Modeling the Relationship of Groundwater Salinity to Neonatal and Infant Mortality From the Bangladesh Demographic Health Survey 2000 to 2014

We evaluated the relationship of drinking water salinity to neonatal and infant mortality using Bangladesh Demographic Health Surveys of 2000, 2004, 2007, 2011, and 2014. Point data of groundwater electrical conductivity (EC)— a measure of salinity—were collated from the Bangladesh Water Development Board and digitizing salinity contour map. Data for groundwater dissolved elements (sodium, calcium, magnesium, and potassium) data came from a national hydrochemistry survey in Bangladesh. Point EC and dissolved minerals data were then interpolated over entire Bangladesh and extracted to each cluster location, the primary sampling unit of Bangladesh Demographic Health Surveys. We used restricted cubic splines and survey design‐specific logistic regression models to determine the relationship of water salinity to neonatal and infant mortality. A U‐shaped association between drinking water salinity and neonatal and infant mortality was found, suggesting higher mortality when salinity was very low and high. Compared to mildly saline (EC ≥0.7 and < 2 mS/cm) water drinkers, freshwater (EC < 0.7 mS/cm) drinkers had 1.37 (95% CI: 1.01, 1.84) times higher neonatal mortality and 1.43 (95% CI: 1.08, 1.89) times higher infant mortality. Compared to mildly saline water drinkers, severe‐saline (EC ≥10 mS/cm) water drinkers had 1.77 (95% CI: 1.17, 2.68) times higher neonatal mortality and 1.93 (95% CI: 1.35, 2.76) times higher infant mortality. We found that mild‐salinity water had a high concentration of calcium and magnesium, whereas severe‐salinity water had a high concentration of sodium. Freshwater had the least concentrations of salubrious calcium and magnesium.