Abstract
BACKGROUND: Religious leaders are trusted in Bangladesh and can promote maternal and child health (MCH)in underserved communities. We assessed the population-level impact of a national training program for religious leaders on knowledge, attitudes, practices, and behaviors (KAPB) related to antenatal care (ANC), facility-delivery, and essential newborn care (ENC). METHODS: We conducted repeated cross-sectional household surveys among imams, female religious-school teachers, and parents of children aged 0-2 years in 17 clusters across eight districts in Bangladesh at baseline (2019, n = 3,808) and endline following the training program (2021, n = 3,746). Standardized questionnaires captured domain-specific KAPB indicators. Multivariable logistic regression compared endline with baseline, adjusting for age, sex, education, and occupation. RESULTS: Baseline and endline groups were socio-demographically similar, with small differences in education. Compared with baseline, endline participants showed about two-fold increase in ANC knowledge (aOR 2.3, 95% CI 2.1-2.6), positive attitudes (aOR 1.5, 95% CI 1.4-1.6), and reported ANC use (aOR 2.3, 95% CI 2.1-2.6). Knowledge of home-delivery risks increased markedly (aOR 11.9, 95% CI 9.5-14.9), alongside modest improvements in facility-delivery (aOR 1.5, 95% CI 1.4-1.6). Knowledge of essential newborn care improved greatly (aOR 8.9, 95% CI 7.8-10.1), with better early breastfeeding and bathing practices (aOR 1.8, 95% CI 1.7-2.0). Religious leaders also reported sharing health messages more frequently at endline. CONCLUSION: Despite being a quasi-experimental study that limits causal inference, training religious leaders was found to be associated with improved community knowledge and practices across the MCH care continuum. Findings suggests that integrating such engagement into national health communication platforms may enhance coverage in underserved settings. A key limitation of this study is its quasi-experimental design, which limits causal inference, and the reliance on self-reported outcomes, which may be subject to recall and social desirability bias.