Dietary recommendations for healthy adults include adequate hydration (males: 125 ounces/day, females: 91 ounces/day) from drinking water, other beverages, and water contained in food. Bottled water sales in U.S. exceeded 11.7 billion gallons in 2015 (36.5 gallons per capita). Low water intake, as well as some chemicals in water sources, are associated with hyperglycemia. However, the association between drinking water sources and hyperglycemia/diabetes has not been studied. In a cohort of overweight/obese Puerto Ricans, we evaluated the association between water sources (bottled and filtered compared to tap water) and prediabetes/diabetes. In 2014-2016, 1023 participants reported their primary source of water as bottled (52%), filtered (25%) and tap (23%); diabetes status was assessed using ADA criteria for fasting and 2-hour post load glucose, and HbA1c. Logistic regression models controlled for age, gender, waist circumference, smoking, alcohol intake, physical activity, and hypertension. Bottled water consumers (compared to tap water) had higher prediabetes (OR=1.43; 95% CI: 1.02, 2.01), and diabetes (OR=1.93; 95% CI: 1.08, 3.46). Further controlling for amount of water, sugar sweetened beverages, or canned food and drinks did not change the associations, but controlling for education or income strengthened the association with diabetes. Results were stronger when restricted to San Juan city residents (43%): prediabetes (OR=1.68; 95% CI: 1.01, 2.79), and diabetes (OR=4.05; 95% CI: 1.48, 11.1). Filtered water users had somewhat higher prediabetes/diabetes (OR=1.38; 95% CI: 0.93, 2.04) compared to tap water users. Results suggest that drinking bottled water may be associated with higher prevalence of diabetes compared to tap water, potentially mediated by endocrine disruptors in plastic bottles. We cannot determine time sequence or causality, given the cross-sectional analyses. The impact of water sources on diabetes risk needs to be further evaluated longitudinally.


K.J. Joshipura: None. M.A. Trak: None. F.J. Muñoz-Torres: None.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at