Assessing Regional Disparities in Bangladesh: A Comparative Cluster Analysis of Health, Education, and Demographic Indicators across Districts

Background: For the development of evidence-based health policies and public health research, representative health information is essential. Often, in developing nations, studies extrapolate data from a small number of communities to the entire population, potentially leading to inaccuracies. This study utilises multivariate cluster analysis to examine regional disparities within a developing country using health indicators from the Bangladesh Multiple Indicator Cluster Survey (MICS) 2019 and demographic variables from the Bangladesh Population Census Report 2022.

Objective: The study aims to analyze disparities in socio-economic indicators across Bangladesh's districts to guide balanced development policy-making.

Methods: Indicators for the study were selected through a two-phase evaluation, retaining only those with significant variations within the dataset. The study focused on maternal, infant, and socio-demographic characteristics at a district level. The data analysis was conducted using hierarchical, kmeans, and pam clustering techniques, with the optimal number of clusters determined using a silhouette diagram. The cluster selection was validated through internal validation and stability tests.

Results: Two distinct clusters of districts showed significant disparities in health, education, and demographic indicators. The first cluster (21 districts) had lower literacy rates (45% vs 73%), school attendance (65% vs 85%), and early childhood education enrollment (25% vs 58%). This cluster also had higher rates of child stunting (40% vs 23%), wasting (16% vs 9%), maternal mortality (239 vs 140 per 100,000 live births), and unemployment (12% vs 6%) compared to the second cluster (43 districts). These findings highlight the need for targeted interventions.

Conclusion: The study demonstrates the potential for unsupervised learning techniques like cluster analysis in identifying regional disparities in developing countries. It emphasises the importance of individual district-level data in policy planning and underscores the need for targeted interventions to address specific regional health challenges.