Beverage Consumption Habits and Association with Total Water and Energy Intakes in the Spanish Population: Findings of the ANIBES Study

Background: Inadequate hydration is a public health issue that imposes a significant economic burden. In Spain, data of total water intake (TWI) are scarce. There is a clear need for a national study that quantifies water and beverage intakes and explores associations between the types of beverages and energy intakes. Methods: The Anthropometry, Intake and Energy Balance Study ANIBES is a national survey of diet and nutrition conducted among a representative sample of 2285 healthy participants aged 9–75 years in Spain. Food and beverage intakes were assessed in a food diary over three days. Day and time of beverage consumption were also recorded. Results: On average, TWI was 1.7 L (SE 21.2) for men and 1.6 L (SE 18.9) for women. More than 75% of participants had inadequate TWI, according to European Food Safety Authority (EFSA) recommendations. Mean total energy intake (EI) was 1810 kcal/day (SE 11.1), of which 12% was provided by beverages. Water was the most consumed beverage, followed by milk. The contribution of alcoholic drinks to the EI was near 3%. For caloric soft drinks, a relatively low contribution to the EI was obtained, only 2%. Of eight different types of beverages, the variety score was positively correlated with TWI (r = 0.39) and EI (r = 0.23), suggesting that beverage variety is an indicator of higher consumption of food and drinks. Conclusions: The present study demonstrates that well-conducted surveys such as the ANIBES study have the potential to yield rich contextual value data that can emphasize the need to undertake appropriate health and nutrition policies to increase the total water intake at the population level promoting a healthy Mediterranean hydration pattern.


Introduction
Dehydration occurs when the body loses more water than is taken in. It is often accompanied by disturbances in the body's mineral salt or electrolyte balance, especially in concentrations of sodium and potassium. Populations at particular risk of hypohydration are children, those engaged in professions where fluid homeostasis is regularly challenged, and older adults [1]. Limited data are available on the prevalence of hypohydration, but there is evidence to suggest that it may be relatively common among older populations [2]. The percentage of the population with inadequate water intake varies from 5% to 35% among European countries [3]. Whereas the burden of disease from inadequate water intake is well known, its consequences in Europe are far from being well understood. Recent research into the risk of disease (from falls and accidents and bowel, metabolic, and kidney diseases), disability (cognitive function, physical performance, and headaches), and death has confirmed the importance of poor hydration with respect to the overall disease burden and quality of life in Europe [4]. Therefore, the significant economic burden dehydration represents makes it an important public health issue. Depending on the degree or magnitude of dehydration in hospitalized patients, costs may be increased by 7%-8.5%. Dehydration represents a potential target for intervention to reduce healthcare expenditures and improve patient quality of life [5].
On the other hand, scientific literature recognizes that the "adequate intake" value of beverages (AI) is a variable event, in which differences are in part due to the inter-individual variation for water needs in response to different health status, metabolism, and environmental factors such as ambient temperature and humidity, as well as individual factors such as age, body size, and level of physical activity. Furthermore, the water needs also depend partially on overall diet and the water contained in food. About 80% of the required daily intake is provided by drinks, including water; the rest is acquired through solid food. Identifying the variety of drinks consumed and establishing the percentage of energy provided by each beverage to the diet allows us to set the pattern of individual or population-level beverage consumption. Understanding the contribution of each fluid type to the total fluid intake will allow us to draw conclusions about the adequacy of drinking habits.
In Spain, the influence of the Mediterranean Diet is widespread. This pattern of consumption also includes the hydration pattern of the population. Traditionally, this pattern included water as a main drink, along with daily but moderate consumption of wine or beer with the principal meal, and the intake of a group of beverages elaborated with fresh vegetables (Gazpacho, Salmorejo, etc.). However, in the last decades [6], the adherence to the Mediterranean pattern has been decreasing, especially in children and young people. The actual beverage pattern also included the occasional consumption of soft drinks, which was associated with leisure time. Furthermore, total water intake (TWI) data of Spain are scarce. There are no recent epidemiological studies that focus exclusively on beverage intake. Most available hydration data focuses on alcohol consumption. Apart from the Spanish National Survey on Dietary Intake (ENIDE) [7] in 2011, we are unaware of other research investigating beverage intake among the Spanish population. According to ENIDE data, the average beverage consumption was 1646.5 mL/day, which reflects insufficient fluid intake for that study population.
Given the extent of the problem and its poor recognition, there is the clear need for a national study to update the existing data. The aim of this study was to quantify the total water and beverage intake, and to explore associations between the types of beverage consumed and energy intake. The ANIBES study, a national survey of diet and nutrition conducted in 2013 among a representative random sample of 2285 healthy participants aged 9-75 years, provides us the likely best source of detailed information on the diet of normal individuals in Spain. Characteristics of ANIBES include a food and beverages record list (in grams) of each item consumed per participant and a record of the time of consumption. These study attributes provide a rich resource for exploring patterns of consumption according to age and gender, thereby permitting us to reach conclusions about whether or not drinking a variety of beverages helps to increase fluid intake to a level that meets current guidelines.

