Validity and Reliability of a Water Frequency Questionnaire to Estimate Daily Total Water Intake in Adults

The purpose of this investigation was to assess the validity and reliability of a seven-day water frequency questionnaire (TWI-FQ) to estimate daily total water intake (TWI) in comparison to a water turnover objective reference value via deuterium oxide (D(2)O). Data collection occurred over 3 weeks, with a wash-out period during week two. Healthy adults (n = 98; 52% female; 41 ± 14 y; BMI, 26.4 ± 5.5 kg·m(−2)) retrospectively self-reported consumption frequencies of 17 liquids and 35 foods with specified volumes/amounts for weeks one and three via TWI-FQ. Standard water content values were utilized to determine the volume of water consumed from each liquid and food for calculation of mean daily TWI for each week. Diet records were completed daily during week two to estimate metabolic water production. To assess validity of the TWI-FQ, participants consumed D(2)O at the start of each week and provided urine samples immediately before ingestion, the following day, and at the end of the week to calculate water turnover. Metabolic water was subtracted from water turnover to estimate TWI. TWI-FQ validity was assessed via Bland-Altman plot for multiple observations. Reliability was assessed via intraclass correlation and Pearson's correlation between weeks. TWI-FQ significantly underestimated D(2)O TWI by −350 ± 1,431 mL·d(−1) (95% confidence interval (CI): −551, −149 mL·d(−1)). TWI-FQ TWI was significantly correlated (r = 0.707, P <0.01) and not different (198 ± 1,180 mL·d(−1), 95% CI: −38, 435 mL·d(−1)) between weeks. TWI-FQ intraclass correlation = 0.706 was significant [95% CI: 0.591, 0.793; F((97, 98)) = 5.799], indicating moderate test-retest reliability. While this tool would not be suitable for individual TWI assessment, the magnitude of bias may be acceptable for assessment at the sample-level.