Abstract Electronic health information systems, including electronic medical records (EMRs), have the potential to improve access to information and quality of care, among other things. Success factors and challenges for novel EMR implementations in low-resource settings have increasingly been studied, although less is known about maturing systems and sustainability. One systematic review identified seven categories of implementation success factors: ethical, financial, functionality, organizational, political, technical and training. This case study applies this framework to iSanté, Haiti’s national EMR in use in more than 100 sites and housing records for more than 750 000 patients. The author group, consisting of representatives of different agencies within the Haitian Ministry of Health (MSPP), funding partner the Centers for Disease Control and Prevention (CDC) Haiti, and implementing partner the International Training and Education Center for Health (I-TECH), identify successes and lessons learned according to the seven identified categories, and propose an additional cross-cutting category, sustainability. Factors important for long-term implementation success of complex information systems are balancing investments in hardware and software infrastructure upkeep, user capacity and data quality control; designing and building a system within the context of the greater eHealth ecosystem with a plan for interoperability and data exchange; establishing system governance and strong leadership to support local system ownership and planning for system financing to ensure sustainability. Lessons learned from 10 years of implementation of the iSanté EMR system are relevant to sustainability of a full range of increasingly interrelated information systems (e.g. for laboratory, supply chain, pharmacy and human resources) in the health sector in low-resource settings. Haiti, health information system, HIV, primary care, implementation, international health Key Messages While functionality and technical factors continue to be relevant as electronic medical records (EMRs) in low-resource settings mature, other factors also become significant over time, including governance and leadership, ongoing user capacity, data quality, integration within a larger eHealth framework and financing. Other complex information systems, such as those for laboratory management, procurement, pharmacy or human resources management, are increasingly interdependent and may face similar challenges after implementation. Further evidence based on case studies of systems which have achieved both sustainable continuity and transition to local ownership is needed, specifically on the approaches to health information system development, implementation, financing and governance that contribute to system sustainability and health improvement in low-resource settings. Introduction Access to timely, complete and accurate information about patients is critical in health care delivery. Electronic health information systems, including electronic medical records (EMRs), have the potential to improve access to information, and in high-resource settings have been shown to improve patient safety outcomes and quality, efficiency and cost-effectiveness of care (Chaudhry et al. 2006; Goldzweig et al. 2009; Holroyd-Leduc et al. 2011; Straub et al. 2013; Jones et al. 2014; Raymond et al. 2015). EMR implementation in low-resource settings dates back to the early 2000s, usually to support the scale up of tuberculosis (TB) or Human immunodeficiency virus (HIV) vertical programs in countries, such as Peru, Haiti, Kenya and Malawi (Fraser et al. 2005). Documented benefits in these settings include improved quality of care, reduction in loss-to-follow-up and greater efficiency (Oluoch et al. 2012; Driessen et al. 2013). Requirements for successful implementation of novel electronic systems in low-resource settings have been increasingly studied, while factors contributing to system sustainability in these settings (here referring to successful ongoing use of systems as they mature over time) have received less attention (Kimaro and Nhampossa 2007; Soar et al. 2012). Various models have drawn on experience in high-resource settings to characterize successful systems in general, such as DeLone and McLean’s Information Systems Success Model, which evolved to include six interrelated dimensions: three quality dimensions (system, information and service); system use/intention to use; user satisfaction; and net benefits (DeLone and McLean 2003). In low-resource settings, success factors for implementation include system quality factors, such as ‘fit factor’, ease of use, responsiveness and security features (Gruber et al. 2009; Raymond et al. 2015; Tilahun and Fritz 2015; Jawhari et al. 2016). In studies of implementations in high-resource settings, management or environmental factors impact system use and system quality, including leadership support; user training and ongoing support (including addressing limited computer literacy); team approach to system design, development and implementation; attention to policies, procedures and role changes; and system evaluation (Gruber et al. 2009; Ludwick and Doucette 2009; Rizer et al. 2015). A non-systematic review of the literature on sustainable health informatics implementations in developing countries identified six categories of challenges, which focused mostly on technical issues [need for development of health information technology (IT) agendas; lack of common interoperability standards, trained workforce, and regional integration] as compared with environmental, financial or political issues (resource and infrastructure limitations; uncertainty, ethics and legal considerations) (Luna et al. 2014). Focusing on implementation experience specific to low-resource settings, Fritz et al. conducted a systematic review and identified seven categories of factors contributing to system success: ethical, financial, functionality, organizational, political, technical and training (Fritz et al. 2015). They found that approximately three-quarters of the 381 success criteria identified fell into the functionality (29%), organizational (23.5%) and technical (21.5%) categories. The review focused on the design, planning, rollout and initial use of system, and many of the systems reviewed were donor-funded, which perhaps explains the apparent lesser role of financing and political factors. Nevertheless, the categories provide a useful framework that includes both system-intrinsic factors (system functionality, technical structure) and extrinsic or environmental factors that influence institutions’ and individuals’ abilities to implement and maintain a system (training, ethical, political, financial). Success factors as systems move from initial implementation to sustained use and local country ownership and management are not well-explored. The purpose of this article is to apply