Materials and Methods
The design, protocol, and methodology of the ANIBES study have been described in detail elsewhere [8][9][10].

Sample
The ANIBES study is a cross-sectional study conducted using stratified multistage sampling. To guarantee better coverage and representativeness, the fieldwork was performed at 128 sampling points across Spain. The design of the ANIBES study aims to define a sample size that is representative of all individuals living in Spain, aged 9-75 years, and residing in municipalities of at least 2000 inhabitants. The initial potential sample consisted of 2634 individuals. For all analyses, we eliminated participants with anomalous values of energy intake (EI) (men, <800 or >4000 kcal/day; women, <500 or >3500 kcal/day) [11] to avoid introducing bias to the analysis. The final sample comprised 2007 individuals (1011 men, 50%; 996 women, 50%). In addition, for the youngest age groups (9-12, 13-17, and 18-24 years), an "augment sample" was included to provide at least n = 200 per age group (error˘6.9%). The augment sample is the process of increasing the amount of interviews for a particular subgroup within the population in order to achieve an adequate number of interviews to allow analysis of population subgroups or segments that wouldn't normally yield a sufficient number of interviews in a main random survey, without the expense of increasing the sample size for the whole survey. Therefore, the random sample plus augment sample comprised 2285 participants.
The sample quotas according to the following variables were: age groups (9-12, 13-17, 18-64, and 65-75 years), gender (men/women), and geographical distribution (Northeast, Levant, Southwest, North-Central, Barcelona, Madrid, Balearic, and Canary Islands). Additionally, other factors for sample adjustment were considered: unemployment rate, percentage of foreigners (immigrant population), physical activity level assessed by The International Physical Activity questionnaire (IPAQ) [12], tobacco use and education or economic level. Finally, participants' weight, height, and waist circumference were measured and body mass index was also calculated.
The fieldwork for the ANIBES study was conducted from mid-September 2013 to mid-November 2013, and two previous pilot studies were also performed. To equally represent all days of the week, study subjects participated during two weekdays and one weekend day. The final protocol was approved by the Ethical Committee for Clinical Research of the Region of Madrid, Spain [9].

Food and Beverage Record
Study participants were provided with a tablet device (Samsung Galaxy Tab 2 7.0, Samsung Electronics, Suwon, South Korea) and trained in recording information by taking photos of all food and drinks consumed during the 3 days of the study, both at home and outside the home. Photos were to be taken before beginning to eat and drink, and again after finishing, so as to record the actual intake. Additionally, a brief description of meals, recipes, brands, and other information was recorded using the tablet. Participants who declared or demonstrated that they were unable to use the tablet device were offered other options, such as using a digital camera and paper record and/or telephone interviews. A total 79% of the sample used a tablet, 12% a digital camera, and 9% opted for a telephone interview. As no differences in the percentage of misreporting were found according to the type of device used to assess dietary intake, we used the measurements of the three assessment methods in the analysis. In addition to details of what and how much was eaten, for each eating/drinking event, participants recorded where they were, who they were eating with, and whether they were watching television and/or sitting at a table. After each survey day, participants recorded if their intake was representative for that day (or the reason why if it was not), and details of any dietary supplements taken. The survey also contained a series of questions about participants' customary eating habits (e.g., the type of milk usually consumed) to facilitate further coding. Food records were returned from the field in real time, to be coded by trained coders who were supervised by dieticians. An ad hoc central server software/database was developed for this purpose, to work in parallel with the codification and verification processes. Food, beverage, and energy and nutrient intakes were calculated from food consumption records using VD-FEN 2.1 software, a Dietary Evaluation Program from the Spanish Nutrition Foundation (FEN), Spain, which was newly developed for the ANIBES study by the FEN and is based mainly on Spanish food composition tables [13], with several expansions and updates. Data obtained from food manufacturers and nutritional information provided on food labels were also included. A food photographic atlas was used to assist in assigning gram weights to portion sizes. The VD-FEN 2.1 software was developed to receive information from field tablets every 2 s, and the database was updated every 30 min. Energy distribution objectives for the Spanish population were used to analyze the overall quality of the diet [9,14].

