quality of hospitals, and patients are likely to seek care in a high-quality hospital with a good reputation (Robinson and Luft, 1985). Hence, hospitals have strong incentives to invest in high-tech medical equipment because both patient volume and hospital revenue are likely to increase. In the meantime, since technology adoption may be an opportunistic response to regulatory constraints on personnel and physical expansion, we attempt to analyse the nuanced interaction among these strategies. The patient’s perspective is also essential since their welfare is most directly affected by clinical behaviours and the monetary costs of the medical arms race are ultimately paid by them. Therefore, we pose the following hypotheses for empirical testing: Hypothesis 1a:Hospital inpatient/outpatient revenue is increased with the value of high-tech medical equipment. Hypothesis 1b:Hospital drug sales revenue/diagnosis revenue is increased with the value of high-tech medical equipment. Hypothesis 1c: Increased high-tech medical equipment value is associated with increased service volume for both inpatient and outpatient care. A related hypothesis would be that the effectiveness of increasing investment in medical technologies increases alongside other dimensions of hospital capacity, especially the number of beds and physicians. In other words, investing in medical equipment may complement other strategies to increase hospitals’ financial interests. Under the Chinese context of bed and personnel regulation, this trend may be more nuanced, so we test the following hypothesis using the same outcomes from hypotheses 1a and 1b but adding in interaction terms: Hypothesis 1d:The effect of investing in high-tech medical equipment is increased with hospital capacity (i.e. hospital beds and personnel). We are also concerned with the cost implications of the medical arms race. In particular, does the adoption of new technologies increase patients’ financial burden? We thus propose the following hypothesis: Hypothesis 2:Increased high-tech medical equipment value is associated with higher inpatient/outpatient patient bills. We assume that the adoption of new medical technologies improves a hospital’s operational efficiency and clinical outcomes through higher diagnostic accuracy and better treatment outcomes (Hurlen et al., 2010). We take two commonly used indicators in the Chinese health statistical reporting system to represent operational efficiency: bed turnover rate and length of stay (LOS). Hospitals may benefit from new technologies through accelerated patient turnover that increases efficiency. Clinical outcomes are equally important since patients are the ultimate purchasers of service and their well-being should be considered. Two other commonly used indicators are used as proxy of clinical outcomes: improvement rate (of clinical condition) and death rate in a hospital. We, therefore, propose two more research hypotheses: Hypothesis 3:Bed turnover rate of a hospital is increased and length of stay is decreased with increased value of high-tech medical equipment. Hypothesis 4:Patients’ improvement rate in a hospital is increased and death rate is decreased with increased value of high-tech medical equipment. Methodology This study is set in the city of Shenzhen in China’s prosperous Guangdong Province. As the first ‘special economic zone’ (SEZ) in the country, Shenzhen has pioneered China’s economic reforms. It is now one of the fastest growing cities in the world and home to giant high-technology companies such as Tencent and Huawei. Shenzhen has a robust healthcare system, with a total of 134 hospitals equipped with 38 124 beds, 29 300 licenced physicians and 34 065 nurses (Shenzhen Statistical Bureau, 2017). It is observed that in China, the density of medical technologies is closely associated with the economic status of a locality (He et al., 2013). Therefore, given Shenzhen’s wealth and growing demand for quality care from the large middle- and upper-income classes, we expect that investment in cutting-edge medical technologies would be more pronounced in this city. For example, between 2007 and 2015, the number of MRI systems jumped from 32 to 61 in Shenzhen, and their monetary value almost tripled (see Figure 1). Shenzhen’s density (per million people) of MRI systems is even greater than that of Shanghai in 2009 (3.5 vs 3.2; He et al., 2013). These characteristics make Shenzhen an interesting case study to assess the effects of technology adoption in hospitals. Figure 1. View largeDownload slide The number of MRI systems, total monetary value and density in Shenzhen, 2007–2015. Source: Data on the number and total monetary value of MRI systems were collected from Shenzhen Health Statistical Yearbook (various years). The per-million people density was calculated based on population data collected from Shenzhen Statistical Yearbook (various years) Figure 1. View largeDownload slide The number of MRI systems, total monetary value and density in Shenzhen, 2007–2015. Source: Data on the number and total monetary value of MRI systems were collected from Shenzhen Health Statistical Yearbook (various years). The per-million people density was calculated based on population data collected from Shenzhen Statistical Yearbook (various years) Data are collected from the Shenzhen Health Statistical Yearbook spanning the years between 2007 and 2014 (i.e. 8 years in total). All public hospitals are required to report—on a yearly basis—the total value of any ‘major medical equipment’ with a unit price of one million RMB or above. This data source enables us to calculate the total monetary value of high-tech medical equipment in all public hospitals as an indication of their possession of new technologies. The panel nature of this dataset allows us to trace any new acquisition of medical equipment, and this longitudinal trend is able to capture the ‘race’ to medical arms. Our sample comprises all 80 public hospitals in Shenzhen. The revenue and expenditure related variables are deflated by the consumer price index to capture the price changes. Table 1 summarizes the variable definitions. Table 1. Variable definition Variable Definition Measurement Equipment value Total monetary value of ‘major medical equipment’ of a hospital Million RMB Out-test revenue Total revenue earned from diagnostic tests in a hospital’s outpatient divisions of the year Million RMB Out-drug revenue Total revenue earned from pharmaceuticals in a hospital’s outpatient divisions of the year Million RMB Out-service revenue Total revenue earned from service provision in a hospital’s outpatient divisions of the year Million RMB In-test revenue Total revenue earned from diagnostic tests in a hospital’s inpatient divisions of the year Million RMB In-drug revenue Total revenue earned from pharmaceuticals in a hospital’s inpatient divisions of the year Million RMB In-service revenue Total revenue earned from service provision in a hospital’s inpatient divisions of the year Million RMB Out-volume Total number of outpatient visits in a hospital of the year 1000 visits In-volume Total number of outpatient admissions in a hospital of the year 1000 admissions LOS Average length of inpatient stay in a hospital of the year Days Bed turnover Bed turnover rate in a hospital of the year % Improvement rate The percentage of patients with a disease who have seen improvement during an inpatient stay in a hospital of the year % Death rate The percentage of patients decreased during an inpatient stay in a hospital of the year % Physicians The total number of licenced physicians in a hospital of the year 100 Nurses The total number of registered nurses in a hospital of the year 100 Bed The total number of inpatient beds in a hospital of the year 100 In-bill Average cost per inpatient admission in a hospital of the year RMB Out-bill Average cost per outpatient visit in a hospital of the year RMB Large hospital Hospitals with 500 beds or more Dummy Variable Definition Measurement Equipment value Total monetary value of ‘major medical equipment’ of a hospital Million RMB Out-test revenue Total revenue earned from diagnostic tests in a hospital’s outpatient divisions of the year Million RMB Out-drug revenue Total revenue earned from pharmaceuticals in a hospital’s outpatient divisions of the year Million RMB Out-service revenue Total revenue earned from service provision in a hospital’s outpatient divisions of the year Million RMB In-test revenue Total revenue earned from diagnostic tests in a hospital’s inpatient divisions of the year Million RMB In-drug revenue Total revenue earned from pharmaceuticals in a hospital’s inpatient divisions of the year Million RMB In-service revenue Total revenue earned from service provision in a hospital’s inpatient divisions of the year Million RMB Out-volume Total number of outpatient visits in a hospital of the year 1000 visits In-volume Total number of outpatient admissions in a hospital of the year 1000 admissions LOS Average length of inpatient stay in a hospital of the year Days Bed turnover Bed turnover rate in a hospital of the year % Improvement rate The percentage of patients with a disease who have seen improvement during an inpatient stay in a hospital of the year % Death rate The percentage of patients decreased during an inpatient stay in a hospital of the year % Physicians The total number of licenced physicians in a hospital of the year 100 Nurses The total number of registered nurses in a hospital of the year 100 Bed The total number of inpatient beds in a hospital of the year 100 In-bill Average cost per inpatient admission in a hospital of the year RMB Out-bill Average cost per outpatient visit in a hospital of the year RMB Large hospital Hospitals with 500 beds or more Dummy View Large Table 1. Variable definition Variable Definition Measurement Equipment value Total monetary value of ‘major medical equipment’ of a hospital Million RMB Out-test revenue Total revenue earned from diagnostic tests in a hospital’s outpatient divisions of the year Million RMB Out-drug revenue Total revenue earned from pharmaceuticals in a hospital’s outpatient divisions of the year Million RMB Out-service revenue Total revenue earned from service provision in a hospital’s outpatient divisions of the year Million RMB In-test revenue Total revenue earned from diagnostic tests in a hospital’s inpatient divisions of the year Million RMB In-drug revenue Total revenue earned from pharmaceuticals in a hospital’s inpatient divisions of the year Million RMB In-service revenue Total revenue earned from service provision in a hospital’s inpatient divisions of the year Million RMB Out-volume Total number of outpatient visits in a hospital of the year 1000 visits In-volume Total number of outpatient admissions in a hospital of the year 1000 admissions LOS Average length of inpatient stay in a hospital of the year Days Bed turnover Bed turnover rate in a hospital of the year % Improvement rate The percentage of patients with a disease who have seen improvement during an inpatient stay in a hospital of the year % Death rate The percentage of patients decreased during an inpatient stay in a hospital of the year % Physicians The total number of licenced physicians in a hospital of the year 100 Nurses The total number of registered nurses in a hospital of the year 100 Bed The total number of inpatient beds in a hospital of the year 100 In-bill Average cost per inpatient admission in a hospital of the year RMB Out-bill Average cost per outpatient visit in a hospital of the year RMB Large hospital Hospitals with 500 beds or more Dummy Variable Definition Measurement Equipment value Total monetary value of ‘major medical equipment’ of a hospital Million RMB Out-test revenue Total revenue earned from diagnostic tests in a hospital’s outpatient divisions of the year Million RMB Out-drug revenue Total revenue earned from pharmaceuticals in a hospital’s outpatient divisions of the year Million RMB Out-service revenue Total revenue earned from service provision in a hospital’s outpatient divisions of the year Million RMB In-test revenue Total revenue earned from diagnostic tests in a hospital’s inpatient divisions of the year Million RMB In-drug revenue Total revenue earned from pharmaceuticals in a hospital’s inpatient divisions of the year Million RMB In-service revenue Total revenue earned from service provision in a hospital’s inpatient divisions of the year Million RMB Out-volume Total number of outpatient visits in a hospital of the year 1000 visits In-volume Total number of outpatient admissions in a hospital of the year 1000 admissions LOS Average length of inpatient stay in a hospital of the year Days Bed turnover Bed turnover rate in a hospital of the year % Improvement rate The percentage of patients with a disease who have seen improvement during an inpatient stay in a hospital of the year % Death rate The percentage of patients decreased during an inpatient stay in a hospital of the year % Physicians The total number of licenced physicians in a hospital of the year 100 Nurses The total number of registered nurses in a hospital of the year 100 Bed The total number of inpatient beds in a hospital of the year 100 In-bill Average cost per inpatient admission in a hospital of the year RMB Out-bill Average cost per outpatient visit in a hospital of the year RMB Large hospital Hospitals with 500 beds or more Dummy View Large We estimate the following model: ln yi,t=β ln⁡(EquipmentValuei,t)+δBedsi,t+γ(ln⁡(EquipmentValuei,t)*Bedsi,t)+ρXi,t+μi+ωt+ei,t, where yi,t refers to the outcomes of hospital i in year t. The outcomes include financial outcomes, such as revenues earned from diagnostic test, drugs and medical services for both inpatient and outpatient care. To show the relative changes of revenue, we transform the monetary variables by using the logarithm of those variables. The estimated coefficient in the results thus reflects the marginal effect of the changes of the outcome variables in percentage terms. Measurements for cost (such as patient bill sizes), operational efficiency (LOS and bed turnover rate) and clinical outcomes (improvement rate and death rate) are also considered as outcomes. β, γ, ρ, and δ are the parameters for the corresponding variables in the model. Covariates in the model are represented by Xi,t ⁠, including the number of physicians and nurses (under full-time permanent contract) as well as the number of beds in a hospital. These variables measure the capacity of a hospital in providing services. A dummy variable for large hospitals is also added. A medical institution with 500 beds or above is considered as a ‘large hospital’, according to the central government guideline on health resource planning that was promulgated in 2015.1 In practice, health resource planning regulations are typically set based on the number of beds. µi denotes hospital-specific effects that we control for. ωt corresponds to the year dummy variable, while ei,t is the error term in the model. In the analysis, it is possible that investing in high-tech medical equipment is a substitute for or complement to other strategies to increase hospital capacity. To account for this possibility, we test the interaction of equipment with other capacity indicators (i.e. number of beds for inpatient services and number of physicians for outpatient services) in the regression. In the model, EquipmentValuei,t thus refers to the value of the high-tech equipment in hospital i in year t. ln⁡(EquipmentValuei,t)*Bedsi,t is the interaction term between beds and total equipment value. If investing in medical equipment is a substitute for/complement to expanding hospital beds, the marginal effect of this interaction term will offset/enhance the marginal effect of hospital beds. All the regression models used in this study control for hospital-level fixed effect to address omitted variable bias at the hospital level. In the data analysis, effects from attributes of hospitals such as hospital levels are addressed by adding the hospital dummy. Results Table 2 reports the descriptive statistics of the dataset. The key independent variable—total value of a hospital’s major medical equipment—varies from RMB1.2 million to more than 590 million (US$1≈RMB6.63). Other resources also exhibit high variation across hospitals: the number of physicians and nurses varies from a dozen to more than 1000, and the number of hospital beds varies from 30 to more than 20 000. Utilization rates also vary, with the number of outpatient visits ranging from 18 000 to more than 3.5 million per year. The same pattern is noted for inpatient admissions, which range from 700 to 222 000. Table 2. Sample description Variable Obs. Mean SD Min. Max. Equipment value 411 47.9 68.3 1.2 590.2 Out-test revenue 411 25.8 31.8 0.1 224.6 Out-drug revenue 411 61.9 70.3 0.5 487.0 Out-service revenue 409 55.5 63.9 0.9 440.9 In-test revenue 352 9.5 13.7 0.1 111.6 In-drug revenue 352 31.2 47.8 0.2 324.9 In-service revenue 352 20.8 91.1 0.1 1655.3 In-bill 352 5826.4 4676.2 328 39493.0 Out-bill 411 140.1 79.3 8 479.5 Out-volume 411 1054.4 1083 18.2 3595.8 In-volume 352 16.4 17.5 0.7 222.2 LOS 345 8.1 4.8 4.2 56 Bed turnover 352 44.3 14.6 9.9 89.7 Improvement rate 228 30.1 16.5 2 88.9 Death rate 228 0.7 0.7 0 4.3 Physicians 411 2.6 2.1 0.1 10.8 Nurses 411 3 2.6 0.1 13.4 Beds 352 3.8 3.3 0.3 23.3 Large hospital 411 0.3 0.4 0 1 Variable Obs. Mean SD Min. Max. Equipment value 411 47.9 68.3 1.2 590.2 Out-test revenue 411 25.8 31.8 0.1 224.6 Out-drug revenue 411 61.9 70.3 0.5 487.0 Out-service revenue 409 55.5 63.9 0.9 440.9 In-test revenue 352 9.5 13.7 0.1 111.6 In-drug revenue 352 31.2 47.8 0.2 324.9 In-service revenue 352 20.8 91.1 0.1 1655.3 In-bill 352 5826.4 4676.2 328 39493.0 Out-bill 411 140.1 79.3 8 479.5 Out-volume 411 1054.4 1083 18.2 3595.8 In-volume 352 16.4 17.5 0.7 222.2 LOS 345 8.1 4.8 4.2 56 Bed turnover 352 44.3 14.6 9.9 89.7 Improvement rate 228 30.1 16.5 2 88.9 Death rate 228 0.7 0.7 0 4.3 Physicians 411 2.6 2.1 0.1 10.8 Nurses 411 3 2.6 0.1 13.4 Beds 352 3.8 3.3 0.3 23.3 Large hospital 411 0.3 0.4 0 1 View Large Table 2. Sample description Variable Obs. Mean SD Min. Max. Equipment value 411 47.9 68.3 1.2 590.2 Out-test revenue 411 25.8 31.8 0.1 224.6 Out-drug revenue 411 61.9 70.3 0.5 487.0 Out-service revenue 409 55.5 63.9 0.9 440.9 In-test revenue 352 9.5 13.7 0.1 111.6 In-drug revenue 352 31.2 47.8 0.2 324.9 In-service revenue 352 20.8 91.1 0.1 1655.3 In-bill 352 5826.4 4676.2 328 39493.0 Out-bill 411 140.1 79.3 8 479.5 Out-volume 411 1054.4 1083 18.2 3595.8 In-volume 352 16.4 17.5 0.7 222.2 LOS 345 8.1 4.8 4.2 56 Bed turnover 352 44.3 14.6 9.9 89.7 Improvement rate 228 30.1 16.5 2 88.9 Death rate 228 0.7 0.7 0 4.3 Physicians 411 2.6 2.1 0.1 10.8 Nurses 411 3 2.6 0.1 13.4 Beds 352 3.8 3.3 0.3 23.3 Large hospital 411 0.3 0.4 0 1 Variable Obs. Mean SD Min. Max. Equipment value 411 47.9 68.3 1.2 590.2 Out-test revenue 411 25.8 31.8 0.1 224.6 Out-drug revenue 411 61.9 70.3 0.5 487.0 Out-service revenue 409 55.5 63.9 0.9 440.9 In-test revenue 352 9.5 13.7 0.1 111.6 In-drug revenue 352 31.2 47.8 0.2 324.9 In-service revenue 352 20.8 91.1 0.1 1655.3 In-bill 352 5826.4 4676.2 328 39493.0 Out-bill 411 140.1 79.3 8 479.5 Out-volume 411 1054.4 1083 18.2 3595.8 In-volume 352 16.4 17.5 0.7 222.2 LOS 345 8.1 4.8 4.2 56 Bed turnover 352 44.3 14.6 9.9 89.7 Improvement rate 228 30.1 16.5 2 88.9 Death rate 228 0.7 0.7 0 4.3 Physicians 411 2.6 2.1 0.1 10.8 Nurses 411 3 2.6 0.1 13.4 Beds 352 3.8 3.3 0.3 23.3 Large hospital 411 0.3 0.4 0 1 View Large Hospital revenue Tables 3 and 4 demonstrate the impact of total equipment value on revenues for outpatient and inpatient services, respectively. In Table 3, Column (1) reveals a statistically significant association between diagnostic revenue and equipment value for outpatient services. A 1% increase in equipment value is associated with an increase in ∼0.31% in diagnostic test revenue for outpatient services. Column (2) suggests that equipment value remains significant, with a smaller yet still high magnitude (0.20%) for drug sales revenue. Column (3) shows that equipment value also serves as a very powerful predictor for hospitals’ outpatient service revenues. In all the models (1–3), the number of physicians is not statistically significant and positive. Interestingly, the interaction term between the number of physicians and equipment value is negative and significant in Column (3). This result suggests that investing on high-tech medical equipment and increasing the number of physicians are likely to be substitute rather than complementary strategies for hospitals to increase outpatient revenues. Yet, the coefficient of the interaction term in Column (3) is marginally significant. Table 3. Regression results: outpatient revenue and total equipment value (1) (2) (3) Log(Out-test revenue) Log(Out-drug revenue) Log(Out-service revenue) Log(Equipment value) 0.310*** (0.0876) 0.203*** (0.0550) 0.283*** (0.0630) Large hospital*Log(Equipment value) −0.00602 (0.00468) −0.00599 (0.00593) −0.00454 (0.00509) Physicians 0.860 (0.689) 0.500 (0.470) 0.726 (0.447) Physicians*Log(Equipment value) −0.0393 (0.0351) −0.0221 (0.0235) −0.0375* (0.0224) Nurses −0.00416 (0.0443) −0.00287 (0.0396) 0.000330 (0.0406) Constant 10.51*** (1.533) 13.40*** (0.953) 11.98*** (1.082) Hospital fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 411 411 409 Adj. R2 0.279 0.303 0.424 (1) (2) (3) Log(Out-test revenue) Log(Out-drug revenue) Log(Out-service revenue) Log(Equipment value) 0.310*** (0.0876) 0.203*** (0.0550) 0.283*** (0.0630) Large hospital*Log(Equipment value) −0.00602 (0.00468) −0.00599 (0.00593) −0.00454 (0.00509) Physicians 0.860 (0.689) 0.500 (0.470) 0.726 (0.447) Physicians*Log(Equipment value) −0.0393 (0.0351) −0.0221 (0.0235) −0.0375* (0.0224) Nurses −0.00416 (0.0443) −0.00287 (0.0396) 0.000330 (0.0406) Constant 10.51*** (1.533) 13.40*** (0.953) 11.98*** (1.082) Hospital fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 411 411 409 Adj. R2 0.279 0.303 0.424 Clustered robust standard error at hospital level in parentheses; * P< 0.1, *** P< 0.01, ‘large hospital’ denotes hospitals with 500 beds or more. View Large Table 3. Regression results: outpatient revenue and total equipment value (1) (2) (3) Log(Out-test revenue) Log(Out-drug revenue) Log(Out-service revenue) Log(Equipment value) 0.310*** (0.0876) 0.203*** (0.0550) 0.283*** (0.0630) Large hospital*Log(Equipment value) −0.00602 (0.00468) −0.00599 (0.00593) −0.00454 (0.00509) Physicians 0.860 (0.689) 0.500 (0.470) 0.726 (0.447) Physicians*Log(Equipment value) −0.0393 (0.0351) −0.0221 (0.0235) −0.0375* (0.0224) Nurses −0.00416 (0.0443) −0.00287 (0.0396) 0.000330 (0.0406) Constant 10.51*** (1.533) 13.40*** (0.953) 11.98*** (1.082) Hospital fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 411 411 409 Adj. R2 0.279 0.303 0.424 (1) (2) (3) Log(Out-test revenue) Log(Out-drug revenue) Log(Out-service revenue) Log(Equipment value) 0.310*** (0.0876) 0.203*** (0.0550) 0.283*** (0.0630) Large hospital*Log(Equipment value) −0.00602 (0.00468) −0.00599 (0.00593) −0.00454 (0.00509) Physicians 0.860 (0.689) 0.500 (0.470) 0.726 (0.447) Physicians*Log(Equipment value) −0.0393 (0.0351) −0.0221 (0.0235) −0.0375* (0.0224) Nurses −0.00416 (0.0443) −0.00287 (0.0396) 0.000330 (0.0406) Constant 10.51*** (1.533) 13.40*** (0.953) 11.98*** (1.082) Hospital fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 411 411 409 Adj. R2 0.279 0.303 0.424 Clustered robust standard error at hospital level in parentheses; * P< 0.1, *** P< 0.01, ‘large hospital’ denotes hospitals with 500 beds or more. View Large Table 4. Regression result: inpatient revenue and total equipment value (1) (2) (3) Log(In-test revenue) Log(In-drug revenue) Log(In-service revenue) Log(Equipment value) 0.321*** (0.111) 0.148 (0.0963) 0.237** (0.106) Large hospital*Log(Equipment value) 0.00480 (0.00786) −0.00591 (0.00699) 0.00361 (0.00992) Beds 1.393*** (0.405) 1.018*** (0.356) 1.484*** (0.500) Beds*Log(Equipment value) −0.0721*** (0.0205) −0.0515*** (0.0179) −0.0752*** (0.0247) Physicians 0.187** (0.0926) 0.126* (0.0682) 0.101 (0.135) Nurses 0.0689 (0.0662) 0.0661 (0.0458) 0.0245 (0.0595) Constant 8.187*** (1.998) 12.52*** (1.741) 10.61***(1.737) Hospital fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 352 352 352 Adj. R2 0.541 0.512 0.191 (1) (2) (3) Log(In-test revenue) Log(In-drug revenue) Log(In-service revenue) Log(Equipment value) 0.321*** (0.111) 0.148 (0.0963) 0.237** (0.106) Large hospital*Log(Equipment value) 0.00480 (0.00786) −0.00591 (0.00699) 0.00361 (0.00992) Beds 1.393*** (0.405) 1.018*** (0.356) 1.484*** (0.500) Beds*Log(Equipment value) −0.0721*** (0.0205) −0.0515*** (0.0179) −0.0752*** (0.0247) Physicians 0.187** (0.0926) 0.126* (0.0682) 0.101 (0.135) Nurses 0.0689 (0.0662) 0.0661 (0.0458) 0.0245 (0.0595) Constant 8.187*** (1.998) 12.52*** (1.741) 10.61***(1.737) Hospital fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 352 352 352 Adj. R2 0.541 0.512 0.191 Clustered robust standard error at hospital level in parentheses; * P< 0.1, ** P< 0.05, *** P< 0.01, ‘large hospital’ denotes hospitals with 500 beds or more. View Large Table 4. Regression result: inpatient revenue and total equipment value (1) (2) (3) Log(In-test revenue) Log(In-drug revenue) Log(In-service revenue) Log(Equipment value) 0.321*** (0.111) 0.148 (0.0963) 0.237** (0.106) Large hospital*Log(Equipment value) 0.00480 (0.00786) −0.00591 (0.00699) 0.00361 (0.00992) Beds 1.393*** (0.405) 1.018*** (0.356) 1.484*** (0.500) Beds*Log(Equipment value) −0.0721*** (0.0205) −0.0515*** (0.0179) −0.0752*** (0.0247) Physicians 0.187** (0.0926) 0.126* (0.0682) 0.101 (0.135) Nurses 0.0689 (0.0662) 0.0661 (0.0458) 0.0245 (0.0595) Constant 8.187*** (1.998) 12.52*** (1.741) 10.61***(1.737) Hospital fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 352 352 352 Adj. R2 0.541 0.512 0.191 (1) (2) (3) Log(In-test revenue) Log(In-drug revenue) Log(In-service revenue) Log(Equipment value) 0.321*** (0.111) 0.148 (0.0963) 0.237** (0.106) Large hospital*Log(Equipment value) 0.00480 (0.00786) −0.00591 (0.00699) 0.00361 (0.00992) Beds 1.393*** (0.405) 1.018*** (0.356) 1.484*** (0.500) Beds*Log(Equipment value) −0.0721*** (0.0205) −0.0515*** (0.0179) −0.0752*** (0.0247) Physicians 0.187** (0.0926) 0.126* (0.0682) 0.101 (0.135) Nurses 0.0689 (0.0662) 0.0661 (0.0458) 0.0245 (0.0595) Constant 8.187*** (1.998) 12.52*** (1.741) 10.61***(1.737) Hospital fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 352 352 352 Adj. R2 0.541 0.512 0.191 Clustered robust standard error at hospital level in parentheses; * P< 0.1, ** P< 0.05, *** P< 0.01, ‘large hospital’ denotes hospitals with 500 beds or more. View Large Table 4 presents the effect of total equipment value on hospitals’ inpatient care revenues. The positive effect of equipment value on hospital revenue appears similar for inpatient care vis-à-vis outpatient care. Column (1) shows that a 1% increase in equipment value is associated with a 0.32% increase in diagnostic test revenue. In other words, high-tech equipment plays an even bigger role in revenue generation in the inpatient division, arguably because of the more frequent tests during hospitalization. The equipment value is also positively and significantly associated with the inpatient service revenue in Column (3). There is no significant difference between large hospitals and other hospitals in terms of generating inpatient revenues by procuring high-tech medical equipment. The number of beds is significantly and positively associated with all the revenue indicators, suggesting that increasing the number of beds could be a very important strategy to earn inpatient revenues for hospitals. Interestingly, the interaction term between hospital beds and total equipment value yields a negative significant coefficient in all inpatient revenue categories [Columns (1)–(3)], suggesting that while both the number of beds and the value of equipment are positive and significant determinants of inpatient revenues, increasing the former and investing in the latter are substitute rather than complementary strategies for revenue generation as far as inpatient divisions are concerned. Patient volume and medical bills In Table 5, we investigate how total equipment value is associated with size of the bill paid by patients and examine whether there is an actual increase in patient volumes. Column (1) and Column (2) suggest that a 1% increase in total equipment value is associated with a 0.12% increase in inpatient service volume and a 0.21% increase in outpatient service volume, respectively. Yet, a 1% increase in equipment value has no effect on inpatient or outpatient bill sizes [Columns (3) and (4)]. There is no significant difference between large hospitals and other hospitals in the effect of equipment value in service volume or bill size. Table 5. Regression result: inpatient and outpatient bill sizes and volume, and total equipment value (1) (2) (3) (4) Log(In-volume) Log(Out-volume) Log(In-bill) Log(Out-bill) Log(Equipment value) 0.119* (0.0642) 0.211*** (0.0579) 0.0134 (0.0448) 0.0428 (0.0412) Large hospital*Log(Equipment value) −0.00125 (0.00526) −0.00131 (0.00552) −0.000320 (0.00180) −0.00871 (0.00701) Beds 0.803** (0.311) 0.123 (0.118) Beds*Log(Equipment value) −0.0396** (0.0155) −0.00674 (0.00588) Physicians 0.0514 (0.0510) 0.739 (0.474) 0.0823*** (0.0304) 0.241 (0.295) Physicians*Log(Equipment value) −0.0353 (0.0239) −0.00764 (0.0145) Nurses 0.0347 (0.0304) −0.0440 (0.0388) −0.0194 (0.0159) 0.108 (0.0657) Constant 6.488*** (1.210) 9.368*** (1.038) 7.827*** (0.711) 3.385*** (0.765) Hospital fixed effect Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes N 352 411 352 411 Adj. R2 0.426 0.244 0.211 0.253 (1) (2) (3) (4) Log(In-volume) Log(Out-volume) Log(In-bill) Log(Out-bill) Log(Equipment value) 0.119* (0.0642) 0.211*** (0.0579) 0.0134 (0.0448) 0.0428 (0.0412) Large hospital*Log(Equipment value) −0.00125 (0.00526) −0.00131 (0.00552) −0.000320 (0.00180) −0.00871 (0.00701) Beds 0.803** (0.311) 0.123 (0.118) Beds*Log(Equipment value) −0.0396** (0.0155) −0.00674 (0.00588) Physicians 0.0514 (0.0510) 0.739 (0.474) 0.0823*** (0.0304) 0.241 (0.295) Physicians*Log(Equipment value) −0.0353 (0.0239) −0.00764 (0.0145) Nurses 0.0347 (0.0304) −0.0440 (0.0388) −0.0194 (0.0159) 0.108 (0.0657) Constant 6.488*** (1.210) 9.368*** (1.038) 7.827*** (0.711) 3.385*** (0.765) Hospital fixed effect Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes N 352 411 352 411 Adj. R2 0.426 0.244 0.211 0.253 Clustered robust standard error at hospital level in parentheses; * P< 0.1, ** P< 0.05, *** P< 0.01, ‘large hospital’ denotes hospitals with 500 beds or more. View Large Table 5. Regression result: inpatient and outpatient bill sizes and volume, and total equipment value (1) (2) (3) (4) Log(In-volume) Log(Out-volume) Log(In-bill) Log(Out-bill) Log(Equipment value) 0.119* (0.0642) 0.211*** (0.0579) 0.0134 (0.0448) 0.0428 (0.0412) Large hospital*Log(Equipment value) −0.00125 (0.00526) −0.00131 (0.00552) −0.000320 (0.00180) −0.00871 (0.00701) Beds 0.803** (0.311) 0.123 (0.118) Beds*Log(Equipment value) −0.0396** (0.0155) −0.00674 (0.00588) Physicians 0.0514 (0.0510) 0.739 (0.474) 0.0823*** (0.0304) 0.241 (0.295) Physicians*Log(Equipment value) −0.0353 (0.0239) −0.00764 (0.0145) Nurses 0.0347 (0.0304) −0.0440 (0.0388) −0.0194 (0.0159) 0.108 (0.0657) Constant 6.488*** (1.210) 9.368*** (1.038) 7.827*** (0.711) 3.385*** (0.765) Hospital fixed effect Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes N 352 411 352 411 Adj. R2 0.426 0.244 0.211 0.253 (1) (2) (3) (4) Log(In-volume) Log(Out-volume) Log(In-bill) Log(Out-bill) Log(Equipment value) 0.119* (0.0642) 0.211*** (0.0579) 0.0134 (0.0448) 0.0428 (0.0412) Large hospital*Log(Equipment value) −0.00125 (0.00526) −0.00131 (0.00552) −0.000320 (0.00180) −0.00871 (0.00701) Beds 0.803** (0.311) 0.123 (0.118) Beds*Log(Equipment value) −0.0396** (0.0155) −0.00674 (0.00588) Physicians 0.0514 (0.0510) 0.739 (0.474) 0.0823*** (0.0304) 0.241 (0.295) Physicians*Log(Equipment value) −0.0353 (0.0239) −0.00764 (0.0145) Nurses 0.0347 (0.0304) −0.0440 (0.0388) −0.0194 (0.0159) 0.108 (0.0657) Constant 6.488*** (1.210) 9.368*** (1.038) 7.827*** (0.711) 3.385*** (0.765) Hospital fixed effect Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes N 352 411 352 411 Adj. R2 0.426 0.244 0.211 0.253 Clustered robust standard error at hospital level in parentheses; * P< 0.1, ** P< 0.05, *** P< 0.01, ‘large hospital’ denotes hospitals with 500 beds or more. View Large Importantly, Column (1) reveals that the interactive effect between hospital beds and equipment value is negative and significant. This is consistent with the results presented above that increasing the number of beds and investing in high-tech equipment are substitute rather than complementary strategies for expansion-minded hospitals. However, in Columns (2) and (4), neither the number of physicians nor the interaction between physicians and equipment value are statistically significant, implying that the regulatory constraints in personnel may not be rigidly binding in reality. Operational efficiency Moving on to operational efficiency indicators, Table 6 suggests that different from the case in generating revenue, investing in medical equipment does not accelerate the bed turnover rate or reduce LOS for inpatient services. The equipment effect is insignificant in both Columns (1) and (2). It is obvious that the number of beds is significant in improving the bed turnover rate (i.e. negative sign). However, the interaction term between the number of beds and total equipment value offsets the effect of the number of beds, which once again reinforces our finding that investing in high-tech medical equipment is a substitute for other hospital expansion strategies, such as increasing beds. This result may be interpreted as follows: in their keen pursuit of revenues, hospitals are able to offset infrastructural constraints by employing high-tech equipment rather than by improving their operational efficiency. Table 6. Regression result: bed turnover and service quality and total equipment value (1) Bed turnover (2) LOS (3) Improvement rate (4) Death rate Log(Equipment value) −0.856 (1.335) −0.150 (0.679) 1.771 (1.312) −0.468*** (0.171) Large hospital*Log(Equipment value) −0.191* (0.103) −0.00172 (0.0247) −0.155* (0.0862) −0.00621 (0.0117) Beds −17.40*** (5.109) 0.962 (2.018) 12.98** (4.916) −1.768*** (0.628) Beds*Log(Equipment value) 0.859*** (0.258) −0.0541 (0.0995) −0.662** (0.276) 0.103*** (0.0356) Physicians −0.458 (1.473) 0.219 (0.321) −2.656 (1.781) 0.0747 (0.134) Nurses −0.657 (1.256) 0.162 (0.191) 0.797 (0.922) −0.00395 (0.113) Constant 68.29*** (23.93) 9.728 (11.46) −0.912 (23.98) 9.149*** (2.866) Hospital fixed effect Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes N 352 345 228 228 Adj. R2 0.147 0.028 0.263 0.650 (1) Bed turnover (2) LOS (3) Improvement rate (4) Death rate Log(Equipment value) −0.856 (1.335) −0.150 (0.679) 1.771 (1.312) −0.468*** (0.171) Large hospital*Log(Equipment value) −0.191* (0.103) −0.00172 (0.0247) −0.155* (0.0862) −0.00621 (0.0117) Beds −17.40*** (5.109) 0.962 (2.018) 12.98** (4.916) −1.768*** (0.628) Beds*Log(Equipment value) 0.859*** (0.258) −0.0541 (0.0995) −0.662** (0.276) 0.103*** (0.0356) Physicians −0.458 (1.473) 0.219 (0.321) −2.656 (1.781) 0.0747 (0.134) Nurses −0.657 (1.256) 0.162 (0.191) 0.797 (0.922) −0.00395 (0.113) Constant 68.29*** (23.93) 9.728 (11.46) −0.912 (23.98) 9.149*** (2.866) Hospital fixed effect Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes N 352 345 228 228 Adj. R2 0.147 0.028 0.263 0.650 Clustered robust standard error at hospital level in parentheses; * P< 0.1, ** P< 0.05, *** P< 0.01, ‘large hospital’ denotes hospitals with 500 beds or more. View Large Table 6. Regression result: bed turnover and service quality and total equipment value (1) Bed turnover (2) LOS (3) Improvement rate (4) Death rate Log(Equipment value) −0.856 (1.335) −0.150 (0.679) 1.771 (1.312) −0.468*** (0.171) Large hospital*Log(Equipment value) −0.191* (0.103) −0.00172 (0.0247) −0.155* (0.0862) −0.00621 (0.0117) Beds −17.40*** (5.109) 0.962 (2.018) 12.98** (4.916) −1.768*** (0.628) Beds*Log(Equipment value) 0.859*** (0.258) −0.0541 (0.0995) −0.662** (0.276) 0.103*** (0.0356) Physicians −0.458 (1.473) 0.219 (0.321) −2.656 (1.781) 0.0747 (0.134) Nurses −0.657 (1.256) 0.162 (0.191) 0.797 (0.922) −0.00395 (0.113) Constant 68.29*** (23.93) 9.728 (11.46) −0.912 (23.98) 9.149*** (2.866) Hospital fixed effect Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes N 352 345 228 228 Adj. R2 0.147 0.028 0.263 0.650 (1) Bed turnover (2) LOS (3) Improvement rate (4) Death rate Log(Equipment value) −0.856 (1.335) −0.150 (0.679) 1.771 (1.312) −0.468*** (0.171) Large hospital*Log(Equipment value) −0.191* (0.103) −0.00172 (0.0247) −0.155* (0.0862) −0.00621 (0.0117) Beds −17.40*** (5.109) 0.962 (2.018) 12.98** (4.916) −1.768*** (0.628) Beds*Log(Equipment value) 0.859*** (0.258) −0.0541 (0.0995) −0.662** (0.276) 0.103*** (0.0356) Physicians −0.458 (1.473) 0.219 (0.321) −2.656 (1.781) 0.0747 (0.134) Nurses −0.657 (1.256) 0.162 (0.191) 0.797 (0.922) −0.00395 (0.113) Constant 68.29*** (23.93) 9.728 (11.46) −0.912 (23.98) 9.149*** (2.866) Hospital fixed effect Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes N 352 345 228 228 Adj. R2 0.147 0.028 0.263 0.650 Clustered robust standard error at hospital level in parentheses; * P< 0.1, ** P< 0.05, *** P< 0.01, ‘large hospital’ denotes hospitals with 500 beds or more. View Large Interestingly, the statistical result suggests that the bed turnover rate is slightly decreased (with statistical significance) in large hospitals rather than in small hospitals [Column (1)]. It seems that hospitals’ operational efficiency marginally declines in this case. In small hospitals, high-tech equipment may facilitate the treatment process as they tend to have a smaller number of patients in severe conditions. However, after upgrading medical technologies, the case-mix in large hospitals could change as more severe cases may be treated. We must acknowledge that both the local health administration and social health insurance agency usually impose control on a hospital’s standard LOS and, therefore, the results presented here may be complicated, subject to alternative interpretation. Clinical outcomes Table 6 also presents the impact of equipment value on clinical outcomes measured by in-hospital improvement rate and death rate. Apparently, increasing the number of beds can improve the clinical outcomes [lower death rate in Column (4)]. The effect of total equipment value is significantly negative in death rate [0.47% in Column (4)]. Interestingly, in large hospitals, the equipment value is negatively associated with the improvement rate. As suggested above, this result could be interpreted as that the case-mix in large hospitals may change as more severe cases may be treated after upgrading medical technologies. A patient’s in-hospital health outcome is a complex function of many clinical and non-clinical factors that cannot simply be explained by the utilization of medical technologies, especially when these technologies are mostly diagnostic equipment. But nevertheless, the statistical results still reveal a discernible positive effect in this regard. Again, echoing the results shown above, the interaction term between the number of beds and equipment value offsets the effect of the number of beds for both indexes shown in Columns (3) and (4). Summary of empirical results and robustness check On the basis of the statistical results presented above, Hypotheses 1a, 1b and 1c are well supported. High-tech medical equipment indeed contributes a great deal to hospital revenues and patient volumes in both inpatient and outpatient settings. Hypothesis 2 is not supported. Increasing investment in medical equipment produces no significant impact on bill size. Hypothesis 1d is not supported, revealing more nuanced effects of the medical arms race. Increasing the number of hospital beds and investing in medical equipment are substitute rather than complementary strategies for generating revenue. In other words, investing in high-tech medical equipment may offset some effects of expanding hospitals’ infrastructural capacity. Hypothesis 3 is not supported either. Investment in medical technologies does not lead to improved operational efficiency of hospitals. Hypothesis 4 is partially supported in the sense that investment in high-tech medical equipment does make some positive contribution to patients’ clinical outcomes in terms of reducing the death rate. Several other major healthcare reforms were introduced in China’s public hospital system in the past years that might have complicated the results presented in our study. The most significant intervention was the zero mark-up policy that has been implemented since 2012 in all public hospitals in Shenzhen. This intervention has removed the 15% profit margin of pharmaceutical products dispensed in hospitals that used to constitute a substantive proportion of hospitals’ revenues. Because the change in drug profit may affect hospitals’ operational behaviours, we thus performed an additional robustness check by controlling the effect of equipment value in post-2012 (including 2012) on revenue-related dependent variables (Table 7). The results are consistent with the patterns reported in Table 4. Moreover, the result suggests that the equipment effect is not significantly different before and after 2012, further corroborating our findings. Table 7. Regression results: outpatient revenue and total equipment value with 2012 effect (1) (2) (3) Log(In-test revenue) Log(In-drug revenue) Log(In-service revenue) Log(Equipment value) 0.321*** (0.112) 0.147 (0.0960) 0.238** (0.107) Log(Equipment value)*after2012 −0.0246 (0.0716) 0.0276 (0.0591) −0.0424 (0.100) Large hospital*Log(Equipment value) 0.00512 (0.00773) −0.00626 (0.00681) 0.00415 (0.0100) Beds 1.344*** (0.385) 1.073*** (0.346) 1.399*** (0.525) Beds*Log(Equipment value) −0.0693*** (0.0193) −0.0546*** (0.0173) −0.0704*** (0.0263) Physicians 0.190** (0.0955) 0.121* (0.0712) 0.108 (0.130) Nurses 0.0720 (0.0688) 0.0626 (0.0466) 0.0300 (0.0601) Constant 8.165*** (2.048) 12.54*** (1.749) 10.57*** (1.777) Hospital fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 352 352 352 Adj. R2 0.540 0.511 0.190 (1) (2) (3) Log(In-test revenue) Log(In-drug revenue) Log(In-service revenue) Log(Equipment value) 0.321*** (0.112) 0.147 (0.0960) 0.238** (0.107) Log(Equipment value)*after2012 −0.0246 (0.0716) 0.0276 (0.0591) −0.0424 (0.100) Large hospital*Log(Equipment value) 0.00512 (0.00773) −0.00626 (0.00681) 0.00415 (0.0100) Beds 1.344*** (0.385) 1.073*** (0.346) 1.399*** (0.525) Beds*Log(Equipment value) −0.0693*** (0.0193) −0.0546*** (0.0173) −0.0704*** (0.0263) Physicians 0.190** (0.0955) 0.121* (0.0712) 0.108 (0.130) Nurses 0.0720 (0.0688) 0.0626 (0.0466) 0.0300 (0.0601) Constant 8.165*** (2.048) 12.54*** (1.749) 10.57*** (1.777) Hospital fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 352 352 352 Adj. R2 0.540 0.511 0.190 Clustered robust standard error at hospital level in parentheses; * P< 0.1, ** P< 0.05, *** P< 0.01, ‘large hospital’ denotes hospitals with 500 beds or more. After 2012 refers to year 2012 and years after. View Large Table 7. Regression results: outpatient revenue and total equipment value with 2012 effect (1) (2) (3) Log(In-test revenue) Log(In-drug revenue) Log(In-service revenue) Log(Equipment value) 0.321*** (0.112) 0.147 (0.0960) 0.238** (0.107) Log(Equipment value)*after2012 −0.0246 (0.0716) 0.0276 (0.0591) −0.0424 (0.100) Large hospital*Log(Equipment value) 0.00512 (0.00773) −0.00626 (0.00681) 0.00415 (0.0100) Beds 1.344*** (0.385) 1.073*** (0.346) 1.399*** (0.525) Beds*Log(Equipment value) −0.0693*** (0.0193) −0.0546*** (0.0173) −0.0704*** (0.0263) Physicians 0.190** (0.0955) 0.121* (0.0712) 0.108 (0.130) Nurses 0.0720 (0.0688) 0.0626 (0.0466) 0.0300 (0.0601) Constant 8.165*** (2.048) 12.54*** (1.749) 10.57*** (1.777) Hospital fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 352 352 352 Adj. R2 0.540 0.511 0.190 (1) (2) (3) Log(In-test revenue) Log(In-drug revenue) Log(In-service revenue) Log(Equipment value) 0.321*** (0.112) 0.147 (0.0960) 0.238** (0.107) Log(Equipment value)*after2012 −0.0246 (0.0716) 0.0276 (0.0591) −0.0424 (0.100) Large hospital*Log(Equipment value) 0.00512 (0.00773) −0.00626 (0.00681) 0.00415 (0.0100) Beds 1.344*** (0.385) 1.073*** (0.346) 1.399*** (0.525) Beds*Log(Equipment value) −0.0693*** (0.0193) −0.0546*** (0.0173) −0.0704*** (0.0263) Physicians 0.190** (0.0955) 0.121* (0.0712) 0.108 (0.130) Nurses 0.0720 (0.0688) 0.0626 (0.0466) 0.0300 (0.0601) Constant 8.165*** (2.048) 12.54*** (1.749) 10.57*** (1.777) Hospital fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 352 352 352 Adj. R2 0.540 0.511 0.190 Clustered robust standard error at hospital level in parentheses; * P< 0.1, ** P< 0.05, *** P< 0.01, ‘large hospital’ denotes hospitals with 500 beds or more. After 2012 refers to year 2012 and years after. View Large Discussion The medical arms race and its effects have been studied extensively in the Western context, but little work has been done in Asia. In China, given the unique setting of regulatory constraints on both personnel and infrastructure for capacity expansion, one less regulated avenue for hospital competition has been to aggressively invest in medical technologies. Despite the desirability of technological advancement in healthcare, its actual utilization must consider a broader set of factors given the various associated risks, especially cost explosion. The race to state-of-the-art medical technologies is not solely driven by material incentives but is also explained by the regulatory framework and its enforcement in a health system. The case of China offers an excellent example to examine this medical arms race because Chinese public hospitals operate in a peculiar regulatory environment under which personnel, infrastructure and equipment—three essential medical sources—are subject to varying degrees of regulation. Therefore, a quantitative analysis is able to analyse not only the effect of technology adoption but also its interaction with other aspects of hospital capacity. The empirical results reported above lead us to several crucial issues warranting further discussion. First, our empirical results underscore the viability of technology adoption as a method to earn revenue for hospitals. Given that local governments in China provide little subsidization for public hospitals, acquiring and utilizing more cutting-edge technologies is an effective way to compensate for this financial shortfall. Contrary to what one might expect, the acquisition of high-tech equipment has not led to discernible increase in average inpatient or outpatient costs, as far as our sample in Shenzhen is concerned. Contrary to what we had expected, this study has found no significant evidence that hospitals over-use these new technologies which may escalate unit cost. This paradoxical finding might be explained by two reasons. First, the operational strategy of hospitals in Shenzhen seems to rely on higher patient volume rather than increasing unit cost, for the purpose of revenue generation. This result echoes the findings in the West that technology investment mainly acts as a non-price quality indicator to attract patients (Noether, 1988). The second possibility results from other interventions that are exogenous to our study design. For instance, the selective introduction of case-mix payment may alter hospitals’ strategies in utilizing certain technologies, which have, in turn, compounded our statistical results. Yet, this possibility could not be substantiated due to data unavailability. In short, the exact impact of technology adoption on medical costs may present more nuanced patterns that require closer scrutiny. Second, as explained earlier on, the medical arms race is, in part, a rational response of profit-seeking hospitals to regulated infrastructural expansion. This empirical study endorses this nuanced substitutive effect because the investment in high-tech medical equipment reduces the marginal effect of the number of hospital beds. In other words, to some degree, the medical arms race could be understood as resulting from an undersupply of hospital beds. This potentially leads to the saturation of technology and the oversupply of services such as MRI and CT (He et al., 2013; Yip and Hsiao, 2014). Third, despite the evidence that medical arms race may—to a certain extent—improve clinical outcomes, hospitals’ operational efficiency is not significantly improved by the acquisition of more valuable equipment. This reveals a tricky situation: While new technologies may have contributed to clinical outcomes through higher diagnostic accuracy and better treatment effects, these positive effects have not been translated into reduced LOS or faster bed turnover, as might be expected. One plausible explanation is that the prevalent incentive structure still essentially encourages hospitals—even major tertiary hospitals—to keep patients in rather than refer them out. Even when a discharge is clinically necessary, or is mandated by the health administration (or social health insurance agency) due to LOS control, hospitals in reality may still circumvent by readmitting the same patient. When this explanation is in effect, it defeats the desired positive contribution of new medical technologies. Concluding remarks Analysing a secondary dataset collected from the city of Shenzhen, this study has examined the wide adoption of medical technologies in Chinese public hospitals and its effects. We argue that despite some positive effects, this medical arms race must be understood in the broader health policy perspective. This race is made possible by the loopholes in the regulatory framework in which profit-driven and expansion-minded hospitals navigate to generate revenues through over-utilization of medical technologies, especially expensive diagnostic services. Several useful policy implications are in order. First, the weak regulatory enforcement of medical technology acquisition has provided possible room for opportunistic behaviours. Fortunately, the Chinese government seems to be increasingly alert to this problem. A laudable policy move has been the enactment of a new regulation governing the purchase of expensive equipment (National Health Commission, 2018). Provincial health authorities have been granted with considerable autonomy in the planning and approval of acquisition. Given the poor record of enforcement of similar policies in the past decade, bona fide enforcement must be exercised should the new regulation wish to avoid another failure and effectively prevent the oversaturation of medical technology. Second, the tricky interaction among the various regulations governing personnel, infrastructure and equipment illustrates the poor coordination of the regulatory framework which has essentially made hospitals’ opportunistic behaviours possible. Health policy makers must recognize the complexities and possible interactions in the regulatory regime and take an incentive-compatible view in its design (Qian, 2015). Specifically, regulations with regard to various types of medical resources must be compatible and provide a coherent set of incentives to hospitals. Third, one possible long-term solution to this chronic problem is a move away from fee-for-service in paying providers. Alternative payment methods, such as capitation and case-mix, have been experimented with in many localities and have produced largely positive outcomes (Gao et al., 2014). These prospective payment methods are associated with in-built cost containment incentives and may be more effective in curbing hospitals’ desire to join the medical arms race. This study is certainly not without limitations. First, the findings are all associative and the pathways are not clearly defined, and so some speculation has been introduced as to the means by which increased investment in medical technology affects the outcomes. Second, the sample size only covers public hospitals in Shenzhen, an affluent SEZ that may not be representative of most Chinese cities. Therefore, we acknowledge a modest ambition of generalizability. Third, we assume that the investment in medical technology means that these services will be utilized or over-utilized, which may not always be true. Last, there may still be a couple of imperfections in measurement. For instance, the data on physicians and nurses may not include those staff hired with non-permanent status. The operationalization of a hospital’s operational efficiency into LOS or bed turnover rate may also be subject to debate. While the use of secondary hospital-level administrative data is fairly common in health policy studies, the possibility of inaccurate reporting could not be completely ruled out. Therefore, the results of this study should be interpreted with caution. Ethical approval The empirical evidence shown in this study is collected solely from open-access secondary data published by the Shenzhen Municipal Government, PR China. The dataset contains no sensitive information, and therefore, the authors believe that no ethical review is necessary. Funding This study was funded by the Departmental Research Grant of the Department of Asian and Policy Studies, The Education University of Hong Kong. Conflict of interest statement. None declared. Footnotes 1 The State Council, National Health Service Planning Guideline. 2015–20, see http://www.gov.cn/zhengce/content/2015-03/30/content_9560.htm, accessed 11 December 2018. Acknowledgements The authors would like to thank the following persons for their professional research assistance: Mandy Sin-Man Wong, Jiayi Min, Kenneth Kin-Long Wong, Sunny Kwan-Yiu Wong and Luke Yuan-Kun Luo. References Allen P , Cao Q , Wang H. 2014 . Public hospital autonomy in China in an international context . The International Journal of Health Planning and Management 29 : 141 – 59 . Google Scholar Crossref Search ADS PubMed Anand S , Fan VY , Zhang J et al. 2008 . China’s human resources for health: quantity, quality, and distribution . Lancet (London, England) 372 : 1774 – 81 . Google Scholar Crossref Search ADS PubMed Baker LC , Brown ML. 1999 . Managed care, consolidation among health care providers, and health care: evidence from mammography . The RAND Journal of Economics 30 : 351 – 74 . Google Scholar Crossref Search ADS PubMed Baker LC , Wheeler SK. 1998 . Managed care and technology diffusion: the case of MRI . Health Affairs (Project Hope) 17 : 195 – 207 . Google Scholar Crossref Search ADS PubMed Barber SL , Borowitz M , Bekedam H , Ma J. 2014 . The hospital of the future in China: China’s reform of public hospitals and trends from industrialized countries . Health Policy and Planning 29 : 367 – 78 . Google Scholar Crossref Search ADS PubMed Barros PP. 1998 . The black box of health care expenditure growth determinants . Health Economics 7 : 533 – 44 . Google Scholar Crossref Search ADS PubMed Bech M , Christiansen T , Dunham K et al. 2009 . The influence of economic incentives and regulatory factors on the adoption of treatment technologies: a case study of technologies used to treat heart attacks . Health Economics 18 : 1114 – 32 . Google Scholar Crossref Search ADS PubMed Berenson RA , Bodenheimer T , Pham HH. 2006 . Specialty-service lines: salvos in the new medical arms race . Health Affairs 25 , doi:10.1377/hlthaff.25.w337. Bhattacharyya O , Yin D , Wong ST , Chen B. 2011 . Evolution of primary care in China 1997–2009 . Health Policy (Amsterdam, Netherlands) 100 : 174 – 80 . Google Scholar Crossref Search ADS PubMed Chou SY , Liu JT , Hammitt JK. 2004 . National Health Insurance and technology adoption: evidence from Taiwan . Contemporary Economic Policy 22 : 26 – 38 . Google Scholar Crossref Search ADS Cooper Z , Gibbons S , Jones S , McGuire A. 2011 . Does hospital competition save lives? Evidence from the English NHS patient choice reforms . The Economic Journal 121 : 228 – 60 . Google Scholar Crossref Search ADS Devers KJ , Brewster LR , Casalino LP. 2003 . Changes in hospital competitive strategy: a new medical arms race? Health Services Research 38 : 447 – 69 . Google Scholar Crossref Search ADS PubMed Eggleston K , Ling L , Qingyue M et al. 2008 . Health service delivery in China: a literature review . Health Economics 17 : 149 – 65 . Google Scholar Crossref Search ADS PubMed Eggleston K , Yip W. 2004 . Hospital competition under regulated prices: application to urban health sector reforms in China . International Journal of Health Care Finance and Economics 4 : 343 – 68 . Google Scholar Crossref Search ADS PubMed Fu Q , Li L , Yip W. 2018 . Intended and unintended impacts of price changes for drugs and medical services: evidence from China . Social Science and Medicine 211 : 114 – 22 . Google Scholar Crossref Search ADS PubMed Gao C , Xu F , Liu GG. 2014 . Payment reform and changes in health care in China . Social Science & Medicine (1982) 111 : 10 – 6 . Google Scholar Crossref Search ADS PubMed He D , Yu H , Chen Y. 2013 . Equity in the distribution of CT and MRI in China: a panel analysis . International Journal for Equity in Health 12 : 39. Google Scholar Crossref Search ADS PubMed He JA , Qian J. 2013 . Hospitals’ responses to administrative cost-containment policy in urban China: the case of Fujian Province . The China Quarterly 216 : 946 – 69 . Google Scholar Crossref Search ADS Hillman BJ , Goldsmith JC. 2010 . The uncritical use of high-tech medical imaging . The New England Journal of Medicine 363 : 4 – 6 . Google Scholar Crossref Search ADS PubMed Hsiao W. 1995 . The Chinese health care system: lessons for other nations . Social Science & Medicine (1982) 41 : 1047 – 55 . Google Scholar Crossref Search ADS PubMed Hsiao W. 2007 . The political economy of Chinese health reform . Health Economics, Policy, and Law 2 : 241 – 9 . Google Scholar Crossref Search ADS PubMed Hu J , Cai Z. 2014 . The debt analysis of provincial hospitals in Zhejiang . Chinese Health Economics Journal 10 : 87 – 9 (in Chinese). Hurlen P , Østbye T , Borthne AS , Gulbrandsen P. 2010 . Does improved access to diagnostic imaging results reduce hospital length of stay? A retrospective study . BMC Health Services Research 10 : 262. Google Scholar Crossref Search ADS PubMed Kanyijie. 2018 . Number of beds ranking in China’s big hospitals, 2017. http://www.sohu.com/a/227317751_456062, accessed 30 June 2018. Kessler DP , McClellan MB. 2000 . Is hospital competition socially wasteful? The Quarterly Journal of Economics 115 : 577 – 615 . Google Scholar Crossref Search ADS Li H , Wang Z , Tian Q et al. 2005 . The analysis of large medical equipment in Henan Province . China Medical Equipment 2 : 9 – 12 (in Chinese). Liu M , Qin Y , Pan J. 2017 . Does medical equipment expansion lead to more diagnostic services? Evidence from China’s Sichuan Province . Emerging Markets Finance and Trade 53 : 1289 – 300 . Google Scholar Crossref Search ADS Liu X , Liu Y , Chen N. 2000 . The Chinese experience of hospital price regulation . Health Policy and Planning 15 : 157 – 63 . Google Scholar Crossref Search ADS PubMed Liu X , Zhao S , Zhang M et al. 2015 . The development of rural primary health care in China’s health system reform . Journal of Asian Public Policy 8 : 88 – 101 . Google Scholar Crossref Search ADS Ma J , Lu M , Quan H. 2008 . From a national, centrally planned health system to a system based on the market: lessons from China . Health Affairs 27 : 937 – 48 . Google Scholar Crossref Search ADS PubMed Mas N , Seinfeld J. 2008 . Is managed care restraining the adoption of technology by hospitals? Journal of Health Economics 27 : 1026 – 45 . Google Scholar Crossref Search ADS PubMed National Commission of Health and Family Planning. 2015 . The Commission’s Notice on Mandatory Reporting of Public Hospitals’ Bed Capacity. National Health Commission. 2018 . Policy interpretation of the new inventory of major medical equipment (2018). http://www.nhfpc.gov.cn/zwgk/jdjd/201804/da87c8aa190e4620bab727b88d39cc00.shtml, accessed 30 May 2018. Newhouse JP. 1992 . Medical care costs: how much welfare loss? The Journal of Economic Perspectives 6 : 3 – 21 . Google Scholar Crossref Search ADS PubMed Noether M. 1988 . Competition among hospitals . Journal of Health Economics 7 : 259 – 84 . Google Scholar Crossref Search ADS PubMed Okunade A , Murthy VNR. 2002 . Technology as a “major driver” of health care costs: a cointegration analysis of the Newhouse conjecture . Journal of Health Economics 21 : 147 – 59 . Google Scholar Crossref Search ADS PubMed Qian J. 2015 . Reallocating authority in the Chinese health system: an institutional perspective . Journal of Asian Public Policy 8 : 19 – 35 . Google Scholar Crossref Search ADS Qian J , He JA. 2018 . The bonus scheme, motivation crowding-out and quality of the doctor-patient encountered in Chinese public hospitals . Public Organization Review 18 : 143 – 58 . Google Scholar Crossref Search ADS Robinson JC , Luft HS. 1985 . The impact of hospital market structure on patient volume, average length of stay, and the cost of care . Journal of Health Economics 4 : 333 – 56 . Google Scholar Crossref Search ADS PubMed Romeo AA , Wagner JL , Lee RH. 1984 . Prospective reimbursement and the diffusion of new technologies in hospitals . Journal of Health Economics 3 : 1 – 24 . Google Scholar Crossref Search ADS PubMed Selder A. 2005 . Physician reimbursement and technology adoption . Journal of Health Economics 24 : 907 – 30 . Google Scholar Crossref Search ADS PubMed Shain M , Roemer MI. 1959 . Hospital costs relate to the supply of beds . Modern Hospital 92 : 71 – 3 . Google Scholar PubMed Shenzhen Statistical Bureau. 2017 . Shenzhen Statistical Yearbook 2017. Smith S , Newhouse JP , Freeland MS. 2009 . Income, insurance, and technology: why does health spending outpace economic growth? Health Affairs 28 : 1276 – 84 . Google Scholar Crossref Search ADS PubMed Song Y , Bian X , Petzold M et al. 2014 . The impact of China’s national essential medicine system on improving rational drug use in primary health care facilities: an empirical study in four provinces . BMC Health Services Research 14 : 507. Google Scholar Crossref Search ADS PubMed The Economic Observer. 2015 . Procurement of medical equipment and pharmaceuticals: a heavily corrupt area, May 29 2015. World Bank, World Health Organization, Chinese Government. 2016 . Deepening Health Reform in China: Building High-Quality and Value-Based Service Delivery. http://www.worldbank.org/en/country/china/publication/healthy-china-deepening-health-reform-in-china, accessed 2 July 2018. Wu D , Lam TP. 2016 . Underuse of primary care in China: the scale, causes, and solutions . Journal of the American Board of Family Medicine 29 : 240 – 7 . Google Scholar Crossref Search ADS PubMed Yang Y. 2016 . Research on China’s public hospital governance: from perspective of implementation issues . Chinese Medical Journal 129 : 357 – 60 . Google Scholar Crossref Search ADS PubMed Yip W , Hsiao W. 2014 . Harnessing the privatization of China’s fragmented health-care delivery . Lancet (London, England) 384 : 805 – 18 . Google Scholar Crossref Search ADS PubMed Yip WC-M , Hsiao W , Meng Q et al. 2010 . Realignment of incentives for health-care providers in China . Lancet (London, England) 375 : 1120 – 30 . Google Scholar Crossref Search ADS PubMed Zhou Z , Su Y , Campbell B et al. 2015 . The impact of China’s zero-markup drug policy on county hospital revenue and government subsidy levels . Journal of Asian Public Policy 8 : 102 – 16 . Google Scholar Crossref Search ADS © The Author(s) 2019. 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)

Ravi, Sanjana J; Snyder, Michael R; Rivers, Caitlin

doi: 10.1093/heapol/czy102pmid: 30624680

Abstract The 2014–16 West Africa Ebola epidemic was a watershed moment for global health. The outbreak galvanized global action around strengthening infectious disease prevention, detection and response capabilities. We examined the nascent landscape of international programmes, initiatives and institutions established in the aftermath of the 2014–16 Ebola outbreak with the aim of assessing their progress to date to illustrate the current state of the world’s global health security architecture. We also compare these efforts with shortcomings in epidemic management documented during the epidemic, and underscore remaining gaps in regional and global epidemic response capabilities that might benefit from additional programmatic and financial support. Notably, most of the post-Ebola initiatives considered in this analysis have yet to meet their financial goals. Operational progress has also been limited, revealing a need for continued investments to improve outbreak surveillance and detection capabilities specifically. Furthermore, our review highlighted the dominance of the USA and Europe in leading and financing efforts to coordinate long-term recovery efforts in West Africa, strengthen health systems across the continent, and enhance global preparedness for future epidemics, raising important questions about ownership of global health security efforts in non-Western regions of the world. Finally, the lack of transparency and available data on these initiatives’ activities and budgets also complicate efforts to project their impacts on the global health security landscape. Ebola, epidemics, health systems, global health security, policy Key Messages Most of the multilateral health security-strengthening initiatives established in the wake of the 2014–16 West Africa Ebola epidemic have yet to meet their funding targets. Additional investments are needed to strengthen global outbreak surveillance and detection capabilities. The United States and Europe currently lead and finance the majority of post-Ebola initiatives identified in this review, highlighting the need for greater low- and middle-income countries ownership of health security initiatives. Introduction On 18 March 2014, Médecins Sans Frontières initiated emergency response operations to contain a small outbreak of Ebola virus disease in Guéckédou, Guinea. Just 2 weeks later, MSF declared the outbreak ‘of a magnitude never before seen’, and warned that the disease had spread too far to be easily contained (Médecins Sans Frontières, 2015a,b). This warning was borne out when, after months of rapid escalation, the World Health Organization (WHO) declared the epidemic a Public Health Emergency of International Concern (PHEIC) in August of 2014. It was not until 17 months later, in March of 2016, that the PHEIC designation was lifted, after 28 616 cases and 11 310 deaths were recorded in Guinea, Sierra Leone and Liberia (World Health Organization, 2016d). The initial, protracted failure to contain what was previously thought to be an easily controlled disease was a grave lesson for the international public health community. Many of the systems and institutions responsible for preventing, detecting and responding to outbreaks were largely unprepared to operate effectively, each in their own way. Although this observation had been made many times prior to 2014, it was not until those systems were truly tested during the Ebola outbreak that the international health community collectively assessed the real limits of existing response capabilities, and the implications thereof. WHO declared the Ebola epidemic in Liberia to be over on 9 June 2016, which was followed by 90 days of heightened surveillance for additional cases (World Health Organization, 2016a). The severity of the epidemic compelled numerous multilateral bodies and institutions—including the World Health Assembly, the World Bank and the European Union, among others—to reimagine global health security-strengthening efforts in a post-Ebola context, create new policies and programmes to help counter the perennial threats posed by epidemic disease, and develop new strategies for accelerating country progress towards the health security benchmarks articulated in the International Health Regulations (IHR). However, given growing membership in the Global Health Security Agenda (GHSA) and GHSA’s emphasis on monitoring and evaluating progress across its 11 focus areas (i.e. action packages), it remains imperative that these still-nascent, post-Ebola health security initiatives align with and supplement existing efforts to strengthen global health security. Ultimately, monitoring the emergence and progress of these new initiatives will help ensure that vulnerable countries can access resources to develop and sustain the capacities required to counter catastrophic epidemics and achieve positive health outcomes during health crises. Materials and methods We undertook a non-systematic, targeted review of major programmes, institutions and initiatives that were launched after the 2014–16 Ebola outbreak ended, and assessed their progress to date to illustrate the current state of the world’s international health security architecture. A comprehensive, systematic review of the scholarly literature was not feasible, given that many of these post-Ebola efforts are relatively new, and that details of their strategic aims, capacities, funding levels and operations are published predominantly in the grey literature and news media. We began our review by examining strategic policy documents for initiatives identified from a priori knowledge and previously conducted global health security research. Next, we used forward- and backward-snowballing methods (i.e. electronic citation tracking and parsing the references of initially identified sources, respectively) to identify additional programmes, initiatives and organizations established in the wake of the Ebola epidemic. We limited our search to English-language documents describing multilateral initiatives that are