Primary health care in Nigeria: a systematic subnational analysis of levels and trends in maternal and child interventions 2000–13

Abstract

Background: Nigeria is a large and diverse country with considerable social and economic variation, yet little is definitively known about the status of health intervention coverage and health outcomes at a subnational level. To address this knowledge gap, we developed methods to produce a comparable set of estimates that describe the trends and levels of key health indicators at the state level for all years between 2000 and 2013. Methods: We first systematically sought out all data sources that measure indicators of interest. We sourced unit-level data where possible, but in some cases used published estimates. We extracted data on all-cause under-5 mortality, maternal and child health interventions, health system access, and socioeconomic factors for Nigeria's 37 states from 1998 to 2013. With the exception of mortality, which we estimated using methods described elsewhere, we developed a systematic approach to generate comparable time trend estimates. First, we tested more than 100 different model specifications for each indicator, accounting for time, covariates, and space in various ways, including linear models, natural cubic spline models, and spatially linked random effects. The best model for each indicator was chosen on the basis of predictive validity in a cross-validation framework incorporating a number of potential data missingness patterns. Finally, we estimated a mean posterior function with uncertainty using Gaussian process regression. In total, we produced complete time-series estimates for 35 coverage indicators, eight outcome indicators for under-5 health including mortality, and 16 indicators of socioeconomic status for every state in Nigeria. Findings: At the national level, most indicators have shown modest trends towards increased coverage of primary care interventions, concurrent with improved indicators of socioeconomic status and improved health outcomes. However, there is a great deal of variation within Nigeria. While all states saw a declining trend in under-5 mortality, we estimated a difference of 137 deaths per 1000 births between the state with the lowest level of child mortality (Edo State with 72 deaths per 1000 births) and the state with the highest level of child mortality (Zamfara with 209 deaths per 1000 births) in 2013. Results for intervention coverage were even more mixed, where some states saw trends decreasing and others increasing. For example in Kebbi state, skilled antenatal care coverage increased by 10% in the 13 year period, while Plateau saw a 15% decrease in coverage. Interpretation: Although Nigeria overall has made progress in the delivery of major primary care interventions, we note substantial within-country variation in coverage and health outcomes, suggesting high levels of inequality within Nigeria. Our estimates allow health policymakers and donors to prioritise interventions in areas with the greatest need: to this end, an online, publically available data visualisation tool for our results is available. Funding: Bill and Melinda Gates Foundation.