Pathway 2: How did Rwanda foster a culture of accountability and information use?

Establishing a culture of accountability and oversight

After Rwanda’s genocide in the 1990s, restoring confidence in public officials was essential. Decentralization, which boosted the authority of mayors and other local officials, was a key part of these efforts. Improving the quality and delivery of key services to people in their communities was another.

Performance-based financing

In the mid-1990s, immediately after the genocide, the new Rwandan government temporarily abolished user fees for health services. As support from development partners started to decline, however, officials reintroduced the fees. Subsequently, utilization rates fell.^1,2 At the same time, salaries for health workers in the public sector were low, and they were not linked to performance.³ Consequently, health workers had no incentive to try and improve the quality or increase the quantity of the services they provided.

To boost utilization of key health services and improve the quality of care in the aftermath of this crisis, in 2002, two Dutch NGOs adapted an experimental Cambodian scheme for PBF in two Rwandan districts, Cyangugu and Butare. Both aimed to increase the use of basic health services by calculating staff payments based, in part, on the quantity of care they delivered. At health centers, quality was defined in terms of adherence to protocols. Also, both programs encouraged managers, providers, and other stakeholders at the local level to devise innovative ways to increase and improve service delivery.^3,4

Between 2001 and 2004, one study found that the PBF pilots improved care coverage, quality, and patient impact.³ In 2005, the Belgian Technical Cooperation supported Rwanda’s MoH as it expanded this system of PBF, known as l’approche contractuelle, to 74 health centers covering 2 million people. The next year, it extended the program to 85 more health centers covering 3.8 million people.⁵

The results of Rwanda’s experiment with PBF have been mostly positive. Between 2005 and 2008, according to Demographic and Health Surveys data, PBF significantly improved the quality of care for children by ensuring they received essential medicines and that facilities had appropriate stock of supplies,⁶ but these improvements did not increase the propensity to seek care. Another evaluation showed similar improvements in the quality of prenatal care, as well as increases in institutional deliveries and child preventive care visits.⁷

Building on the success of the early PBF experiments, Rwanda has leveraged two other mechanisms to incentivize the better performance of the health system: community performance-based financing and imihigo.

Community performance-based financing

In 2009, the Rwandan MoH piloted further incentives in 31 health centers in the nation’s poorest localities with a “second generation” of PBF, community performance-based financing (CPBF). CPBF had two types of incentives: demand side for patients and supply side for workers. For the demand-side incentives, CPBF aimed to encourage more mothers and children to seek care by providing in-kind incentives to women (e.g., clothing and water treatment tablets at health center visits during the first four months of pregnancy, or baby soap, clothing, and bed sheets for women who delivered at health centers). For the supply-side incentives, CPBF provided financial rewards for community health worker (CHW) cooperatives.⁸ At first, 30% of these supply-side financial rewards went to individual CHWs and 70% to cooperative accounts, but the distribution structure changed in 2018. Since then, CHWs receive 100% of the reward, while the cooperatives are funded separately by the Ministry of Finance and Economic Planning.⁹

In the first nine months of the pilot, the average number of prenatal visits to the pilot health centers increased by 77% over the previous year, and institutional deliveries increased by 16%. Subsequently, the MoH scaled up CPBF nationally. In 2010, 69% of Rwanda’s pregnant women delivered in health facilities, up from 30% in 2005, and the gap in institutional deliveries between urban and rural areas was narrowing.⁸

Research has demonstrated that PBF in Rwanda enhanced PHC performance. For instance, one impact evaluation showed that children residing in a PBF district were less likely to be stunted than those in districts where the PBF intervention did not yet exist.¹⁰ Other studies suggest that PBF incentivizes demand for care such as giving birth in a health facility. ¹¹

Imihigo

Rwanda’s pioneering experience with PBF in its health sector inspired officials in other sectors to adopt a similar approach. In a 2006 meeting, President Paul Kagame challenged district mayors from across Rwanda to establish ambitious goals for their districts.¹² This challenge was rooted in a pre-colonial Rwandan practice known as imihigo, derived from the Kinyarwanda word for “vow to deliver,” in which leaders would make public promises—and face public humiliation if they failed to follow through. Mayors incorporated their goals, along with action plans for their implementation and metrics for measuring their success, into the Ministry of Local Administration (MINILOC)’s formal performance-management system. Within a few years, officials and staff at all levels of government were also required to sign imihigo performance contracts.¹³

• Over time, its design has evolved to include more rigorous assessments and greater citizen participation, aligning closely with national development goals. The approach emphasizes sustainability and inclusivity by involving communities in shaping and implementing solutions. Despite disruptions from the COVID-19 pandemic in 2019/2020, Imihigo remains central to achieving Rwanda’s long-term development vision, including Vision 2050.¹⁴

Official imihigo goals are also often linked to health system performance. These include building new health centers, equipping existing ones, and increasing health insurance coverage.¹² Health facilities also sign performance contracts with district mayors that link funding to performance on key health-related indicators, including morbidity and mortality outcomes and access to care. These are the bedrocks of a strong PHC system.¹⁵

Codifying and maintaining quality of care

Rwanda has also implemented measures to measure, codify, and maintain quality of care—a key element of a strong PHC system.