international in scope, were established directly in response to the 2014–16 Ebola epidemic, and focused on improving infectious disease outbreak prevention, detection and response capacities and capabilities. National after-action reports, reports describing country-level post-Ebola initiatives, and documents describing sub-national Ebola response and recovery efforts were excluded. We also excluded efforts spearheaded by philanthropic foundations and non-governmental organizations if they did not involve multilateral partnerships between countries. We deemed that the review reached saturation once our snowballing strategy failed to uncover any new initiatives. Upon finalizing a list of initiatives to include in our study (see Table 1), we selectively parsed the grey literature, the scholarly literature and news media articles to collect background information on each initiative, identify their strategic aims and participating stakeholders, determine current levels of funding, and summarize their impacts on global public health to date. In cases where information was incomplete or publicly unavailable, we attempted to contact a member of the initiative in question via email. Next, with the aim of identifying remaining gaps in global preparedness and response capacities, we also considered the stated goals of these efforts in the context of global health security capacities documented in the scholarly literature during and after the Ebola outbreak (see Table 2), as well as in the context of each component of the Prevent-Detect-Respond paradigm articulated in the GHSA (see Figure 1). Figure 1. Open in new tabDownload slide The role of high-level initiatives in enhancing the three main phases of global outbreak preparedness. Table 1. High-level initiatives established since the Ebola crisis (in alphabetical order) Name of initiative . Sponsoring entities . Launch date . Funding goal (US$) . % Funded . Other progress . Intervening phase . Africa Centres for Disease Control and Prevention (Africa CDC) African Union January 2017 $35 million (for 2017 and 2018) Unknown 3 of 5 Regional Collaborating Centres opened in central, southern, and western Africa Prevention, Detection, Response Coalition for Epidemic Preparedness Innovations (CEPI) Norway, Germany, Japan, India, Belgium, Canada, Australia, European Commission, World Economic Forum, Gates Foundation, Wellcome Trust January 2017 $1 billion 54% 2 calls for proposals to develop new vaccine platform technologies Prevention, Response European Medical Corps (EMC) European Union February 2016 Unknown Unknown 11 of 28 EU countries have contributed experts or equipment, with 1 deployment Response REDISSE (The World Bank) The World Bank, Gates Foundation, Foundation Merieux June 2016 $110 million Unknown Detection The WHO Global Health Emergency Workforce WHO member states May 2016 Unknown Unknown 64 EMTs from 25 countries and NGOs have registered Response The WHO Health Emergencies Programme (WHE) WHO member states January 2016 $500 million for 2017 73% Prevention, Detection, Response The WHO Contingency Fund for Emergencies (CFE) Germany, Japan, UK May 2015 $100 million 67% 21 health emergencies have received CFE funding Response The WHO R&D Blueprint Core funding for the Blueprint is provided by the UK Department for International Development; the full list of partners involved in developing the Blueprint may be found below this table. May 2015 Unknown Unknown 9 priority diseases identified for R&D Prevention The World Bank Pandemic Emergency Financing Facility (PEF) The World Bank, Germany, Japan May 2017 $500 million insurance market ∼100% backed by bonds and credit 6 priority pathogens covered under insurance window Response Name of initiative . Sponsoring entities . Launch date . Funding goal (US$) . % Funded . Other progress . Intervening phase . Africa Centres for Disease Control and Prevention (Africa CDC) African Union January 2017 $35 million (for 2017 and 2018) Unknown 3 of 5 Regional Collaborating Centres opened in central, southern, and western Africa Prevention, Detection, Response Coalition for Epidemic Preparedness Innovations (CEPI) Norway, Germany, Japan, India, Belgium, Canada, Australia, European Commission, World Economic Forum, Gates Foundation, Wellcome Trust January 2017 $1 billion 54% 2 calls for proposals to develop new vaccine platform technologies Prevention, Response European Medical Corps (EMC) European Union February 2016 Unknown Unknown 11 of 28 EU countries have contributed experts or equipment, with 1 deployment Response REDISSE (The World Bank) The World Bank, Gates Foundation, Foundation Merieux June 2016 $110 million Unknown Detection The WHO Global Health Emergency Workforce WHO member states May 2016 Unknown Unknown 64 EMTs from 25 countries and NGOs have registered Response The WHO Health Emergencies Programme (WHE) WHO member states January 2016 $500 million for 2017 73% Prevention, Detection, Response The WHO Contingency Fund for Emergencies (CFE) Germany, Japan, UK May 2015 $100 million 67% 21 health emergencies have received CFE funding Response The WHO R&D Blueprint Core funding for the Blueprint is provided by the UK Department for International Development; the full list of partners involved in developing the Blueprint may be found below this table. May 2015 Unknown Unknown 9 priority diseases identified for R&D Prevention The World Bank Pandemic Emergency Financing Facility (PEF) The World Bank, Germany, Japan May 2017 $500 million insurance market ∼100% backed by bonds and credit 6 priority pathogens covered under insurance window Response The full list of partners involved in developing the R&D Blueprint include: the US Biomedical Advanced Research & Development Authority, CEPI, the Gates Foundation, Bernhard Nocht Institute for Tropical Medicine, CDC, Chatham House, European Medicines Agency, EpiCentre, US Food & Drug Administration, Health Canada, Harvard School of Public Health, Imperial College London, Institut Pasteur, Johns Hopkins School of Medicine, Kenya Medical Research Institute, London School of Hygiene & Tropical Medicine, Medical Research Council Unit The Gambia, MSF, UK National Institute for Biological Standards and Control, National Institutes of Health, National Institute of Public Health, Northeastern University, World Organisation for Animal Health, Paul-Ehrlich-Institut, Public Health England, University of California Los Angeles, University of Florida, University of Georgia, Université Laval, University of Minnesota, University of Oxford, University of Pennsylvania, University of Texas Medical Branch, University of Washington, UNICEF, Wellcome Trust. Open in new tab Table 1. High-level initiatives established since the Ebola crisis (in alphabetical order) Name of initiative . Sponsoring entities . Launch date . Funding goal (US$) . % Funded . Other progress . Intervening phase . Africa Centres for Disease Control and Prevention (Africa CDC) African Union January 2017 $35 million (for 2017 and 2018) Unknown 3 of 5 Regional Collaborating Centres opened in central, southern, and western Africa Prevention, Detection, Response Coalition for Epidemic Preparedness Innovations (CEPI) Norway, Germany, Japan, India, Belgium, Canada, Australia, European Commission, World Economic Forum, Gates Foundation, Wellcome Trust January 2017 $1 billion 54% 2 calls for proposals to develop new vaccine platform technologies Prevention, Response European Medical Corps (EMC) European Union February 2016 Unknown Unknown 11 of 28 EU countries have contributed experts or equipment, with 1 deployment Response REDISSE (The World Bank) The World Bank, Gates Foundation, Foundation Merieux June 2016 $110 million Unknown Detection The WHO Global Health Emergency Workforce WHO member states May 2016 Unknown Unknown 64 EMTs from 25 countries and NGOs have registered Response The WHO Health Emergencies Programme (WHE) WHO member states January 2016 $500 million for 2017 73% Prevention, Detection, Response The WHO Contingency Fund for Emergencies (CFE) Germany, Japan, UK May 2015 $100 million 67% 21 health emergencies have received CFE funding Response The WHO R&D Blueprint Core funding for the Blueprint is provided by the UK Department for International Development; the full list of partners involved in developing the Blueprint may be found below this table. May 2015 Unknown Unknown 9 priority diseases identified for R&D Prevention The World Bank Pandemic Emergency Financing Facility (PEF) The World Bank, Germany, Japan May 2017 $500 million insurance market ∼100% backed by bonds and credit 6 priority pathogens covered under insurance window Response Name of initiative . Sponsoring entities . Launch date . Funding goal (US$) . % Funded . Other progress . Intervening phase . Africa Centres for Disease Control and Prevention (Africa CDC) African Union January 2017 $35 million (for 2017 and 2018) Unknown 3 of 5 Regional Collaborating Centres opened in central, southern, and western Africa Prevention, Detection, Response Coalition for Epidemic Preparedness Innovations (CEPI) Norway, Germany, Japan, India, Belgium, Canada, Australia, European Commission, World Economic Forum, Gates Foundation, Wellcome Trust January 2017 $1 billion 54% 2 calls for proposals to develop new vaccine platform technologies Prevention, Response European Medical Corps (EMC) European Union February 2016 Unknown Unknown 11 of 28 EU countries have contributed experts or equipment, with 1 deployment Response REDISSE (The World Bank) The World Bank, Gates Foundation, Foundation Merieux June 2016 $110 million Unknown Detection The WHO Global Health Emergency Workforce WHO member states May 2016 Unknown Unknown 64 EMTs from 25 countries and NGOs have registered Response The WHO Health Emergencies Programme (WHE) WHO member states January 2016 $500 million for 2017 73% Prevention, Detection, Response The WHO Contingency Fund for Emergencies (CFE) Germany, Japan, UK May 2015 $100 million 67% 21 health emergencies have received CFE funding Response The WHO R&D Blueprint Core funding for the Blueprint is provided by the UK Department for International Development; the full list of partners involved in developing the Blueprint may be found below this table. May 2015 Unknown Unknown 9 priority diseases identified for R&D Prevention The World Bank Pandemic Emergency Financing Facility (PEF) The World Bank, Germany, Japan May 2017 $500 million insurance market ∼100% backed by bonds and credit 6 priority pathogens covered under insurance window Response The full list of partners involved in developing the R&D Blueprint include: the US Biomedical Advanced Research & Development Authority, CEPI, the Gates Foundation, Bernhard Nocht Institute for Tropical Medicine, CDC, Chatham House, European Medicines Agency, EpiCentre, US Food & Drug Administration, Health Canada, Harvard School of Public Health, Imperial College London, Institut Pasteur, Johns Hopkins School of Medicine, Kenya Medical Research Institute, London School of Hygiene & Tropical Medicine, Medical Research Council Unit The Gambia, MSF, UK National Institute for Biological Standards and Control, National Institutes of Health, National Institute of Public Health, Northeastern University, World Organisation for Animal Health, Paul-Ehrlich-Institut, Public Health England, University of California Los Angeles, University of Florida, University of Georgia, Université Laval, University of Minnesota, University of Oxford, University of Pennsylvania, University of Texas Medical Branch, University of Washington, UNICEF, Wellcome Trust. Open in new tab Table 2. Coverage of high-level initiatives in addressing key gaps in global outbreak preparedness identified during the Ebola outbreak Gaps Clinical and public health workforce surge capacity Formal mechanisms for crisis funding Pipelines for the development of medical countermeasures Greater community engagement and support Clear and empowered leadership Emphasis on early containment of zoonotic threats Initiatives Africa CDC WHO Contingency Fund for Emergencies Coalition for Epidemic Preparedness Innovations Africa CDC WHO Health Emergencies Programme Africa CDC WHO Health Emergencies Programme Pandemic Emergency Financing Facility WHO R&D Blueprint WHO Health Emergencies Programme WHO Global Health Emergency Workforce REDISSE European Medical Corps Gaps Clinical and public health workforce surge capacity Formal mechanisms for crisis funding Pipelines for the development of medical countermeasures Greater community engagement and support Clear and empowered leadership Emphasis on early containment of zoonotic threats Initiatives Africa CDC WHO Contingency Fund for Emergencies Coalition for Epidemic Preparedness Innovations Africa CDC WHO Health Emergencies Programme Africa CDC WHO Health Emergencies Programme Pandemic Emergency Financing Facility WHO R&D Blueprint WHO Health Emergencies Programme WHO Global Health Emergency Workforce REDISSE European Medical Corps Open in new tab Table 2. Coverage of high-level initiatives in addressing key gaps in global outbreak preparedness identified during the Ebola outbreak Gaps Clinical and public health workforce surge capacity Formal mechanisms for crisis funding Pipelines for the development of medical countermeasures Greater community engagement and support Clear and empowered leadership Emphasis on early containment of zoonotic threats Initiatives Africa CDC WHO Contingency Fund for Emergencies Coalition for Epidemic Preparedness Innovations Africa CDC WHO Health Emergencies Programme Africa CDC WHO Health Emergencies Programme Pandemic Emergency Financing Facility WHO R&D Blueprint WHO Health Emergencies Programme WHO Global Health Emergency Workforce REDISSE European Medical Corps Gaps Clinical and public health workforce surge capacity Formal mechanisms for crisis funding Pipelines for the development of medical countermeasures Greater community engagement and support Clear and empowered leadership Emphasis on early containment of zoonotic threats Initiatives Africa CDC WHO Contingency Fund for Emergencies Coalition for Epidemic Preparedness Innovations Africa CDC WHO Health Emergencies Programme Africa CDC WHO Health Emergencies Programme Pandemic Emergency Financing Facility WHO R&D Blueprint WHO Health Emergencies Programme WHO Global Health Emergency Workforce REDISSE European Medical Corps Open in new tab Results Based on the aforementioned criteria, we identified eight major, internationally focused initiatives aiming to strengthen global health security following the 2014–16 Ebola epidemic. Africa centres for disease control and prevention The Africa Centres for Disease Control and Prevention (Africa CDC) was first considered in July 2013 at the African Union (AU) Special Summit on HIV and AIDS. On 31 January 2017, the Africa CDC was formally established. The objectives of the organization include (1) establishing surveillance systems; (2) engaging in preparedness and response activities; (3) bringing member states up to compliance with the IHR; (4) conducting risk assessments and (5) establishing laboratory networks. The Africa CDC is part of a three-tier system. The first tier is the Africa CDC operating at the continental level. The second tier includes five Collaborating Regional Centres based on Egypt, Nigeria, Gabon, Zambia and Kenya (Africa CDC, 2017a,b; African Union 2017). Finally, there are plans for a third tier of National Public Health Institutes that will be established or strengthened in each country. The future of the Africa CDC will depend in large part on the state of its funding in the coming years. The budget for its first 18 months was set at USD$5.9M, which was to be raised from AU member states. The AU Commission has allocated 0.5% of its budget, or about $1.5 M, towards start-up costs (Peyton, 2017). Information on receipt of those funds could not be found. A press release announcing the development of a 5-year strategic plan announced a requirement of $34.4 M for 2017 and 2018, although this document is currently not publicly available (Africa CDC, 2017c). We attempted to contact the organization to obtain up-to-date financial information, but our request was not fulfilled. If sustainable funding—backed by a clearer picture of its planning and fundraising strategy and expected timeline for establishing its subsidiaries—does not materialize, then the vision of an African Public Health Network will be difficult to realize. Coalition for epidemic preparedness innovations The coalition for epidemic preparedness innovations (CEPI) was formally launched at the World Economic Forum in January 2017 as a public–private–philanthropic partnership to accelerate the development of vaccines for diseases of public health relevance (Coalition for Epidemic Preparedness Innovations, 2017b). CEPI describes itself as an end-to-end player in the vaccine development cycle (Coalition for Epidemic Preparedness Innovations, 2016). It plans to fund the development and licensure of vaccines specifically (the stage ranging from late preclinical studies to safety and proof of concept), but aims to also facilitate work from discovery to research, manufacturing and stockpiling. CEPI has the backing of several industry representatives, such as Merck and GlaxoSmithKline, which both hold seats on CEPI’s board. The recommended initial diseases of focus are MERS, Lassa and Nipah, each of which is designated as WHO priority pathogens and has a vaccine candidate in development (Coalition for Epidemic Preparedness Innovations, 2016). Although CEPI has accepted over $540 M in funding (as of July 2017) from the governments of Norway, Germany, Japan, the Bill & Melinda Gates Foundation and the Wellcome Trust, the organization is short of its $1B funding goal (Coalition for Epidemic Preparedness Innovations, 2017c). CEPI has also faced a long runway from when the idea for international collaboration around adaptive clinical trial design, common protocols for randomized clinical trials, and product development and advanced manufacturing was first raised during the Ebola epidemic in September 2015 (Borio et al., 2015). As of this writing, no awards have been publicly announced, but it has launched two calls for proposals for new vaccine development (Coalition for Epidemic Preparedness Innovations, 2017a). A detailed business plan that outlines clear objectives for the next 5 years is available online, and the organization has articulated a desired end state and a clear path forward. European medical corps In February 2016, the European Union (EU) launched the European Medical Corps (EMC) to rapidly deploy human and technological resources for disaster preparedness, response and recovery. As the culmination of the ‘White Helmets’ initiative proposed by France and Germany in 2014, the EMC represents the first major attempt by a regional organization to build a reserve medical corps of international emergency responders (Haussig et al., 2017). As of December 2016, the EMC comprised eight medical teams, two mobile biosafety laboratories, three medical evacuation teams and five logistics/coordination experts (European Commission, 2018). In May 2016, the EMC deployed for the first time to Angola to advise the government on containment strategies for an outbreak of yellow fever (European Commission, 2018). The mission deployed within 4 months of the declaration of the outbreak and comprised eight officials, including two epidemiologists, an infectious disease specialist and a public health expert (European Centre for Disease Prevention and Control and European Union Humanitarian Aid and Civil Protection, 2016). While the Angola mission demonstrated the EU’s ability to rapidly deploy trained experts and equipment in an emergency, no other missions have been reported to date. The voluntary nature of the EMC—in which EU members may elect to opt out of missions on a case-by-case basis—could potentially lead to critical insufficiencies during a larger pandemic. Only 11 of the 28 EU countries have contributed experts as of December 2016, according to the latest information available at the time of this writing. A review of European Commission and European Civil Protections and Humanitarian Aid Operations strategic documents covering 2016 and 2017 provides little mention of the EMC, and does not outline strategic objectives, funding, or plans for growth (European Commission, 2016). In the absence of greater financial and human commitment from EU members, the sustainability of the ‘White Helmets’ experiment appears uncertain. REDISSE In June 2016, the World Bank announced a new initiative designed to strengthen disease surveillance systems in West Africa, known as the Regional Disease Surveillance Systems Enhancement Program (REDISSE). REDISSE represents the first significant source of funding to develop disease surveillance capacity in West Africa, including boosting laboratory capacity and epidemiological surveillance (The World Bank, 2016a). It will provide US$110 million in financing from the International Development Association (IDA) ‘to address systemic weaknesses within the human and animal health sectors that hinder effective disease surveillance and response’ (The World Bank, 2016b). West Africa was selected for this initiative due to its increased susceptibility to infectious disease outbreaks as demonstrated by the Ebola outbreak. Guinea, Sierra Leone and Senegal are expected to receive the first multi-million instalments; however, there are plans to eventually expand to all 15 countries in the Economic Community of West African States (ECOWAS) with an eye towards developing an interconnected, regional surveillance network. In addition to strengthening capacity for disease surveillance, REDISSE also includes a response component to improve national response capabilities in the event of an emergency. It remains to be seen whether the REDISSE model will be effective and scalable to other regional or sub-regional arrangements beyond ECOWAS countries. The WHO global health emergency workforce During the 69th World Health Assembly in May 2016, the WHO inaugurated its new Global Health Emergency Workforce to provide rapid surge capacity during a crisis. The initiative responds to several post-Ebola recommendations calling on the WHO to ‘establish significant operational capabilities [including] rapidly deployable human resource assets’ to respond to health crises (United Nations, 2016). The Workforce is a global registry of emergency medical teams (EMTs) from national, regional and global networks, which join following a quality assurance and verification process (World Health Organization, 2017b). As of July 2016, there were approximately 64 EMTs from 25 countries and international NGOs undergoing or having completed registration in the Global Health Emergency Workforce. Australia, China, Costa Rica, Ecuador, Germany, Israel, Japan, New Zealand, Russia and the UK are among the countries that have registered EMTs (Pan American Health Organization, 2017). Over 200 teams are expected to join, representing a global workforce of an estimated 100 000 trained experts across a variety of health fields (Burkle, 2016). These include clinicians, public health experts, laboratory specialists, epidemiologists, operations coordinators and incident managers. Peer-to-peer training and mentorship for EMTs is provided, with an emphasis on building domestic EMTs and national response capacity. The key challenge will be to effectively operationalize this new workforce model. The WHO has expressed concern that unverified teams will bypass the Workforce by showing up unannounced on a country’s doorstep without adequate training or specialized skillsets, as occurred during the response to the Haiti earthquake in 2010–11 (World Health Organization, 2016b). This could lead to well-intentioned but unhelpful or duplicative efforts on the ground. The WHO health emergencies programme and contingency fund In January 2016, WHO's Global Policy Group announced new reforms in support of WHO's commitment to enhancing its emergency response capacities (World Health Organization, 2016c). The WHO Health Emergencies (WHE) Programme is holistic in its approach, designed to address the full range of preparedness, response and recovery considerations associated with all hazards, from traditional outbreaks as well as natural disasters and humanitarian crises. With the support of a dedicated workforce and budget, the WHE focuses on six major areas of work: infectious hazard management, WHO Member State preparedness, risk assessment and health emergency information management, emergency operations, management and administration, and external relations. Efforts by the WHE since the Ebola crisis are illustrative of its potential to become an important component of the international community's emergency response architecture. In 2016, for example, the WHE activated its Incident Management Systems to coordinate WHO's response to outbreaks in Angola, Democratic Republic of the Congo and Uganda, as well as to the Zika virus epidemic, which was declared a PHEIC (World Health Organization, 2016e). Internal assessments of reforms to the WHE thus far underscore its success in incident management and note improvements in responses to complex health crises (World Health Organization, 2018b). However, critical gaps in funding, workforce management, policy implementation, and monitoring and accountability remain. The Independent Oversight and Advisory Committee for the WHE reports, for example, that as of August 2017, only 73% of its core 2017 appeals budget of approximately $500 million has been funded (World Health Organization, 2017c). A September 2017 report of the Independent Oversight and Advisory Committee evaluating WHE’s progress in Pakistan found that: The WHE Programme [in Pakistan] is yet to be fully developed in terms of staff recruitment and skills, adoption of the revised standard operating procedures (SOPs) on delegation of authority for decision making, and improved communication between the three levels of WHO. This is limiting the capacity of the WHO country office to meet the country’s needs (World Health Organization, 2017d). Preceding the establishment of the WHE was the creation of the WHO Contingency Fund for Emergencies (CFE) in May 2015. The CFE is designed to support the WHE response component; it provides rapid funding during the critical window after a crisis is identified, but prior to the disbursement of traditional funding sources (World Health Organization, 2015c). The CFE has allocated a total of $19.1 million directed towards 21 health emergencies since its first disbursement in November 2015. Despite its wide utility, the CFE reports a 67% funding gap of its target capitalization of $100 million, as of May 2017. Reimbursements have not kept pace with funding demands. While the voluntary nature of the fund has contributed to its flexibility, this has also led to cycles of fundraising and donor fatigue (World Health Organization, 2015c). The WHO R&D blueprint In May 2015, the 68th World Health Assembly released a resolution in favour of ‘accelerating research and development in epidemics or health emergency situations where there are no, or insufficient, preventive and curative solutions’ (World Health Organization, 2015b). In support of this resolution, WHO convened a coalition of Member State representatives and international stakeholders to prepare a blueprint explicating a novel R&D model for emerging pathogens with the potential to cause devastating outbreaks, and for which few or no medical countermeasures exist. The primary aims of the Blueprint are to assist stakeholders in identifying pathogens of international concern, facilitate alignment of research agendas to tackle priority threats, and incentivize greater investment in research and development among the public, private and philanthropic sectors (World Health Organization, 2018a). We were unable to obtain funding estimates for R&D Blueprint activities. However, we found that the R&D Blueprint currently consists of five work streams: prioritization of pathogens and operational plans; identification of research priorities; coordination of stakeholders and expansion of capacity; assessment of preparedness and impact of intervention; and exploration of funding models for R&D preparedness and response. During the most recent review of the Blueprint in January 2017, stakeholders identified nine priority diseases using a Delphi process: several haemorrhagic fevers (e.g. Ebola/Marburg, Crimean Congo, Lassa); coronavirus infections (MERS, SARS); and other viral infections, including Zika virus disease, Nipah virus disease, Severe Fever with Thrombocytopenia Syndrome and Rift Valley fever (World Health Organization, 2017a). World Bank pandemic emergency financing facility In May of 2017, the World Bank announced a new US$500 million initiative to combat pandemics by shortening the time between when an outbreak is recognized and when response funding is mobilized (The World Bank, 2017a). In collaboration with WHO, the World Bank has instituted a two-prong Pandemic Emergency Financing Facility (PEF). The primary mechanism is pandemic insurance, which has been operational as of July 2017 and which covers low income client countries (i.e. those eligible to borrow from the IDA). The insurance can be paid out for outbreaks of six viruses evaluated by the World Bank as most likely to cause a pandemic: orthomyxovirus (e.g. H1N1 and other influenza virus A subtypes), coronavirus (e.g. SARS, MERS), filovirus (e.g. Ebola, Marburg), Crimean Congo haemorrhagic fever, Rift Valley fever or Lassa fever. Disbursement of funds occurs after the outbreak reaches a certain severity threshold in an eligible country. To date, these activation thresholds have not been publicly detailed. The PEF is funded through a combination of bonds, credit derivatives and cash contributions totalling $425 million in risk. For outbreak emergencies that do not qualify for coverage, a discretionary cash-based fund can be mobilized on a case-by-case basis. Germany has contributed 50 million euros to this ‘cash window’, which is expected to open in 2018 (The World Bank, 2017b). The World Bank hopes that the PEF will create a new market for pandemic insurance, in which preparedness is incentivized and risks are mitigated. Early indicators suggest that the PEF may indeed spark such a market—the initial bond offerings were oversubscribed by 200% (The World Bank, 2017b). However, time will tell if the market is capable of long-term stability, especially if only six viruses are covered under the insurance window. Discussion Though these programmes bear considerable potential to bolster international outbreak preparedness and response capabilities, several significant gaps remain. In response to the unprecedented 2014–16 Ebola outbreak in West Africa, many institutions and thinkers in global public health published important reflections on the lessons that were learned from the catastrophe. We reviewed several such peer-reviewed publications and reports and found a number of themes that were dominant in those reflections, which collectively form an important roadmap to strengthening outbreak preparedness and response (Médecins Sans Frontières, 2015a). Chief among those themes were the following gaps: (1) clinical and public health workforce surge capacity; (2) formal mechanisms for crisis funding; (3) pipelines for the development of medical countermeasures; (4) greater community engagement and support; (5) clear and empowered leadership and (6) emphasis on early containment of zoonotic threats. To assess progress made towards improving global outbreak preparedness and response, we illustrate how the new initiatives announced since the Ebola outbreak addresses these gaps in Table 2. In Figure 1, we map the aims of these programmes against the ‘Prevent-Detect-Respond’ paradigm articulated in the GHSA, which has become an important guiding principle for conceptualizing global efforts to mitigate infectious disease threats (Centers for Disease Control & Prevention, 2016). Although these initiatives are necessarily still in their early days, they face significant challenges. Although it was difficult to obtain data on financial status, we note that most of the initiatives we reviewed have not yet met their funding targets. These funding challenges are representative of persistent underfunding in public health that complicate effective response operations. Furthermore, the difficulty in obtaining accurate and up-to-date financial and operational information is emblematic of a lack of transparency that characterize many actors in global health governance. Significantly, we also find several thematic areas where the literature indicates a need, yet few new programmes have been announced. For example, we note that while many programmes are focused on the prevention and response phases of the outbreak, far fewer are working at the detection phase—the chief exception being the World Bank’s REDISSE initiative. Considering the difficulty of detecting and diagnosing the Ebola virus in the initial months of the outbreak, this reveals a need for programmatic innovation to improve global surveillance and detection capabilities. Notably, though all the initiatives discussed here involve participation from various national governments, their foci remain regional and global in scope. As such, the sponsors of these initiatives should consider the trade-offs associated with investing in national and local health service delivery systems that operate sustainably without the assistance of supranational health bodies (Mackey, 2016). In this vein, efforts to strengthen in-country capacities for engaging more effectively with communities affected by emerging crises would complement ongoing efforts to enhance global and regional health security capabilities. We also observed significant disparities in the level of ownership granted to high-income nations vs low- and middle-income countries (LMIC) among the initiatives we identified. With the exception of Africa CDC, which is financed and led predominantly by member states of the AU; WHO’s Global Emergency Workforce, which includes a handful of emergency medical teams representing middle-income countries (Costa Rica, Ecuador, Russia and China); and the WHO R&D Blueprint, which includes Kenya Medical Research Institute and Medical Research Council Unit The Gambia as partners, the majority of the initiatives identified in our review are sponsored largely by high-income and Western nations, philanthropic groups and universities (see Table 1). Though we did not consider the GHSA in our analysis, we note that its model of vesting LMICs with leadership roles across several action packages may provide a useful blueprint for future global health security-strengthening efforts. These initiatives are supplemented by numerous national and subnational programmes which have not been reviewed here. We were not able to obtain up to date or complete information for all the programmes we reviewed, so it is possible that progress has made beyond what we have reported. We also note that there are several programmes that were established prior to the Ebola outbreak but have since expanded, including the GHSA and the Global Outbreak Alert and Response Network (2000). These programmes are critical to strengthening emergency response and public health preparedness, and together create a more resilient outbreak management system. Our analysis had several limitations, some of which are linked to our literature search strategy. Because our search was largely restricted to English-language grey literature, we may have unintentionally omitted relevant documents describing health security-strengthening efforts helmed by non-Western countries, or analyses of post-Ebola reform efforts authored by researchers from countries directly affected by the epidemic. However, this limitation also reflects a real-world bias, as most of the major post-Ebola health security efforts identified in this study are being led and financed predominantly by Western and high-income countries. Additionally, our review was also restricted to documents made available in the public domain. As a result, we did not have access to financial records, internal memoranda and reports, or other non-public materials that might account for some of the publicly documented funding, workforce, technical and administrative gaps identified in this review Our ultimate goal is to determine whether the measures enacted by the global community in the wake of the Ebola epidemic result in sustainable advances in public health preparedness and response with respect to catastrophic infectious disease events. The findings from this investigation comprise an initial step in monitoring and evaluating progress towards this goal. As the initiatives identified in our investigation mature, a future analysis of their impacts on global health security might benefit from additional, complementary modes of data collection, such as interviews with public health leaders in LMIC and systematic reviews of scholarly literature covering these initiatives. As additional countries undergo the Joint External Evaluation process, the results of their assessments might also provide valuable insights into the effectiveness of global health security-strengthening initiatives. Immediate next steps in conducting such an analysis might include identifying relevant points of contact in each of the organizations and initiatives that emerged from this review, as well as examining impact and outcome evaluations of each effort. Conclusion This investigation represents an initial step in monitoring and evaluating the landscape of internationally-focused, multilateral health security-strengthening efforts launched in response to the 2014–16 Ebola epidemic. Our review of these efforts revealed critical gaps in global detection capabilities and programmatic funding. Additionally, we found that LMICs at risk of experiencing catastrophic epidemics have largely not been granted ownership of global health security-strengthening efforts. Monitoring the progress of these initiatives could help ensure that prevention, detection and response efforts facilitate post-Ebola recovery and preparedness for future epidemics in the most sustainable and equitable manner possible. Conflict of interest statement. None declared. Acknowledgement The authors did not receive funding to conduct the research described in this manuscript. References Africa CDC. 2017a . Africa Centres for Disease Control and Prevention’s Regional Collaborating Centres in Africa agree on a strategic plan and roadmap for disease prevention and response in Africa. https://au.int/en/pressreleases/20170320/africa-centres-disease-control-and-prevention’s-regional-collaborating. Africa CDC. 2017b . Central Africa establishes the Africa Centres for Disease Control and Prevention Regional Collaborating Centre to Improve Surveillance, Preparedness and Response to Infectious and Non-communicable Diseases. https://au.int/en/pressreleases/20170803/central-africa-establishes-africa-centres-disease-control-and-prevention. Africa CDC. 2017c . Africa Centres for Disease Control and Prevention Launches Five-Year Strategic Plan to Improve Surveillance, Emergency Response and Prevention of Infectious Diseases. https://www.au.int/web/en/pressreleases/20170324/africa-centres-disease-control-and-prevention-launches-five-year-strategic. African Union. 