Household Water and Sanitation Practices and Their Association With Diarrheal and Acute Respiratory Infections Among Children Under Five in Ghana: A Cross-Sectional Predictive Model and Spatial Analysis.

BACKGROUND AND AIMS: Diarrhea and acute respiratory infections (ARI) remain leading causes of under-five morbidity in low- and middle-income countries. Although Ghana achieved improved drinking water and sanitation coverage, significant regional and socioeconomic inequities persist. This study examined the associations between household water, sanitation, and hygiene (WASH) practices and diarrhea, ARI, and their comorbidity among under-five children in Ghana.

METHODS: A cross-sectional analysis of 6344 children aged 0-59 months from the 2022 Ghana Demographic and Health Survey was conducted. Survey-weighted descriptive statistics and multilevel logistic regression were applied. Predictive performance was assessed using fivefold cluster-preserving cross-validation, Brier scores, calibration slopes, calibration plots, and cross-validated area under the curve (AUC). Spatial patterns were described using choropleth maps, Global Moran's I, Getis-Ord Gi*, and Local Moran's I (LISA). Counterfactual WASH reductions were estimated with bootstrap confidence intervals (200 replicates). A complementary random-effects meta-analysis quantified between-region heterogeneity (I2).

RESULTS: Two-week prevalence of diarrhea, ARI, and comorbidity was 14.4% (95% CI: 12.7%-16.3%), 5.6% (95% CI: 4.7%-6.6%), and 1.8% (95% CI: 1.3%-2.5%), respectively, with a pronounced north-south gradient. No WASH variable reached significance in adjusted multilevel models. Diarrhea was associated with child age 6-35 months, maternal secondary and higher education, and household wealth. ARI was associated with co-occurring diarrhea and stunting. Stunting was the sole significant predictor of comorbidity. Community-level factors explained 21%-35% of variance. Cross-validated AUC was 0.787 (diarrhea), 0.896 (ARI), and 0.577 (comorbidity); calibration slopes were 3.68-11.68 (all poor). Global Moran's I showed no significant spatial autocorrelation for all three outcomes. Improving cooking fuel universally was the only scenario associated with a statistically significant absolute reduction in diarrhea prevalence: 1.4% points (relative reduction 13.9%; 95% CI: 6.2%-21.6%).

CONCLUSIONS: Household WASH infrastructure showed no significant adjusted association with childhood diarrhea, ARI, or comorbidity after controlling for socioeconomic and nutritional confounders. A pronounced north-south disease gradient calls for geographically targeted, multisectoral interventions. Predictive models demonstrated adequate discrimination but poor calibration, limiting their use to population-level risk stratification.