Where and how can WASH work? Understanding limited impacts from a randomized control trial of water, sanitation, and hygiene interventions in a high burden setting

BackgroundDespite their strong theoretical basis, water, sanitation, and hygiene (WASH) interventions have had inconsistent benefits on diarrhea in low- and middle-income settings. The WASH Benefits (WASH-B) Kenya randomized controlled trial evaluated a set of WASH interventions targeted at children under age 2 and found no effect on diarrheal prevalence. ObjectivesWe explored whether and how changes to intervention and contextual factors could impact health gains as a means to inform future WASH interventions. MethodsWe implemented a compartmental transmission model with environmental pathways and water (W), sanitation (S), and hygiene (H) interventions fit to WASH-B Kenya trial data (n=11,856) using a Bayesian sampling approach. We developed counterfactual simulations to predict how a trial might perform given improved 1) local contextual factors (i.e., reduced transmission, and increased completeness of transmission pathways targeted) and/or 2) intervention factors (i.e., increased intervention efficacy, compliance, and community coverage). We considered change in intervention effectiveness in the counterfactual scenarios for each intervention alone and in combination (WSH). We determined what combinations would be needed to achieve 50% reduction in child diarrhea compared to the control arm. ResultsWe found that high diarrheal prevalence in the WASH-B Kenya trial was likely the primary reason for ineffectiveness. While none of the tested counterfactual factors greatly impacted intervention effectiveness in isolation, we estimated that 50% reduction in child diarrhea in the combined WSH treatment arm could be achieved through a combination of substantial intervention improvements (i.e., 50% intervention efficacy, 100% compliance, and 60% community coverage), but could not be attained for the single intervention arms. With improvements to contextual factors (consistent, 7.5% diarrheal prevalence, 50% increase in completeness) coupled with more modest increases in intervention factors (i.e., 50% efficacy and 100% compliance but only 20% community coverage), could achieve a 50% reduction in diarrhea in the combined WSH arm. ConclusionsIn settings with high enteric pathogen prevalence, WASH interventions must be used by a substantial fraction the population and block all main transmission routes to achieve substantial reductions in diarrheal disease burden, including those over age 2. The WASH interventions and targeting strategy for the WASH-B Kenya trial were unlikely to appreciably reduce diarrheal disease because of the high burden. In settings with more modest transmission, there are intervention factor targets that could result in measurable reductions in diarrhea. Application of this simulation-based approach could inform WASH policies and programs, as well as the design of future trials.