2017 . Southern Africa Establishes the Africa Centres for Disease Control and Prevention Regional Collaborating Centre to Improve Surveillance, Emergency Response and Prevention of Infectious and Non-communicable Diseases. https://au.int/en/pressreleases/20170721/southern-africa-establishes-africa-centres-disease-control-and-prevention. Borio L , Cox E, Lurie N. 2015 . Combating emerging threats – accelerating the availability of medical therapies . New England Journal of Medicine 11 : 993 – 5 . Google Scholar Crossref Search ADS WorldCat Burkle FM. 2016 . The World Health Organization global health emergency workforce: what role will the United States Play? Disaster Medicine and Public Health Preparedness 4 : 531 – 5 . Google Scholar Crossref Search ADS WorldCat Centers for Disease Control & Prevention. 2016 . The Global Health Security Agenda. https://www.cdc.gov/globalhealth/security/ghsagenda.htm. Coalition for Epidemic Preparedness Innovations. 2016 . Preliminary Business Plan 2017–2021 . Oslo, Norway : Coalition for Epidemic Preparedness Innovations . http://cepi.net/sites/default/files/CEPI%20Preliminary%20Business%20Plan%20011116.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Coalition for Epidemic Preparedness Innovations. 2017a . Coalition for Epidemic Preparedness Innovations Newsletter—30 September 2017. Coalition for Epidemic Preparedness Innovations. 2017b . Global Partnership Launched to Prevent Epidemics with New Vaccines. http://cepi.net/cepi-officially-launched. Coalition for Epidemic Preparedness Innovations. 2017c . Norway Strengthens Its Commitment to CEPI. http://cepi.net/norway-strengthens-its-commitment-cepi. European Centre for Disease Prevention and Control and European Union Humanitarian Aid and Civil Protection. 2016 . Mission Report: Assessing the Yellow Fever Outbreak in Angola . Solna : European Centre for Disease Prevention and Control . https://ecdc.europa.eu/sites/portal/files/media/en/publications/Publications/yellow-fever-angola-joint-ecdc-mission-report-2016.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC European Commission. 2016 . General Guidelines on Operational Priorities for Humanitarian Aid in 2017 . Brussels : European Commission . http://ec.europa.eu/echo/sites/echo-site/files/ggopha_2017.pdf#view=fit. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC European Commission. 2018 . European Medical Corps. http://ec.europa.eu/echo/files/aid/countries/factsheets/thematic/European_Medical_Corps_en.pdf. Haussig JM , Severi E, Baum JHJ et al. 2017 . The European Medical Corps: first public health team mission and future perspectives . Eurosurveillance 37 . Google Scholar OpenURL Placeholder Text WorldCat Lupel A , Snyder M. 2017 . The Mission to Stop Ebola: Lessons for UN Crisis Response. New York : International Peace Institute . https://www.ipinst.org/wp-content/uploads/2017/02/IPI-Rpt-Mission-to-Stop-Ebola.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mackey TK. 2016 . The Ebola outbreak: catalyzing a “shift” in global health governance? BMC Infectious Diseases 16 : 699. Google Scholar Crossref Search ADS PubMed WorldCat Médecins Sans Frontières. 2015a . An Unprecedented Year: Médecins Sans Frontières’ Response to the Largest Ever Ebola Outbreak . Geneva : Médecins Sans Frontières . http://www.msf.org/sites/msf.org/files/ebola_accountability_report_final_july_low_res.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Médecins Sans Frontières. 2015b . Pushed to the Limit and beyond: A Year into the Largest Ever Ebola Outbreak . Geneva : Médecins Sans Frontières . https://www.msf.org.uk/sites/uk/files/ebola_-_pushed_to_the_limit_and_beyond.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Moon S , Sridhar D, Pate MA et al. 2015 . Will Ebola change the game? Ten essential reforms before the next pandemic. The report of the Harvard-LSHTM Independent Panel on the Global Response to Ebola . The Lancet 10009 : 2204 – 21 . Google Scholar Crossref Search ADS WorldCat Pan American Health Organization. 2017 . Ecuador’s Emergency Medical Team Classified as a WHO EMT . Washington, DC : Pan American Health Organization . https://www.paho.org/hq/index.php? option=com_content&view=article&id=13735%3Aequipo-medico-de-emergencia-de-ecuador-recibio-clasificacion-como-emt-de-la-oms&catid=9470%3A29psc-events&Itemid=42439&lang=en. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Peyton N. 2017 . New Africa Disease Agency Aims to Prevent Fresh Ebola Crisis. https://www.reuters.com/article/us-africa-health-ebola/new-africa-disease-agency-aims-to-prevent-fresh-ebola-crisis-idUSKBN15U227. The World Bank. 2016a . West Africa Regional Disease Surveillance Systems Enhancement Project (REDISSE) P154807 . Washington, DC : The World Bank . http://pubdocs.worldbank.org/en/769681467208334446/REV5-English-WEST-AFRICA-REGIONAL-DISEASE-SURVEILLANCE-Project-2pager.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC The World Bank. 2016b . World Bank Contributes to Improved Disease Surveillance and Health Systems in West Africa Following Ebola Epidemic. http://www.worldbank.org/en/news/press-release/2016/06/29/world-bank-contributes-to-improved-disease-surveillance-and-health-systems-in-west-africa-following-ebola-epidemic. The World Bank. 2017a . From Panic and Neglect to Investing in Health Security: Financing Pandemic Preparedness at a National Level . Washington, DC : The World Bank . http://documents.worldbank.org/curated/en/979591495652724770/pdf/115271-REVISED-FINAL-IWG-Report-3-5-18.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC The World Bank. 2017b . World Bank Launches First-Ever Pandemic Bonds to Support $500 Million Pandemic Emergency Financing Facility. http://treasury.worldbank.org/cmd/htm/World-Bank-Launches-First-Ever-Pandemic-Bonds-to-Support-500-Million-Pandemic-Emergenc.html. United Nations. 2016 . Protecting Humanity from Future Health Crises: Report of the High-Level Panel on the Global Response to Health Crises . New York : United Nations . http://www.un.org/News/dh/infocus/HLP/2016-02-05_Final_Report_Global_Response_to_Health_Crises.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. 2015a . Decisions and List of Resolutions . Geneva : World Health Organization . http://apps.who.int/gb/ebwha/pdf_files/WHA68/A68_DIV3-en.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. 2015b . Report of the Ebola Interim Assessment Panel . Geneva : World Health Organization . http://www.who.int/csr/resources/publications/ebola/report-by-panel.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. 2015c . WHO Contingency Fund for Emergencies . Geneva : World Health Organization . http://www.who.int/about/finances-accountability/funding/financing-dialogue/Presentation-on-Contigency-Fund-for-Emergencies.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. 2016a . End of the Most Recent Ebola Virus Disease Outbreak in Liberia. http://www.who.int/en/news-room/detail/09-06-2016-end-of-the-most-recent-ebola-virus-disease-outbreak-in-liberia. World Health Organization. 2016b . Building a Global Health Emergency Workforce for a Better Response. http://www.who.int/hac/emergency_workforce_february_2016/en/. World Health Organization. 2016c . Global Policy Group Statement on Reforms of WHO Work in Outbreaks and Emergencies . Geneva : World Health Organization . http://www.who.int/dg/speeches/2016/reform-statement/en/. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. 2016d . Situation Report: Ebola Virus Disease, 10 June 2016 . Geneva : World Health Organization . http://apps.who.int/iris/bitstream/handle/10665/208883/ebolasitrep_10Jun2016_eng.pdf? sequence=1. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. 2016e . WHO’s New Health Emergencies Programme. http://www.who.int/features/qa/health-emergencies-programme/en/. World Health Organization. 2017a . Annual Review of Diseases Prioritized under the Research and Development Blueprint . Geneva : World Health Organization . http://www.who.int/blueprint/what/research-development/2017-Prioritization-Long-Report.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. 2017b . Emergency Medical Teams: World Health Organization EMT Initiative . Geneva : World Health Organization . https://extranet.who.int/emt/sites/default/files/EMT_Updatednews_19.01.2017.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. 2017c . Independent Oversight and Advisory Committee for the WHO Health Emergency Programme – Draft Meeting Report . Geneva : World Health Organization . http://www.who.int/about/who_reform/emergency-capacities/oversight-committee/report-11th-meeting.pdf? ua=1. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. 2017d . Pakistan Mission Report . Geneva : World Health Organization . http://www.who.int/about/who_reform/emergency-capacities/oversight-committee/pakistan-mission-report-ioac-visit-6-8september2017.pdf. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC World Health Organization. 2018a . A Research and Development Blueprint for Action to Prevent Epidemics. http://www.who.int/blueprint/en/. World Health Organization. 2018b . Report of the Independent Oversight and Advisory Committee for the WHO Health Emergencies Programme . Geneva : World Health Organization . http://apps.who.int/gb/ebwha/pdf_files/EB142/B142_8-en.pdf? ua=1. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC © The Author(s) 2019. 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) © The Author(s) 2019. 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]

Mishra, Shiva, Raj;Lygidakis,, Charilaos;Neupane,, Dinesh;Gyawali,, Bishal;Uwizihiwe, Jean, Paul;Virani, Salim, S;Kallestrup,, Per;Miranda, J, Jaime

doi: 10.1093/heapol/czy099pmid: 30668690

Abstract The use of community health workers (CHWs) has been explored as a viable option to provide home health education, counselling and basic health care, notwithstanding their challenges in training and retention. In this manuscript, we review the evidence and discuss how the digitalization affects the CHWs programmes for tackling non-communicable diseases (NCDs) in low- and middle-income countries (LMICs). We conducted a review of literature covering two databases: PubMED and Embase. A total of 97 articles were abstracted for full text review of which 26 are included in the analysis. Existing theories were used to construct a conceptual framework for understanding how digitalization affects the prospects of CHW programmes for NCDs. The results are divided into two themes: (1) the benefits of digitalization and (2) the challenges to the prospects of digitalization. We also conducted supplemental search in non-peer reviewed literature to identify and map the digital platforms currently in use in CHW programmes. We identified three benefits and three challenges of digitalization. Firstly, it will help improve the access and quality of services, notwithstanding its higher establishment and maintenance costs. Secondly, it will add efficiency in training and personnel management. Thirdly, it will leverage the use of data generated across grass-roots platforms to further research and evaluation. The challenges posed are related to funding, health literacy of CHWs and systemic challenges related to motivating CHWs. Several dozens of digital platforms were mapped, including mobile-based networking devices (used for behavioural change communication), Web-applications (used for contact tracking, reminder system, adherence tracing, data collection and decision support), videoconference (used for decision support) and mobile applications (used for reminder system, supervision, patients’ management, hearing screening and tele-consultation). The digitalization efforts of CHW programmes are afflicted by many challenges, yet the rapid technological penetration and acceptability coupled with the gradual fall in costs constitute encouraging signals for the LMICs. Both CHWs interventions and digital technologies are not inexpensive, but they may provide better value for the money when applied at the right place and time. Community health workers, non-communicable diseases, low-income countries, community health programmes Key Messages The technological boom in low- and middle-income countries has opened up plethora of avenues where digital technologies can become part of community health worker (CHW) programmes. There are some benefits, of the digitalization process whereby CHW programme would help expanding access to services and their quality, catalyses training and supervision of CHWs for personnel management, as well as benefiting evaluation and research processes. The challenges related to the prospects of digitalization are related to funding of CHW programmes in general as well as funding for uptake of digitalization process, digital literacy of CHWs and other systematic challenges that are present in the programme even before the digitalization process could began. Background Community Health Workers (CHWs) constitute the backbone of health delivery systems globally (Mishra et al., 2015). Recently, it has been proposed to train an additional million CHWs to speed up the strides towards universal health coverage in rural sub-Saharan Africa (One Million Community Health Workers Campaign). The World Health Organization (WHO) defines CHWs as men and women, young or old, literate or illiterate, who are members of the community, and pertinently trained to deal with the health problems of individuals and the community (Rifkin, 2008). The term CHW is broad with more than 30 different titles named worldwide (Bhattacharyya et al., 2001); CHWs have been classified as health associate professionals (Beard and Redmond, 1979; International Labour Office, 2012). The activities and roles of CHWs are tailored to meet the unique health needs of their communities, and depend on factors such as, whether they work in the health care or social service sectors (Rural Health Information Hub), the type and duration of training and previous education, and the size and geographical spread of the covered population. CHWs have been largely used in programmes that target infectious diseases (e.g. HIV/AIDS and tuberculosis), as well as in those related to maternal and child health (Tsolekile et al., 2014). This focus has been driven by the Millennium Development Goals, resulting in less attention to other conditions such as non-communicable diseases (NCDs) (Lewin et al., 2010; Perry and Zulliger, 2012). According to the 2015 Global Burden of Disease Study, NCDs accounted for more than 70% of global deaths (Reddy, 2016). Responding to this challenge, the United Nations’ Member States included a target in the sustainable development goals for all countries to reduce premature mortality from NCDs by one-third by 2030 (United Nations, 2015). Recent evidence has demonstrated that CHWs may be effective in tackling the burden of cardiovascular diseases (CVDs) and other NCDs in low- and middle-income countries (LMICs) (Khetan et al., 2016), where there is a perennial shortage of health professionals. Digital technologies have changed the way consumers (the patients) and providers (health professionals) interact with each other, especially in LMICs, such as India and China. Internet connectivity currently covers 85% of habitable areas in the world, far greater than the average coverage of electrical grid (Kay et al., 2011; International Telecommunication Union, 2017). A 2009 study found 83% of the 114 investigated countries offered at least 1 mobile Health (mHealth) service, with high-income countries showing higher usage than low-income ones. The increasing use of mobile phones and expansion in internet connectivity are expected to be the driving forces for the electronic Health (eHealth) expansion in LMICs. It is inevitable that the CHW programmes will be affected in this process. It is unclear, however, how technology will benefit the existing CHW interventions for NCDs. Previous reviews discussed the benefits of digitalization with focus on mHealth (Källander et al., 2013), particularly covering regions such as Africa (Opoku et al., 2017). A review by Källandar et al. identified six main areas in which mobile technologies can be used: education and awareness, data access, monitoring and compliance, disease and emergency tracking, health information systems, and diagnosis and consultation (Källander et al., 2013). The aim of this review is to explore the digital opportunities and challenges for CHW programmes in LMICs, particularly focusing on NCDs. Two themes have been used to support this discussion: (1) how digitalization will support CHWs interventions for NCDs and (2) what are the challenges for the CHW programmes with digitalization in the future. Methods A review of literature was conducted in PubMed/Medline using the MeSH terms ‘Telemedicine’ and ‘Community Health Workers’ from 1 January 2000 to 31 December 2016. The MeSH term in PubMed for telemedicine entails the following entry terms: ‘Mobile Health’, ‘Health, Mobile’, ‘mHealth’, ‘Telehealth’ and ‘eHealth’. Articles and abstracts in English were only taken into consideration. A total of 33 articles were identified and reviewed. Similarly, a search for studies was performed in Embase, replacing ‘community health workers in PubMed with ‘health auxiliary’ and using a combination of entry terms in Emtree (equivalent of MeSH in Embase) such as ‘telehealth’, ‘mobile health’, ‘mHealth’ and ‘telemedicine’, as done in PubMed. A total of 70 papers were obtained from 1 January 2000 to 31 December 2016. Further searches were conducted in Google Scholar for grey literature and 69 papers were identified. Individual papers (e.g. reviews, opinion pieces and commentaries) were considered for full text review. A total of 97 papers were accessed for full text (observational studies n = 20, interventional studies n = 27, review articles n = 15, comments/letters/opinion pieces n = 6, conference abstracts n = 29), and of them, 26 papers that laid down uses of at least 1 digital device and/or platform were abstracted for information (see Supplementary data S1) and included in the review. To facilitate the analysis, a framework centred on digitalization was developed based on earlier work (Labrique et al., 2013,; Opoku et al., 2017). There are only limited deliberations of the impact of digitalization on CHW programmes in NCDs context, and it is also relatively unclear whether digitalization is in itself a process or an outcome in this context. Therefore, we presented this paper as a structured narrative review, following some components of the preferred reporting of systematic review and meta-analysis guidelines (Figure 1). Figure 1 View largeDownload slide Flow diagram of the review process Figure 1 View largeDownload slide Flow diagram of the review process An adapted framework for NCDs According to Gartner, digitalization is: ‘the use of digital technologies to change a business model and provide new revenue and value-producing opportunities; simply said it is the process of moving to a digital business’ (Gartner). Building on this definition, our interpretation is of ‘an ongoing process by which information and communicating technologies (ICT) are adopted to improve the value of services’ both at the public and private sector. Contrary to ‘digitization’, ‘digitalization’ goes beyond the mere introduction of ICT, as it builds upon the experience of multiple sectors, and catalyses the convergence of processes to improve their efficiency. The current digital solutions for NCD services are presented in Supplementary Table S1, and cover such areas as: networking devices (used for behavioural change communication), Web-applications (used for contact tracking, reminder systems, adherence tracing, data collection and decision support), videoconferencing (used for decision support) and mobile applications (used as a reminder system and for supervision, patient management, hearing screening and tele-consultation). In an earlier review, Opoku described how patients and providers interact in a complex system involving mHealth, and laid down the factors that influence them (Opoku et al., 2017). Figure 2 of our ‘digital-temple model’ illustrates the impact of digitalization on health systems involving complex interactions among patients, providers and CHWs. This framework may be applicable to other health care areas beyond those related to NCDs. How such interactions take place is discussed in previous works (Agarwal et al., 2016,; Opoku et al., 2017). The outcomes of such interactions would have impact on: (1) expanding access and quality of services; (2) training and supervision of personnel, and (3) research and evaluation. Figure 2 View largeDownload slide ‘Digital-temple model’ for understanding the impact of digitalization and complex interaction among the patients, providers and CHWs in health system. The model shows digitalization as a catalyst in the system. The interactions between patients and providers, as well as CHWs and patients have been shown to be reciprocal, similar to previous studies (Opoku et al., 2017; Agarwal et al., 2016). This model is based on earlier work (Opoku et al., 2017). The 12 core functions presented are adaption of Labrique (2013) work on 12 key functions of mHealth platforms in reproductive, child and maternal health context (Labrique et al., 2013). The model presents three stepping stones as (1) research and evaluation, (2) training, supervision on personnel management and (3) access and quality of services. Figure 2 View largeDownload slide ‘Digital-temple model’ for understanding the impact of digitalization and complex interaction among the patients, providers and CHWs in health system. The model shows digitalization as a catalyst in the system. The interactions between patients and providers, as well as CHWs and patients have been shown to be reciprocal, similar to previous studies (Opoku et al., 2017; Agarwal et al., 2016). This model is based on earlier work (Opoku et al., 2017). The 12 core functions presented are adaption of Labrique (2013) work on 12 key functions of mHealth platforms in reproductive, child and maternal health context (Labrique et al., 2013). The model presents three stepping stones as (1) research and evaluation, (2) training, supervision on personnel management and (3) access and quality of services. Opportunities for CHW programmes in LMICS Increasing service access and quality The CHW programmes can become more efficient, expand their range of services and increase their quality. CHWs use mobile phones, tablets and other digital devices (Källander et al., 2013), which have become increasingly affordable, in a task-specific manner. For example, digital blood pressure monitoring devices, glucometers and spirometers have been adapted to fit the local context increasing service access (Agarwal et al., 2016). Risk assessment cards can be used in addition to glucometers to track and report progress on diabetes and to increase effectiveness in risk prediction (Ranslow et al., 2015). Such devices which readily provide interpretable visual results are highly useful for CHWs with low literacy (Ranslow et al., 2015). For maternal and child health, mobile applications have been in use for a long time (Little et al., 2013). In a study in India, a mobile application for scheduling (electronic registries, prompts and reminders for home visits) is used as a job aid to increase the frequency of CHW’s home visits (Modi et al., 2015). The inclusion of checklists and video clips can help with the behavioural change communication activities of CHWs. Additionally, algorithms recommend relevant examinations for diagnosis and treatment plan customization. Some applications are also linked with the primary care centres, so that the clinicians can supervise CHWs, monitor them for performance-based incentives, follow high-risk patients and their vital statistics, and track items across the supply chain (Modi et al., 2015). Other solutions can also enable central monitoring. For example, in Rajasthan State in India, the CHWs use digital devices to record vaccination data (Mishra et al., 2016), helping to identify children who drop out from the regular vaccination schedule. The data from CHWs are later transferred to a cloud-based dashboard and analysed centrally. Alerts, notifications, reminders, checklists and decision-support tools facilitate compliance to protocols, support evidence-based practice and ultimately, improve the quality of the care provided by the CHWs (Svoronos et al., 2010). For example, DeRenzi et al. (2012) reported improvement in the performance of CHWs by integrating text reminders into the regular practice and aiding supervisors to be aware of overdue visits. Florez-Arango et al. (2011) noticed an increase in protocol compliance and a reduction of errors when conducting an mHealth intervention with CHWs using point-of-care guidelines in an experimental setting. Furthermore, sophisticated platforms can be used for peer-to-peer communication, electronic decision support and imparting positive education and behaviour in the community (Table 1). Additionally, the use of such platforms can be on supply chain management to provide records of their stocks (Labrique et al., 2013; Agarwal et al., 2016). Table 1 provides the summary of various platforms in use today and where they fit in a complex system involving patients, providers and CHWs, and the interaction thereof. Table 1 Functions, characteristics and examples of digital platforms in relation to their interaction with patients, providers and health system Interaction Theme Functions (Labrique et al., 2013) Characteristics Patients CHWs Providers Health system Examples (Funes et al., 2012; Braun et al., 2013; Labrique et al., 2013; Agarwal et al., 2016; Basavarajappa and Chand, 2017) Health system goals Opportunities for NCDs Challenges Increasing service access and quality Patient’s education and communication One-way or two-way communication gateway for BCC + + + Kunji (BBC world trust), Dimagi, FreeSwitch, FrontlineSMS, RapidPro, txtAlert, Verboice, Voto, Vumi, SMS, SAMHSA, PRIORI -Impart knowledge, attitude and behaviour -Increase health literacy of the population Bulk or individual SMS, audio, video schemes for health promotion, a continuous communication may lead to establishing and enhancing the therapeutic alliance between HWs and patients Illiteracy, low mobile network coverage, power shortages, fragmentation of solutions, increased costs for adaptation of educational material and contextualization of the solutions, user experience and usability concerns Point-of-care decision support system Provides point by point support to diagnosis and management of illness, based on a guideline for diagnosis and treatment + + e-IMCI, GuideView, MDconsults, Mezzanine, Mobenzi-outreach, OpenSRP -Support decision-making -Evidenced-based care -Appropriate use of resources -Preventing adverse reactions and errors Management of HTN, diabetes (e.g. foot ulcers) following simple algorithms, management of multimorbidity and multiple drug regimes, prioritization of health needs No treatment and management guidelines in LMICs for most NCDs, decision-support systems are regulated as medical devices is some nations requiring more time and resources, user experience and usability concerns Systems, sensors at point of care diagnostics Extends the delivery points, uses sensors that records and transmits patients’ health records/information, on real time or for later during counselling + + AliveCor, Ubiquitous health care −Patient-based data collection at real time off the facility Tracking blood pressure, glucose, physical activity Integration of data deriving from sensors into eHealth records to facilitate case management Increase health literacy of the population and by extent, encourage health behaviour change Digital literacy is low, costs associated with devices and connectivity, limited mobile coverage, durability of the devices, privacy and security concerns, regulatory and legislative frameworks are needed Supply chain and logistics management Checks, tracks the stock of supplies or medicines, to ensure supply of commodities are maintained year around + + + Life mHealth, Sproxil, Dimagi(commcare supply), mSupply, OpenLMIS -Improve stock management and availability, and decrease waste - Ensure appropriate medicine conservation and patient safety Tracking stocks of HTN, antidiabetic medications at community level when CHWs and HWs fills out their stocks Digital literacy of CHWs, user experience and usability concerns, high burden on the providers, initial costs of the implementation, durability of the devices Research and evaluation Repositories and databases, and built in system for vital events tracking Tracks unique identifiers or indicators for later use, for monitoring, evaluation and research + + + Mother and Child tracking System (MCTS) in India, Rapid SMS Uganda, OpenSRP, mTika, Cell PREVEN -Tacking indicators, storage for future use - data aggregation Storing blood pressure, glucose, medication use data for compliance reporting Costs of registries, and maintenance Data collection, harmonization, harvesting and reporting Reduces the lag between data collection and reporting to go to authorities, improves the completeness and accuracy of data. + + Open Data Kit (ODK),FrontlineSMS,Magpi,Fromhub,DHIS2,Dimagi(commcare),Data Winners,iFormBuilderEnketo,Mango platform,MedicMobile,Mobenzi-research,ONA,OpenXData,PoimapperKhusibaby(necklace & app),GSM,PDA PREVEN  -Improve accuracy, completeness and timeliness of data collection, harvest more data points on real time  - Data aggregation  - Better intelligence for targeted interventions Collection of WHO core modifiable NCD risk factors in existing LMICs surveys Network coverage for reporting or sinking collected data Health care data platform From one office to community set up, routinely collected data can be added to the data pool longitudinally + OpenMRS, Childcount+, OpenEMR −Systematize data reporting and storage process with ability to tack patients longitudinally Blood pressure, glucose, lipids, GFR, creatinine levels from local PHC, or blood pressure, glucose from CHWs Identification of at-risk patients Costs, and maintenance of platform, fragmentation of solutions, digital literacy Training, supervision on personnel management CHW to CHW or CHW to patients/providers interaction Exchange of voice, images, or texts for instant communication at/for remote consultation and decision support + + Switchboard (Closed User Group), Case.io -Improve CHW to CHW, CHW to provide communication for decision support -Access to otherwise unavailable resources and expertise -Interprofessional collaboration leading to better quality of care -Enhance work satisfaction of HWs Decision support for HTN, diabetes management Digital literacy, mobile coverage costs CHW’s work planning, organization and scheduling Reminder-based system, to increase compliance and accountability + txtAlert,MoTech −Improve productivity at work Reminders to make regular home visitation Increase treatment adherence Digital literacy CHW’s training and education Provides continued opportunity for enhancing skills via quizzes and case-based learning to inforce in-person training + eMOCHA, Moodle, OppiaMobile; Leap, Tele-ECHO, Mobile Kunji, mTrack, MOTECH, IQCare, Pesinet Mali, Nompilo, Uganda HIN, Mozambique HIN -Continued education and skills enhancement -Enhance work satisfaction of HWs Simple point-point algorithm for HTN, diabetes, diagnosis, management, counselling Recruit and retain through training Digital literacy, Existing burden on CHWs for other duties, need for updated material, demand for contextualized material, blended learning seems more effective but may be more expensive Human resource and personnel management Monitoring of field activities, and that helps to set up performance-based incentive scheme + mUbuzima, Rapid SMS, iHRIS, MOTECH −Performance tracking GPS tracking of CHWs to be tied with monthly follow ups Network coverage Financial transaction, incentive management and disbursement Mobile transaction scheme as payment or transfer or cash voucher + MTN, m-Pesa −Easing performance-based incentives (CHWs, HWs), payment at point of care (patients) Patients can pay for services or medicines upfront, and CHWs getting performance-based incentives Mobile network partner limited in most LMICs, business model not clear Interaction Theme Functions (Labrique et al., 2013) Characteristics Patients CHWs Providers Health system Examples (Funes et al., 2012; Braun et al., 2013; Labrique et al., 2013; Agarwal et al., 2016; Basavarajappa and Chand, 2017) Health system goals Opportunities for NCDs Challenges Increasing service access and quality Patient’s education and communication One-way or two-way communication gateway for BCC + + + Kunji (BBC world trust), Dimagi, FreeSwitch, FrontlineSMS, RapidPro, txtAlert, Verboice, Voto, Vumi, SMS, SAMHSA, PRIORI -Impart knowledge, attitude and behaviour -Increase health literacy of the population Bulk or individual SMS, audio, video schemes for health promotion, a continuous communication may lead to establishing and enhancing the therapeutic alliance between HWs and patients Illiteracy, low mobile network coverage, power shortages, fragmentation of solutions, increased costs for adaptation of educational material and contextualization of the solutions, user experience and usability concerns Point-of-care decision support system Provides point by point support to diagnosis and management of illness, based on a guideline for diagnosis and treatment + + e-IMCI, GuideView, MDconsults, Mezzanine, Mobenzi-outreach, OpenSRP -Support decision-making -Evidenced-based care -Appropriate use of resources -Preventing adverse reactions and errors Management of HTN, diabetes (e.g. foot ulcers) following simple algorithms, management of multimorbidity and multiple drug regimes, prioritization of health needs No treatment and management guidelines in LMICs for most NCDs, decision-support systems are regulated as medical devices is some nations requiring more time and resources, user experience and usability concerns Systems, sensors at point of care diagnostics Extends the delivery points, uses sensors that records and transmits patients’ health records/information, on real time or for later during counselling + + AliveCor, Ubiquitous health care −Patient-based data collection at real time off the facility Tracking blood pressure, glucose, physical activity Integration of data deriving from sensors into eHealth records to facilitate case management Increase health literacy of the population and by extent, encourage health behaviour change Digital literacy is low, costs associated with devices and connectivity, limited mobile coverage, durability of the devices, privacy and security concerns, regulatory and legislative frameworks are needed Supply chain and logistics management Checks, tracks the stock of supplies or medicines, to ensure supply of commodities are maintained year around + + + Life mHealth, Sproxil, Dimagi(commcare supply), mSupply, OpenLMIS -Improve stock management and availability, and decrease waste - Ensure appropriate medicine conservation and patient safety Tracking stocks of HTN, antidiabetic medications at community level when CHWs and HWs fills out their stocks Digital literacy of CHWs, user experience and usability concerns, high burden on the providers, initial costs of the implementation, durability of the devices Research and evaluation Repositories and databases, and built in system for vital events tracking Tracks unique identifiers or indicators for later use, for monitoring, evaluation and research + + + Mother and Child tracking System (MCTS) in India, Rapid SMS Uganda, OpenSRP, mTika, Cell PREVEN -Tacking indicators, storage for future use - data aggregation Storing blood pressure, glucose, medication use data for compliance reporting Costs of registries, and maintenance Data collection, harmonization, harvesting and reporting Reduces the lag between data collection and reporting to go to authorities, improves the completeness and accuracy of data. + + Open Data Kit (ODK),FrontlineSMS,Magpi,Fromhub,DHIS2,Dimagi(commcare),Data Winners,iFormBuilderEnketo,Mango platform,MedicMobile,Mobenzi-research,ONA,OpenXData,PoimapperKhusibaby(necklace & app),GSM,PDA PREVEN  -Improve accuracy, completeness and timeliness of data collection, harvest more data points on real time  - Data aggregation  - Better intelligence for targeted interventions Collection of WHO core modifiable NCD risk factors in existing LMICs surveys Network coverage for reporting or sinking collected data Health care data platform From one office to community set up, routinely collected data can be added to the data pool longitudinally + OpenMRS, Childcount+, OpenEMR −Systematize data reporting and storage process with ability to tack patients longitudinally Blood pressure, glucose, lipids, GFR, creatinine levels from local PHC, or blood pressure, glucose from CHWs Identification of at-risk patients Costs, and maintenance of platform, fragmentation of solutions, digital literacy Training, supervision on personnel management CHW to CHW or CHW to patients/providers interaction Exchange of voice, images, or texts for instant communication at/for remote consultation and decision support + + Switchboard (Closed User Group), Case.io -Improve CHW to CHW, CHW to provide communication for decision support -Access to otherwise unavailable resources and expertise -Interprofessional collaboration leading to better quality of care -Enhance work satisfaction of HWs Decision support for HTN, diabetes management Digital literacy, mobile coverage costs CHW’s work planning, organization and scheduling Reminder-based system, to increase compliance and accountability + txtAlert,MoTech −Improve productivity at work Reminders to make regular home visitation Increase treatment adherence Digital literacy CHW’s training and education Provides continued opportunity for enhancing skills via quizzes and case-based learning to inforce in-person training + eMOCHA, Moodle, OppiaMobile; Leap, Tele-ECHO, Mobile Kunji, mTrack, MOTECH, IQCare, Pesinet Mali, Nompilo, Uganda HIN, Mozambique HIN -Continued education and skills enhancement -Enhance work satisfaction of HWs Simple point-point algorithm for HTN, diabetes, diagnosis, management, counselling Recruit and retain through training Digital literacy, Existing burden on CHWs for other duties, need for updated material, demand for contextualized material, blended learning seems more effective but may be more expensive Human resource and personnel management Monitoring of field activities, and that helps to set up performance-based incentive scheme + mUbuzima, Rapid SMS, iHRIS, MOTECH −Performance tracking GPS tracking of CHWs to be tied with monthly follow ups Network coverage Financial transaction, incentive management and disbursement Mobile transaction scheme as payment or transfer or cash voucher + MTN, m-Pesa −Easing performance-based incentives (CHWs, HWs), payment at point of care (patients) Patients can pay for services or medicines upfront, and CHWs getting performance-based incentives Mobile network partner limited in most LMICs, business model not clear The 12 functions presented are adaption of the Labrique et al. (2013) work on 12 key functions of mHealth platforms (Labrique et al., 2013). The list of platforms are adaption of Agarwal et al. (2016) work on lists of mHealth platforms for CHWs (Agarwal et al., 2016) which is then divided into 12 core-functions and 3 themes, with additional searches online (Funes et al., 2012; Braun et al., 2013; Labrique et al., 2013; Basavarajappa and Chand 2017). CHWs, community health workers; HIN, health information network; HTN, hypertension; HWs, health workers; LMICs, low- and middle-income countries; NCDs, non-communicable diseases; BCC, behaviour change communication. View Large Table 1 Functions, characteristics and examples of digital platforms in relation to their interaction with patients, providers and health system Interaction Theme Functions (Labrique et al., 2013) Characteristics Patients CHWs Providers Health system Examples (Funes et al., 2012; Braun et al., 2013; Labrique et al., 2013; Agarwal et al., 2016; Basavarajappa and Chand, 2017) Health system goals Opportunities for NCDs Challenges Increasing service access and quality Patient’s education and communication One-way or two-way communication gateway for BCC + + + Kunji (BBC world trust), Dimagi, FreeSwitch, FrontlineSMS, RapidPro, txtAlert, Verboice, Voto, Vumi, SMS, SAMHSA, PRIORI -Impart knowledge, attitude and behaviour -Increase health literacy of the population Bulk or individual SMS, audio, video schemes for health promotion, a continuous communication may lead to establishing and enhancing the therapeutic alliance between HWs and patients Illiteracy, low mobile network coverage, power shortages, fragmentation of solutions, increased costs for adaptation of educational material and contextualization of the solutions, user experience and usability concerns Point-of-care decision support system Provides point by point support to diagnosis and management of illness, based on a guideline for diagnosis and treatment + + e-IMCI, GuideView, MDconsults, Mezzanine, Mobenzi-outreach, OpenSRP -Support decision-making -Evidenced-based care -Appropriate use of resources -Preventing adverse reactions and errors Management of HTN, diabetes (e.g. foot ulcers) following simple algorithms, management of multimorbidity and multiple drug regimes, prioritization of health needs No treatment and management guidelines in LMICs for most NCDs, decision-support systems are regulated as medical devices is some nations requiring more time and resources, user experience and usability concerns Systems, sensors at point of care diagnostics Extends the delivery points, uses sensors that records and transmits patients’ health records/information, on real time or for later during counselling + + AliveCor, Ubiquitous health care −Patient-based data collection at real time off the facility Tracking blood pressure, glucose, physical activity Integration of data deriving from sensors into eHealth records to facilitate case management Increase health literacy of the population and by extent, encourage health behaviour change Digital literacy is low, costs associated with devices and connectivity, limited mobile coverage, durability of the devices, privacy and security concerns, regulatory and legislative frameworks are needed Supply chain and logistics management Checks, tracks the stock of supplies or medicines, to ensure supply of commodities are maintained year around + + + Life mHealth, Sproxil, Dimagi(commcare supply), mSupply, OpenLMIS -Improve stock management and availability, and decrease waste - Ensure appropriate medicine conservation and patient safety Tracking stocks of HTN, antidiabetic medications at community level when CHWs and HWs fills out their stocks Digital literacy of CHWs, user experience and usability concerns, high burden on the providers, initial costs of the implementation, durability of the devices Research and evaluation Repositories and databases, and built in system for vital events tracking Tracks unique identifiers or indicators for later use, for monitoring, evaluation and research + + + Mother and Child tracking System (MCTS) in India, Rapid SMS Uganda, OpenSRP, mTika, Cell PREVEN -Tacking indicators, storage for future use - data aggregation Storing blood pressure, glucose, medication use data for compliance reporting Costs of registries, and maintenance Data collection, harmonization, harvesting and reporting Reduces the lag between data collection and reporting to go to authorities, improves the completeness and accuracy of data. + + Open Data Kit (ODK),FrontlineSMS,Magpi,Fromhub,DHIS2,Dimagi(commcare),Data Winners,iFormBuilderEnketo,Mango platform,MedicMobile,Mobenzi-research,ONA,OpenXData,PoimapperKhusibaby(necklace & app),GSM,PDA PREVEN  -Improve accuracy, completeness and timeliness of data collection, harvest more data points on real time  - Data aggregation  - Better intelligence for targeted interventions Collection of WHO core modifiable NCD risk factors in existing LMICs surveys Network coverage for reporting or sinking collected data Health care data platform From one office to community set up, routinely collected data can be added to the data pool longitudinally + OpenMRS, Childcount+, OpenEMR −Systematize data reporting and storage process with ability to tack patients longitudinally Blood pressure, glucose, lipids, GFR, creatinine levels from local PHC, or blood pressure, glucose from CHWs Identification of at-risk patients Costs, and maintenance of platform, fragmentation of solutions, digital literacy Training, supervision on personnel management CHW to CHW or CHW to patients/providers interaction Exchange of voice, images, or texts for instant communication at/for remote consultation and decision support + + Switchboard (Closed User Group), Case.io -Improve CHW to CHW, CHW to provide communication for decision support -Access to otherwise unavailable resources and expertise -Interprofessional collaboration leading to better quality of care -Enhance work satisfaction of HWs Decision support for HTN, diabetes management Digital literacy, mobile coverage costs CHW’s work planning, organization and scheduling Reminder-based system, to increase compliance and accountability + txtAlert,MoTech −Improve productivity at work Reminders to make regular home visitation Increase treatment adherence Digital literacy CHW’s training and education Provides continued opportunity for enhancing skills via quizzes and case-based learning to inforce in-person training + eMOCHA, Moodle, OppiaMobile; Leap, Tele-ECHO, Mobile Kunji, mTrack, MOTECH, IQCare, Pesinet Mali, Nompilo, Uganda HIN, Mozambique HIN -Continued education and skills enhancement -Enhance work satisfaction of HWs Simple point-point algorithm for HTN, diabetes, diagnosis, management, counselling Recruit and retain through training Digital literacy, Existing burden on CHWs for other duties, need for updated material, demand for contextualized material, blended learning seems more effective but may be more expensive Human resource and personnel management Monitoring of field activities, and that helps to set up performance-based incentive scheme + mUbuzima, Rapid SMS, iHRIS, MOTECH −Performance tracking GPS tracking of CHWs to be tied with monthly follow ups Network coverage Financial transaction, incentive management and disbursement Mobile transaction scheme as payment or transfer or cash voucher + MTN, m-Pesa −Easing performance-based incentives (CHWs, HWs), payment at point of care (patients) Patients can pay for services or medicines upfront, and CHWs getting performance-based incentives Mobile network partner limited in most LMICs, business model not clear Interaction Theme Functions (Labrique et al., 2013) Characteristics Patients CHWs Providers Health system Examples (Funes et al., 2012; Braun et al., 2013; Labrique et al., 2013; Agarwal et al., 2016; Basavarajappa and Chand, 2017) Health system goals Opportunities for NCDs Challenges Increasing service access and quality Patient’s education and communication One-way or two-way communication gateway for BCC + + + Kunji (BBC world trust), Dimagi, FreeSwitch, FrontlineSMS, RapidPro, txtAlert, Verboice, Voto, Vumi, SMS, SAMHSA, PRIORI -Impart knowledge, attitude and behaviour -Increase health literacy of the population Bulk or individual SMS, audio, video schemes for health promotion, a continuous communication may lead to establishing and enhancing the therapeutic alliance between HWs and patients Illiteracy, low mobile network coverage, power shortages, fragmentation of solutions, increased costs for adaptation of educational material and contextualization of the solutions, user experience and usability concerns Point-of-care decision support system Provides point by point support to diagnosis and management of illness, based on a guideline for diagnosis and treatment + + e-IMCI, GuideView, MDconsults, Mezzanine, Mobenzi-outreach, OpenSRP -Support decision-making -Evidenced-based care -Appropriate use of resources -Preventing adverse reactions and errors Management of HTN, diabetes (e.g. foot ulcers) following simple algorithms, management of multimorbidity and multiple drug regimes, prioritization of health needs No treatment and management guidelines in LMICs for most NCDs, decision-support systems are regulated as medical devices is some nations requiring more time and resources, user experience and usability concerns Systems, sensors at point of care diagnostics Extends the delivery points, uses sensors that records and transmits patients’ health records/information, on real time or for later during counselling + + AliveCor, Ubiquitous health care −Patient-based data collection at real time off the facility Tracking blood pressure, glucose, physical activity Integration of data deriving from sensors into eHealth records to facilitate case management Increase health literacy of the population and by extent, encourage health behaviour change Digital literacy is low, costs associated with devices and connectivity, limited mobile coverage, durability of the devices, privacy and security concerns, regulatory and legislative frameworks are needed Supply chain and logistics management Checks, tracks the stock of supplies or medicines, to ensure supply of commodities are maintained year around + + + Life mHealth, Sproxil, Dimagi(commcare supply), mSupply, OpenLMIS -Improve stock management and availability, and decrease waste - Ensure appropriate medicine conservation and patient safety Tracking stocks of HTN, antidiabetic medications at community level when CHWs and HWs fills out their stocks Digital literacy of CHWs, user experience and usability concerns, high burden on the providers, initial costs of the implementation, durability of the devices Research and evaluation Repositories and databases, and built in system for vital events tracking Tracks unique identifiers or indicators for later use, for monitoring, evaluation and research + + + Mother and Child tracking System (MCTS) in India, Rapid SMS Uganda, OpenSRP, mTika, Cell PREVEN -Tacking indicators, storage for future use - data aggregation Storing blood pressure, glucose, medication use data for compliance reporting Costs of registries, and maintenance Data collection, harmonization, harvesting and reporting Reduces the lag between data collection and reporting to go to authorities, improves the completeness and accuracy of data. + + Open Data Kit (ODK),FrontlineSMS,Magpi,Fromhub,DHIS2,Dimagi(commcare),Data Winners,iFormBuilderEnketo,Mango platform,MedicMobile,Mobenzi-research,ONA,OpenXData,PoimapperKhusibaby(necklace & app),GSM,PDA PREVEN  -Improve accuracy, completeness and timeliness of data collection, harvest more data points on real time  - Data aggregation  - Better intelligence for targeted interventions Collection of WHO core modifiable NCD risk factors in existing LMICs surveys Network coverage for reporting or sinking collected data Health care data platform From one office to community set up, routinely collected data can be added to the data pool longitudinally + OpenMRS, Childcount+, OpenEMR −Systematize data reporting and storage process with ability to tack patients longitudinally Blood pressure, glucose, lipids, GFR, creatinine levels from local PHC, or blood pressure, glucose from CHWs Identification of at-risk patients Costs, and maintenance of platform, fragmentation of solutions, digital literacy Training, supervision on personnel management CHW to CHW or CHW to patients/providers interaction Exchange of voice, images, or texts for instant communication at/for remote consultation and decision support + + Switchboard (Closed User Group), Case.io -Improve CHW to CHW, CHW to provide communication for decision support -Access to otherwise unavailable resources and expertise -Interprofessional collaboration leading to better quality of care -Enhance work satisfaction of HWs Decision support for HTN, diabetes management Digital literacy, mobile coverage costs CHW’s work planning, organization and scheduling Reminder-based system, to increase compliance and accountability + txtAlert,MoTech −Improve productivity at work Reminders to make regular home visitation Increase treatment adherence Digital literacy CHW’s training and education Provides continued opportunity for enhancing skills via quizzes and case-based learning to inforce in-person training + eMOCHA, Moodle, OppiaMobile; Leap, Tele-ECHO, Mobile Kunji, mTrack, MOTECH, IQCare, Pesinet Mali, Nompilo, Uganda HIN, Mozambique HIN -Continued education and skills enhancement -Enhance work satisfaction of HWs Simple point-point algorithm for HTN, diabetes, diagnosis, management, counselling Recruit and retain through training Digital literacy, Existing burden on CHWs for other duties, need for updated material, demand for contextualized material, blended learning seems more effective but may be more expensive Human resource and personnel management Monitoring of field activities, and that helps to set up performance-based incentive scheme + mUbuzima, Rapid SMS, iHRIS, MOTECH −Performance tracking GPS tracking of CHWs to be tied with monthly follow ups Network coverage Financial transaction, incentive management and disbursement Mobile transaction scheme as payment or transfer or cash voucher + MTN, m-Pesa −Easing performance-based incentives (CHWs, HWs), payment at point of care (patients) Patients can pay for services or medicines upfront, and CHWs getting performance-based incentives Mobile network partner limited in most LMICs, business model not clear The 12 functions presented are adaption of the Labrique et al. (2013) work on 12 key functions of mHealth platforms (Labrique et al., 2013). The list of platforms are adaption of Agarwal et al. (2016) work on lists of mHealth platforms for CHWs (Agarwal et al., 2016) which is then divided into 12 core-functions and 3 themes, with additional searches online (Funes et al., 2012; Braun et al., 2013; Labrique et al., 2013; Basavarajappa and Chand 2017). CHWs, community health workers; HIN, health information network; HTN, hypertension; HWs, health workers; LMICs, low- and middle-income countries; NCDs, non-communicable diseases; BCC, behaviour change communication. View Large Training and supervision on personnel management Digital devices can be used for training, personalized supervision of CHWs, and communication between CHWs and their supervisors at a higher institution (Mishra et al., 2015; White et al., 2016). In Table 1, we present various platforms, which can be used for CHW’s work planning and scheduling, CHW’s training and education, overall human resource management, financial transaction and incentive scheme, all aiming to improve the management and performance of CHWs (Labrique et al., 2013; Agarwal et al., 2016). Digital applications can provide a supportive channel between CHWs and their supervisors. Blended learning can combine digital delivery of the content and remote interaction with in-person training (Diedhiou et al., 2015,; Agarwal et al., 2016; Lygidakis et al., 2016). Digital solutions have also been able to support peer communication in the form of virtual informal groups and learning networks that are able to create educational opportunities from the daily practice (Braun et al., 2013,; Lygidakis et al., 2016). The virtual communities of practice can be employed to foster team-building and collaborative training, as they enable CHWs to share interests, ask questions, raise concerns and exchange rich content, such as photos, with each other (Barnett et al., 2014). CHW programmes can capitalize on tools with an already broad user base, like WhatsApp (Henry et al., 2016). The 6-month review of WhatsApp chat logs from 41 CHWs showed that the multi-modal communication and information systems were used by CHWs for one-to-one, group and peer-to-peer forms of coaching, counselling and supervision, with minimal training (Henry et al., 2016). Finally, technology enables a participatory approach in the design and implementation of contextualized solutions. Svoronos et al. (2010) described how employing a participatory approach for the development of their mHealth module validated its appropriateness and fostered a shared ownership of the solution. Evaluation and research Digitalization increases the opportunities for data collection, analysis and research. Examples of use of mobile devices in maternal and child health provide early indications of the platforms that can be built for in-country and intercountry data-sharing and research collaboration (Agarwal et al., 2016). For example, user-friendly and personally tailored devices, connected to a cloud-based dashboard and a central database, will further boost research capacity in LMICs (Mishra et al., 2016). Three kinds of platforms are in use today: data collection and reporting; tracking vital events; building or maintaining eHealth records (Table 1) (Labrique et al., 2013; Agarwal et al., 2016). Evidence supports that the employment of digital tools is cost-effective, improves record completeness and decreases the time for the transmission of data from the source to the coordination centres (Manda and Herstad, 2009; Ngabo et al., 2012,; Agarwal et al., 2015). For instance, RapidSMS-MCH is a text-message-based system in Rwanda that assists CHWs reporting real-time information on pregnant women and following them up till the post-partum period in their communities (Ngabo et al., 2012). The system has also helped pregnant women seeking urgent care, as CHWs can alert the health facilities promptly. Further, in a study on diabetes medication decision tools among Latino and African American Adults, CHWs used an interactive and web-based tablet-/computer-delivered tool (iDecide) for reporting (Heisler et al., 2014). Challenges to prospect of digitalization Funding for CHW programmes The increased use of digital means in CHW programmes is in stark contrast with the countries’ capacity to scale up their use. Deployment and initial operational costs are considerable. With significant financial constraints and funding cuts (Atun et al., 2017), supporting digitalization in LMICs is challenged. To sustainably pool funds from diverse sectors for CHW programmes, partnership with the private sector may be needed, though it should be taken cautiously. The private sector can help mainly by technological expertise and know-how, and human and financial resources. The critical question is how we can leverage benefit from such collaboration. A new model for corporate social responsibility needs to be introduced. Mobile connectivity is spreading rapidly and there are IT companies in every country, which may contribute (Kay et al., 2011). However, balance between corporate motives and the health needs of the population have to be guarded. Digital health literacy of the CHWs CHWs may be either illiterate or semi-literate, as there are no formal literacy requirements for CHWs in many countries (Rubinstein et al., 2016; Mishra et al., 2016). Training CHWs to different health programmes especially with NCD components can be difficult and increase the costs. Due to diverse literacy, socio-economic backgrounds and country health needs, there is no doubt that CHW interventions need to be country- and context-specific, and modulated according to the digital inputs the intervention requires. Our previous study from Nepal has shown that CHWs possess a positive attitude towards taking on future tasks to help tackling NCDs despite their lack of required technical competences and digital health literacy (Neupane et al., 2015). Future CHW programmes, however, need to introduce appropriate training to help CHWs acquire new skills, as this is one of the most important non-financial incentives that can prevent isolation, burn out and attrition (Bhattacharyya et al., 2001). Systemic challenges of CHW programmes Some perennial challenges to CHW programmes will also affect the prospect of digitalization, as they affect structure, performance and motivation of the workforce. Most of the CHW programmes run under a loose and informal structure, where there are inadequate formal recruitment processes, no job description and no provision for retirement (Singh et al., 2015). CHWs should be regularly supervised and routinely monitored (Mishra et al., 2015). Due to lack of health workers at the primary health care level and their high work load, adequate supervision for CHWs in future NCD programmes seems challenging during their early years of implementation when the supervisory need is the most. There is increasing expectation of financial incentives (salaried position, daily allowance, transportation support and logistics) among CHWs. This will even go further, if they are to use digital devices, which demand higher skills and literacy. However, other non-monetary incentives, such as a desire to better their family health, change from routine work, gain of health knowledge, social recognition and religious merits, are also important determinants (Singh et al., 2015). A study from Nepal showed that despite being minimally trained and having low literacy in hypertension (HTN), CHWs were intrinsically motivated to take part in future HTN management programmes at the community level (Neupane et al., 2015). A precautionary step should be taken not to overburden CHWs in future NCD programmes, as previous studies have shown burnout and stress reduces motivation in CHWs as with other health cadres (Tien et al., 2000). Discussion This review is an expansion of earlier reviews by Labrique et al. (2013) and Agrawal et al. (2016) in NCD context. Labrique et al. (2013) pointed out the 12 core functions of mHealth and proposed a visual framework for understanding their relationship in health system—specific to reproductive, maternal and child health. Our review discussed the three benefits of digitalization, including: (1) expanding coverage and quality; (2) training and supervision of personnel and (3) research and evaluation. We also mapped digital platforms currently in use, building on the work of Agrawal et al. (2016) and Labrique et al. (2013), and deliberated over the NCD context for CHWs (Labrique et al., 2013; Agarwal et al., 2016). Finally, we have identified relevant challenges, namely funding for the sustainability of CHW programmes, digital health literacy, and those existing and systematic related to motivation and myths around CHWs. The growing prevalence of NCDs will increase the burden on health care workers at primary care level. CHWs will be required to spend a significant amount of their time in counselling, conducting home visits and referring patients to health centres. Not limited to primary prevention, there is some evidence to suggest CHWs can play a crucial role in secondary prevention (reducing the risk of recurrent events) of CVDs and many other chronic diseases. For example, in a study conducted in India (Xavier et al., 2016), CHWs were found effective in reducing the cardiovascular risk factors and improving adherence to evidence-based drugs among patients with CVD who were linked to their health centres in periphery. In the countries with workforce shortages, digitalization may reduce the workload of the health professionals, but cannot completely take over their role. The best way forward is to train the future health workforce in digital skills. Ability to use devices rationally is of paramount importance—more important rather than the coverage of devices itself, as authors in a report suggest: ‘it is not technology, it is the carefully planned home bred solutions, the key to success’ (Levy et al., 2012). The discussed core functions of digitalization are greatly dependent on having a reliable mobile network coverage and a source of electricity. Even though mobile subscriptions have already outnumbered people, the LMICs are yet to benefit from their greater expansion. Internet connectivity remains poor in most LMICs, especially in rural settings, where the majority of health care providers still do not have access to the internet at work. The cost of service can pose significant limitations too (Safaie et al., 2006; Wallis et al., 2017). Mobile devices used in some implementations do not require a stable mobile connection, allowing for offline use and database synchronization when internet is available; such are the cases of KhushiBaby (Mishra et al., 2016; Khusibaby) and the Nyitaho mobile app for diabetic patients in the D2Rwanda clinical trial (Lygidakis, 2017). With power shortages being a perennial problem in LMICs, solar powered devices and stations for connectivity, such as those provided by ‘Africa Renewable Energy Distributor (ARED)’ in Uganda and Rwanda, have the potential to help CHWs in some locations (Africa Renewable Energy Distributor). Furthermore, the high turnover on digital platforms is worrying. Many switches between platforms or abandon them within the first 6 months of use (Levy et al., 2012). This turnover raises questions of maintenance, cost-effectiveness and investments in retraining the workforce. Service integration of existing fragmented digital and non-digital CHW interventions is critical. Different programmes and components should be brought under a single umbrella, and