The value of mortality data to advance learning and decision-making dynamics in Mozambique’s health system

Abstract Introduction There is a lack of evidence on whether and how health systems stakeholders in Mozambique learn from routine health data. Despite the recurring claim by the stakeholders themselves that data in general should be the centre of informed public health policy, it remains unclear how decision-makers and managers access, interpret, and use mortality data for learning and decision-making. Using the Learning Health System (LHS) lens, we examined how mortality data are accessed, interpreted, and used across system levels in Mozambique. Methods We conducted a qualitative case study from September to November 2024, involving 20 key informants from the Ministry of Health at both national and subnational levels, non-governmental organisations, data producers, and users. Data were collected in the districts of Manhiça and Quelimane in Mozambique, through semi-structured interviews, transcribed, coded, and thematically analysed using NVivo12, guided by the LHS dimensions: learning levels, learning loops, and means of learning. Results Participants recognised the value of mortality data for learning and decision-making, but routine use remained uneven. At the central Ministry of Health level, mortality data informed strategic adjustment and emerging double-loop learning, particularly through reprioritisation, planning, and resource reallocation discussions. At the subnational level, mortality data mainly supported single-loop learning, where routine reports and outbreak signals triggered immediate adjustments in clinical practice and public health response. Delays in data access, low data literacy among decision-makers, fragmented information systems, and centralised decision-making appear to be the greatest challenges to the effective use of mortality data. Budget authority also remains centralised, and allocation decisions are largely locked into the wage bill rather than the system’s needs in terms of functionality. Heavy dependence on central directives limits local autonomy and its ability to act, preventing new ways of learning and working. Conclusion Mortality data already support operational and strategic learning in Mozambique, particularly through single- and double-loop processes. Systemic transformation in how the system learns remains limited and is exacerbated by structural barriers within the health system. Priorities include improving timely access to mortality information across levels, strengthening analytical and interpretive capacity, integrating fragmented systems, and widening delegated decision space, which will ultimately enhance evidence-based planning, local action, and health system learning.