Our findings indicated that nearly half of respondents drank less than 4 cups per day of water (ie, bottled or tap water) and that 56% of respondents reported drinking 4 or more cups of water daily. These results are consistent with those based on 2005–2008 NHANES data, which indicated that US adults consumed an average of 4.3 cups of water per day (14
). The biologic requirement for water may be met with plain water or via foods and other beverages. Results from previous epidemiologic studies indicate that water intake may be inversely related to volume of calorically sweetened beverages and other fluid intake (4
Our results indicated that low drinking water intake was associated with many demographic characteristics, including older age. Despite being susceptible to dehydration due to increased prevalence of chronic diseases and the use of multiple medications, older adults have lower fluid consumption primarily due to a decrease in thirst (1
). Previous studies indicate that water consumption decreases with age; a study of 4,112 US adults by Kant et al found lower plain water intake among older US adults (15
). Kant et al reported no significant differences in water intake by race/ethnicity (15
), whereas we found significantly higher intake among respondents in the “other” race/ethnicity category than among whites. The reasons for this association are unclear (FAB was not powered to detect differences among subgroups in this diverse category). In a study of 4,292 Florida students in grades 6 through 8, Park et al found significantly lower odds of low drinking water intake among Hispanic/Latino or “other”/non-Hispanic adolescents than among white adolescents (adjusted OR = 0.79 and OR = 0.76, respectively), results that are similar to those we obtained among adults (23
). Although our study found no association between drinking water intake and education or household income in multivariable models, previous studies reported that plain water intake is positively associated with years of education but not associated with poverty-income ratio (15
). An analysis of the US Department of Agriculture Nationwide Food Consumption Survey of 1977 found lower tap water intake in the Northeast (1.2 L/d) than in other regions (1.4 L/d), possibly due to greater need for water among residents in regions with warm or humid climates (24
Our findings of associations between water intake and certain behaviors were similar to those found in previous research. Meeting the national recommendation for 150 minutes per week of moderate physical activity was associated with significantly higher drinking water intake in this and a previous study (15
), which is not surprising given that physical activity leads to increased hydration needs due to sweating (1
). The results of our multivariable regression analysis showed no association between water intake and time spent watching television, which is consistent with results of a study among 3,867 US children and adolescents (25
). Our finding that former smokers were likely to drink more water than those who never smoked might be explained by the common practice of encouraging participants in tobacco cessation programs to increase their water intake (26
Low fruit and vegetable intake, which epidemiologic studies link to higher risk of chronic disease (11
), was associated with drinking significantly less water in multivariable regression models. In addition, in models controlled for sociodemographic variables, respondents with unhealthful eating behaviors and attitudes (eg, high fast-food intake) drank significantly less water, whereas healthful eating behaviors and attitudes (eg, shopping at farmers markets) were related to drinking more water. These results, which are consistent with findings from previous epidemiologic studies (14
), add to a growing body of evidence that drinking water intake is associated with healthful dietary practices and attitudes. Whether drinking water supports these healthful dietary patterns or simply coexists with them is unclear. Nonetheless, this evidence suggests that health educators or health care practitioners aiming to promote increased water intake should keep in mind that low water consumption may be closely tied to other unhealthful behaviors.
In our study, respondents trying to lose weight consumed significantly more water than those trying to gain weight; however, results of a previous study (15
) showed no significant difference in water intake among respondents trying to lose weight in the previous year than among those not trying to lose weight. Although there is a known significant negative association between energy intake and water consumption, evidence is less clear about the relationship between BMI and water intake. In our study, BMI and water intake levels were unrelated after models controlled for sociodemographics and health-related variables. There are at least 3 plausible explanations for this lack of association: 1) the self-reported BMI values of survey participants may be lower than the true values because survey respondents underestimated weight and overestimated height (27
), thus decreasing our ability to detect an association; 2) the cross-sectional data did not allow us to assess whether prior behaviors of survey participants may have contributed to weight gain; and 3) our adjustments for factors closely associated with obesity, such as physical activity level or fruit and vegetable intake, may have masked the bivariate association we found between water intake and BMI.
FAB data are cross-sectional and the survey results can show only an association between factors, not a causal relationship. The FAB sample was selected from a consumer opinions panel rather than the US population (due to declining response to random–digit-dial telephone surveys); this method is commonly used in other nutrition and health studies such as Styles (18
). The response rate of 57% is similar to other random–digit-dial and consumer opinion mailed surveys; information on nonresponders was not available. To minimize bias, households from the larger consumer opinions panel pool selected for FAB were similar to the US population (by age, household income, geographic region, population density, and household size), and data were weighted using US Census estimates; however, these efforts do not guarantee lack of residual bias due to sample selection or nonresponse (18
). FAB oversampled African Americans as part of the study design, but the sample size for “other” racial/ethnic groups was not sufficient for subgroup analyses. Dietary intake estimates were self-reported and may have been over- or underestimated and less accurate than data from surveys such as NHANES. Although the validity of the water intake question in FAB has not been determined, a recent study among adults found no significant difference between water intake that was self-reported on a questionnaire (on which the question was worded similarly to the question on the FAB) and water intake determined through 4-day food intake records (r
= 0.7) (29
). BMI data in FAB are determined on the basis of self-reported weight and height and subject to reporting bias; however, measured and self-reported BMI are highly correlated among adults (r
> 0.9) (30
). Finally, because the FAB data set did not include data about intake of calorically or artificially sweetened beverages, milk, or alcohol, we were unable to assess the relationship between intake of water and these beverages.
Approximately 7% of respondents reported drinking no water daily, and nearly half reported drinking less than 4 cups per day. Low drinking water intake was associated with various unhealthful behaviors, including low levels of physical activity and low levels of fruit and vegetable intake. Models controlling for sociodemographics indicated that attitudes about eating and health, as well as food-related behaviors such as eating meals while watching television, were also related to low drinking water intake. Further studies of population samples with greater ability to assess differences in water intake by race/ethnicity subgroups (eg, Hispanics, Asians) are needed, as is research to learn where people consume drinking water, such as homes, worksites, or community venues. Our results suggest that low drinking water intake is common and is associated with known unhealthful behaviors. Clinical and public health practitioners aiming to help people drink more water should consider low water intake as part of a group of unhealthful behaviors and attitudes.