either liaise or merge. Integrating CHWs programmes with mHealth potentially benefits hard-to-reach populations that are otherwise not covered by CHW programmes in LMICs (Källander et al., 2013). Similarly, many interventions can be combined. For example, interventions for diabetes and HTN can be brought together under a single umbrella, as they share common risk factors. Service integration will help create comprehensive digital platforms targeting many NCDs, even though there is still a paucity of evidence to what level the synergy works, and what components of the interventions should be brought together due to the diverse socio-economic, cultural and environmental conditions (Rubinstein et al., 2016; Mishra et al., 2016). Bridging the fragmented market of digital-device companies is also necessary. As many LMICs may lack resources and know-how, a systematic approach for database creation and maintenance is needed to enable data accessibility and tailoring for researchers, health professionals and other frontline workers (Mishra et al., 2016). The growth in digitalization has also spurred concerns over patient’s safety and privacy. Digitalization may lead to a fragmentation of health system by limiting patient–provider interaction and shifting decision-making on information deriving from applications [that can lead to the risk of wrong treatment given to the patients and the risk of an adverse drug reaction (ADE) in extreme cases] (Smart Conculting). Two million of such ADE happen every year in high-income countries such as Australia (Smart Conculting); therefore, an unregulated digital health industry and absence of coordination (between patients, providers and CHWs) is more likely to complicate the problem. Personal and sensitive information must always be collected after patient consent and protected accordingly. Greater use of multiple applications may create challenges in privacy of information shared across such platforms (Kay et al., 2011). To ensure patient safety and privacy, digital health applications must be regulated in LMICs in a similar way as in nations where strict legislation already exists, such as in the USA and the European Union. Countries need to collaborate to harmonize communication platforms, set up ethical standards for data sharing between research agencies and IT companies, and increase security and affordability of data exchange (Kay et al., 2011). To achieve this, engagement of the private sector, including tele-communication companies, is also necessary (Wallis et al., 2017). Finally, from the end-users’ perspective, a multidisciplinary approach is needed to improve the user experience, including usability and adaptability (Wallis et al., 2017). A digital health application which does not integrate with the everyday activities and the context and becomes a burden for the end user, whether he/she is a CHW or a patient, will most likely fail in its implementation. Some questions remain despite our deliberations. How can we monitor the digital platforms and guarantee data privacy and security? How can we balance the role of providers and the ‘digital inputs’ in decision making? How can we make CHW interventions more cost-effective and sustainable? Is it worth investing in unforeseen gains from digitalization? Can partnerships with IT companies bring good investments in LMICs and increase funding for future CHW interventions? Acknowledging the ethical dilemmas in balancing corporate motives, a line of equilibrium should be drawn with a clear demarcation where interests of both parties are met. Conclusions The prospects of digitization in CHW programmes are encouraging, but are equally coupled with challenges. The benefits range from an expansion in the coverage and quality of services, to adding efficiency to recording, reporting and personnel management, and finally catalysing research and evaluation processes. Digitally equipped CHWs may provide new services particularly around NCD prevention in the future. A substantial investment in capacity building, training, improving digital literacy, workforce management, supervision and logistics is however required. Sustainability issues of CHW programmes are critically important. Initial roll-out of the digital devices and maintenance incur significant costs, yet a surge in the availability of inexpensive digital devices and improved internet connectivity can accelerate the digitalization of CHW programmes and generate cascading benefits in the long run. Conflict of interest statement. None declared. Acknowledgements We would like to thank Nepal Development Society, the civil society NGO working in NCD prevention in Nepal, for encouragement and support in carrying out the study. References Africa Renewable Energy Distributor. About ARED. http://www.webcitation.org/71ubvU66k, accessed 12 August 2018. Agarwal S , Perry HB , Long LA , Labrique AB. 2015 . Evidence on feasibility and effective use of mHealth strategies by frontline health workers in developing countries: systematic review . Tropical Medicine & International Health 20 : 1003 – 14 . Google Scholar Crossref Search ADS Agarwal S , Rosenblum L , Goldschmidt T , Carras M , Goel N , Labrique A. 2016 . Mobile technology in support of front-line health workers . Johns Hopkins University Global mHealth Initiative . Atun R , Silva S , Knaul FM. 2017 . Innovative financing instruments for global health 2002–2015: a systematic analysis . The Lancet Global Health 5 : e720 – 6 . Google Scholar Crossref Search ADS PubMed Barnett S , Jones SC , Caton T , Iverson D , Bennett S , Robinson L. 2014 . Implementing a virtual community of practice for family physician training: a mixed-methods case study . Journal of Medical Internet Research 16 : e83 . Doi: 10.2196/jmir.3083. Google Scholar Crossref Search ADS PubMed Basavarajappa C , Chand PK. 2017 . Digital platforms for mental health-care delivery . Indian Journal of Psychological Medicine 39 : 703 . Google Scholar Crossref Search ADS PubMed Beard T , Redmond S. 1979 . Declaration of alma-ata . The Lancet 313 : 217 – 8 . Google Scholar Crossref Search ADS Bhattacharyya K , LeBan K , Winch P , Tien M. 2001 . Community Health Workers: Incentives and Disincentives: How They Affect Motivation, Retention, and Sustainability . Arlington, VA : Basic Support for Institutionalizing Child Survival Project (Basics II) for the United States Agency for International Development . Braun R , Catalani C , Wimbush J , Israelski D. 2013 . Community health workers and mobile technology: a systematic review of the literature . PLoS One 8 : e65772 . Google Scholar Crossref Search ADS PubMed DeRenzi B , Findlater L , Payne J et al. 2012 . Improving community health worker performance through automated SMS. In: Proceedings of the Fifth International Conference on Information and Communication Technologies and Development, (ICTD’12) Atlanta, GA: ACM , 25 – 34 . Diedhiou A , Gilroy KE , Cox CM et al. 2015 . Successful mLearning pilot in senegal: delivering family planning refresher training using interactive voice response and SMS . Global Health: Science and Practice 3 : 305 – 21 . Google Scholar Crossref Search ADS Florez-Arango JF , Iyengar MS , Dunn K , Zhang J. 2011 . Performance factors of mobile rich media job aids for community health workers . Journal of the American Medical Informatics Association 18 : 131 – 7 . Google Scholar Crossref Search ADS PubMed Funes R , Hausman V , Rastegar A . 2012 . Preparing the next generation of community health workers: the power of technology for training. New York: Dalberg Global Development Advisors. http://hetv.org/pdf/power-of-technology-for-training.pdf, accessed 23 December 2017. Gartner IT. Glossary. Digitalization. https://www.gartner.com/it-glossary/digitalization/, accessed 11 August 2018. Heisler M , Choi H , Palmisano G et al. 2014 . Comparison of community health worker–led diabetes medication decision-making support for low-income Latino and African American adults with diabetes using E-health tools versus print materials: a randomized, controlled trial . Annals of Internal Medicine 161 : S13 – 22 . Google Scholar Crossref Search ADS PubMed Henry JV , Winters N , Lakati A et al. 2016 . Enhancing the supervision of community health workers with WhatsApp mobile messaging: qualitative findings from 2 low-resource settings in Kenya . Global Health: Science and Practice 4 : 311 – 25 . Google Scholar Crossref Search ADS ILO. International standard classification of occupations 2008 (ISCO-08): Structure, group definitions and correspondence tables. 2012 . ILO publishing, pp. 12 – 14 . http://www.ilo.org/global/publications/ilo-bookstore/order-online/books/WCMS_172572/lang--en/index.htm, accessed 12 January 2018. International Telecommunication Union. 2017 . ICT Facts and Figures. https://www.itu.int/en/ITU-D/Statistics/Documents/facts/ICTFactsFigures2017.pdf, accessed 12 November 2018. Källander K , Tibenderana JK , Akpogheneta OJ et al. 2013 . Mobile health (mHealth) approaches and lessons for increased performance and retention of community health workers in low- and middle-income countries: a review . Journal of Medical Internet Research 15 : e17 . Google Scholar Crossref Search ADS PubMed Kay M , Santos J , Takane M. 2011 . mHealth: new horizons for health through mobile technologies . World Health Organization 64 : 66 – 71 . Khetan AK , Purushothaman R , Chami T et al. 2016 . The Effectiveness of community health workers for CVD prevention in LMIC . Global heart 12 : 233 – 43 . Google Scholar Crossref Search ADS PubMed Khusibaby. The Problem. https://www.khushibaby.org/, accessed 12 August 2018. Labrique AB , Vasudevan L , Kochi E , Fabricant R , Mehl G. 2013 . mHealth innovations as health system strengthening tools: 12 common applications and a visual framework . Global Health: Science and Practice 1 : 160 – 71 . Google Scholar Crossref Search ADS Levy D , Wasden C , DiFilippo D , Sur P. 2012 . Emerging mHealth: paths for growth . PwC M-Health , 1 – 44 . https://www.pwc.com/gx/en/healthcare/mhealth/assets/pwc-emerging-mhealth-exec-summary.pdfsummary.pdf. Lewin S , Munabi‐Babigumira S , Glenton C 2010 . Lay health workers in primary and community health care for maternal and child health and the management of infectious diseases . Cochrane Database Systematic Review CD004015 . Little A , Medhanyie A , Yebyo H , Spigt M , Dinant G-J , Blanco R. 2013 . Meeting community health worker needs for maternal health care service delivery using appropriate mobile technologies in Ethiopia . PLoS One 8 : e77563. Google Scholar Crossref Search ADS PubMed Lygidakis C , McLoughlin C , Patel K. 2016 . Achieving Universal Health Coverage: Technology for Innovative Primary Health Care Education. World Organization of Family Doctors (WONCA), Iheed. Lygidakis C. 2017 . Community- and mHealth-Based Integrated Management of Diabetes in Primary Healthcare in Rwanda (D2Rwanda): Study Record Detail . https://clinicaltrials.gov/ct2/show/NCT03376607, accessed 9 November 2018. Manda TD , Herstad J. 2009 . Implementing mobile phone solutions for health in resource constrained areas: understanding the opportunities and challenges. In: International Conference on e-Infrastructure and e-Services for Developing Countries . Berlin : Springer , 95 – 104 . Mishra SR , Neupane D , Briffa TG , Kallestrup P. 2016 . mHealth plus community health worker interventions: the future research agenda . The Lancet. Diabetes & Endocrinology 4 : 387 – 8 . Google Scholar Crossref Search ADS PubMed Mishra SR , Neupane D , Preen D , Kallestrup P , Perry HB. 2015 . Mitigation of non-communicable diseases in developing countries with community health workers . Globalization and Health 11 : 43. Google Scholar Crossref Search ADS PubMed Mishra SR , Sharma A , Kaplan WA , Adhikari B , Neupane D. 2016 . Liberating data to combat NCDs . The Lancet Diabetes & Endocrinology 4 : 482 – 3 . Google Scholar Crossref Search ADS PubMed Modi D , Gopalan R , Shah S et al. 2015 . Development and formative evaluation of an innovative mHealth intervention for improving coverage of community-based maternal, newborn and child health services in rural areas of India . Global Health Action 8 : 26769 . Google Scholar Crossref Search ADS PubMed Neupane D , Mclachlan CS , Gautam R et al. 2015 . Literacy and motivation for the prevention and control of hypertension among female community health volunteers: a qualitative study from Nepal . Global Health Action 8 : 28254 . Google Scholar Crossref Search ADS PubMed Ngabo F , Nguimfack J , Nwaigwe F et al. 2012 . Designing and implementing an innovative SMS-based alert system (RapidSMS-MCH) to monitor pregnancy and reduce maternal and child deaths in Rwanda . Pan African Medical Journal 13 : 1 – 16 . One Million Community Health Workers Campaign. 2019 . First National Planning Workshop-Success. http://1millionhealthworkers.org/tag/unicef/, accessed 12 August 2018. Opoku D , Stephani V , Quentin W. 2017 . A realist review of mobile phone-based health interventions for non-communicable disease management in sub-Saharan Africa . BMC Medicine 15 : 24. Google Scholar Crossref Search ADS PubMed Perry H , Zulliger R. 2012 . How Effective Are Community Health Workers. An Overview of Current Evidence with Recommendations for Strengthening Community Health Worker Programs to Accelerate Progress in Achieving the Health-Related Millennium Development Goals . Baltimore : Johns Hopkins Bloomberg School of Public Health . Ranslow A , Crompton D , Mehta K , Butler P , Adair J. 2015 . Empowering community health workers with inkjet-printed diagnostic test strips . Procedia Engineering 107 : 205 – 14 . Google Scholar Crossref Search ADS Reddy KS. 2016 . Global Burden of Disease Study 2015 provides GPS for global health 2030 . Lancet (London, England) 388 : 1448 – 9 . Google Scholar Crossref Search ADS PubMed Rifkin S. 2008 . Community Health Workers . In: Kirch W (ed). Encyclopedia of Public Health . Berlin : Springer Reference , 773 – 81 . Rubinstein A. 2016 . mHealth plus community health worker interventions: the future research agenda—author's reply . The Lancet Diabetes & Endocrinology 4 : 388 . Google Scholar Crossref Search ADS PubMed Rural Health Information Hub. Roles of Community Health Workers. https://www.ruralhealthinfo.org/community-health/community-health-workers/1/roles, accessed 12 August 2018. Safaie A , Mousavi SM , LaPorte RE , Goya MM , Zahraie M. 2006 . Introducing a model for communicable diseases surveillance: cell phone surveillance (CPS) . European Journal of Epidemiology 21 : 627 – 32 . Google Scholar Crossref Search ADS PubMed Singh D , Negin J , Otim M , Orach CG , Cumming R. 2015 . The effect of payment and incentives on motivation and focus of community health workers: five case studies from low-and middle-income countries . Human Resources for Health 13 : 58 . Google Scholar Crossref Search ADS PubMed Smart Conculting. Why Digital Health Needs Standards and Health Informatics More Than Ever. http://www.semanticconsulting.com.au/2017/01/05/digital-health-needs-standards-health-informatics-ever/, accessed 12 August 2018. Svoronos T , Mjungu D , Dhadialla P , Luk R , Zue C. 2010 . CommCare: automated quality improvement to strengthen community-based health the need for quality improvement for CHWs . http://d-tree.org/wp.../Svoronos-Medinfo-CommCare-safe-pregnancy1.pdf, accessed 26 December 2017. Tien M , LeBan K , Winch P. 2000 . Community Health Workers. Incentives and Disincentives: How They Affect Motivation, Retention, and Sustainability . Calverton, MD : Child Survival Technical Support . Tsolekile LP , Puoane T , Schneider H , Levitt NS , Steyn K. 2014 . The roles of community health workers in management of non-communicable diseases in an urban township . African Journal of Primary Health Care & Family Medicine 6 : 1 – 8 . Google Scholar Crossref Search ADS United Nations. Resolution Adopted by the General Assembly on 25 September 2015 . http://www.un.org/ga/search/view_doc.asp? symbol=A/RES/70/1&Lang=E, accessed 11 August 2018. Wallis L , Blessing P , Dalwai M , Shin SD. 2017 . Integrating mHealth at point of care in low-and middle-income settings: the system perspective . Global Health Action 10 : 1327686. Google Scholar Crossref Search ADS PubMed White A , Thomas DS , Ezeanochie N , Bull S. 2016 . Health worker mHealth utilization: a systematic review . Computers, Informatics, Nursing 34 : 206 . Google Scholar Crossref Search ADS Xavier D , Gupta R , Kamath D et al. 2016 . Community health worker-based intervention for adherence to drugs and lifestyle change after acute coronary syndrome: a multicentre, open, randomised controlled trial . The Lancet Diabetes & Endocrinology 4 : 244 – 53 . Google Scholar Crossref Search ADS PubMed © The Author(s) 2019. 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)

Smith, Stephanie, L

doi: 10.1093/heapol/czy109pmid: 30668676

Abstract Civil society actors have substantially increased their participation in global and national health policymaking processes since the 1970s. Civil society roles in shaping such significant global health milestones as the Doha Declaration on Intellectual Property Rights, the Framework Convention on Tobacco Control and the recently adopted United Nations Sustainable Development Goals are well documented, but knowledge of civil society actor influence on health policy processes in low- and middle-income countries remains fragmented. This study analyses 24 peer-reviewed research articles published between 2007 and 2016 to identify factors affecting civil society influence in the pre-implementation stages of the policy process. The articles reviewed span 13 health issues and more than 50 countries in four regions of the world. This body of work focuses on civil society as represented by formal groups, primarily domestic and to some extent international non-governmental organizations, but also social movements, professional associations and faith-based organizations, among others. The studies document several actor-centred and contextual factors that affect civil society actor power, commonly across stages of the policy process. Crucially, civil society actors were challenged to impact the process in countries that lacked participative norms and governing structures. When repressive conditions existed, regime changes and donors sometimes helped to open doors to participation. The power of civil society actors was enhanced when they joined strong epistemic networks and broader coalitions of stakeholders, were resourced, and framed issues in ways that resonated with national policies and political priorities. The synthesis offers guidance to practitioners on factors to consider in strategy development and points to several issues for further investigation by health policy analysis scholars, including the implications of issue (non)adoption by civil society actors and contestation dynamics among those with differing perspectives. Civil society, NGO, health policy, LMIC, review, narrative synthesis Introduction Key Messages Civil society is under-defined in recent research on health policy processes in low- and middle-income countries. Civil society is primarily represented by formal, especially non-governmental, organizations that are active in the implementation stage of the policy process. Politically oriented representations and stages are deemphasized. The reviewed research suggests that robust financial resources, technical expertise and coalitions are primary sources of civil society actor power, though their influence hinges on factors in disparate political contexts. Key factors in political contexts, including existing political priorities and government openness to collaboration, condition civil society influence. Future research should draw upon more established bodies of civil society scholarship to bolster theory development, with particular attention to issue adoption, sources of power relative to other actors and variation across political contexts. Roles for civil society actors in health policy processes at global and national levels have grown alongside the ‘global associational revolution’ of the past half-century (Salamon, 1994). Civil society actors have played critical roles in shaping such significant global health milestones as the 1981 International Code of Marketing of Breast-milk Substitutes, 1994 International Conference on Population and Development and its Programme of Action, the Doha Declaration on Intellectual Property Rights, the Framework Convention on Tobacco Control and the recently adopted United Nations Sustainable Development Goals (Joachim, 2003; Mamudu and Glantz, 2009; ‘t Hoen et al., 2011; Lencucha et al., 2011; Gneiting, 2016). They have also played prominent roles in the HIV/AIDS policy arena, effectively advocating for government action and the rights of those affected, mobilizing against stigma, and providing community-based support and treatment access (Zuniga, 2006; Birdsall and Kelly, 2007; Parker, 2011; Kippax et al., 2013; van Pletzen et al., 2014). Nonetheless, knowledge concerning the power of civil society actors to influence health policy processes in low- and middle-income countries—where global disease burdens are concentrated—remains fragmented. This is perhaps unsurprising given that assessments of ‘the terrain of health policy analysis’ in low- and middle-income countries indicate that it remained underdeveloped through at least 2007, giving rise to calls for more systematic and explanatory studies and to engage the full range of policy actors, including civil society organizations (Gilson and Raphaely, 2008; Walt et al., 2008). This study seeks to illuminate the terrain of health policy analysis research with respect to factors affecting civil society actor influence since these calls were made. Hence, this study reviews a recent decade’s worth of systematically selected peer-reviewed publications reporting on such research, with aims to synthesize findings and inform future research on the topic. The following research question is thus investigated: what factors affect the power of civil society actors to influence the pre-implementation stages of the health policy process in low- and middle-income countries? The study focuses on the agenda setting, formulation and adoption stages of the policy process, excluding research on the implementation stage due to its theoretical, methodological and empirical diversity (Winter, 2012). Synthesis methods previously used to review health policy analysis research in low- and middle-income countries (Gilson and Raphaely, 2008; Berlan et al., 2014; Erasmus, 2014; Erasmus et al., 2014; Gilson, 2014; Gilson et al., 2014; Walt and Gilson, 2014) and global health diplomacy (Ruckert et al., 2016) are employed to answer the research question. Twenty-four mostly qualitative studies published in peer-reviewed academic journals between 2007 and 2016, a period during which publications meeting search criteria in the Web of Science Core Collection academic database more than tripled (from 39 between 2007 and 2009 to 135 between 2014 and 2016), were analysed for this study. Two key analytical concepts that are elaborated in the next section of this article guide the inquiry: civil society and the policy process. Though several prominent policy analysis frameworks draw attention to actors as an analytical category (Sabatier and Jenkins Smith, 1993; Walt and Gilson, 1994; Kingdon, 1995; Shiffman and Smith, 2007), <7% of the articles meeting search criteria met selection criteria for specifically analysing civil society actors and their influence on the pre-implementation stages of health policy processes in low- and middle-income countries. Civil society actors frequently inform the broader body of policy analysis research (as through interviews) and calls for their greater engagement in the health policy process are often made, but the subset of work employing robust research methods to contribute to knowledge of the subject at hand is relatively circumscribed. Encompassing 24 studies spanning 13 health issues and >50 countries, this review synthesizes findings from a subset of recently published policy process research to advance knowledge of this important category of actors and the contexts that shape their influence. The article begins by defining civil society and reviewing contributions of some prominent theories of the policy process, including frameworks developed for the express purpose of informing health policy analysis in low- and middle-income countries and at the global level. A description of procedures for data collection and analysis and characteristics of included articles follows. The core of the article documents findings by policy process stage, highlighting significant factors and the degree of success civil society actors found in influencing outcomes. The results, synthesized along the lines of actor-centred and contextual factors in the discussion section (see also Tables 4 and 5), offer guidance to practitioners on factors to consider in strategy development and point to several issues for further investigation by health policy analysis scholars, including the implications of issue (non)adoption by civil society actors and contestation dynamics among those with differing perspectives. Analytical concepts: civil society and the policy process This review takes civil society and the policy process in the context of low- and middle-income countries as its central analytical concepts. Civil society is conceptually defined in terms of ‘the broad spectrum of voluntary associations that are entirely or largely independent of government and that are not primarily motivated by commercial concerns’ (Najam, 2000, p. 378; see also Hall, 1987; Muukkonen, 2009). Civil society encompasses but is not limited to several distinct organizational forms, such as domestic and international non-governmental organizations (NGOs or non-profits), civil society organizations (CSOs), trade unions, faith-based organizations, social movements, advocacy groups, philanthropic foundations, community groups, think tanks, epistemic/knowledge communities (Haas, 1992), professional associations and groups of activists, such as Brazil’s sanitaristas. Civil society actors are generally thought to represent societal interests, though their social, political and economic origins and operations are complex (Salamon and Anheier, 1998). This study focuses on civil society actors that seek to influence health policymaking processes as independent entities and as members of more or less formally defined networks or social movements bound by shared concerns for social issues (Shiffman et al., 2016; VanDyck, 2017). The stages heuristic is a widely used conceptual framework that identifies a series of stages in the policy process. Drawing upon Sabatier (2007), Kingdon (1995) and Berlan et al. (2014), the stages and their associated outcomes include: agenda setting, with agenda status indicated by the allocation of policy attention and resources; policy formulation, which involves the specification and narrowing of alternative solutions; adoption, which is marked by an authoritative decision; and implementation and evaluation, involving the delivery and assessment of services. A pre-agenda setting stage, issue adoption by networks and other gatekeepers, may form an important component of the process for health issues (Carpenter, 2007). Some studies focus on a single stage, but many investigate the overlapping nature of stages in the policy process, recognizing the existence of feedback loops and addressing a key criticism of the heuristic as overly linear (Sabatier, 2007). Theoretically based frameworks that help to identify factors affecting dynamics and outcomes of the policy process should also or alternatively be employed, and several are pertinent to policy analysis in low- and middle-income countries (Sabatier, 2007; Walt et al., 2008; Smith and Shiffman, 2018). For instance, Kingdon’s (1995) multiple streams model poses that when problem, policy and politics streams are coupled, an issue’s prospects for gaining agenda status are greatly improved; coupling is facilitated by entrepreneurial leaders and other elite actors. The Advocacy Coalition Framework (Sabatier and Jenkins Smith, 1993, 1999) draws analytical attention to networks of elite actors that are motivated by shared beliefs and act primarily within substantively and geographically bound policy systems. Theories of social construction draw attention to factors such as international norms (Katzenstein, 1996), transnational advocacy networks (Keck and Sikkink, 1998) and issue frames (Snow et al., 1986; Stone, 1989), including major international development frames (McInnes and Lee, 2012), in shaping the policy process. Established frameworks developed for the express purpose of guiding health policy analysis in low- and middle-income countries and at the global level include: Walt and Gilson’s (1994) policy triangle, which directs analytical attention to roles of actors, context and process alongside policy content; Reich’s (1995) framework for analysing political dimensions of health policy reform; Shiffman’s (2007) factors influencing national policy agendas, which identifies factors related to domestic advocacy, national political environments and transnational influences; and Shiffman and Smith’s (2007) determinants of political priority, which draws attention to four analytical categories, including actor power, ideas (frames), political contexts and issue characteristics. The last of these frameworks is the only to explicitly incorporate civil society actors, suggesting grassroots mobilization that presses international and national political authorities to act may affect issue ascendance. Overall, these frameworks highlight the ways in which key actors and the dynamics of specific political contexts interact to shape the policy process. Methodology and characteristics of selected articles Following methodological guidance from Lucas et al. (2007) on narrative synthesis, Gilson (2014) and colleagues’ (Gilson and Raphaely, 2008; Berlan et al., 2014; Erasmus, 2014; Erasmus et al., 2014; Gilson et al., 2014; Walt and Gilson, 2014) work synthesizing health policy analysis research in low- and middle-income countries and Ruckert et al.’s (2016) critical review of research on global health diplomacy, the processes for the systematic search, selection and analysis, and characteristics of the articles included in this study are detailed in this section. Data collection and analysis Data are drawn from research published in peer-reviewed journals between 2007 and 2016, a period beginning with calls for systematic research (Gilson and Raphaely, 2008; Walt et al., 2008) and during which the number of articles meeting search criteria grew significantly. Topic searches were conducted in the Web of Science Core Collection (August 2017) and Scopus (September 2017) academic databases using the following terms: Health AND Policy AND civil society OR non-governmental OR non-governmental OR NGO OR activist OR activism OR transnational advocacy network OR movement OR grassroot* AND low-income OR middle-income OR ‘develop* countr*’ OR Africa OR Asia OR Latin America OR Caribbean OR Pacific OR Middle East OR East* Europe Searches were limited to articles published between 2007 and 2016 in the English language. The searches produced 410 results in Web of Science and 233 results in Scopus. Articles published in journals featuring more than one publication that met search criteria during the 10-year period were included in the next stage of the review process; the purpose was to focus on publications in journals most likely to be consulted to learn about civil society roles in health policy processes. Totals of 287 (Web of Science) and 130 (Scopus) remained. Forty-five duplicates were removed from the dataset, leaving 372 articles to screen. The author, a scholar who studies health policy processes globally and in low- and middle-income countries, then reviewed abstracts for conformance with inclusion criteria. Inclusion criteria focussed on pertinent, rigorous research and included: (1) analysed factors affecting civil society actor influence on the policy process; (2) analysed one or more stages of the policy process; (3) focussed on one or more health policy issues in one or more low- or middle-income countries; (4) empirically based, systematic research; (5) published in English. Opinion, viewpoint and largely conceptual or theoretical papers were excluded. When most criteria appeared met or indicators were present to suggest more extensive review was warranted, articles were selected for further review. Ninety-eight abstracts met the initial screening criteria. Of the 98 articles subjected to a more extensive review process, 57 were eliminated for the following reasons: civil society and its influence were little- or not-at-all analysed (20); the policy process was not directly analysed (16); neither civil society nor the policy process was analysed (6); did not report on systematic research (13); and reported on high-income countries (2). A further 17 articles focussing on the implementation stage were later eliminated due to difficulties systematically analysing and forming conclusions based on this highly diverse group of studies. In total 24 articles were deemed to fully meet inclusion criteria (full references in Table 1). Table 1. Studies and types of civil society actors analysed Study no. Reference Civil society (CS) actor type 1 Agyepong and Adjei (2008) Professional/labour association 2 Babis (2014) Social movement, professional association 3 Gómez and Harris (2016) CS broadly 4 Hirsch et al. (2015) CSOs 5 Jat et al. (2013) CSOs, CS broadly 6 Llamas and Mayhew (2016) Social movement, domestic NGOs 7 O’Brien (2015) TANs, social movements 8 Odoch et al. (2015) Domestic NGOs, faith-based, traditional leaders 9 Okello et al. (2015) Domestic NGOs, professional associations 10 Omar et al. (2010) Grassroots 11 Oronje (2013) Social movement, rights groups 12 Pedregal et al. (2015) International NGOs 13 Pelletier et al. (2012) Domestic NGOs, international NGOs 14 Pick et al. (2008) Domestic NGOs 15 Piscopo (2014) Women's movement, issue networks, TANs, faith based 16 Powers (2016) Domestic NGOs, activists, CS broadly 17 Robinson (2015) International NGOs 18 Romero and Quental (2014) Domestic NGOs, CS broadly 19 Shearer et al. (2016) Domestic NGOs 20 Smith (2014) Social movement 21 Spicer et al. (2016) Domestic NGOs, international NGOs 22 Spicer et al. (2014) Domestic NGOs, international NGOs 23 Spicer et al. (2011) CSOs 24 Tkatchenko Schmidt et al. (2008) Domestic NGOs, faith based Study no. Reference Civil society (CS) actor type 1 Agyepong and Adjei (2008) Professional/labour association 2 Babis (2014) Social movement, professional association 3 Gómez and Harris (2016) CS broadly 4 Hirsch et al. (2015) CSOs 5 Jat et al. (2013) CSOs, CS broadly 6 Llamas and Mayhew (2016) Social movement, domestic