Fritz et al.’s framework to a case study of a more mature system: Haiti’s national EMR. We describe the system’s development, performance on indicators of success and lessons learned. We propose modifications to reflect the experience of a national-scale and long-standing implementation (Table 1). We identify areas for future exploration to achieve scalable, sustainable and locally-owned health information systems. Table 1. Categories of EMR implementation success factors (modified from Fritz et al. 2015) Category Descriptions Functionality System features and functions, including data handling, reports Technical Infrastructure, software architecture, data standards, privacy/security Organizational Managerial circumstances within the organization itself, including human resources/skilled staff/local buy-in, leadership and governance, project management and commitment to implementation, data usea Training Skills training as well as computer literacy and educational background, ongoing user supporta Political Health policies and country-wide circumstances, including health care infrastructure, characteristics, such as trust and willingness to change Ethical Regulatory and cultural issues, concerns of privacy/security Financial Resources (including human) and funding, including need for efficiency Sustainabilitya Transition from external/stakeholder to local management across all categories including financing Category Descriptions Functionality System features and functions, including data handling, reports Technical Infrastructure, software architecture, data standards, privacy/security Organizational Managerial circumstances within the organization itself, including human resources/skilled staff/local buy-in, leadership and governance, project management and commitment to implementation, data usea Training Skills training as well as computer literacy and educational background, ongoing user supporta Political Health policies and country-wide circumstances, including health care infrastructure, characteristics, such as trust and willingness to change Ethical Regulatory and cultural issues, concerns of privacy/security Financial Resources (including human) and funding, including need for efficiency Sustainabilitya Transition from external/stakeholder to local management across all categories including financing a Denotes this article’s additions to Fritz et al. categories and success factors. Haiti’s national EMR, iSanté Approximately three-quarters of Haiti’s 10 million people live on less than US$2 per day (Cayemittes et al. 2013). The health workforce is overburdened (0.8 doctors and 2.8 nurses per 10 000 population) and weak infrastructure, such as poor roads, lack of electricity, running water and equipment hinder patient access and quality of care (Ministère de la Santé Publique et de la Population 2015). In 2005, in the face of the highest HIV prevalence in the Caribbean region (2.2%), the Haitian Ministry of Health (MSPP) and US Centers for Disease Control and Prevention (CDC), with the support of the International Training and Education Center for Health (I-TECH), partnered to develop and launch an EMR, known as iSanté, for the national HIV program (Cayemittes et al. 2007). Its early history is described elsewhere (Lober et al. 2008; Matheson et al. 2012). iSanté is the largest of three multi-site EMRs in Haiti. It is seen as a cornerstone of the broader planned national eHealth architecture, eventually feeding into an overarching system for aggregate health indicator reporting (called Système d’Information Sanitaire National Unique), set as a priority by MSPP in 2013 (MSPP 2013). Figure 1 shows the annual growth in number of sites using iSanté (defined as entering data into the system during that year) and number of patients in iSanté. As of December 2015, 107 sites were using iSanté. As context, a 2013 census survey of Haiti’s health system counted 905 health facilities (342 public and the rest private for profit, non-profit or mixed), of which 588 had stable electricity (defined as having a power interruption of no more than 2 h during working hours in the prior week, or having an alternative energy source such as a backup generator) and 127 offered antiretroviral treatment (ART) (IHE & ICF International 2014). Therefore, iSanté is used at ∼12% of all sites, 18% of sites with electricity and 84% of sites offering ART. The number of sites using iSanté in recent years has fluctuated, with new sites being equipped for and implementing iSanté as other sites discontinued use. This discontinuation was primarily due to funding cut-backs and a reduction in the number of health facilities offering HIV services (though sites have the option to continue using iSanté for primary care patients). iSanté serves as the primary medical record for ∼63% of people enrolled in HIV-support services, and houses records for >750 000 unique patients since 2005 (Monitoring Evaluation et Surveillance Intégrée (MESI) database 2015). Figure 1. View largeDownload slide iSanté health care site usage and patient coverage, 2005–15. Figure 1. View largeDownload slide iSanté health care site usage and patient coverage, 2005–15. Methods We used the Fritz et al. framework to describe and analyse the iSanté implementation over time, drawing out key themes and lessons learned in each success category based on the direct experience of the author group in supporting the project over the past decade. The author group includes representatives from the principal agencies responsible for iSanté financing, governance and implementation. Specifically, the Haitian MSPP Management Unit (Unité de Gestion de Projet, UGP/MSPP) is responsible for IT infrastructure and maintenance, user support and participates in system governance. The MSPP’s Research and Programming Unit leads use of iSanté data analysis and provides input on software requirements. The Haiti office of the CDC (CDC Haiti) is responsible for iSanté project financing, shared system governance, and provides input on software design and development. I-TECH conducts software design and development, provides user training and support, and participates in system governance. The author group referenced existing project documentation, as well as performance indicators based upon queries from the iSanté system itself. These indicators included point-of-care (POC) system usage, frequency of data replication from sites to a central server, number of system users, average users per site and forms created per user. Data quality indicators included data completeness for patient age (paediatric and adult), height, weight, sex, TB screening results for HIV-positive patients, ART eligibility and pregnancy status. Project documentation consisted of project reports, meeting minutes, software release notes and training records. Results A description of iSanté’s evolution and quantitative indicators are presented according to each category per Fritz et