Objective Our objective was to assess the feasibility of using the routine health information system data source, District Health Information System (DHIS2) to measure the effective coverage of selected health service indicators in Ethiopia and to explore stakeholder perceptions of those measures. Design We conducted a mixed-methods study. We mapped the availability of data elements in DHIS2 between July 2022 and June 2023 for five indicators (four or more antenatal care visits (ANC4+), skilled birth attendance, postnatal care, sick child care and child nutrition care). We defined effective coverage cascade steps for each indicator, assessed data quality and analysed data using STATA V.17. Finally, qualitative interviews were conducted with 15 key stakeholders, and the data were analysed thematically for reflections on the DHIS2 output. Setting The data were captured from all public health facilities of 11 regions and 2 administrative cities in Ethiopia. Results There was better availability of data elements for maternal healthcare than for child healthcare. It was possible to estimate the intervention-adjusted coverage of ANC4+ (16% nationally) and the process-quality-adjusted coverage of skilled birth attendance (19% nationally). Postnatal care, sick child care and child nutrition indicators lacked data across multiple cascade steps. The quality of data for effective coverage measurement differed by region. The key informants expressed concerns about the adequacy and appropriateness of DHIS2 data for this analysis. While all acknowledged its potential for decision-making, respondents emphasised the need for standardised methods and data sources to enhance comparability and acceptability of the findings. Conclusions The findings underscore the need for system-level improvement of data availability and quality, and adoption of a standardised approach to calculating effective coverage using DHIS2. There was a concern that the findings may not be accepted by policymakers; however, the local level granularity made possible through DHIS2 was appreciated.
Lemma et al. (Thu,) studied this question.