NGOs 7 O’Brien (2015) TANs, social movements 8 Odoch et al. (2015) Domestic NGOs, faith-based, traditional leaders 9 Okello et al. (2015) Domestic NGOs, professional associations 10 Omar et al. (2010) Grassroots 11 Oronje (2013) Social movement, rights groups 12 Pedregal et al. (2015) International NGOs 13 Pelletier et al. (2012) Domestic NGOs, international NGOs 14 Pick et al. (2008) Domestic NGOs 15 Piscopo (2014) Women's movement, issue networks, TANs, faith based 16 Powers (2016) Domestic NGOs, activists, CS broadly 17 Robinson (2015) International NGOs 18 Romero and Quental (2014) Domestic NGOs, CS broadly 19 Shearer et al. (2016) Domestic NGOs 20 Smith (2014) Social movement 21 Spicer et al. (2016) Domestic NGOs, international NGOs 22 Spicer et al. (2014) Domestic NGOs, international NGOs 23 Spicer et al. (2011) CSOs 24 Tkatchenko Schmidt et al. (2008) Domestic NGOs, faith based View Large Table 1. Studies and types of civil society actors analysed Study no. Reference Civil society (CS) actor type 1 Agyepong and Adjei (2008) Professional/labour association 2 Babis (2014) Social movement, professional association 3 Gómez and Harris (2016) CS broadly 4 Hirsch et al. (2015) CSOs 5 Jat et al. (2013) CSOs, CS broadly 6 Llamas and Mayhew (2016) Social movement, domestic NGOs 7 O’Brien (2015) TANs, social movements 8 Odoch et al. (2015) Domestic NGOs, faith-based, traditional leaders 9 Okello et al. (2015) Domestic NGOs, professional associations 10 Omar et al. (2010) Grassroots 11 Oronje (2013) Social movement, rights groups 12 Pedregal et al. (2015) International NGOs 13 Pelletier et al. (2012) Domestic NGOs, international NGOs 14 Pick et al. (2008) Domestic NGOs 15 Piscopo (2014) Women's movement, issue networks, TANs, faith based 16 Powers (2016) Domestic NGOs, activists, CS broadly 17 Robinson (2015) International NGOs 18 Romero and Quental (2014) Domestic NGOs, CS broadly 19 Shearer et al. (2016) Domestic NGOs 20 Smith (2014) Social movement 21 Spicer et al. (2016) Domestic NGOs, international NGOs 22 Spicer et al. (2014) Domestic NGOs, international NGOs 23 Spicer et al. (2011) CSOs 24 Tkatchenko Schmidt et al. (2008) Domestic NGOs, faith based Study no. Reference Civil society (CS) actor type 1 Agyepong and Adjei (2008) Professional/labour association 2 Babis (2014) Social movement, professional association 3 Gómez and Harris (2016) CS broadly 4 Hirsch et al. (2015) CSOs 5 Jat et al. (2013) CSOs, CS broadly 6 Llamas and Mayhew (2016) Social movement, domestic NGOs 7 O’Brien (2015) TANs, social movements 8 Odoch et al. (2015) Domestic NGOs, faith-based, traditional leaders 9 Okello et al. (2015) Domestic NGOs, professional associations 10 Omar et al. (2010) Grassroots 11 Oronje (2013) Social movement, rights groups 12 Pedregal et al. (2015) International NGOs 13 Pelletier et al. (2012) Domestic NGOs, international NGOs 14 Pick et al. (2008) Domestic NGOs 15 Piscopo (2014) Women's movement, issue networks, TANs, faith based 16 Powers (2016) Domestic NGOs, activists, CS broadly 17 Robinson (2015) International NGOs 18 Romero and Quental (2014) Domestic NGOs, CS broadly 19 Shearer et al. (2016) Domestic NGOs 20 Smith (2014) Social movement 21 Spicer et al. (2016) Domestic NGOs, international NGOs 22 Spicer et al. (2014) Domestic NGOs, international NGOs 23 Spicer et al. (2011) CSOs 24 Tkatchenko Schmidt et al. (2008) Domestic NGOs, faith based View Large In an attempt to more fully capture representations of ‘civil society’, the following terms were also searched alongside the other parameters in September 2018: third sector, independent sector, charity, charitable org*, social ent* (entrepreneurship), non-profit, community-based org*. The searches returned 63 results in Web of Science Core Collection and 86 in Scopus. The abstracts were reviewed. None met inclusion criteria, suggesting these terms are infrequently used in systematic research specific to health policy processes in low- and middle-income countries, though other areas of scholarship may employ them frequently. These results were not incorporated into overall study reporting. A standardized table including some descriptive items and several items corresponding to the research question was used to guide the collection and thematic analysis of textual data from each of the 24 articles informing this study (Table 2). A space for elaboration or for observations that did not fit into established analytical categories was also included. Once data were extracted from all articles and entered into tables (one for each article), data were synthesized using inductive and deductive approaches to inform the research question. It should be noted that though civil society is often defined by way of contrast with government and commercial actors, it was beyond the scope of this review to delve deeply into comparisons with actors in other sectors. Table 2. Data collection and analysis guide Descriptive data Author Year Title Journal title Health issue Geographic scope Author positionality: institutional location Methodological foundations Descriptive, analytical or combination Method: experiment, survey, archival analysis, history, case study etc. N= Type of data collected and how analysed Time period studied Quality and relevance of study (clear framework and methods, limitations identified) Which civil society actors receive analytical attention? Civil society actor definition and/or portrayal (e.g. NGO, faith based, etc.) Civil society actor analysis: domestic, transnational or both What frameworks are used to analyse civil society roles in the policy process? Analytical approach/framework Stage of the policy process examined What is the state of knowledge? Research question(s) Findings on civil society actor roles in the policy process Findings on civil society actor sources of power/influence Policy process outcomes affected by civil society actors Descriptive data Author Year Title Journal title Health issue Geographic scope Author positionality: institutional location Methodological foundations Descriptive, analytical or combination Method: experiment, survey, archival analysis, history, case study etc. N= Type of data collected and how analysed Time period studied Quality and relevance of study (clear framework and methods, limitations identified) Which civil society actors receive analytical attention? Civil society actor definition and/or portrayal (e.g. NGO, faith based, etc.) Civil society actor analysis: domestic, transnational or both What frameworks are used to analyse civil society roles in the policy process? Analytical approach/framework Stage of the policy process examined What is the state of knowledge? Research question(s) Findings on civil society actor roles in the policy process Findings on civil society actor sources of power/influence Policy process outcomes affected by civil society actors View Large Table 2. Data collection and analysis guide Descriptive data Author Year Title Journal title Health issue Geographic scope Author positionality: institutional location Methodological foundations Descriptive, analytical or combination Method: experiment, survey, archival analysis, history, case study etc. N= Type of data collected and how analysed Time period studied Quality and relevance of study (clear framework and methods, limitations identified) Which civil society actors receive analytical attention? Civil society actor definition and/or portrayal (e.g. NGO, faith based, etc.) Civil society actor analysis: domestic, transnational or both What frameworks are used to analyse civil society roles in the policy process? Analytical approach/framework Stage of the policy process examined What is the state of knowledge? Research question(s) Findings on civil society actor roles in the policy process Findings on civil society actor sources of power/influence Policy process outcomes affected by civil society actors Descriptive data Author Year Title Journal title Health issue Geographic scope Author positionality: institutional location Methodological foundations Descriptive, analytical or combination Method: experiment, survey, archival analysis, history, case study etc. N= Type of data collected and how analysed Time period studied Quality and relevance of study (clear framework and methods, limitations identified) Which civil society actors receive analytical attention? Civil society actor definition and/or portrayal (e.g. NGO, faith based, etc.) Civil society actor analysis: domestic, transnational or both What frameworks are used to analyse civil society roles in the policy process? Analytical approach/framework Stage of the policy process examined What is the state of knowledge? Research question(s) Findings on civil society actor roles in the policy process Findings on civil society actor sources of power/influence Policy process outcomes affected by civil society actors View Large Characteristics of selected articles Diverse issues, countries and regions are represented in the dataset. Two-thirds of the articles focus on HIV/AIDS (7), maternal health (5) and sexual or reproductive health (4, including sexuality/reproductive health education policy). Single articles focus on several health (mental health, nutrition, newborn survival, child survival, malaria, zoonotic diseases, violence against women) and health systems (national health insurance, indigenous medicine, health research system, general health policy) issues. Methodologically, most of the articles (13) report on single case studies conducted at the national or sub-national level. Nine report on multiple (two or more) case studies. Two studies use statistical methods, including one mixed methods study. Geographically, the 22 qualitative and one mixed-method study analyse 23 countries representing four regions at the national or sub-national level (Table 3). One study uses quantitative methods to analyse population policy dynamics in 42 countries in sub-Saharan Africa (Robinson, 2015). Table 3. Regions and countries represented Regions N Countries Africa 8 Burkina Faso, *Ethiopia, *Ghana, Kenya, *Nigeria, *South Africa, *Uganda, Zambia Central Asia/Eastern Europe 4 Georgia, Kyrgystan, *Russia, Ukraine South and East Asia 4 Cambodia, China, *India, Vietnam The Americas 7 Argentina, Bolivia, Brazil, Ecuador, *Mexico, Panama, Peru Total 23 Regions N Countries Africa 8 Burkina Faso, *Ethiopia, *Ghana, Kenya, *Nigeria, *South Africa, *Uganda, Zambia Central Asia/Eastern Europe 4 Georgia, Kyrgystan, *Russia, Ukraine South and East Asia 4 Cambodia, China, *India, Vietnam The Americas 7 Argentina, Bolivia, Brazil, Ecuador, *Mexico, Panama, Peru Total 23 * Indicates a country is represented in more than one study. N = 23, not including Robinson's (2015) 42-country study or Pelletier et al. (2012) beyond Peru because their other cases do not have findings pertinent to this review. View Large Table 3. Regions and countries represented Regions N Countries Africa 8 Burkina Faso, *Ethiopia, *Ghana, Kenya, *Nigeria, *South Africa, *Uganda, Zambia Central Asia/Eastern Europe 4 Georgia, Kyrgystan, *Russia, Ukraine South and East Asia 4 Cambodia, China, *India, Vietnam The Americas 7 Argentina, Bolivia, Brazil, Ecuador, *Mexico, Panama, Peru Total 23 Regions N Countries Africa 8 Burkina Faso, *Ethiopia, *Ghana, Kenya, *Nigeria, *South Africa, *Uganda, Zambia Central Asia/Eastern Europe 4 Georgia, Kyrgystan, *Russia, Ukraine South and East Asia 4 Cambodia, China, *India, Vietnam The Americas 7 Argentina, Bolivia, Brazil, Ecuador, *Mexico, Panama, Peru Total 23 * Indicates a country is represented in more than one study. N = 23, not including Robinson's (2015) 42-country study or Pelletier et al. (2012) beyond Peru because their other cases do not have findings pertinent to this review. View Large Two-thirds of articles were published in 2014 (5), 2015 (6) and 2016 (5). Half of the articles appeared in Health Policy and Planning (5), Social Science & Medicine (5) and Health Research Policy and Systems (2). Nearly half of authors were based at institutions located in high-income countries (11), though some represent institutions based in low- and middle-income countries (4) and many featured co-authors based in both (9). Results The results incorporate studies that focus on a combination of primarily domestic civil society actors (Table 1), particularly NGOs (11 studies), but also social movements (6), professional associations (3), faith-based organizations (3), civil society broadly (3), CSOs (2), rights groups (1), traditional leaders (1), activists (1) and broad grassroots demand (1). About a quarter include or focus on international actors, including NGOs (5) and TANs (2). Findings are organized under the earliest policy process stage analysed and by the degree of success civil society actors found in influencing outcomes, with close attention to the types of civil society actors studied, their sources of power and ways in which context matters. Results are synthesized along the lines of actor-centred and contextual factors in Tables 4 and 5 and in the discussion section. Table 4. Studies documenting actor-centred factors by stage Agenda setting Policy formulation Policy adoption Total studies (N = 24) Total countries (N = 23) Networks 1 Strength of expert networking (epistemic communities) 6, 8 (N=2) 8 (N=1) 7, 15, 19 (N=3) 5 6 2 Strength of relationships with governmental and non-governmental stakeholders beyond epistemic communities 3, 6, 8, 13 (N=4) 2, 4, 8 (N=3) 2, 3, 7, 14, 15, 19, 22 (N=7) 10 15 Issue frames 3 Portrayals and resonance of problems and solutions 6, 8, 23 (N=3) 6, 8 (N=2) 14, 19, 22, 23 (N=4) 6 10 4 Portrayals and resonance of major international development frames 5, 6, 11, 13 (N=4) 11 (N=1) 11 (N=1) 4 4 Resources 5 Knowledge, skills and experience 6, 8 (N=2) 2, 8, 12, 16 (N=4) 2, 7, 12, 19 (N=4) 7 8 6 Financial resources 8, 23 (N=2) 8 (N=1) 23 (N=1) 2 4 7 Evidence from pilot or demonstration projects (N=0) 12, 13 (N=2) 12, 14, 22 (N=3) 4 6 Agenda setting Policy formulation Policy adoption Total studies (N = 24) Total countries (N = 23) Networks 1 Strength of expert networking (epistemic communities) 6, 8 (N=2) 8 (N=1) 7, 15, 19 (N=3) 5 6 2 Strength of relationships with governmental and non-governmental stakeholders beyond epistemic communities 3, 6, 8, 13 (N=4) 2, 4, 8 (N=3) 2, 3, 7, 14, 15, 19, 22 (N=7) 10 15 Issue frames 3 Portrayals and resonance of problems and solutions 6, 8, 23 (N=3) 6, 8 (N=2) 14, 19, 22, 23 (N=4) 6 10 4 Portrayals and resonance of major international development frames 5, 6, 11, 13 (N=4) 11 (N=1) 11 (N=1) 4 4 Resources 5 Knowledge, skills and experience 6, 8 (N=2) 2, 8, 12, 16 (N=4) 2, 7, 12, 19 (N=4) 7 8 6 Financial resources 8, 23 (N=2) 8 (N=1) 23 (N=1) 2 4 7 Evidence from pilot or demonstration projects (N=0) 12, 13 (N=2) 12, 14, 22 (N=3) 4 6 Study numbers cross-reference with Table 1. Pelletier et al.’s (2012) study (#13) is counted as only one country because the only pertinent findings were from Peru. View Large Table 4. Studies documenting actor-centred factors by stage Agenda setting Policy formulation Policy adoption Total studies (N = 24) Total countries (N = 23) Networks 1 Strength of expert networking (epistemic communities) 6, 8 (N=2) 8 (N=1) 7, 15, 19 (N=3) 5 6 2 Strength of relationships with governmental and non-governmental stakeholders beyond epistemic communities 3, 6, 8, 13 (N=4) 2, 4, 8 (N=3) 2, 3, 7, 14, 15, 19, 22 (N=7) 10 15 Issue frames 3 Portrayals and resonance of problems and solutions 6, 8, 23 (N=3) 6, 8 (N=2) 14, 19, 22, 23 (N=4) 6 10 4 Portrayals and resonance of major international development frames 5, 6, 11, 13 (N=4) 11 (N=1) 11 (N=1) 4 4 Resources 5 Knowledge, skills and experience 6, 8 (N=2) 2, 8, 12, 16 (N=4) 2, 7, 12, 19 (N=4) 7 8 6 Financial resources 8, 23 (N=2) 8 (N=1) 23 (N=1) 2 4 7 Evidence from pilot or demonstration projects (N=0) 12, 13 (N=2) 12, 14, 22 (N=3) 4 6 Agenda setting Policy formulation Policy adoption Total studies (N = 24) Total countries (N = 23) Networks 1 Strength of expert networking (epistemic communities) 6, 8 (N=2) 8 (N=1) 7, 15, 19 (N=3) 5 6 2 Strength of relationships with governmental and non-governmental stakeholders beyond epistemic communities 3, 6, 8, 13 (N=4) 2, 4, 8 (N=3) 2, 3, 7, 14, 15, 19, 22 (N=7) 10 15 Issue frames 3 Portrayals and resonance of problems and solutions 6, 8, 23 (N=3) 6, 8 (N=2) 14, 19, 22, 23 (N=4) 6 10 4 Portrayals and resonance of major international development frames 5, 6, 11, 13 (N=4) 11 (N=1) 11 (N=1) 4 4 Resources 5 Knowledge, skills and experience 6, 8 (N=2) 2, 8, 12, 16 (N=4) 2, 7, 12, 19 (N=4) 7 8 6 Financial resources 8, 23 (N=2) 8 (N=1) 23 (N=1) 2 4 7 Evidence from pilot or demonstration projects (N=0) 12, 13 (N=2) 12, 14, 22 (N=3) 4 6 Study numbers cross-reference with Table 1. Pelletier et al.’s (2012) study (#13) is counted as only one country because the only pertinent findings were from Peru. View Large Table 5. Studies documenting domestic and transnational contextual factors by stage Agenda setting Policy formulation Policy adoption Total studies (N = 24) Total countries (N = 23 + 42a) Domestic political context 1 Openness vs repressiveness of governments 3, 23 (N=2) 4 (N=1) 3, 7, 21, 23 (N=4) 5 11 2 Level of support vs opposition from societal leaders, such as traditional, religious and law enforcement leaders (N=0) 8 (N=1) 8, 15, 24 (N=3) 3 3 3 Consultative norms and practices of political and administrative bodies 1, 18 (N=2) 1, 4, 16, 18 (N=4) 1, 18 (N=2) 4 4 4 Strength of social movement politics 6, 20 (N=2) 2, 20 (N=2) 2, 20 (N=2) 3 3 5 Level of social acceptance surrounding the issue 10 (N=1) (N=0) 14, 24 (N=2) 3 6 6 Level of grassroots demand, including by affected groups 10 (N=1) (N=0) 24 (N=1) 2 5 7 Level and type of media coverage 5 (N=1) (N=0) (N=0) 1 1 Domestic policy environment 8 Consistency with existing national policies, such as strategic plans and laws 5 (N=1) 2 (N=1) 2, 21, 22, 24 (N=4) 5 5 9 Health system strength, including leadership stability 18, 20, 23 (N=3) 18, 20 (N=2) 18, 20, 21, 23 (N=4) 4 7 10 Level of domestic investment 18 (N=1) 18 (N=1) 18, 21, 24 (N=3) 3 5 11 Prior implementation experience 1 (N=1) 1 (N=1) 1, 19, 22 (N=3) 3 5 Domestic leadership 12 High-level entrepreneurs/champions 10, 13, 20 (N=3) 20 (N=1) 15, 20, 22 (N=3) 5 8 13 Consistency with existing political commitments (as made by candidates, parties and leaders) 1, 20 (N=2) 1, 2, 20 (N=3) 1, 2, 14, 15, 20, 21 (N=4) 6 7 14 Windows of opportunity presented by leadership/regime change 5, 6, 11 (N=3) 11 (N=1) 11, 19 (N=2) 4 4 Domestic issue frames 15 Knowledge and resonance of the problem's magnitude and feasibility of solutions 5, 8, 10, 13 (N=4) 8 (N=1) 19, 24 (N=2) 6 8 16 Competing portrayals of the issue and its solutions 8, 11 (N=2) 8, 11 (N=2) 8, 11 (N=2) 2 2 Transnational factors 17 Level of international agency/donor prioritization 1, 8, 9, 10 (N=4) 1, 8, 9, 16 (N=4) 1, 9, 19, 21, 24 (N=5) 8 9 18 Degree of dependence on international donors for resources 9 (N=1) 4, 9, 16 (N=3) 9, 19 (N=2) 4 5 19 International norm resonance 5, 9 (N=2) 2 (N=1) 2, 7, 17 (N=3) 5 5(+42a) 20 Policy diffusion dynamics 8 (N=1) 8 (N=1) 8 (N=1) 1 1 Agenda setting Policy formulation Policy adoption Total studies (N = 24) Total countries (N = 23 + 42a) Domestic political context 1 Openness vs repressiveness of governments 3, 23 (N=2) 4 (N=1) 3, 7, 21, 23 (N=4) 5 11 2 Level of support vs opposition from societal leaders, such as traditional, religious and law enforcement leaders (N=0) 8 (N=1) 8, 15, 24 (N=3) 3 3 3 Consultative norms and practices of political and administrative bodies 1, 18 (N=2) 1, 4, 16, 18 (N=4) 1, 18 (N=2) 4 4 4 Strength of social movement politics 6, 20 (N=2) 2, 20 (N=2) 2, 20 (N=2) 3 3 5 Level of social acceptance surrounding the issue 10 (N=1) (N=0) 14, 24 (N=2) 3 6 6 Level of grassroots demand, including by affected groups 10 (N=1) (N=0) 24 (N=1) 2 5 7 Level and type of media coverage 5 (N=1) (N=0) (N=0) 1 1 Domestic policy environment 8 Consistency with existing national policies, such as strategic plans and laws 5 (N=1) 2 (N=1) 2, 21, 22, 24 (N=4) 5 5 9 Health system strength, including leadership stability 18, 20, 23 (N=3) 18, 20 (N=2) 18, 20, 21, 23 (N=4) 4 7 10 Level of domestic investment 18 (N=1) 18 (N=1) 18, 21, 24 (N=3) 3 5 11 Prior implementation experience 1 (N=1) 1 (N=1) 1, 19, 22 (N=3) 3 5 Domestic leadership 12 High-level entrepreneurs/champions 10, 13, 20 (N=3) 20 (N=1) 15, 20, 22 (N=3) 5 8 13 Consistency with existing political commitments (as made by candidates, parties and leaders) 1, 20 (N=2) 1, 2, 20 (N=3) 1, 2, 14, 15, 20, 21 (N=4) 6 7 14 Windows of opportunity presented by leadership/regime change 5, 6, 11 (N=3) 11 (N=1) 11, 19 (N=2) 4 4 Domestic issue frames 15 Knowledge and resonance of the problem's magnitude and feasibility of solutions 5, 8, 10, 13 (N=4) 8 (N=1) 19, 24 (N=2) 6 8 16 Competing portrayals of the issue and its solutions 8, 11 (N=2) 8, 11 (N=2) 8, 11 (N=2) 2 2 Transnational factors 17 Level of international agency/donor prioritization 1, 8, 9, 10 (N=4) 1, 8, 9, 16 (N=4) 1, 9, 19, 21, 24 (N=5) 8 9 18 Degree of dependence on international donors for resources 9 (N=1) 4, 9, 16 (N=3) 9, 19 (N=2) 4 5 19 International norm resonance 5, 9 (N=2) 2 (N=1) 2, 7, 17 (N=3) 5 5(+42a) 20 Policy diffusion dynamics 8 (N=1) 8 (N=1) 8 (N=1) 1 1 Study numbers cross-reference with Table 1. Pelletier et al.’s (2012) study (#13) is counted as only one country because the only pertinent findings were from Peru. a The 42-country analysis by Robinson (2015, study #17). View Large Table 5. Studies documenting domestic and transnational contextual factors by stage Agenda setting Policy formulation Policy adoption Total studies (N = 24) Total countries (N = 23 + 42a) Domestic political context 1 Openness vs repressiveness of governments 3, 23 (N=2) 4 (N=1) 3, 7, 21, 23 (N=4) 5 11 2 Level of support vs opposition from societal leaders, such as traditional, religious and law enforcement leaders (N=0) 8 (N=1) 8, 15, 24 (N=3) 3 3 3 Consultative norms and practices of political and administrative bodies 1, 18 (N=2) 1, 4, 16, 18 (N=4) 1, 18 (N=2) 4 4 4 Strength of social movement politics 6, 20 (N=2) 2, 20 (N=2) 2, 20 (N=2) 3 3 5 Level of social acceptance surrounding the issue 10 (N=1) (N=0) 14, 24 (N=2) 3 6 6 Level of grassroots demand, including by affected groups 10 (N=1) (N=0) 24 (N=1) 2 5 7 Level and type of media coverage 5 (N=1) (N=0) (N=0) 1 1 Domestic policy environment 8 Consistency with existing national policies, such as strategic plans and laws 5 (N=1) 2 (N=1) 2, 21, 22, 24 (N=4) 5 5 9 Health system strength, including leadership stability 18, 20, 23 (N=3) 18, 20 (N=2) 18, 20, 21, 23 (N=4) 4 7 10 Level of domestic investment 18 (N=1) 18 (N=1) 18, 21, 24 (N=3) 3 5 11 Prior implementation experience 1 (N=1) 1 (N=1) 1, 19, 22 (N=3) 3 5 Domestic leadership 12 High-level entrepreneurs/champions 10, 13, 20 (N=3) 20 (N=1) 15, 20, 22 (N=3) 5 8 13 Consistency with existing political commitments (as made by candidates, parties and leaders) 1, 20 (N=2) 1, 2, 20 (N=3) 1, 2, 14, 15, 20, 21 (N=4) 6 7 14 Windows of opportunity presented by leadership/regime change 5, 6, 11 (N=3) 11 (N=1) 11, 19 (N=2) 4 4 Domestic issue frames 15 Knowledge and resonance of the problem's magnitude and feasibility of solutions 5, 8, 10, 13 (N=4) 8 (N=1) 19, 24 (N=2) 6 8 16 Competing portrayals of the issue and its solutions 8, 11 (N=2) 8, 11 (N=2) 8, 11 (N=2) 2 2 Transnational factors 17 Level of international agency/donor prioritization 1, 8, 9, 10 (N=4) 1, 8, 9, 16 (N=4) 1, 9, 19, 21, 24 (N=5) 8 9 18 Degree of dependence on international donors for resources 9 (N=1) 4, 9, 16 (N=3) 9, 19 (N=2) 4 5 19 International norm resonance 5, 9 (N=2) 2 (N=1) 2, 7, 17 (N=3) 5 5(+42a) 20 Policy diffusion dynamics 8 (N=1) 8 (N=1) 8 (N=1) 1 1 Agenda setting Policy formulation Policy adoption Total studies (N = 24) Total countries (N = 23 + 42a) Domestic political context 1 Openness vs repressiveness of governments 3, 23 (N=2) 4 (N=1) 3, 7, 21, 23 (N=4) 5 11 2 Level of support vs opposition from societal leaders, such as traditional, religious and law enforcement leaders (N=0) 8 (N=1) 8, 15, 24 (N=3) 3 3 3 Consultative norms and practices of political and administrative bodies 1, 18 (N=2) 1, 4, 16, 18 (N=4) 1, 18 (N=2) 4 4 4 Strength of social movement politics 6, 20 (N=2) 2, 20 (N=2) 2, 20 (N=2) 3 3 5 Level of social acceptance surrounding the issue 10 (N=1) (N=0) 14, 24 (N=2) 3 6 6 Level of grassroots demand, including by affected groups 10 (N=1) (N=0) 24 (N=1) 2 5 7 Level and type of media coverage 5 (N=1) (N=0) (N=0) 1 1 Domestic policy environment 8 Consistency with existing national policies, such as strategic plans and laws 5 (N=1) 2 (N=1) 2, 21, 22, 24 (N=4) 5 5 9 Health system strength, including leadership stability 18, 20, 23 (N=3) 18, 20 (N=2) 18, 20, 21, 23 (N=4) 4 7 10 Level of domestic investment 18 (N=1) 18 (N=1) 18, 21, 24 (N=3) 3 5 11 Prior implementation experience 1 (N=1) 1 (N=1) 1, 19, 22 (N=3) 3 5 Domestic leadership 12 High-level entrepreneurs/champions 10, 13, 20 (N=3) 20 (N=1) 15, 20, 22 (N=3) 5 8 13 Consistency with existing political commitments (as made by candidates, parties and leaders) 1, 20 (N=2) 1, 2, 20 (N=3) 1, 2, 14, 15, 20, 21 (N=4) 6 7 14 Windows of opportunity presented by leadership/regime change 5, 6, 11 (N=3) 11 (N=1) 11, 19 (N=2) 4 4 Domestic issue frames 15 Knowledge and resonance of the problem's magnitude and feasibility of solutions 5, 8, 10, 13 (N=4) 8 (N=1) 19, 24 (N=2) 6 8 16 Competing portrayals of the issue and its solutions 8, 11 (N=2) 8, 11 (N=2) 8, 11 (N=2) 2 2 Transnational factors 17 Level of international agency/donor prioritization 1, 8, 9, 10 (N=4) 1, 8, 9, 16 (N=4) 1, 9, 19, 21, 24 (N=5) 8 9 18 Degree of dependence on international donors for resources 9 (N=1) 4, 9, 16 (N=3) 9, 19 (N=2) 4 5 19 International norm resonance 5, 9 (N=2) 2 (N=1) 2, 7, 17 (N=3) 5 5(+42a) 20 Policy diffusion dynamics 8 (N=1) 8 (N=1) 8 (N=1) 1 1 Study numbers cross-reference with Table 1. Pelletier et al.’s (2012) study (#13) is counted as only one country because the only pertinent findings were from Peru. a The 42-country analysis by Robinson (2015, study #17). View Large Agenda setting Of the 12 studies focussing on the agenda setting stage of the policy process, seven address sexual and reproductive (including maternal) health and HIV/AIDS, finding that civil society organizations were mostly successful in drawing some attention and/or resources to the issues. Political repression and other constraints limited opportunities for civil society to influence the health policy process. Successes Three studies of agenda setting for sexual and reproductive health documented successes. Jat et al. (2013) employed Kingdon’s (1995) multiple streams framework, finding that domestic CSOs framed maternal mortality as a violation of human rights in prominent public forums, facilitating the emergence of the issue as a political priority nationally and in Madhya Pradesh, India. Facilitating factors in the national political context included growing evidence of the problem’s magnitude and of effective solutions, a political leadership transition, launch of a major national rural health program, the country’s desire to exhibit global leadership, international norms (the United Nations Millennium Development Goals) and increasing media coverage, with spillover effects to the state level. Also employing the multiple streams framework, Llamas and Mayhew (2016) found that a knowledge-based health policy community involving indigenous actors and a range of domestic NGOs used its expertise to identify maternal health problems, frame them as issues of indigenous rights, formulate a feasible solution and consolidate support to advance adoption of a vertical birth policy in Otavalo, Ecuador. Agenda setting was facilitated by a new government’s support for the indigenous rights movement, which opened a window of opportunity to promote intercultural health policies. In another study with framing as a central analytical concept, Oronje (2013) found maternal health frames and technical actors to be more effective than rights frames and organizations in the processes of agenda setting, formulation and adoption of a 2007 sexual and reproductive health policy in Kenya. While political leadership transitions shaped opportunities to advance policy, the dominance of medical and population experts in Kenya’s health policy subsystem tempered the effects of human rights frames and organizations. Turning to other issue areas, Odoch et al. (2015) drew on concepts of power, the policy triangle and multiple streams frameworks, finding that alongside international agencies and donors highly committed civil society actors were among the main drivers of agenda setting and policy formulation for male circumcision for HIV prevention in Uganda. Their primary sources of power included expertise (university-based researchers), financial resources (NGOs) and networking abilities (both). Increasing evidence of the problem and male circumcision’s preventive efficacy helped to convince the health ministry to support the policy. Addressing concerns of traditional and religious leaders, a shift in issue frames (from medical to safe circumcision) helped to bring about support from some who wanted to play a role in offering the procedure. Prior adoption of male circumcision in other sub-Saharan African countries was also facilitative. In a study focussing on under-nutrition, Pelletier et al. (2012) found that a coalition of international NGOs and UN agencies used frames linking the issue with socio-economic disparities to positively influence the issue’s agenda status and gain the president’s support in Peru. Evidence that rates of under-nutrition were no longer declining was also influential. They also found that NGO demonstration projects shaped policy formulation; however, they did not find broader domestic civil society mobilization to be a factor. The 17-country study by Pelletier et al. (2012) did not systematically report on civil society influence across the cases. Mixed results Two comparative studies document mixed results. Smith’s (2014) study of maternal health policymaking in South India found that Tamil Nadu’s non-Brahmin social movement shaped the state’s political context, leading the major political parties to compete to advance social policies. This, along with support from a relatively strong public health system and its leaders, facilitated policymaking on maternal health. Featuring no such unifying movement, policy progress was hindered in neighbouring Karnataka by a lack of leadership and a relatively weak public health system. Gómez and Harris (2016) used the lens of political repression to analyse government relationships with civil society and their responses to AIDS (as indicated by antiretroviral treatment scale-up) in Brazil, Russia, India, China and South Africa (BRICS), hypothesizing that the more repressive governments would be less responsive. They found that strong state-civil society relationships in Brazil contributed to a robust response. Antagonistic state-civil society relations in South Africa and highly bureaucratic systems of governance and resistance to civil society partnership in India delayed responses in both countries. Even stronger resistance and antagonism contributed to an ineffective response in Russia. China, despite its tenuous engagement with civil society, was relatively responsive. Limited influence Three studies spanning a range of issues and countries speak to the constraints posed when governments limit opportunities for civil society engagement. For instance, Romero and Quental (2014) found that civil society, primarily referring to domestic NGOs, was under-represented in the process for setting national health research priorities in Panama between 2006 and 2011, limiting the strength of the agenda that was set. The lack of a structured system, low national investment in health research and limited opportunities for public consultation, review and reinforcement were also constraining factors. In Ghana, political commitments to replace out-of-pocket fees, some limited experience with social health insurance and support from international agencies and donors prompted a new administration to set up a task force to advise the Ministry of Health on the development of a national health insurance scheme in 2001 (Agyepong and Adjei, 2008). When the final bill (developed by a small group of elite actors without public consultation) came before the parliament in 2003, organized labour groups protested, delaying passage of the bill but ultimately having no influence on its content or passage. Spicer et al.’s (2011) comparative study of agenda setting for and adoption of policies related to HIV/AIDS and illicit drugs in three countries in Central Asia and Eastern Europe, including Georgia, Kyrgystan and Ukraine, found that governments that sought to limit the power of CSOs (representing a broad set of actors), public sector inertia, political instability and frequent leadership turnover constrained civil society influence. Competition for resources, lack of consensus on intervention strategies and fears of government reprisals also limited CSO power. Different dynamics were at work in two comparative studies set in sub-Saharan Africa. One indicated that low grassroots demand combined with stigma, a lack of information and low prioritization by donors and domestic authorities to contribute to mental health’s low agenda status in Ghana, South Africa, Uganda and Zambia (Omar et al., 2010). And a study of the policy process for zoonotic diseases in Nigeria and Uganda found that though NGOs and local farmers’ associations provide some input through steering committees, highly resourced international actors and global mandates drive policymaking, crowding out civil society voices (Okello et al., 2015). Donor dependence weighed heavily in Uganda. And a lack of platforms for addressing the intersection of animal and human diseases was a barrier in both countries. Policy formulation Four historical case studies focus on policy formulation, with three documenting civil society successes in shaping content and an ethnographic study finding a lack of influence. Successes The rise of indigenous movements in Latin America, Evo Morales’s election as Bolivia’s first indigenous president of the post-colonial era, the subsequent adoption of a new constitution giving explicit rights to indigenous people and growing international acceptance of alternative medicine provided a supportive political context for indigenous health policy promotion (Babis, 2014). Indigenous health and professional organizations worked with the government, leveraging power derived from their expertise to advocate for a 2013 law integrating indigenous medicine into the National Health System, to identify treatments for inclusion and to shape licensing requirements. In contrast, Pedregal et al. (2015) found that in the