Characterizing performance improvement in primary care systems in Mesoamerica: A realist evaluation protocol.

Background. Improving performance of primary care systems in low- and middle-income countries (LMICs) may be a necessary condition for achievement of universal health coverage in the age of Sustainable Development Goals. The Salud Mesoamerica Initiative (SMI), a large-scale, multi-country program that uses supply-side financial incentives directed at the central-level of governments, and continuous, external evaluation of public, health sector performance to induce improvements in primary care performance in eight LMICs. This study protocol seeks to explain whether and how these interventions generate program effects in El Salvador and Honduras. Methods. This study presents the protocol for a study that uses a realist evaluation approach to develop a preliminary program theory that hypothesizes the interactions between context, interventions and the mechanisms that trigger outcomes. The program theory was completed through a scoping review of relevant empirical, peer-reviewed and grey literature; a sense-making workshop with program stakeholders; and content analysis of key SMI documents. The study will use a multiple case-study design with embedded units with contrasting cases. We define as a case the two primary care systems of Honduras and El Salvador, each with different context characteristics. Data will be collected through in-depth interviews with program actors and stakeholders, documentary review, and non-participatory observation. Data analysis will use inductive and deductive approaches to identify causal patterns organized as 'context, mechanism, outcome' configurations. The findings will be triangulated with existing secondary, qualitative and quantitative data sources, and contrasted against relevant theoretical literature. The study will end with a refined program theory. Findings will be published following the guidelines generated by the Realist and Meta-narrative Evidence Syntheses study (RAMESES II). This study will be performed contemporaneously with SMI's mid-term stage of implementation. Of the methods described, the preliminary program theory has been completed. Data collection, analysis and synthesis remain to be completed.