We investigate the impacts of access to piped water on drinking water quality, sanitation, hygiene and health outcomes in marginalized rural households of north-western Bangladesh, using a quasi-experimental setup. A government organization – the Barindra Multipurpose Development Authority (BMDA) – established a piped water network to connect rural households with the deep ground water resources and improve their access to potable water. Using propensity score matching, the study compares a treatment and a control group of households to identify gains in water-sanitation, hygiene and health outcomes. In terms of water safety, we find no improvement in the quality of drinking water, measured by E. coli count per 100 ml of water at the point of use (i.e. the pots and jars used to store it). Food utensils tested positive for E. coli in both the control and treatment group, thus showing no improvement through the BMDA intervention. Hygiene behavior such as handwashing with soap after defecation or before feeding children also does not improve. Finally, we do not find evidence of health benefits, such as decreased diarrhea incidence of under-five children or improved nutritional outcomes such as stunting, underweight and wasting. Although access to BMDA piped water in the premises is subject to a fee, it seems this incentive mechanism is not strong enough to improve water behavior or its outcomes: treated households are as poor as the non-treated in terms of maintaining hygiene and water quality, possibly because of lack of information.

The major landslide of 2010 in Bududa district in Eastern Uganda is the most catastrophic natural disaster in Uganda's recorded history. An estimated 350 people died and some of the affected were resettled in Kiryandongo district. We assessed housing, water and sanitation practises of affected households and controls to establish possible implications on food insecurity and diet diversity.This cross-sectional study assessed 1078 affected and control households in Bududa and Kiryandongo districts. The head of the households were either affected, selected from landslide disaster-affected communities, or controls from a random sub-county bordering the affected. Structured interviews were used and entries were tested statistically to report crude estimates based on the Pearson x² and ANOVA, while adjusted analysis used multivariate analysis of co-variance (MANCOVA) and Wald's Odds Ratio (OR) of Binary Logistic Regression at the 95% CI.On adjusting for the disaster and covariates, households who had insufficient access to water scored higher mean (SE) on food insecurity than those who reported sufficient access: 13.1 (0.4) vs. 9.5 (0.3) (P < 0.01). Intriguingly, households who owned a toilet and those who reported always using soap to wash hands scored higher food insecurity than their counterparts: 11.1 (0.3) vs. 9.0 (0.6) (P < 0.01) and 11.3 (0.3) vs. 9.1 (0.5) (P < 0.01), respectively. However, not owning a toilet increased the likelihood of being food insecure (OR 3.43; 95% CI 1.31, 8.97; P = 0.02). In Kiryandongo, affected households scored higher food insecurity than controls: 9.5 (0.5) vs. 5.4 (0.5) (P < 0.01) and higher diet diversity scores (DDS): 6.7 (0.2) vs. 6.2 (0.2) (P = 0.04). Type of housing also predicted food insecurity (P < 0.01) and DDS (P = 0.03). Like Bududa, households with insufficient water access scored higher food insecurity while those owning toilet scored higher DDS. Uniquely, only number of rooms was linked to a high likelihood of being food insecure (OR 1.60; 95% CI 1.19, 2.15; P < 0.05).Disaster, food security and diet were sensitive to housing, water and sanitation and were integral to an adequate standard of living of victims of this landslide.