Improvements in diarrheal disease prevalence with point-of-use water filter implementation in the informal settlement of Kibera, Kenya

BackgroundThere is increasing evidence of the efficacy of point-of-use water filters on diarrhea prevalence in numerous global settings, in both observational studies and randomized experiments. Most studies, however, are focused on rural locations. Methods We use self-report household surveys to monitor a set of approximately 10,000 households receiving point-of-use water filters and WASH training in Kibera, Kenya. Twenty-five drinking water sources throughout the 7 neighborhoods were also selected for testing of E. coli, total coliform, bacterial 16S rRNA community sequencing and metals. Albendazole was provided to all households at distribution as part of the standard filter distribution protocol, with a subset of 2,642 households not receiving Albendazole at distribution, instead receiving it at the second follow-up (approximately 5 weeks after filter distribution). ResultsAfter data cleaning, a sample of 6,795 households were analyzed using mixed effects generalized linear models to account for repeated household measurements, geospatial and temporal effects, interviewer and other household covariates. Models predicted self-reported, 2-week prevalence of diarrhea. After accounting for confounding factors, self-reported diarrhea rates dropped from 52.7% to 2.2% after approximately 70 days of filter use. Field testing characterized most water sources (18 out of 25) as unsafe for Total coliforms, many for E. coli (6 out of 25), and one source above WHO health guidelines for arsenic. There was no evidence of a difference in self-reported diarrhea prevalence between households receiving Albendazole at distribution vs. those that didnt (p>0.05). ConclusionsThe introduction of Sawyer filters to households in a densely populated informal settlement reduced diarrhea and other health related problems. Representative water quality testing indicates a high frequency of drinking water source contamination with E. coli and Total Coliforms but a very low frequency of dissolved metals present, above WHO guidelines for drinking water. Anti-parasitic medication distribution had little to no impact on the results. Future randomized controlled studies with objective health measures are needed to ensure cause-effect impact of the filters, and study of filter longevity in the field continues to be a critical need.