RESUMO
Performance management (PM) reforms have been introduced in health systems worldwide to improve accountability, transparency and learning. However, gaps in evidence exist regarding the ways in which PM contributes to organizational-level outcomes. Between 2015 and 2017, the government of El Salvador and the Salud Mesoamerica Initiative (SMI) introduced team-based PM interventions in the country's primary health care (PHC) system including target setting, performance measurement, provision of feedback and in-kind incentives. The programme's evaluation showed widespread improvements in performance for community outreach and service timeliness, quality and utilization. The current study characterizes how the implementation of team-based PM interventions by SMI implementers contributed to PHC system performance improvements. We used a descriptive, single-case study design informed by a programme theory (PT). Data sources included qualitative in-depth interviews and SMI programme documents. We interviewed the members of four PHC teams (n = 13), Ministry of Health (MOH) decision makers (n = 8) and SMI officials (n = 6). Coded data were summarized, and thematic analysis was employed to identify broader categories and patterns. The outcomes chain in the PT was refined based on empirical findings that revealed the convergence of two processes: (1) increased social interactions and relationships among implementers that enhanced communication and created opportunities for social learning and (2) cyclical performance monitoring that generated novel flows of information. These processes contributed to emergent outcomes including the uptake of performance information, altruistic behaviours in service delivery and organizational learning. Through time, the cyclical nature of PM appears to have led to the spread of these behaviours beyond the teams studied here, thus contributing to system-wide effects. Findings illustrate the social nature of implementation processes and describe plausible pathways through which lower-order implementation programme effects can contribute to higher-order changes in system performance.