al.’s framework. Fritz et al. includes sustainability in several of their original seven categories; we added a separate category for sustainability, since we find it an essential cross-cutting issue for long-term implementations. Successes, lessons learned and policy implications are summarized in Table 2. For a more detailed discussion of each point, see Supplementary Material, Web Annex 1 available in the online version of this article. Table 2. Summary of successes, lessons learned and policy implications for national long-term EMR implementations in low-resource settings Category Key iSanté successes Lessons learned and policy implications Functionality Dashboards for basic site information and data quality Clinical decision support features added during switch to point-of-care use Reports integrated into program goals and quality improvement activities User-generated reports and visualization via data warehouse Automated reporting to national health information systems Balance additional functionality with other priorities over time Make data intake forms concise, focus on key data, define required fields sparingly Define the business case for new functionality before investing Focus on data quality, develop DQ indicators, make DQ part of performance monitoring Define data reporting and transformation needs carefully Integrate into the broader eHealth ecosystem, or follow international standards for interoperability Technical System architecture supports flexibility (e.g. local server or ASP use) Establish architecture that mitigates infrastructure weakness Budget for infrastructure maintenance and replacement Plan for software platform updates or upgrades Organizational MSPP involvement from beginning, improved governance structures over time Shared technical support between MSPP/UGP, I-TECH Programmatic and research uses of data Establish and empower representative governance structures early on As system scales up, ensure support staffing and operational plans also scale up Encourage data use for routine analysis and research Training Hands-on training with peers ‘Super user’ strategy Video job aids Plan for shift from user training to user support resources, and consider baseline computer literacy Ensure system administrators and IT support receive sufficient training Review workflow changes and ensure training and support are allocated Use mentors to build staff’s skills in decision-making and quality improvement Political Growing support from MSPP for iSanté as EMR for primary care, not just HIV Engage high-level leadership to understand the system, value and cost Ensure EMRs fit within the national eHealth architecture Ethical User profiles support patient confidentiality Ensure privacy and security standards evolve over time as technology changes Identify responsible entity and develop data security policies and procedures Establish mechanisms to respond to data requests Financial Consistent donor financial support Conduct business planning to identify total cost of ownership, resources and gaps Sustainability Significant steps toward transition of servers, software development, governance, user support Plan for transition and sustainability, including financial, from start of implementation Category Key iSanté successes Lessons learned and policy implications Functionality Dashboards for basic site information and data quality Clinical decision support features added during switch to point-of-care use Reports integrated into program goals and quality improvement activities User-generated reports and visualization via data warehouse Automated reporting to national health information systems Balance additional functionality with other priorities over time Make data intake forms concise, focus on key data, define required fields sparingly Define the business case for new functionality before investing Focus on data quality, develop DQ indicators, make DQ part of performance monitoring Define data reporting and transformation needs carefully Integrate into the broader eHealth ecosystem, or follow international standards for interoperability Technical System architecture supports flexibility (e.g. local server or ASP use) Establish architecture that mitigates infrastructure weakness Budget for infrastructure maintenance and replacement Plan for software platform updates or upgrades Organizational MSPP involvement from beginning, improved governance structures over time Shared technical support between MSPP/UGP, I-TECH Programmatic and research uses of data Establish and empower representative governance structures early on As system scales up, ensure support staffing and operational plans also scale up Encourage data use for routine analysis and research Training Hands-on training with peers ‘Super user’ strategy Video job aids Plan for shift from user training to user support resources, and consider baseline computer literacy Ensure system administrators and IT support receive sufficient training Review workflow changes and ensure training and support are allocated Use mentors to build staff’s skills in decision-making and quality improvement Political Growing support from MSPP for iSanté as EMR for primary care, not just HIV Engage high-level leadership to understand the system, value and cost Ensure EMRs fit within the national eHealth architecture Ethical User profiles support patient confidentiality Ensure privacy and security standards evolve over time as technology changes Identify responsible entity and develop data security policies and procedures Establish mechanisms to respond to data requests Financial Consistent donor financial support Conduct business planning to identify total cost of ownership, resources and gaps Sustainability Significant steps toward transition of servers, software development, governance, user support Plan for transition and sustainability, including financial, from start of implementation Functionality Functionality in iSanté has been steadily added as users and stakeholders have interacted with the system and identified new priorities. New features and functionalities since 2010 include forms and reports for obstetrics and gynaecology (OBGYN) services, TB, malaria and primary care; clinical decision support alerts for malnutrition; and added confidentiality protections such as user profiles. Additional functionalities are described in the following paragraphs. POC use In 2012, the MSPP and CDC Haiti decided to extend iSanté to POC (i.e. during patient consultations) at the larger health facilities. This required additional equipment purchases and installation, infrastructure improvements, such as wiring to support internet access and training of clinicians, since formerly data clerks entered data retrospectively from paper-based clinician forms. As of December 2015, 75 of the 107 sites using iSanté had been equipped to use iSanté at POC, and ∼71% (53) of those were actually using it at POC. Across all sites