Longitudinal Effects of a Sanitation Intervention on Environmental Fecal Contamination in a Cluster-Randomized Controlled Trial in Rural Bangladesh

Abstract

Household latrine access generally is not associated with reduced fecal contamination in the environment, but its long-term effectiveness has not been measured. We conducted an environmental assessment nested within the WASH Benefits Bangladesh randomized controlled trial (NCT01590095). We quantified E. coli and fecal coliforms in samples of stored drinking water, child hands, mother hands, soil, and food among a random sample of households from the sanitation and control arms of the trial. Samples were collected during eight quarterly visits approximately 1-3.5 years after intervention initiation. Overall, there were no substantial differences in environmental fecal contamination between households enrolled in the sanitation and control arms. Statistically significant reductions were found in stored water and child hands after pooling across sampling rounds, but the effects were small and not consistent across rounds. In addition, we assessed potential effect modification of intervention effects by follow-up time, season, wealth, community-level latrine density and coverage, population density, and domestic animal ownership. While the intervention had statistically significant effects within some subgroups, there were no consistent patterns of effect modification. Our findings support a growing consensus that on-site latrines are insufficient to prevent fecal contamination in the rural household environment.

Keywords: E. coli; WASH; child feces management; disease transmission pathways; environmental sampling; fecal indicator bacteria; latrine; potty.

Conflict of interest statement

The authors declare no competing financial interest.

Figures

Figure 1

Prevalence (left column) and mean log10 most probable number (MPN) (right column) of E. coli with 95% confidence intervals by sample type, study round, and intervention arm. Each row contains a separate sample type. Units by sample type are 100 mL of stored water, two child/mother hands, and one dry gram of soil/food. Each study round took place approximately three months after the preceding round.

Figure 2

Effect of the intervention on log-transformed E. coli counts in environmental samples by sample type and round. Unadjusted log10 differences for sanitation vs control arms and 95% confidence intervals are shown for each sampling round and for all rounds pooled. Each row contains results for a separate sample type. Units by sample type are 100 mL of stored water, two child/mother hands, and one dry gram of soil/food. Each study round took place approximately three months after the preceding round.

Figure 3

Effect modification on log-transformed E. coli counts in environmental samples by season, year, household wealth, and hygienic latrine density, hygienic latrine coverage, and population density within 100 m. Unadjusted log10 differences for sanitation vs control arms and 95% confidence intervals are shown for individual categories of each potential effect modifier. Latrine density was defined as the number of hygienic latrines per capita within 100 m, and latrine coverage as the percent of compounds within 100 m with at least one hygienic latrine. Latrine and population categories represent tertiles. Each row contains results for a separate sample type. Units are 100 mL of stored water and two child/mother hands.

Figure 4

Effect modification on log-transformed E. coli counts in environmental samples by domestic animal ownership. Unadjusted log10 differences for sanitation vs control arms and 95% confidence intervals are shown for individual categories by animal type. “Other” includes donkeys, pigs, dogs, and cats. For each animal type, the categories represent those with no animals and tertiles of number of animals owned. Each row contains results for a separate sample type. Units are 100 mL of stored water and two child/mother hands.