Data Preparation and Analysis
The present analysis focused on the TWI of all food and drink, determined from food composition tables with several adaptations and updates [13]. Metabolic water (water derived from oxidation of substrates) was not included so as to focus on comparisons with dietary water requirements.
Beverages were combined into eight categories for further analysis: (1) hot beverages, including hot tea and coffee (iced teas in cans or bottles were considered caloric soft drinks); (2) milk (all types of milk without separation by fat percentage); (3) fruit and vegetable juices (including nectars, juice-milk blends, 100% fruit juices, and some typical Spanish beverages: horchata, gazpacho, salmorejo, and white garlic); (4) caloric soft drinks (including colas, tonic water, sodas, ginger ale, fruit flavored drinks, iced teas in cans or bottles, sports drinks such as isotonic drinks with mineral salts, and caffeinated energy drinks); (5) diet soft drinks (including the same beverages as in the caloric soft drinks group but with artificial sweetener); (6) alcoholic drinks, including two groups: (a) low-alcohol grade (mostly beer, wine, and cider); and (b) high-alcohol grade (including brandies, liqueurs, tequila, vodka, whisky, etc.); (7) water (including tap water and bottled water); and (8) other beverages (including soy-based beverages, non-alcoholic beer and wine, and others).
Additionally, a variety score was created as the sum of the different beverages used in our classification with a minimum value of 0 and a maximum value of 8.
To investigate daily trends, beverage consumption events were aggregated into six time periods, approximately corresponding to breakfast (up to 10:00 a.m.), mid-morning (10:00 a.m. TWI was compared with the European Food Safety Authority (EFSA) Dietary Reference Values (DRV) for the Adequate Intake (AI) of water for men and women from 14 years of age onward (2.5 L and 2.0 L, respectively), and for boys and girls from 9 to 13 years of age (2.1 L and 1.9 L, respectively) [15]. Furthermore, Nordic and German-speaking countries take the approach that water intake is considered inadequate when it is less than 1 g per kilocalorie of energy requirement [15]. Therefore, we used three different approaches to define water intake adequacy to provide a more comprehensive estimate of the proportion of participants who consume low amounts of water [16]. First, a classification based on the AI value, defined by the EFSA as criterion 1. The second (criterion 2), a ratio between TWI (water from food and beverages in grams) and EI in kcal higher than 1; and the combination of both as final criterion.

Statistical Analyses
Owing to reported differences in water consumption and recommended intakes between male and female individuals, all analyses were carried out separately by gender. Crude differences in TWI and beverage consumption between groups were assessed through an analysis of variance test or t-tests with Bonferroni correction for multiple comparisons. Chi-squared tests were used for categorical variables. All analyses were two-tailed, with statistical significance set at p < 0.05.
Partial Correlations between water intake, energy intake, and beverage consumption adjusted for age, gender, body weight, and physical activity were calculated by the use of a variety score. Multivariable linear regression models included adjustment for gender, age, weight, and physical activity level. Multiple linear regression models were fitted to assess the effect of varying the type of beverages consumed (caloric vs. non-caloric) on EI while controlling for the effect of confounders (gender, age, weight, and physical activity). The effect of replacing 100 g of caloric beverages with 100 g of non-caloric beverages was estimated by including caloric beverages (as a percentage of total beverage weight) as the main independent variable, with total beverage weight (g) held constant. This necessarily implies an equal and opposite change in other beverages. A further model included energy from food, thus disallowing compensation (reduction in calories from food). Finally, we used within-person daily consumption data to explore the effect of changes in daily beverage consumption, with each person acting as their own control. The independent variables represent the standard deviation of the mean of the three different measurements (3-day records) for each type of beverage collected for each participant. The outcome (change in EI) was derived using the same methodology (deviation from participants' 3-day mean in energy intake) [16].

