Bayesian Spatio-Temporal Modelling of Reported Terminated Pregnancy Across Nigerian States (2013-2024)

IntroductionThe long-standing disconnection of abortion legislation in Nigeria with the estimated incidence of 1.8 million terminations a year has contributed to systematic gaps in reliable abortion data for health policy. Any subnational monitoring under conditions of legal restraint tends to remain hidden beneath under-reporting and spatial instability such that policy makers are not left with a clear picture of where and why these decisions are being made. Methods and AnalysisTo address this ambiguity, this paper traces the path of state-level evolution of reproductive choices within the 2013, 2018 and 2024 NDHS. We detected the latent socio-demographic causes of terminated pregnancy using a Bayesian spatio-temporal framework, such as wealth, education, literacy, and contraceptive prevalence. ResultsThe rates were highly spatio-temporally intense and polarized in the region, with probabilistic evidence to justify state-specific reproductive health interventions between 2013 and 2024. Southern and coastal states (e.g., Lagos, Bayelsa) demonstrated sustained increases in prevalence in line with a high fertility transition, termination is more reproductive agency, access to services and reporting. Conversely, the unmet contraceptive need and structural vulnerability were the major causes of increased rates in the northern states (e.g., Yobe, Kano). Patterns of determinants also changed with time: in previous surveys, household wealth turned out to be a protective factor, as of 2024, education and literacy had become the strongest predictors. ConclusionsSuch findings affirm a dual reproductive regime in Nigeria--choice based in the South and vulnerability based in the North necessitating a shift from homogenous national approaches to state-specific reproductive health policies. What is already known on this topicStudies have noted the continuous disparities in the maternal and reproductive health indicators between northern and southern Nigerian states. Nevertheless, most of the studies done before were based on cross-sectional analysis and national-level summaries. Not many considered spatial dependence among states or studied how decisions on termination vary over time. What this study addsThrough shared modelling of spatial effects, temporal trends and space-time interactions, it establishes consistent high-risk conditions, arising hotspots and areas with decoding risk. The Bayesian model enhances the accuracy of the estimation since it takes into consideration the spatial correlation and the strength of borrowing on neighboring states. How this study might affect research, practice or policyIn the case of research, the study offers a methodological approach to the analysis of other maternal and public healthcare indicators by small-area estimation methods. Practically, with high-risk and emerging hotspots states identified, reproductive health more focused interventions can be implemented and limited resources can be efficiently allocated. To the policy, the study provides state-specifics evidence to inform subnational reproductive health planning and monitoring.