using iSanté, 84% (1 592 352/1 895 100) of patient visit forms created that year were entered into iSanté on the same day as the visit, a proxy indicator for POC use. Reasons for equipped sites not using iSanté at POC were primarily related to the funding cuts mentioned above (e.g. non-replacement of malfunctioning equipment, lack of tables or proper security for the equipment). The shift to using iSanté at POC prompted the addition of clinical decision support features such as alerts. Data quality As iSanté expanded in usage and reporting capabilities, it was increasingly important to formally assess data quality. Users at times expressed doubts that the report data were correct, which could have been due to calculated indicator definitions being different from what users expected, or simply insufficient documentation of indicator definitions within iSanté. Many sites have data quality protocols, although the extent to which they are followed is not known. Based on a study of iSanté data quality, I-TECH developed a data quality dashboard in 2014 to show the status of key elements identified as important by stakeholders, allow aggregation at a national or site level, and drill down into specific records (Puttkammer et al. 2016). During site visits, I-TECH’s clinical mentors worked with clinicians on data quality and documentation, corresponding to an improvement in data completeness over time (Figure 2). Figure 2. View largeDownload slide Data completeness for select iSanté variables, 2005–15. Figure 2. View largeDownload slide Data completeness for select iSanté variables, 2005–15. Reporting Report functionality in iSanté started as a small number of defined reports. The most widely used reports, for instance the HIVQUAL reports including key HIV indicators, had clear specifications that were linked to program targets. These reports had performance challenges, however, because they were run directly from the iSanté transactional database and involved complex query scripts. Additionally, stakeholders increasingly requested custom or one-time data sets, which required programming. As a technical solution, I-TECH developed a data transformation model (Extract-Transform-Load), in which data are extracted from central database, manipulated into pre-calculated variables and loaded into data visualization software. Starting in 2013, new indicators for malaria, nutrition, OBGYN and TB were developed using this data warehouse model. In 2015, the Sisense business intelligence software was linked to the iSanté database to enable improved data processing, analysis and visualization for routine and ad hoc analysis. iSanté also supports routine reporting to other national health information systems, making it a cornerstone for surveillance and monitoring and evaluation. Specifically, iSanté sends automated monthly case reports to the HIV/acquired immune deficiency syndrome (AIDS) Surveillance System, and generates the MSPP-required report that each health facility submits to the national monitoring and surveillance database for HIV and TB programs. Interoperability and data sharing Several functionalities were added related to data exchange and patient identification. iSanté and the open-source lab information system OpenELIS were modified so that the two systems could exchange test orders and results. As of early 2015 approximately 30 sites used iSanté and OpenELIS interoperably. A funder decision was made to discontinue further OpenELIS adaptation in Haiti (although most of the sites have chosen to continue using OpenELIS), and implement a different system at the National Public Health Laboratory. This new system also requires information exchange with multiple other systems, including iSanté. In the meantime, I-TECH has started developing a Master Patient Index for iSanté, a necessary component of a standards-based Health Information Exchange that facilitates the sharing of health information between EMRs, lab information systems and community-based mobile applications. This updated version of iSanté, called iSantéPlus and expected in fall 2017, fits MSPP’s plan to develop a comprehensive eHealth architecture with information systems feeding data summaries into the central system for reporting of aggregate health indicators. Stakeholders recognized a need for national interoperability standards since at least 2014, but it took time to define and integrate into funding cycles and workplans. In settings like Haiti where a unique national identification number does not exist, health information systems must develop mechanisms for assigning a system identity and de-duplicating or merging records so each patient has a single record. In part to meet this need, fingerprint recognition technology and smartcards were piloted in iSanté between 2011 and 2015. Fingerprinting is currently being expanded in high-volume sites by UGP/MSPP. Smartcards were successfully piloted from a technical standpoint, but will not be scaled up due to cost. Technical Architecture As described elsewhere, iSanté’s distributed processing architecture includes a central longitudinal repository (consolidated server) that allows local sites to host their own server and synchronize (or replicate) data periodically, which is useful in a context of weak infrastructure and inconsistent power and internet access (Matheson et al. 2012). Roughly 8% of all sites using iSanté as of December 2015 (8/107) use an application service provider (ASP) server instead of a local server, and connect to iSanté via a web browser. Availability of the ASP server depends on UGP/MSPP having power and connectivity, since the server is now located there. The consolidated server provides an aggregate data source for public health analysis, and also serves as a secondary backup for data recovery to sites in the event of a disaster where a local backup is not available. iSanté’s off-site replication has enabled data restoration for sites following three significant disasters: a mudslide after Hurricane Jeanne in Gonaïves in 2008, the earthquake in 2010 and Hurricane Matthew in 2016 that devastated much of Haiti’s southwestern peninsula. Platform The aging iSanté software, while still functional, became an increasing concern in recent years. The urgency to add new functionality and the reality of finite development resources meant that underlying code was not always kept up to date, including for the current software stack components (Linux/Apache/MySQL/PHP, or LAMP). Hypertext Preprocessor (PHP) as a programming language is still common, but Java is becoming more of a standard, raising a question about finding future qualified PHP developers. A modernized platform was also needed to support information exchange. In 2014, stakeholders began to outline the status of the underlying system and determine how to maximize iSanté’s efficiency and stability. CDC Haiti and I-TECH agreed to rebuild iSanté on an OpenMRS platform, maintaining