District hospital accreditation

In rich countries, hospital accreditation can often be a useful tool for measuring and maintaining hospital performance against a clearly defined set of standards compared to peer institutions. However, accreditation by international bodies is an extremely expensive process.¹⁶ Because initial performance-based evaluations of Rwanda’s public hospitals in 2005 and 2006 showed significant lapses that limited the quality of care they could provide, Rwanda’s MoH decided to develop an accreditation system of its own.¹⁷

To establish a baseline standard, in 2006 officials invited the Council for Health Service Accreditation of Southern Africa (COHSASA) to evaluate King Faisal Hospital Kigali, one of Rwanda’s three teaching hospitals.¹⁸ Its initial scores were low (41/100), but King Faisal used COHSASA’s findings to improve the quality of services it delivered. It has been COHSASA accredited since 2011.¹⁷

In 2013, the MoH developed a plan to implement a three-tier national accreditation system for all 43 of Rwanda’s district hospitals.¹⁷ The aim of the project, according to officials, is “to use the results of accreditation to inform performance-based financing payments, as a means of recognizing facilities that progress toward meeting the quality standards and ultimately achieve accreditation. Consequently, it is imperative that the standards are clear and that the process for measuring these standards is rigorous, reliable, and unbiased.”¹⁹

Rwanda’s MoH worked with technical experts to develop the Essential Hospital Accreditation Standards framework, adapted (according to International Society for Quality in Healthcare standards) from the Joint Commission International’s International Essentials of Health Care Quality and Safety to address five areas of risk to patients.¹⁷ A new Accreditation Steering Committee oversaw the process. By 2014, four of the five hospitals in the first phase had improved their aggregate scores across the focus areas. As of 2018, 28 of Rwanda’s 43 district hospitals were nationally accredited.¹⁷

The MoH merged this accreditation system with its PBF program and began doing performance accreditation assessments bi-annually at the central and health facility levels across the country.^15,20

As a tool for evaluating and improving the quality of care hospitals provide, performance accreditation boosts the public’s trust and confidence in those facilities. In turn, this can help boost utilization rates for key PHC services. It can also improve health outcomes.¹⁶

Assessing maternal mortality

In 2004, the World Health Organization (WHO) recommended that all countries establish maternal death audit (MDA) systems to collect accurate information about causes of maternal mortality.²¹ Rwanda was among the first low-income countries to do this. In 2008, Rwanda’s MoH implemented a three-part system for MDA—Confidential Enquiry into Maternal Deaths, facility-based death reviews, and community-based death reviews (also called verbal autopsies)—in health facilities nationwide.²² Standard tools for these three approaches were adapted to the local context and health care providers from all hospitals were trained. Rwanda also began conducting audits of neonatal deaths in 2010 and of stillbirths in 2015.¹⁵

This facility-based MDA approach has helped improve PHC—especially maternal and neonatal health services—by facilitating recommendations for better health outcomes and higher quality health care delivery, and is credited with reducing both maternal and neonatal deaths.²²

Between 2009 and 2013, MDA committees audited an average of 93% of maternal deaths each year. Over the same period, deaths attributed to unknown causes dropped by almost 80%.²² This allowed health facilities to evaluate maternal health systems much more accurately.

In 2014, after more specific WHO recommendations, the MDA began a transition to a nationally standardized Maternal Death Surveillance and Response system.²³ In 2016, this system expanded and became the Maternal and Perinatal Death Surveillance and Response system.²⁴ Under the new system, all Rwandan hospitals use the same processes and the same forms to collect the same data, and other health facilities had varying degrees of implementation and adoption.²⁵ Also, training was significantly more involved, growing to include not just hospital physicians but also obstetricians, pediatricians, and midwives. A 2018 assessment found that all hospitals under review were using the system.²¹

Using data for oversight and decision making

During the study period, Rwanda took significant strides to improve the level, flow, and quality of information gathered at various levels of its health system. In turn, this helped improve the responsiveness, efficiency, and effectiveness of the PHC it was able to deliver.

TRACnet and eIDSR

At the beginning of the study period, most of Rwanda’s record-keeping systems for HIV testing and monitoring were paper-based. This made it difficult for providers to track their patients, especially at a distance, and for officials to aggregate national data on the HIV epidemic.

