Background The objective of this study is to compare all-cause mortality and drowning-specific mortality rates and survival times among birth cohorts of children from rural Bangladesh over a 14-year period.

Method Seven birth cohorts of children aged 12–59 months were created using data from the Matlab Health and Demographic Surveillance System in Bangladesh. Each cohort represents children born within a 2-year interval between 1 July 1995 and 30 June 2009, who were then evaluated over a 2-year snapshot. All-cause and drowning-specific mortality rates were compared for each cohort using the oldest cohort (cohort 1) as reference. A Cox proportional model was used to estimate hazard rate ratios (HRR) comparing the cohorts, and adjusted for key independent variables. Kaplan-Meier survivor function was estimated for each birth cohort and compared with cohort 1 using log-rank test.

Results The adjusted HRR for all-cause mortality showed that children in cohorts 4, 5, 6 and 7 had significantly lower hazard rate compared with cohort 1 with a dose–response relationship, the adjusted HRRs were 0.51 (95% CI 0.31, 0.84), 0.53 (95% CI 0.32, 0.87), 0.44 (95% CI 0.26, 0.76) and 0.42 (95% CI 0.24, 0.74), respectively. For drowning, none of the adjusted HRR was statistically significant. Mother's primary education and being married were protective of risk of death from all causes.

Conclusion Whereas child mortality rates from all causes declined remarkably between 1998 and 2012 in rural Bangladesh, drowning-specific mortality rates remained unchanged. This shifting burden of disease underscores the epidemiological transition in the childhood causes of death in Bangladesh and the need for urgent action to review child health interventions.