the existing Extract-Transform-Load reporting model. Organizational For a national EMR, organizational issues are relevant at individual sites for each rollout, but also at regional and national levels where decisions need to be made and policies reinforced. We therefore address national leadership and governance of, and commitment to, the system. Leadership and governance The MSPP participated in iSanté’s early direction, with the National Committee for Health Information Systems approving the standardized data collection forms, and the National AIDS Control Program and Research and Programming Unit involved in developing and monitoring reports. UGP/MSPP, CDC Haiti and I-TECH shared responsibility for implementation support at sites, based on a geographic distribution. The success of system adoption locally has varied depending on the involvement and collaboration of site leadership and other implementing partners. As iSanté scaled up, specific programs wanted to use it for their own data collection and research needs. For instance, the National Malaria Prevention Program requested the addition or revision of data elements that could assist with surveillance efforts. On several occasions, these needed to be reconciled with national guidelines, highlighting the need for clearer governance roles. In response, in 2014 the key iSanté partners formed a Technical Working Group, to improve technical direction-setting, facilitate communication and identify and solve problems, which had previously been handled less formally by UGP/MSPP, CDC and I-TECH. The Technical Working Group monthly meetings have proved valuable in updating stakeholders on software development and implementation challenges, and have been expanded when needed to include senior management input. Human resources The need for sufficient trained personnel cuts across all areas of EMR implementation, and requires organizational planning and investment. User training and support is described further in the Training category below. Specialized IT staff or contractors are needed for system administration, requirements gathering and software design and development for system customization or updates. For iSanté, these personnel have been based both at UGP/MSPP and at I-TECH, although they often have multiple responsibilities. System administration (e.g. routine server maintenance, network troubleshooting) staffing has not scaled up as iSanté’s base expanded, which poses some risk. In cases where a site loses connection with the central server, data being captured on local servers are not included in queries run on the consolidated server during the period of disconnection, and the data are at risk if the sites are not following proper local back-up procedure. A dashboard in iSanté shows last replication date for each site, enabling routine monitoring and follow-up, although this requires adequate staffing. In other words, small technical challenges are exacerbated when organizational structures are not well defined or staffed. The rollout of new software releases is also human-resource dependent. After following an agile development approach for several years, resulting in frequent smaller releases, iSanté changed to a longer cycle, with approximately one new release annually and patches for urgent fixes. This eased the logistical burden of upgrading each site several times per year (absent a mechanism to automatically push upgrades to sites), and site staff only needed to absorb one set of changes. Data use The use of data by clinicians and health system managers is critical to improving quality of care. As part of Haiti’s adoption in 2007 of the HIVQUAL methodology for monitoring HIV services, 10 HIV-related performance indicators were programmed into iSanté. The HIVQUAL reports (later expanded to include primary care indicators and renamed HEALTHQUAL) were routinely used at sites for quality monitoring and improvement, and illustrated the potential of focused data analysis on a core set of indicators. More recently, custom queries of the system data have been used by the MSPP to evaluate retention on ART following adoption of the Option B + policy, and evaluate implementation and outcomes of the ‘test and start’ approach to HIV treatment in Haiti, adopted in 2016. Training As the number of sites using iSanté has increased, so has the number of users: as of December 2015, 1825 unique iSanté users had entered data within the prior 12 months (Figure 3). The decrease from 2014 to 2015 is due to the net decrease in sites providing HIV services and using iSanté. The average number of users per site has increased (from 4.21 to 21.24) as more sites are using iSanté at POC. Users are also making more entries, from an average of 300 forms per user per year in 2007 to 1081 in 2015. Initially user training for iSanté was targeted at data clerks and receptionists for data entry, and HIV program clinicians. For POC implementation, more clinicians were trained as direct users. I-TECH conducted large multi-site hands-on training sessions for sites adopting iSanté, with one to four people trained per site, just before or while the sites were being equipped with the necessary hardware and internet connection. I-TECH or UGP/MSPP clinical teams then visited sites to support staff in integrating iSanté into their workflow, and to mentor managers during routine visits on using data for quality improvement. Sites using iSanté only for HIV participated in refresher training when they switched to primary care use and/or POC use. At some sites, after initial training sessions and visits, staff ‘super users’ trained new staff. Figure 3. View largeDownload slide Unique iSanté users and average number of forms per user, 2007–15. Figure 3. View largeDownload slide Unique iSanté users and average number of forms per user, 2007–15. Over time, user support has become a higher priority than initial user training, and one that we propose adding to this category. Most sites have a Disease Reporting Officer, an MSPP employee who serves as a point person for iSanté data questions and supports use by staff. I-TECH and UGP/MSPP staff are consulted for more difficult questions or to report bugs or feature requests. Remote support is provided by I-TECH using TeamViewer software. Digital content in the form of how-to videos were added to recent iSanté releases to reduce the need for costly in-person training sessions or support visits related to new functionality. I-TECH also conducted several training sessions for system administrators from UGP/MSPP and other units between 2010 and 2012. While these were generally successful in increasing technical skills for system monitoring and response, the related roles and procedures were not operationalized and therefore had limited impact. Political Despite significant upheavals (e.g. the 2010 earthquake, cholera outbreak and subsequent disruptions such as strikes), MSPP and CDC Haiti have continued