To address this problem, in late 2004 the MoH’s Treatment and Research AIDS Center introduced TRACnet, an electronic system that enabled practitioners caring for patients undergoing antiretroviral therapy to submit reports and receive test results via a toll-free telephone number or a website that operated in French and English. Rwanda Tel and MTN-Rwanda Cel donated toll-free numbers and free network time.²⁶ The American mobile health developer Voxiva donated IT support.²⁷

The TRACnet system, which uses solar-charged mobile phones, is simple and takes less than 30 minutes of training to learn. It allows practitioners in remote areas to monitor and replenish antiretroviral (ARV) drug stocks in real time and see test results as soon as they are available. It also enables health officials to monitor patterns of transmission and manage outbreaks.²⁸ By the end of 2007, TRACnet covered all 168 facilities that treated Rwanda’s 43,000 patients on ARV therapy.²⁹ As of 2010, TRACnet showed that ARV coverage of eligible patients had increased from 13% in 2005 to 79%.³⁰ As of 2016, the time to reserve PCR results had dropped from 144 days to 23 days, and the time between PCR sampling and receipt of the results by health facilities decreased from 90 days to 5 days. In turn, this reduced the time it took for patients to begin ARV therapy.²⁷

In 2011, Rwandan health officials used the TRACnet system to develop and deploy an electronic Integrated Disease Surveillance and Response system (eIDSR) to track 23 communicable diseases and automatically summarize data in user-friendly electronic dashboards.³¹ Researchers argue that this surveillance system has made disease reporting in Rwanda more timely and complete, and has enabled officials to quickly detect and respond to outbreaks.³¹

The eIDSR and other systems for epidemic preparedness and prevention are a key element of PHC in Rwanda. They can keep some people from getting sick in the first place, and they can help steer others to the treatments and health services they need to avert serious illness.

RapidSMS and SISCom

In 2009, in an attempt to prevent maternal deaths and promote maternal and neonatal health, the MoH’s Community Health Desk piloted a free, open-source mobile health monitoring system called RapidSMS in Rwanda’s Musanze District.³² RapidSMS enabled community health workers (CHWs) to track antenatal care visits, refer at-risk patients to health facilities, and communicate with those facilities in case of emergency.³³ Within a year, prenatal care visits had risen by 25%, home deliveries had fallen by 54%, health facility deliveries had increased by 26%, and under-five mortality had dropped by 48%.³³

In 2011, developers integrated the système d’information sanitaire des communautés, or SISCom (a routine data collection and reporting software), into RapidSMS. The combined system facilitated timely decision making, improved monitoring of health trends, and enhanced responses to health emergencies.³⁴ In 2012, officials expanded RapidSMS to track the health of pre- and postpartum women as well as newborns and some young children, and in 2013 the program was scaled nationwide.³⁵ By 2018, research suggested that CHWs in Rwanda sent more than 9.3 million messages using RapidSMS, suggesting the program was successfully implemented and also correlated with a 100% increase on postnatal care visits per catchment population after one year of implementation.³⁶ However, subsequent surveys show that RapidSMS had little effect on the uptake of maternal and newborn health services in Rwanda, likely because existing uptake for these services was already quite high.³⁵

Health Management Information Systems

Rwanda initially started with fragmented systems to facilitate reporting of health-related data across the country. These systems included the open-source, modular OpenMRS (open medical records system) used for tracking patient medical information, which was useful on its own but was not integrated into other systems. In some cases , Rwanda’s 450 health centers manually aggregated data by sending flash drives to 40 district hospitals monthly. Separate web-based systems tracked CHWs, HIV services, and human resources, but these databases remained siloed and could not interact.³⁷ The reporting systems were not interoperable with one another.³⁸ As a result, though they gathered a considerable quantity of data, much of it had limited utility systemwide.³⁸

In 2009, the US Agency for International Development and Rwanda’s MoH launched the Integrated Health System Strengthening Project (IHSSP) to integrate, harmonize, and improve the efficiency of the country’s electronic health systems.³⁹ Since 2012, the country has used a single, web-based Rwanda Health Management Information System (R-HMIS) that collects data from more than 700 health facilities using the free, customizable web-based DHIS2 platform.⁴⁰ Subsequently, the IHSSP and the MoH began to train data managers to use all this data to make decisions.³⁷

As a result of the project, Rwanda saw improvements to leadership and governance at national and district levels, more sustainable health financing measures and private sector engagement, quality of care improvements, more decisions made by using evidence, and a better mobilized and skilled workforce.³⁷

Because the initial R-HMIS did not include key measures for high-impact interventions in maternal and child health, Rwanda’s MoH began to implement a process for HMIS review in 2016.^39,41 Since 2018, several other modules have also been added to the R-HMIS to measure quarterly tuberculosis program reporting, monthly HIV program reporting, case-based reporting on neonatal, child, and maternal death audits, and other indicators.³⁷

National Health Research Agenda

With the foundation of investments in data and information systems established, Rwanda has worked over time to increase efficiency and reduce redundancy in its health care and research ecosystem. In 2012 the country established a national Health Research Agenda, renewed every five years, to guide medical research in the country and encourage researchers to align their priorities with the health sector’s overall strategic plans.^42,43

Exemplars research found that all study districts were equipped with electronic information systems and used evidence from routine data sources and research to inform decision making. An interview respondent said: “One very important policy change is the digitalization of the different services and accountability mechanisms. This helps us to measure, follow up, improve, and make continuous improvements on whatever accountability mechanisms that are in place.”

Figure 11 shows the breadth of key health information services and utilization in each of the study districts. The most common information systems leveraged across study districts included electronic medical records/individual-level data collection systems and systems for staff management. Data use characteristics across study districts included incorporation of health facility-level data in HMIS and processes for routine data review at the facility level. Together, these systems better enable districts to serve their populations and manage essential staff.