Effectiveness of a large-scale handwashing promotion intervention on handwashing behaviour in Dhaka, Bangladesh

Objective: The behavioural effect of large-scale handwashing promotion programmes has been infrequently evaluated, and variation in the effect over time has not been described. We assess the effect of a large-scale handwashing promotion programme on handwashing outcomes in a community setting in Dhaka, Bangladesh. Methods: We analysed data from a cluster-randomised trial that included three arms: vaccine-and-behaviour-change intervention (VBC), vaccine-only (V) and no intervention (Control). Data collectors randomly selected different subsets of households each month during the study period and assessed: (i) temporal variation in availability of soap and water at handwashing place; (ii) the use of water and soap by participants when asked to demonstrate handwashing, and; (iii) handwashing behaviour according to structured observation. We used log-binomial regression analyses to calculate prevalence ratios (PRs) and 95% confidence intervals and compare outcomes by study arms. Results: Data collectors surveyed 9325 households over 28 months. In VBC, there was a significant positive trend on availability of water and soap from baseline to 9 months after the start of the intervention (P-for-trends <0.001), and no significant trend during months 10–28 (P-for-trend = 0.297). In the entire study period, availability of water and soap was higher in VBC (43%) than in V (23%) (PR = 1.92; CI = 1.72, 2.15) and Control (28%) (PR = 1.53; CI = 1.38, 1.69) households. There were no differences between study arms with regard to use of soap during handwashing demonstrations. Observed handwashing with soap after toilet use was higher in VBC (17%) than in V (8%) (PR = 1.47, CI = 0.58, 3.75) and Control (2%) (PR = 3.47, CI = 0.48, 23.33) groups. At other possible pathogen transmission events, the prevalence of handwashing with soap was ≤3%. Conclusion: VBC households maintained soap and water for handwashing, but the prevalence of observed handwashing was low in all study arms. The results underscore the need to strengthen scalable behaviour change approaches.