to champion health information systems, and iSanté in particular. As governments have changed, new MSPP directors have been appointed, requiring time and effort to re-build awareness and involvement in iSanté’s implementation. Political factors have particular influence for a national-scale information system such as iSanté. A new government can make decisions that impact an established system, although to some extent a mature system has the benefit of being well integrated into the health system and included in national health planning documents and eHealth architecture. Political factors can be related to governance as well, such as balancing priorities or overall strategic direction between different Ministry of Health divisions, funding partners or other stakeholders. A key lesson learned is to engage leadership as necessary to understand the system, its value and total cost of ownership, especially so that investments are not seen as one-time expenses. Ethical Privacy and security Various policies and functionalities have been added to iSanté to increase patient privacy and data security, such as user profiles to enable different levels of access to patient records. Local servers are backed up daily and encrypted, and site-specific encryption keys are stored on the consolidated server. Paradoxically, strong data security can lead to data loss: one site that had stopped replicating to the consolidated server during an internet disruption needed to restore data from its backup due to a server failure. The site had no locally stored copy of the decryption key, nor was it available on the consolidated server because the key had been changed since the last replication; the unreplicated data were lost. As a result, the policy was changed to use a standard national decryption key format common across all sites. Other policies have been developed over time, such as confidentiality agreements for staff who work with patient data, although resource constraints have prevented comprehensive audits of policy adherence. Ethical review A Bioethical Review Board within MSPP reviews protocols for research using data from public health facilities in Haiti. This board therefore provides oversight of iSanté data management for research only, while UGP/MSPP maintains a log for partners requesting anonymized EMR data for analysis. Such structures serve as the de facto governance system for data security, which should be better integrated into overall system governance, and ultimately based on future national standards. Financial iSanté has been primarily funded by CDC Haiti through the United States President’s Emergency Plan for AIDS Relief (PEPFAR) funds, with increasing contribution by MSPP through UGP staff and time. Post-earthquake, Canada’s International Development Research Center funded updates to improve iSanté interoperability with other systems. Significant development and expansion, particularly expanding from an HIV system to include primary care, occurred in a period when PEPFAR funds supported broader health system strengthening. The advent of PEPFAR’s third cycle of funding in 2015 and its re-focus to more targeted HIV programming may impact the extent to which iSanté can be modified to include MSPP priorities, including for in-patient and emergency care. Significant future financial needs will also include replacement of computer equipment purchases in 2013–14 for the POC scale-up. A lesson learned is to conduct business planning to identify total cost of ownership, resources and gaps. This is often suggested but rarely accomplished. From both a technical and political perspective, initial implementation costs are easier to estimate than total cost of ownership, which includes ongoing expenses, such as equipment attrition, software updates and IT and user support costs. Sustainability Each of the categories above has related sustainability issues. With iSanté, which started as an externally-funded and -supported system, the issues are usually framed in terms of transition to local governance and management, or ‘country ownership’. The transition in Haiti has been ongoing. For instance, a technical and organizational transition issue is server hosting and software development, which took place outside of Haiti until approximately 2014. Despite earlier efforts to plan transition of these activities, co-ordination and follow-through was poor until more robust governance structures were established as described above, staff were hired locally, and concrete milestone dates were set. Four full-time Haitian software developers, based at I-TECH Haiti and CDC Haiti, have assumed all development tasks, with support from a US-based developer. A clinical informatics specialist based at I-TECH Haiti, with input from stakeholders, leads the process of requirements gathering, design, feature and functionality prioritization and release scheduling. The consolidated server and ASP server were moved to UGP/MSPP in 2014–15. The eventual goal is for UGP/MSPP to fully manage the iSanté system. Similar planning has been or will be undertaken for each aspect of the system, from governance to equipment purchasing to training. A lesson learned is that implementation planning should include plans for the envisioned transition and/or sustainability of the system. This could include timelines and targets for both short- and medium-term and describe how each category of system operation will be staffed and supported. This then needs to be matched to financial planning and mobilization, and periodically revisited with partners, including costing for research and development, depreciation and operational costs that funders may initially cover. Discussion Today’s global health priorities rely on health information systems that support collection, maintenance and accessibility of patient data over time. The example of iSanté in Haiti suggests ways to refine identified categories of success for EMR implementation, making them more relevant for national and sustainable long-term implementations. Many of the key challenges and lessons learned are also relevant for other complex health information systems, such as systems for laboratory management, procurement, pharmacy or human resources tracking and allocation. Much attention is usually placed on functionality and technical details when implementing an EMR, which was also the case for iSanté. The idea that users would be more likely to adopt a system that responded to their needs drove a focus on software additions, as did the perceived benefits to capturing all outpatient services in the system, not just HIV. As with any complex information system, in a context of limited resources these needs must be weighed against other factors, such as infrastructure and user capacity to actually use the software. Still, important functionality and technical priorities remain in Haiti as the MSPP develops a more comprehensive eHealth architecture