absence of domestic civil society mobilization, an international NGO had used its strong technical expertise to pilot a social health insurance program that the Cambodian government later embraced through legislation. In neighbouring Vietnam, despite a legal environment restricting the establishment of CSOs and a large degree of state control over them, CSOs engaged in community-based HIV/AIDS programming and advocacy helped shape the formulation of the country’s 2006 Law on HIV/AIDS Prevention and Control (Hirsch et al., 2015). Representing civil society, former government officials, intellectuals and activists used their connections with government officials and extensive consultations to influence policy formulation. International donors and health initiatives (particularly PEPFAR and the Global Fund) bolstered civil society influence by providing support to the government that was contingent on CSO involvement. Limited influence Civil society, including domestic NGOs and activists, in Western Cape Province, South Africa, had little direct impact on the formulation of provincial guidelines for implementing the National Strategic Plan for HIV/AIDS and STIs (Powers, 2016). They were limited by a lack of technical knowledge and direct access to the consultative process for developing the guidelines. Powers (2016) found that consultations had been held for purposes of acquiring future financial support from the Global Fund rather than to gain genuine input to the guidelines. The study offers a critique of international health initiative influence on the policy process. Policy adoption Of the eight studies focussing on policy adoption, two employing statistical analysis and four case studies show positive results. The remaining case studies document mixed results. Successes Robinson’s (2015) statistical analysis of population policy adoption in 42 countries in sub-Saharan Africa showed that normative ties (relationships with international NGOs) were not significant prior to the 1994 International Conference on Population and Development (which helped to establish a global right to reproductive health), but they were after. The study suggests that international NGOs are more likely to influence health policy when supportive norms are established. A mixed methods study of subnational policies addressing violence against women in Mexico and Nigeria found that transnational feminist networks (a type of TAN) diffused international norms and drew strength from legal expertise and relationships with political elites to affect policy change. O’Brien (2015, p. 289) concludes that, ‘Issue-specific expert networking improves the likelihood of subnational policy responsiveness to an international norm on violence against women in developing and democratizing nations, where a weak state capacity and repressive contexts can hinder local activism’. Three studies suggest civil society–government partnership and issue alignment with existing political priorities are facilitative. In Mexico, a domestic NGO that developed a sexuality education program and documented its efficacy effectively partnered with the Ministry of Education, resulting in national level adoption and scale-up of the program (Pick et al., 2008). Alignment of the program with political priority for preventing teen pregnancy and documenting societal approval were facilitating factors. A study of factors influencing government adoption and scale-up of maternal and newborn health innovations in Ethiopia, Nigeria and Uttar Pradesh, India, suggests NGOs and other implementers are more likely to see success when they: (1) involve government in all phases of pilots or demonstration projects; (2) gain support from high-profile champions, such as senior officials and first ladies; (3) align with existing policy priorities; (4) use evidence to message effectively; (5) and advocate on an ongoing basis with the pertinent authorities (Spicer et al., 2014). A companion study (Spicer et al., 2016) focussed on contextual factors influencing government decisions to adopt and finance maternal and newborn health policy innovations at scale in the same settings found several factors related to the ways in which policies are made (e.g. responsiveness to civil society, evidence-based decision-making and turnover of officials), issue prioritization (e.g. national policy frameworks, economic resources and powerful international and country actor priorities), and development partner harmonization (e.g. communication and mechanisms) were influential. Both studies suggested that the Uttar Pradesh and Nigerian governments were increasingly responsive to civil society, while the Ethiopian government was less open but not completely closed. Shearer et al. (2016) employed process tracing methods to investigate reasons for policy reform across three health issues—HIV/AIDS, malaria and child health—in Burkina Faso. Donor rules requiring that civil society actors play significant roles alongside the governments their grants supported brought new actors and ideas into the child health and malaria policy reform processes, resulting in adoption of home management of malaria and community case management of childhood illness reforms. In the HIV/AIDS case, a cohesive and experienced civil society network and a leadership change contributed to the adoption of a policy removing user fees for antiretroviral treatment in 2009. In the malaria and HIV/AIDS cases, legacies of home management and civil society participation in treatment provision were also facilitative. In the child health and HIV/AIDS cases, evidence demonstrating failures of existing approaches and successes with the proposed approaches were facilitative, while technical expertise was facilitative in the malaria case. Mixed results Piscopo (2014) used historical process tracing methods to identify factors shaping early failures and later successes in sexual health policy reform in Argentina, asking what if any difference female legislators made. Policy gains were facilitated by female legislators, but also depended on strong coalitions of civil society actors and the support of high-level leaders (which varied across administrations). The Catholic Church played an obstructive role throughout, helping to defeat reform efforts in the 1990s. In Russia, the Ministry of Health and international donors were supporting dozens of harm reduction pilot projects to address the HIV/AIDS crisis among injecting drug users by the mid-2000s. Civil society and government actors identified six primary constraints to the further adoption and scale-up of harm reduction policies (Tkatchenko et al., 2008): limited financial resources and contextually relevant data on intervention effectiveness; perceptions that harm reduction was culturally unacceptable; opposition from the Russian Orthodox Church and law enforcement agencies; unclear legal regulations; and resistance from injecting drug users. Discussion This study draws upon a subset of health policy analysis scholarship, focussing on a diverse set of civil society actors and their power to influence the pre-implementation stages of the health policy process in low- and middle-income countries. Though the set of studies is limited in scope, many of the same factors are documented as affecting the power of a range of civil society actors to exert influence in the policy process across a range of national settings. Mirroring contributions of the policy frameworks that frame this synthesis, the 24 articles analysed highlight the ways in which actor-centred factors (Table 4) and the dynamics of specific political contexts (Table 5) interact to shape the policy process. These factors, study limitations and directions for future research are discussed in this and the concluding section. Contextual factors As suggested in reviews of the broader health policy analysis terrain in low- and middle-income countries (Gilson and Raphaely, 2008; Walt et al., 2008), and by several pertinent analytical frameworks (including but not limited to the policy triangle, multiple streams and factors influencing national agendas) and theories of social construction (Snow et al., 1986; Stone, 1989; Katzenstein, 1996; Keck and Sikkink, 1998), a range of contextual domestic and transnational factors are found to shape the likelihood of civil society influence (Table 5). In the domestic political context (factors 1–7), the degree of openness vs repressiveness of governments received the most analytical attention and empirical support, with five studies encompassing 11 country contexts suggesting that repressive governments limit the likelihood of civil society influence across all three pre-implementation stages. Other limiting factors include: opposition from societal leaders, including those representing traditional, religious and law enforcement institutions (three studies/three countries); exclusion from consultative processes (four studies/four countries); and low social acceptance (three studies/six countries) and grassroots demand (two studies/five countries). In contrast, high social acceptance facilitated sexuality education policy adoption in Mexico (Pick et al., 2008). Three studies documented the facilitative effects of broader, rights-based social movements. And a study in India suggested increasing media coverage paved the way for civil society to influence health agenda setting (Jat et al., 2013). Several factors in the domestic policy environment appear to make a difference across the three stages (factors 8–11). Consistency of proposals with existing national policies (five studies/five countries) and prior implementation experience (three studies/five countries) may increase opportunities for civil society influence. Weak health systems (four studies/seven countries) and low levels of domestic investment in an issue (three studies/five countries) may decrease the likelihood civil society actors will make a significant difference. Domestic leadership (factors 12–14) is associated with civil society influence, with support from high-level champions (five studies, eight countries), windows of opportunity presented by regime change (four studies/four countries) and consistency of issues with existing political commitments (six studies/seven countries) found facilitative. Domestic issue frames (factors 15 and 16) also play a role, with knowledge of a problem’s magnitude and feasible solutions supportive during agenda setting (four studies/six countries) and other stages (three studies/five countries). Competing portrayals of issues and their solutions pose challenges, limiting influence from rights organizations on sexual and reproductive health policymaking in Kenya (Oronje, 2013) and on a male circumcision proposal in Uganda until a compromise frame that satisfied key religious leaders was identified (Odoch et al., 2015). Several transnational factors have the potential to affect domestic policy processes (factors 17–20). Issue prioritization by donors and international agencies is found facilitative while neglect is a hindering factor across the stages (eight studies/nine countries). Four studies with findings in a similar vein but taking a more specific economic interest-based perspective also offer insights. Two studies found that donor dependence dynamics crowded out civil society influence (Okello et al., 2015; Powers, 2016). In two other studies, donor requirements that governments give civil society actors a seat at the table increased their influence (Hirsch et al., 2015; Shearer et al., 2016). Five studies, including a 42-country study of population policy adoption in sub-Saharan Africa (Robinson, 2015), suggest that when international norms resonate in national contexts, the civil society actors advancing synergistic issues are more likely to become influential. Lastly, one study suggested policy diffusion dynamics (True and Mintrom, 2001) were at work, with earlier adoption by peer nations facilitating later adoption of a male circumcision policy in Uganda (Odoch et al., 2015). Actor-centred factors Civil society actors lack the authoritative power of governments and the economic power of corporations, but they gain influence in the pre-implementation stages of the policy process via certain kinds of relationships, resources and framing strategies (Table 4). Strong expert networks (factors 1 and 2) or epistemic communities (Haas, 1992), are likely to exert more influence than weaker ones that disagree about intervention strategies, for instance (five studies/six countries). The strongest actor-centred finding concerns the power civil society actors derive from forming and maintaining strong relationships with governmental and other stakeholders, with 10 studies covering 15 countries suggesting it is important to expand beyond epistemic networking to affect change outside technical circles. Civil society actors also gain strength via issue frames (factors 3 and 4) that effectively convey the efficacy and feasibility of health policy solutions (6 studies/10 countries). Interestingly, this group of studies does not report findings on problem portrayals as a source of actor strength, though problem frames appear as a contextual factor. This may be because problem frames are more important in motivating concerned actors to adopt health policy issues during a pre-agenda setting stage unanalysed by this set of studies (Carpenter, 2007) and solution frames are more important to concerned actors in later stages. In addition, as found in the global health arena (McInnes and Lee, 2012), the studies show that civil society actors may gain influence by employing major international development frames, such as human (including indigenous) rights (Jat et al., 2013; Babis, 2014; Llamas and Mayhew, 2016), though rights frames were not successfully employed in all cases (Oronje, 2013). Lastly, in terms of resources (factors 5–7), strong and relevant knowledge, skills and experience were found to be sources of power while weaknesses diminished civil society actor influence (seven studies/eight countries). The same dynamics were documented with respect to financial strengths and weaknesses (two studies/four countries). The ability to provide evidence from pilot or demonstration projects proved influential in the policy formulation and adoption stages (four studies/six countries). Conclusion This study reviews a subset of health policy analysis research in low- and middle-income countries over a recent decade, with aims to synthesize findings and inform future research. The study has several limitations. It is confined to analysis of 24 peer-reviewed research articles published between 2007 and 2016 that met search and selection criteria. It does not incorporate the broader body of articles, books, book chapters, grey literature, reports or theoretical contributions that exists on the topic of civil society and its relationship to health policymaking. Thus, some pertinent work is likely under-represented, with historical contributions from scholarship on HIV/AIDS and tobacco control efforts perhaps the most notable. In addition, the study incorporates analysis of a diverse array of civil society actors. These actors may be more meaningfully distinguished in a study with a larger sample size—this would likely mean incorporating work beyond that focussing on the health policy process in low- and middle-income countries given the limited extent to which the >400 studies evaluated for this endeavour systematically analysed civil society actors and causal dynamics. Some factors and dynamics are likely under-represented in this study, such as issue adoption in a pre-agenda setting stage (Carpenter, 2007), policy entrepreneurs (Kingdon, 1995), implications of stigma (Omar et al., 2010), grassroots demand (Shiffman and Smith, 2007; Parker, 2011) and domestic economic interests (Reich, 1995; Author, 2018). A longer timeframe, less restrictive inclusion criteria and other synthesis methods might yield more robust results. Nonetheless, the synthesis offers insights to several actor-centred and contextual factors affecting the power of civil society actors to influence health policy processes in low- and middle-income countries. The most strongly supported are consistent with some of the most prominent policy analysis frameworks, including the multiple streams (Kingdon, 1995), advocacy coalition (Sabatier and Jenkins Smith, 1993, 1999), policy triangle (Walt and Gilson, 1994), factors influencing national policy agendas (Shiffman, 2007) and determinants of political priority (Shiffman and Smith, 2007) frameworks. The studies suggest that civil society actors are more likely to influence policy processes when: They are resourced in terms of technical expertise, financial resources and practical implementation experience. They form or join strong epistemic networks and broader coalitions. They frame issues in alignment with existing policies and political priorities. And, in the absence of governing structures and norms that facilitate their participation, powerful actors intervene to provide civil society actors a seat at the table. The second through fourth points emphasize the centrality of contextual factors. Civil society actors must be granted some degree of access to participate in national and subnational policy processes. Those that form common cause with well-positioned stakeholders and tailor their strategies to the dynamics of specific political contexts increase their chances of shaping the policy process and its outcomes. This study raises several questions for future research that draw attention to the ways in which agency and agency-structural dynamics shape the policy process in low- and middle-income countries: Given the diverse array of actors representing civil society, how do differing interests and ideas among them affect their power and influence? How does contestation among civil society actors and other stakeholders affect the health policy process and its outcomes? How does network composition affect civil society voice in the policy process? Is participation in strong epistemic networks and broader coalitions equally important across all stages of the policy process? Do civil society actors privilege some types of issues for adoption over others? How likely are they to adopt and influence policies on non-communicable diseases (ovarian cancer), risk factors (obesity), interventions (micronutrients) and health systems issues (universal health care) compared with rights-based issues such as HIV/AIDS and reproductive health? Why and with what implications (Carpenter, 2007)? In repressive contexts, what are potential pathways to civil society influence? How effective (and desirable) is the donor intervention model documented by Hirsch et al. (2015) and Shearer et al. (2016)? Is the Vietnamese NGO model of building expertise, credibility and organizational capacity (Pallas and Nguyen, 2018) applicable to other contexts? Do international actors and funding play more significant roles in empowering or crowding out domestic civil society actors and their voice in the policy process? What are the implications for the outcomes of health policy processes in low- and middle-income countries? Scholars and practitioners should also be looking to the intersection of civil society and media. How, during which stages of the policy process and to what effects are civil society actors leveraging advances in traditional and social media technologies? What is effective and what are the key barriers to employing these potentially democratizing technological advances? Comparative case studies, medium-N and large-N studies like those included in this review (see especially Smith, 2014; Robinson, 2015; Gómez and Harris, 2016; Shearer et al., 2016) that analyse reasons for differing processes and outcomes across contexts, issues and time would be particularly helpful in sorting out answers to these and other important questions. Acknowledgements I would like to thank Hans Peter Schmitz, Sabith Khan and anonymous reviewers for their insightful comments and suggestions on unpublished drafts of this manuscript. Errors and limitations are mine alone. Conflict of interest statement. None declared. References Agyepong IA , Adjei S. 2008 . Public social policy development and implementation: a case study of the Ghana National Health Insurance scheme . Health Policy and Planning 23 : 150 – 160 . Google Scholar Crossref Search ADS PubMed Babis D. 2014 . The role of civil society organizations in the institutionalization of indigenous medicine in Bolivia . Social Science & Medicine 123 : 287 – 294 . Google Scholar Crossref Search ADS Berlan D , Buse K , Shiffman J , Tanaka S. 2014 . The bit in the middle: a synthesis of global health literature on policy formulation and adoption . Health Policy and Planning 29 : iii23 – 34 . Google Scholar Crossref Search ADS PubMed Birdsall K , Kelly K. 2007 . Pioneers, Partners, Providers: The Dynamics of Civil Society and AIDS Funding in Southern Africa. Johannesburg: CADRE/OSISA. https://hivhealthclearinghouse.unesco.org/library/documents/pioneers-partners-providers-dynamics-civil-society-and-aids-funding-southern, accessed 19 December 2018. Carpenter RC. 2007 . Setting the advocacy agenda: theorizing issue emergence and nonemergence in transnational advocacy networks . International Studies Quarterly 51 : 99 – 120 . Google Scholar Crossref Search ADS Erasmus E. 2014 . The use of street-level bureaucracy theory in health policy analysis in low- and middle-income countries: a meta-ethnographic synthesis . Health Policy and Planning 29 : iii70 – 8 . Google Scholar Crossref Search ADS PubMed Erasmus E , Orgill M , Schneider H , Gilson L. 2014 . Mapping the existing body of health policy implementation research in lower income settings: what is covered and what are the gaps? Health Policy and Planning 29 : iii35 – 50 . Google Scholar Crossref Search ADS PubMed Gilson L. 2014 . Qualitative research synthesis for health policy analysis: what does it entail and what does it offer? Health Policy and Planning 29 : iii1 – 5 . Google Scholar Crossref Search ADS PubMed Gilson L , Raphaely N. 2008 . The terrain of health policy analysis in low and middle income countries: a review of published literature 1994–2007 . Health Policy and Planning 23 : 294 – 307 . Google Scholar Crossref Search ADS PubMed Gilson L , Schneider H , Orgill M. 2014 . Practice and power: a review and interpretive synthesis focused on the exercise of discretionary power in policy implementation by front-line providers and managers . Health Policy and Planning 29 : iii51 – 69 . Google Scholar Crossref Search ADS PubMed Gneiting U. 2016 . From global agenda-setting to domestic implementation: successes and challenges of the global health network on tobacco control . Health Policy and Planning 31 : i74 – 86 . Google Scholar Crossref Search ADS PubMed Gómez EJ , Harris J. 2016 . Political repression, civil society and the politics of responding to AIDS in the BRICS nations . Health Policy and Planning 31 : 56 – 66 . Google Scholar Crossref Search ADS PubMed Haas PM. 1992 . Introduction: epistemic communities and international policy coordination . International Organization 46 : 1 – 35 . Google Scholar Crossref Search ADS Hall PD. 1987 . A historical overview of the private nonprofit sector. In: Powell WW (ed). The Nonprofit Sector . New Haven : Yale University Press , 32 – 65 . Hirsch JS , Giang LM , Parker RG , Duong LB. 2015 . Caught in the middle: the contested politics of HIV/AIDS and health policy in Vietnam . Journal of Health Politics, Policy and Law 40 : 13 – 40 . Google Scholar Crossref Search ADS PubMed Jat TR , Deo PR , Goicolea I , Hurtig AK , San Sebastian M . 2013 . The emergence of maternal health as a political priority in Madhya Pradesh, India: a qualitative study. BMC Pregnancy and Childbirth 13 : 181 . Joachim J. 2003 . Framing issues and seizing opportunities: the UN, NGOs, and women’s rights . International Studies Quarterly 47 : 247 – 74 . Google Scholar Crossref Search ADS Katzenstein PJ. 1996 . The Culture of National Security: Norms and Identity in World Politics . New York : Columbia University Press . Keck ME , Sikkink K. 1998 . Activists beyond Borders: Advocacy Networks in International Politics . Ithaca, NY : Cornell University Press . Kingdon JW. 1995 . Agendas, Alternatives, and Public Policies , 2 nd edn. New York : Longman . Kippax S , Stephenson N , Parker RG , Aggleton P. 2013 . Between individual agency and structure in HIV prevention: understanding the middle ground of social practice . American Journal of Public Health 103 : 1367 – 75 . Google Scholar Crossref Search ADS PubMed Lencucha R , Kothari A , Labonté R. 2011 . The role of non-governmental organizations in global health diplomacy: negotiating the framework convention on tobacco control . Health Policy and Planning 26 : 405 – 12 . Google Scholar Crossref Search ADS PubMed Llamas A , Mayhew S. 2016 . The emergence of the vertical birth in Ecuador: an analysis of agenda setting and policy windows for intercultural health . Health Policy and Planning 31 : 683 – 690 . Google Scholar Crossref Search ADS PubMed Lucas PJ , Baird J , Arai L , Law C , Roberts HM. 2007 . Worked examples of alternative methods for the synthesis of qualitative and quantitative research in systematic reviews . BMC Medical Research Methodology 7 . Mamudu HM , Glantz SA. 2009 . Civil society and the negotiation of the framework convention on tobacco control . Global Public Health 4 : 150 – 68 . Google Scholar Crossref Search ADS PubMed McInnes C , Lee K. 2012 . Framing and global health governance: key findings . Global Public Health 7 : S191 – 8 . Google Scholar Crossref Search ADS PubMed Muukkonen M. 2009 . Framing the field: civil society and related concepts . Nonprofit and Voluntary Sector Quarterly 38 : 684 – 700 . Google Scholar Crossref Search ADS Najam A. 2000 . The four-c’s of third sector-government relations: cooperation, confrontation, complementarity, and co-optation . Nonprofit Management and Leadership 10 : 375 – 96 . Google Scholar Crossref Search ADS O’Brien C. 2015 . Transnational issue-specific expert networking: a pathway to local policy change . Social Science & Medicine 146 : 285 – 291 . Google Scholar Crossref Search ADS Odoch WD , Kabali K , Ankunda R , Zulu JM , Tetui M. 2015 . Introduction of male circumcision for HIV prevention in Uganda: analysis of the policy process . Health Research Policy and Systems 13 : 1 – 13 . Google Scholar Crossref Search ADS PubMed Okello A , Welburn S , Smith J. 2015 . Crossing institutional boundaries: mapping the policy process for improved control of endemic and neglected zoonoses in sub-Saharan Africa . Health Policy and Planning 30 : 804 – 812 . Google Scholar Crossref Search ADS PubMed Omar MA , Green AT , Bird PK et al. 2010 . Mental health policy process: a comparative study of Ghana, South Africa, Uganda and Zambia . International Journal of Mental Health Systems 4 : 24 . Google Scholar PubMed Oronje RN. 2013 . The Kenyan national response to internationally agreed sexual and reproductive health and rights goals: a case study of three policies . Reproductive Health Matters 21 : 151 –– 160 . Google Scholar Crossref Search ADS PubMed Pallas CL , Nguyen L. 2018 . Transnational advocacy without northern NGO partners: Vietnamese NGOs in the HIV/AIDS sector . Nonprofit and Voluntary Sector Quarterly 47 : 159S – 76S . Google Scholar Crossref Search ADS Parker R. 2011 . Grassroots activism, civil society mobilization, and the politics of the global HIV/AIDS epidemic . The Brown Journal of World Affairs 17 : 21 – 37 . Pedregal V , Destremau B , Criel B. 2015 . Health policy in Cambodia: to what extent is an aid-dependent country able to determine its own policy? Journal of Social Policy 44 : 171 – 187 . Google Scholar Crossref Search ADS Pelletier DL , Menon P , Ngo T , Frongillo EA , Frongillo D. 2012 . The nutrition policy process: the role of strategic capacity in advancing national nutrition agendas . Food and Nutrition Bulletin 32 : S59 – S69 . Google Scholar Crossref Search ADS Pick S , Givaudan M , Reich MR. 2008 . NGO–government partnerships for scaling-up: sexuality education in Mexico . Development in Practice 18 : 164 – 175 . Google Scholar Crossref Search ADS Piscopo JM. 2014 . Female leadership and sexual health policy in Argentina . Latin American Research Review 49 : 104 – 127 . Google Scholar Crossref Search ADS Powers T , 2016 . HIV/AIDS policy, transnational donor capital, and localized political formations in South Africa . Human Organization 75 , 239 . Google Scholar Crossref Search ADS Reich M. 1995 . The politics of health sector reform in developing countries: three cases of pharmaceutical policy . Health Policy (Amsterdam, Netherlands) 32 : 47 – 77 . Google Scholar Crossref Search ADS PubMed Robinson RS. 2015 . Population policy in sub-Saharan Africa: a case of both normative and coercive ties to the world polity . Population Research and Policy Review 34 : 201 – 221 . Google Scholar Crossref Search ADS Romero LI , Quental C. 2014 . Research for better health: the Panamanian priority-setting experience and the need for a new process . Health Research Policy and Systems 12 : 38 . Google Scholar Crossref Search ADS PubMed Ruckert A , Labonté R , Lencucha R , Runnels V , Gagnon M. 2016 . Global health diplomacy: a critical review of the literature . Social Science & Medicine (1982) 155 : 61 – 72 . Google Scholar Crossref Search ADS PubMed Sabatier PA. 2007 . Theories of the Policy Process . Boulder, CO : Westview Press . Sabatier PA , Jenkins Smith HC. 1993 . Policy Change and Learning: An Advocacy Coalition Approach . Boulder, CO : Westview Press . Sabatier PA , Jenkins Smith HC. 1999 . The advocacy coalition framework: an assessment. In: Sabatier PA (ed). Theories of the Policy Process . Boulder, CO : Westview Press , 117 – 68 . Salamon LM. 1994 . The rise of the nonprofit sector . Foreign Affairs 73 : 109 – 22 . Google Scholar Crossref Search ADS Salamon LM , Anheier HK. 1998 . Social origins of civil society: explaining the nonprofit sector cross-nationally . Voluntas: International Journal of Voluntary and Nonprofit Organizations 9 : 213 – 48 . Google Scholar Crossref Search ADS Shearer JC , Abelson J , Kouyaté B , Lavis JN , Walt G. 2016 . Why do policies change? Institutions, interests, ideas and networks in three cases of policy reform . Health Policy and Planning 31 : 1200 – 1211 . Google Scholar Crossref Search ADS PubMed Shiffman J. 2007 . Generating political priority for maternal mortality reduction in 5 developing countries . American Journal of Public Health 97 : 796 – 803 . Google Scholar Crossref Search ADS PubMed Shiffman J , Quissell K , Schmitz HP et al. 2016 . A framework on the emergence and effectiveness of global health networks . Health Policy and Planning 31 : i3 – i16 . Google Scholar Crossref Search ADS PubMed Shiffman J , Smith S. 2007 . Generation of political priority for global health initiatives: a framework and case study of maternal mortality . Lancet (London, England) 370 : 1370 – 9 . Google Scholar Crossref Search ADS PubMed Smith SL. 2014 . Political contexts and maternal health policy: insights from a comparison of south Indian states . Social Science & Medicine 100 : 46 – 53 . Google Scholar Crossref Search ADS Smith SL , Shiffman J. 2018 . The politics of global health agenda setting. In: McInnes C, Lee K, Youde J (eds). The Oxford Handbook of Global Health Politics. DOI: 10.1093/oxfordhb/9780190456818.013.20. Snow DA , Rochford EB Jr , Worden SK , Benford RD. 1986 . Frame alignment processes, micromobilization, and movement participation . American Sociological Review 51 : 464 – 81 . Google Scholar Crossref Search ADS Spicer N , Berhanu D , Bhattacharya D et al. 2016 . ‘The stars seem aligned’: a qualitative study to understand the effects of context on scale-up of maternal and newborn health innovations in Ethiopia, India and Nigeria . Globalization and Health 12 . Spicer N , Bhattacharya D , Dimka R et al. 2014 . ‘Scaling-up is a craft not a science’: catalysing scale-up of health innovations in Ethiopia, India and Nigeria . Social Science & Medicine 121 : 30 – 38 . Google Scholar Crossref Search ADS Spicer N , Harmer A , Aleshkina J et al. 2011 . Circus monkeys or change agents? Civil society advocacy for HIV/AIDS in adverse policy environments . Social Science & Medicine 73 : 1748 – 1755 . Google Scholar Crossref Search ADS Stone DA. 1989 . Casual stories and the formation of policy agendas . Political Science Quarterly 104 : 281. Google Scholar Crossref Search ADS ‘t Hoen E , Berger J , Calmy A , Moon S. 2011 . Driving a decade of change: HIV/AIDS, patents and access to medicines for all . Journal of the International Aids Society 14 : 15. Google Scholar Crossref Search ADS PubMed Tkatchenko Schmidt E , Renton A , Gevorgyan R , Davydenko L , Atun R. 2008 . Prevention of HIV/AIDS among injecting drug users in Russia: opportunities and barriers to scaling-up of harm reduction programmes . Health Policy 85 : 162 – 171 . Google Scholar Crossref Search ADS PubMed True J , Mintrom M. 2001 . Transnational networks and policy diffusion: the case of gender mainstreaming . International Studies Quarterly 45 : 27 – 57 . Google Scholar Crossref Search ADS VanDyck CK. 2017 . Concept and Definition of Civil Society Sustainability . Washington, DC : Center for Strategic and International Studies . van Pletzen E , Zulliger R , Moshabela M , Schneider H. 2014 . The size, characteristics and partnership networks of the health-related non-profit sector in three regions of South Africa: implications of changing primary health care policy for community-based care . Health Policy and Planning 29 : 742 – 52 . Google Scholar Crossref Search ADS PubMed Walt G , Gilson L. 1994 . Reforming the health sector in developing countries: the central role of policy analysis . Health Policy and Planning 9 : 353 – 70 . Google Scholar Crossref Search ADS PubMed Walt G , Gilson L. 2014 . Can frameworks inform knowledge about health policy processes? Reviewing health policy papers on agenda setting and testing them against a specific priority-setting framework . Health Policy and Planning 29 : iii6 – 22 . Google Scholar Crossref Search ADS PubMed Walt G , Shiffman J , Schneider H , Murray SF , Brugha R , Gilson L. 2008 . ‘Doing’ health policy analysis: methodological and conceptual reflections and challenges . Health Policy and Planning 23 : 308 – 17 . Google Scholar Crossref Search ADS PubMed Winter SC. 2012 . Implementation perspectives: status and reconsideration. In: Peters BG , Pierre J (eds). The SAGE Handbook of Public Administration . Los Angeles, CA : SAGE , 265 – 78 . Zuniga JM. 2006 . The contribution of civil society. In: Beck EJ , Mays N , Whiteside AW , Zuniga JM (eds). The HIV Pandemic: Local and Global Implications . Oxford : Oxford University Press , 706 – 19 . © The Author(s) 2019. 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)