Enhancing maternal and child health service coverage estimates at the sub-national level in Nigeria: a methodological study using routine health information systems data

Background Reliable and timely data are essential for addressing maternal and child health challenges in Nigeria, where substantial regional disparities persist. Routine health information systems (RHIS) data offer a potential source of frequent subnational coverage and prevalence estimates, but their use depends on the quality of reported events (numerators) and appropriate estimation of the population at risk (denominators). This study evaluated a methodology to improve RHIS-derived numerators and denominators for generating timely subnational coverage and prevalence estimates in Nigeria. Methods RHIS data from three Nigerian states (2017–2022) were used to estimate four or more antenatal care (ANC4) coverage and infant fever prevalence. Unadjusted and adjusted estimates were calculated following a comprehensive data quality assessment and imputation process. Adjustments accounted for private sector service usage, incomplete reporting, non-use of services, and stillbirths and pregnancy loss for ANC4 estimates. Service-based and population-based denominators were used. Estimates were triangulated with population-based survey (PBS) data. Results ANC4 reporting completeness increased from 25.9% to 89.2% after imputation, and infant fever reporting completeness increased from 60.5% to 84.7%. Adjusted ANC4 coverage using service-based denominators aligned more closely with PBS estimates than unadjusted estimates or those using population-based denominators, with about half falling within the 95% confidence interval of PBS estimates. Adjusted ANC4 estimates using service-based denominators in Kebbi (range: 28.9%-33.0%) and Zamfara (range: 19.7%-24.5%) showed smaller discrepancies from PBS estimates than those in Ebonyi (range: 29.4%-36.0%). Quarterly infant fever prevalence estimates demonstrated strong seasonality and good agreement with PBS estimates, with 7 out of 9 values falling within the PBS 95% confidence interval. Adjusted infant fever prevalence using population-based denominators aligned most closely with PBS estimates, followed by those using service-based denominators. Adjusted fever estimates using population-based denominators in Kebbi (range: 7.1%-19.1%) showed the strongest agreement with PBS estimates, followed by Zamfara (range: 13.3%-26.1%) and Ebonyi (range: 4.1%-11.7%). Conclusions In these three Nigerian states, effective adjustment of RHIS data improved the comparability of subnational coverage and prevalence estimates. With further evaluation, these methods could be applied to other indicators and similar low- and middle-income countries, to support program monitoring and health decision-making.