An Inquiry into Health Systems Governance: the Case of Tanzania

Over the last decade, the global movement towards universal health coverage (UHC) gradually shifted its focus. For many low- and middle-income countries (LMICs), the late 1990s and early 2000s have been characterized by strong and long-lasting efforts to eradicate infectious diseases and improve maternal and child health, expanding access to effective healthcare services. Over the following decade, global health institutions started to emphasize the role of health systems governance and the need to address high quality of care across local health systems. Governance has been identified as a building block of all health systems and deemed as conducive of improved social health protection, equity in health and increased quality of healthcare provision. Despite its importance, international institutions and researchers failed to provide a unanimous definition of health systems governance.
This dissertation aims to contribute to the debate about health systems governance in LMICs, reflecting on the case of a large sub-Saharan country: Tanzania. The broad goal of this thesis is to distil some operational governance dimensions and assess their role in health services provision in Tanzania. The introductory chapter provides an overview of the role of health systems governance in LMICs, offering a working definition of governance that is used throughout the manuscript. Part I also presents the specific research objectives, outlining the methods applied in the subsequent parts of the dissertation. Part II introduces the reader to the health system and the health policy context in Tanzania, focusing on the governance challenges that characterize this large East African country. Part III looks at the role of incentives to health workers and supervision for the provision of high-quality healthcare. The study is based on a multilevel regression analysis of secondary data on a large sample of Tanzanian public health facilities from the Service Provision Assessment (SPA) survey conducted in 2014/2015. In part IV, the focus switches on the impact evaluation of social accountability initiatives on healthcare provision. The analysis builds on the implementation of a social accountability monitoring program - by the local NGO Sikika - in the Tanzanian region of Dodoma. Using a difference-in-differences approach, the research aims at identifying the effect of the program on two specific outcomes targeted by Sikika: availability of essential medicines and infrastructure maintenance in government-managed health facilities. Part V addresses the design, management and regulation of patient referral systems in settings characterized by strong focus on primary healthcare. The study is based on primary data collected in the two rural districts of Kilolo (Iringa region, Southern Highlands zone) and Msalala (Shinyanga region, Lake zone). The analytical approach builds on previous work in the field of social network analysis, mapping the available data to networks with health facilities representing nodes and patient referrals representing directed edges. The discussion in part VI summarizes the previous results putting them into perspective, outlining a framework where interrelated governance dimensions contribute to UHC goals. The section concludes touching upon the limitations of the proposed approaches and discussing the implications for health policy.
The findings can be grouped in four main areas, reflecting the operational governance aspects explored empirically across the manuscript. The first area relates to top-down monitoring and supervision arrangements. The Tanzanian case highlights the ineffectiveness of initiatives meant to monitor the activity of local public healthcare providers spread across vast rural areas. Quality of care at point of service shows no association with the intensity of supervision at facility and individual level. Although measurement errors could prevent effective effect identification, the results cast doubts on the implementation of such arrangements. The second area addresses the role of incentive policies to push health workers towards the provision of high-quality healthcare. Financial and non-financial incentives offered to health workers show positive associations with the two measures of quality of care analysed, namely compliance with guidelines for treatment of sick children and patient satisfaction. Salary top-ups improve the otherwise poor salary conditions faced by health workers in public health facilities across the country. The analysis reveals positive effects on both indicators of quality. Among the non-financial incentives included in the analysis, subsidized housing offered to health workers appears to be positively associated to patient satisfaction. The study suggests an interpretation relating increased satisfaction to closeness between communities of health workers living in the health facility compound. The third area is related to bottom-up supervision and social accountability mechanisms. On the one hand, frequency of meetings between health facility staff and local communities is associated with increased patient satisfaction. The result reinforces the finding on subsidized housing, supporting the idea that closeness between providers and communities improves patient satisfaction. A potential pathway for this influence is the increased accessibility of providers (for example in case of emergencies) and improved ability to address the needs of the community. On the other hand, the social accountability monitoring program implemented in the Dodoma region reveals potentially positive effects on availability of essential medicines. Besides favouring effectiveness and appropriateness of treatments, the reduction of stock-outs in essential medicines contributes to improved patient satisfaction and in turn to positive healthcare seeking behaviour. The study does not detect any impact of social accountability monitoring on infrastructure maintenance, possibly suggesting different impact mechanisms related to the need of district level budgeting and resource allocation. Finally, the fourth area discusses the implications of design and regulation of patient referral systems. The results emphasize the importance of appropriate investments in infrastructure, especially in secondary level facilities (i.e. health centres). The study reveals that in Kilolo district most patient are referred directly to the regional referral hospital in Iringa Town, mainly as a result of geographic closeness and infrastructure availability. Health centres have a little role in terms of mediating referrals between primary and tertiary level facilities. A failure in the gatekeeping system may result in negative consequences in terms of increased financial burden for households required to travel to the crowded hospital words in Iringa Town. The analysis also suggests that a strategic orientation towards quality of care at district level may positively influence the functioning of patient referral systems. In Msalala district, besides higher rates of referrals to health centres, the results are consistent with a virtuous cycle of referrals directed towards smaller and more specialized facilities. Although the study is not designed to detect causal effects, these positive results may be associated to the implementation of a system-wide project aimed at strengthening different components of the health systems.
Overall, the dissertation sheds light on the complex interaction between the different health systems governance dimensions analysed. What emerges is a set of mutually connected dimensions that contribute together to the three main goals of UHC: coverage and access, financial protection and the delivery of quality health services. The conclusion offers a number of suggestions to policy makers in relation to the potential outcomes associated to the implementation of different governance tools. On a more general note, the results are consistent with the ongoing efforts to promote systems thinking in health systems research. This dissertation supports the message that health system governance is better addressed with a comprehensive approach rather than with partial policies, unleashing a multiplicative effect of self-reinforcing interactions between different governance functions and tools.