Estimating the Total Water Intake in the Adult US Population

Introduction: The water requirement for individuals is estimated using one of several formulas that are based either on kilocalories consumed or body weight. These formulas have never been validated. Given that total water intake (TWI) per the Dietary Reference Intakes can be used as a proxy for the water requirement, we sought to develop the best formula to predict the water requirement as measured by TWI. Methods: Using 1999-2004 data from the nationally representative National Health and Nutrition Examination Survey (NHANES), we constructed the best fit regression model on individuals ≥19 years of age. We used a priori criteria to produce multiple linear regression equations with variables of suspected or known clinical significance to water metabolism (e.g., plasma osmolality, kilocalories consumed) as well as demographic variables. The TWI was determined by summing water consumed from all dietary sources and metabolic water. Analyses were performed using weighted data appropriate for a multi-stage complex survey sampling design. Results: Significant variables for estimating the TWI were race/ethnicity ( p <0.0001), kilocalories consumed ( p <0.0001), age ( p =0.02), height ( p <0.0001), body mass index ( p <0.0001), and plasma osmolality ( p =0.001). Using these variables, the best fit and most parsimonious final model was significant [ F (9, 44) =166; p <0.0001; RMSE = 1473.8; r ² = 28%]. Discussion: The results suggest that both socio-demographic and clinical factors should be considered when calculating the water requirement in clinical practice. Future research should validate this formula in the general population.