to enable more robust data exchange and reporting, and reduce duplicate data entry into different systems. iSanté’s shift to iSantéPlus, with an OpenMRS platform, Master Patient Index and Health Information Exchange capability, supports this goal. Data quality and data use can work together to increase confidence in and use of an EMR, and need to be supported through both functionality and organizational policies. Data quality is an important factor in EMR adoption and use by clinicians, which feeds back into better data available in the system (Lambooij et al. 2017). There is often a tension between comprehensiveness of data collected, and keeping the system to a minimum data set that will be used and whose quality can be assured. In Haiti, changes are being made to iSanté to improve efficiency and data completeness (e.g. skip logic so that staff can jump to relevant sections when registering a patient, and more required fields). In addition, iSantéPlus will have fewer standard reports, and will allow users to generate specific queries as needed. Data quality and data use initiatives in other low-resource settings should be further examined and shared. Governance questions for any information system become more complex as the system scales up. Who monitors the performance of the system? Who decides what functionality to add? Who has authority to establish and enforce policies on data security, quality and use? For a nationally-implemented system such as iSanté, decision-making about functionality and implementations are made at a national level, although multiple stakeholders have input and there are often competing priorities. Strong leadership was essential to system continuity and expansion. In addition, many developing countries may not have a legal framework that addresses management or protection of health information. More documentation is needed on successful locally-led governance of donor-funded systems, including any capacity building for local responsible entities and joint system design, planning and implementation. Ultimately, the ability to support all activities is a financial question, linked with sustainability. Once beyond initial implementation, a mature and nationally-implemented data system like iSanté contains data for hundreds of thousands of patients, on which health care workers rely to provide care. Such national-scale information systems become critical elements of a functioning health system, and cannot be allowed to lapse. Ensuring continuity and orderly transition to local ownership is essential, and shifting donor strategies can leave systems vulnerable. True transition requires an investment in the local entity assuming responsibility, so that they have sufficient human resources to take over all activities. It is often not realistic to expect that the Ministry of Health will be able to completely assume the full expense of maintaining an EMR in the near term. iSanté partners continue to work together to transition software development and management fully to Haiti, and integrate its future plans with the larger eHealth conversation. In terms of continuity, migrating to the open-source OpenMRS platform allows iSanté to take advantage of a larger global community while also building local capacity. Notably, an emphasis on annual workplans, driven by funding cycles and levels set outside of the local funding office’s control, can be challenging as there is little ability to predict resources or match them with long-term needs or strategy. In the case of iSanté, financial commitments of US donors and the MSPP remain less certain given that funding for PEPFAR is on a year-to-year basis. More discussion is needed among donors and Ministries of Health to become sensitized to the ways that funding shifts can affect the success of health information systems in low-resource settings. Conclusion Descriptions of sustained large-scale EMR deployments in low-resource contexts are rare as compared with those from high-resource settings, likely due to their more recent adoption, slower implementation timelines and abandonment of systems absent stable financing. As a case study of Haiti’s national EMR, this article’s analysis is limited to a specific environment and point of time in the evolution of EMRs, describing one implementation trajectory and highlighting considerations for longer-term maintenance of EMRs and potentially other health information systems. Especially with the increasing demand for system interconnectivity, the issues and investments needed throughout the lifespan of any complex health information system are important for policy makers and funders to consider, including appropriate architecture and integrating with a larger eHealth framework, software platform upgrades to meet advancing software standards, governance structures, realistic resource planning (e.g. equipment replacement, ongoing user support), and an emphasis on data quality and data use. In a context of limited resources, strategic decisions about trade-offs between continued expansion or feature development, and ensuring the quality and stability of the system, may be required. Further evidence based on case studies of systems which have achieved both sustainable continuity and transition to local ownership is needed, specifically on the approaches to health information system development, implementation, financing and governance that contribute to system sustainability and health improvement in low-resource settings. Supplementary Data Supplementary data are available at Health Policy and Planning online. Acknowledgements The authors are grateful to MSPP, partners and clinical staff who have contributed to iSanté’s implementation in Haiti. Funding This article was supported by the Cooperative Agreement Number GH001130 funded by the Centers for Disease Control and Prevention, and the Cooperative Agreement Number U91HA06801 funded by the U.S. Department of Health and Human Services, Health Resources and Services Administration. In addition, this article was supported by the President’s Emergency Plan for AIDS Relief (PEPFAR) through the Centers for Disease Control and Prevention [Grant No. U2GGH001130 and GH001643]. The findings and conclusions in this study are those of the authors and do not necessarily represent the official position of the funding agencies. Conflict of interest statement. None declared. References Cayemittes M, Placide MF, Mariko S et al. 2007. Enquête Mortalité, Morbidité et Utilisation des Services, Haïti, 2005-2006 (EMMUS-IV). Calverton, NY: MSPP, Institut Haïtien de l’Enfance (IHE) and Macro International Inc. https://dhsprogram.com/pubs/pdf/FR192/FR192.pdf, accessed 14 July 2017. Cayemittes M, Busangu MF, Bizimana JD et al. 2013. Enquête Mortalité, Morbidité et Utilisation des Services, Haïti, 2012 (EMMUS-V). Calverton, USA: MSPP, Institut Haïtien de l’Enfance (IHE) and ICF International. https://mspp.gouv.ht/site/downloads/EMMUS%20V%20document%20final.pdf, accessed 