Validation Testing Demonstrates Efficacy of a 7-Day Fluid Record to Estimate Daily Water Intake in Adult Men and Women When Compared with Total Body Water Turnover Measurement

Mean daily water intake from fluids (WATER-FL) has proven to be difficult to measure because of a range of nonvalidated data collection techniques. Few questionnaires have been validated to estimate WATER-FL against self-reported diaries or urinary hydration markers, which may limit their objectivity. The goals of this investigation were ) to assess the validity of a 7-d fluid record (7dFLR) to measure WATER-FL (WATER-FL-7dFLR) through comparison with WATER-FL as calculated by measuring deuterium oxide (D O) disappearance (WATER-FL-D O), and ) to evaluate the reliability of the 7dFLR in measuring WATER-FL. Participants [ = 96; 51% female; mean ± SD age: 41 ± 14 y; mean ± SD body mass index (in kg/m ): 26.2 ± 5.1] completed body water turnover analysis over 3 consecutive weeks. They completed the 7dFLR and food diaries during weeks 2 and 4 of the observation. The records were entered into nutritional software to determine the water content of all foods and fluids consumed. WATER-FL-D O was calculated from water turnover (via the D O dilution method), minus water from food and metabolic water. The agreement between the 2 methods of determining WATER-FL were compared according to a Bland-Altman plot at week 2. The test-retest reliability of 7dFLR between weeks 2 and 4 was assessed via intraclass correlation (ICC). The mean ± SD difference between WATER-FL-7dFLR and WATER-FL-D O was -131 ± 845 mL/d. In addition, no bias was observed ( = 0.484; = 0.006; = 0.488). When comparing WATER-FL-7dFLR from weeks 2 and 4, no significant difference (mean ± SD difference: 71 ± 75 mL/d; = 0.954; = 0.343) and an ICC of 0.85 (95% CI: 0.77, 0.90) was observed. The main findings of this study were that the use of the 7dFLR is an effective and reliable method to estimate WATER-FL in adults. This style of questionnaire may be extremely helpful for collecting water intake data for large-scale epidemiologic studies.