Small-area estimation of district-level fertility in 36 countries in sub-Saharan Africa 2000-2025

Introduction: District-level estimates of fertility are required by policymakers and programme planners in sub-Saharan Africa (SSA) to guide decision making at the local level. Existing approaches to produce fertility estimates at the district level use age-structures or total fertility trends over time from higher administrative levels which prevents representation of district-level heterogeneity in fertility estimates. Methods: We extracted observed numbers of births and person-years stratified by five-year age group, single calendar year, and district from 194 nationally-representative household surveys. We constructed a spatiotemporal Bayesian hierarchical model that reconciles point and areal data and adjusts for non-sampling biases to estimate age-specific (ASFR) and total fertility rates (TFR) in 36 countries between 2000-2025. Results: Fertility rates were calibrated to three million births from two million women over fifteen million person-years. TFR declined in the majority of countries 2000-2020 (median -15%, interquartile range (IQR) -3 to -24%). Substantial subnational heterogeneity in TFR levels was observed in most countries, exceeding that of differences between national fertility levels. ASFR heterogeneity was observed, though to a lesser degree than variation in total fertility. Teenage fertility was unchanged in several countries, with fertility declines driven by older age groups. Conclusion: District-level fertility dynamics can differ greatly from national TFR trends and age-patterns of fertility. Use of national fertility rates will produce inaccurate estimates of births in the majority of SSA districts, leading to inappropriate resource management for key global public health objectives.