14 July 2017. Chaudhry B, Wang J, Wu S et al. 2006. Systematic review: impact of health information technology on quality, efficiency, and costs of medical care. Annals of Internal Medicine 144: 742– 52. Google Scholar CrossRef Search ADS PubMed DeLone WH, McLean E. 2003. The DeLone and McLean model of information systems success: a ten-year update. Journal of Management Information Systems 19: 9– 30. Google Scholar CrossRef Search ADS Driessen J, Cioffi M, Alide N et al. 2013. Modeling return on investment for an electronic medical record system in Lilongwe, Malawi. Journal of the American Medical Informatics Association 20: 743– 8. Google Scholar CrossRef Search ADS PubMed Fraser HSF, Biondich P, Moodley D et al. 2005. Implementing electronic medical record systems in developing countries. Informatics in Primary Care 13: 83– 95. Google Scholar PubMed Fritz F, Tilahun B, Dugas M. 2015. Success criteria for electronic medical record implementations in low-resource settings: a systematic review. Journal of the American Medical Informatics Association 22: 479– 88. Google Scholar CrossRef Search ADS PubMed Goldzweig CL, Towfigh A, Maglione M, Shekelle P. 2009. Costs and benefits of health information technology: new trends from the literature. Health Affairs (Millwood) 28: 282– 93. Google Scholar CrossRef Search ADS Gruber D, Cummings GG, Leblanc L, Smith DL. 2009. Factors influencing outcomes of clinical information systems implementation: a systematic review. Computers, Informatics, Nursing 27: 151– 63. Google Scholar CrossRef Search ADS Holroyd-Leduc JM, Lorenzetti D, Straus SE, Sykes L, Quan H. 2011. The impact of the electronic medical record on structure, process, and outcomes within primary care: a systematic review of the evidence. Journal of the American Medical Informatics Association 18: 732– 7. Google Scholar CrossRef Search ADS PubMed Institut Haïtien de l’Enfance (IHE) and ICF International. 2014. Évaluation de Prestation des Services de Soins de Santé, Haïti, 2013. Rockville, USA. http://mspp.gouv.ht/site/downloads/EPSSS.pdf, accessed 14 July 2017. Jawhari B, Ludwick D, Keenan L, Zakus D, Hayward R. 2016. Benefits and challenges of EMR implementations in low resource settings: a state-of-the-art review. BMC Medical Informatics and Decision Making 16: 116– 27. Google Scholar CrossRef Search ADS PubMed Jones SJ, Rudin RS, Perry T, Shekelle PG. 2014. Health information technology: an updated systematic review with a focus on meaningful use. Annals of Internal Medicine 160: 48– 54. Google Scholar CrossRef Search ADS PubMed Kimaro H, Nhampossa J. 2007. The challenges of sustainability of health information systems in developing countries: comparative case studies of Mozambique and Tanzania. Journal of Health Informatics in Developing Countries 1: 1– 10. Lambooij MS, Drewes HW, Koster F. 2017. Use of electronic medical records and quality of patient data: different reaction patterns of doctors and nurses to the hospital organization. BMC Medical Informatics and Decision Making 17: 17. Google Scholar CrossRef Search ADS PubMed Lober W, Quiles C, Wagner S et al. 2008. Three years experience with the implementation of a networked electronic medical record in Haiti. American Medical Informatics Association Annual Symposium Proceedings, Nov 6, pp. 434–8. Ludwick DA, Doucette J. 2009. Adopting electronic medical records in primary care: lessons learned from health information systems implementation experience in seven countries. International Journal of Medical Informatics 78: 22– 31. Google Scholar CrossRef Search ADS PubMed Luna D, Almerares A, Mayan JC, González Bernaldo de Quirós F, Otero C. 2014. Health informatics in developing countries: going beyond pilot practices to sustainable implementations: a review of the current challenges. Healthcare Informatics Research 20: 3– 10. Google Scholar CrossRef Search ADS PubMed Matheson AI, Baseman JG, Wagner SH et al. 2012. Implementation and expansion of an electronic medical record for HIV care and treatment in Haiti: an assessment of system use and the impact of large-scale disruptions. International Journal of Medical Informatics 81: 244– 56. Google Scholar CrossRef Search ADS PubMed Ministère de la Santé Publique et de la Population (MSPP). 2013. Plan Directeur de Santé, 2012-2022. Port-au-Prince, Haiti. https://mspp.gouv.ht/site/downloads/Plan%20Directeur%20de%20Sante%202012%202022%20version%20web.pdf, accessed 27 July 2017. Ministère de la Santé Publique et de la Population (MSPP), Unité des Etudes et de Programmation. 2015. Rapport Statistique 2014. Port-au-Prince, Haiti. http://mspp.gouv.ht/site/downloads/Rapport%20Statistique%20MSPP%202014%20version%20web.pdf, accessed 14 July 2017. Monitoring Evaluation et Surveillance Intégrée (MESI) database. Data as of 2015. http://www.mesi.ht/presentation/modules/reports.aspx? prg=1, accessed 14 September 2016. Oluoch T, Santas X, Kwaro D et al. 2012. The effect of electronic medical record-based clinical decision support on HIV care in resource-constrained settings: a systematic review. International Journal of Medical Informatics 81: e83– 92. Google Scholar CrossRef Search ADS PubMed Puttkammer N, Baseman JG, Devine EB et al. 2016. An assessment of data quality in a multi-site electronic medical record system in Haiti. International Journal of Medical Informatics 86: 104– 16. Google Scholar CrossRef Search ADS PubMed Raymond L, Paré G, Ortiz de Guinea A et al. 2015. Improving performance in medical practices through the extended use of electronic medical record systems: a survey of Canadian family physicians. BMC Medical Informatics and Decision Making 15: 27. Google Scholar CrossRef Search ADS PubMed Rizer MK, Kaufman B, Sieck CJ, Hefner JL, McAlearney AS. 2015. Top 10 lessons learned from electronic medical record implementation in a large academic medical center. Perspectives in Health Information Management 12: 1– 9. Soar J, Gow J, Caniogo V. 2012. Sustainability of health information systems in developing countries: the case of Fiji. Health Information Management Journal 41: 13– 9. Google Scholar CrossRef Search ADS PubMed Straub HL, Antoniewicz LW, Riggs JW, Plunkett BA, Hollier LM. 2013. Cost-effectiveness analysis of rubella screening strategies using electronic medical records. American Journal of Perinatology 30: 759– 64. Google Scholar CrossRef Search ADS PubMed Tilahun B, Fritz F. 2015. Modeling antecedents of electronic medical record system implementation success in low-resource setting hospitals. BMC Medical Informatics and Decision Making 15: 61. Google Scholar CrossRef Search ADS PubMed © The Author 2017. 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@example.com
Health Policy and Planning – Oxford University Press
Published: Mar 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
All the latest content is available, no embargo periods.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud