Regional trends and forecasting of under-five child mortality in Bangladesh: A mixed method approach of Functional ANOVA and Bayesian spatiotemporal modeling

Background: This study aims to estimate historical under-five mortality rates from 1990 to 2022 across the eight divisions of Bangladesh, analyze regional disparities, and forecast under-five mortality rates trends up to 2030 considering both temporal and spatial dynamics. Methods: A cross-sectional study design was employed, utilizing nationally representative data from the Bangladesh Demographic and Health Survey 2022. The study included 64,697 records of children representing births from eligible women with full birth history across the eight divisions of Bangladesh. The dataset spans 33 years, enabling a detailed examination of under-five mortality rates trends at the sub-national level. The analysis included historical under-five mortality rates estimation and Functional ANOVA was conducted to explore regional disparities, while a Bayesian Spatiotemporal model was employed to forecast under-five mortality rates trends up to 2030. The permutation F-test was used to assess the statistical significance of the regional differences. Results: The permutation F test conclude that while regional differences in child mortality rates were not globally significant over the entire period from 1990 to 2022, significant disparities did emerge during specific years (early 1990s and late 2010s). Barishal, Rangpur, and Sylhet showed high variability in under-five mortality rates reductions in the early 1990s, while Dhaka and Khulna exhibited more consistent progress. ICAR spatial effects indicate that Sylhet and Rangpur, have slower reductions, highlighting persistent challenges in these areas despite national progress. Additionally, spatiotemporal modeling indicates all divisions in Bangladesh show significant reductions from 1990 to 2022, with Sylhet consistently having the highest rates and Khulna the lowest. Projections indicate a plateauing trend, with most divisions nearing the sustainable development goal target of 25 deaths per 1000 live births by 2030, though Sylhet and Rangpur may require additional interventions. Conclusions: Future strategies should focus on reinforcing healthcare infrastructure, enhancing maternal and child health services, and reducing socioeconomic inequalities to achieve faster improvements in high-mortality regions. Targeted interventions in these areas are necessary to achieve the SDG target by 2030, thereby improving child health outcomes nationwide.