Results
The ANIBES sample was of 2285 healthy subjects aged 9 to 75 years, of which 50% of the population was men and the other 50% were women. The study sample reflects the distribution of male and female individuals in the general population of Spain [17] (Statistics National Institute (INE) 2011). A more detailed description of the ANIBES study population is given in Table 1 [13]. There was no statistically significant difference in the education or economic level between men and women. However, men smoked more but engaged in more sport than women. Regarding the measures of overweight and obesity, the data of ANIBES study were representative of the Spanish adult population, according to the Spanish National Health Survey 2011-2012 [18]: 37% and 32% of men and women, respectively, were overweight, and 21% and 19%, respectively, were obese. Figure 1 shows the frequency distribution of TWI (g/day) over a three-day recording period, organized by gender. On average, the TWI was 1.7 L (SE 21.2) for men and 1.6 L (SE 18.9) for women, far less than the EFSA AI recommendations for adults (2.5 L and 2.0 L, respectively). Partial Correlations between water intake, energy intake, and beverage consumption adjusted for age, gender, body weight, and physical activity were calculated by the use of a variety score.
Multivariable linear regression models included adjustment for gender, age, weight, and physical activity level. Multiple linear regression models were fitted to assess the effect of varying the type of beverages consumed (caloric vs. non-caloric) on EI while controlling for the effect of confounders (gender, age, weight, and physical activity). The effect of replacing 100 g of caloric beverages with 100 g of non-caloric beverages was estimated by including caloric beverages (as a percentage of total beverage weight) as the main independent variable, with total beverage weight (g) held constant. This necessarily implies an equal and opposite change in other beverages. A further model included energy from food, thus disallowing compensation (reduction in calories from food). Finally, we used within-person daily consumption data to explore the effect of changes in daily beverage consumption, with each person acting as their own control. The independent variables represent the standard deviation of the mean of the three different measurements (3-day records) for each type of beverage collected for each participant. The outcome (change in EI) was derived using the same methodology (deviation from participants′ 3-day mean in energy intake) [16].

Results
The ANIBES sample was of 2285 healthy subjects aged 9 to 75 years, of which 50% of the population was men and the other 50% were women. The study sample reflects the distribution of male and female individuals in the general population of Spain [17] (Statistics National Institute (INE) 2011). A more detailed description of the ANIBES study population is given in Table 1 [13]. There was no statistically significant difference in the education or economic level between men and women. However, men smoked more but engaged in more sport than women. Regarding the measures of overweight and obesity, the data of ANIBES study were representative of the Spanish adult population, according to the Spanish National Health Survey 2011-2012 [18]: 37% and 32% of men and women, respectively, were overweight, and 21% and 19%, respectively, were obese. Figure 1 shows the frequency distribution of TWI (g/day) over a three-day recording period, organized by gender. On average, the TWI was 1.7 L (SE 21.2) for men and 1.6 L (SE 18.9) for women, far less than the EFSA AI recommendations for adults (2.5 L and 2.0 L, respectively).

Figure 1.
Frequency distribution of total water intake (g/day) over three days by gender. Figure 1. Frequency distribution of total water intake (g/day) over three days by gender.
Percentages of total weight consumed (g/day), daily EI (kcal/day), and water intake (g/day) are presented in Table 2. Beverages were separated by category, with most consumed in similar amounts by participants of both genders. However, men consumed exactly two times more alcoholic beverages than women. The mean total EI was 1809 kcal/day (SE 11.1), and the relative contribution to total EI from beverages was 12% (13% for men, 12% for women). Furthermore, 68% of the TWI came from beverages and 32% from food. Those amounts coincided with EFSA recommendations (70%-80% provided by beverages of all types and the remaining 20%-30% from food). On average, the percentage of total beverage consumption from water over the three-day study period was 46% for women and 41% for men ( Figure 2). Water was the most frequently consumed beverage, followed by milk, for both genders. Among men, the decreasing order of consumption was alcoholic drinks, caloric soft drinks, and hot beverages, with similar percentages (11%, 11%, and 10%, respectively). For women, the decreasing order was hot beverages (12%), caloric soft drinks (8%), and alcohol (5%). Fruit and vegetable juices and diet soft drinks were consumed in lower amounts by both genders. On average, the percentage of total beverage consumption from water over the three-day study period was 46% for women and 41% for men ( Figure 2). Water was the most frequently consumed beverage, followed by milk, for both genders. Among men, the decreasing order of consumption was alcoholic drinks, caloric soft drinks, and hot beverages, with similar percentages (11%, 11%, and 10%, respectively). For women, the decreasing order was hot beverages (12%), caloric soft drinks (8%), and alcohol (5%). Fruit and vegetable juices and diet soft drinks were consumed in lower amounts by both genders. In general (Table 3), the contribution of water intake from food increased with age, from 434 g/day (SE 13.9) among younger participants (13-17 years) to 584 g/day (SE 24.8) among older adults (65-75 years). This finding is likely owing to lower consumption of fruits and vegetables, which are rich in water, for the youngest participants. The water contribution from beverages declined from 1202 g/day (SE 21.6) among adults (18-64 years) to 1002 g/day (SE 44.0) among older adults.
For adolescents, the mean consumption of caloric soft drinks was 167 g/day (SE 19.1) for men, and 139 g/day (SE 15.9) for women (equivalent to about three cans per week for each). However, consumption was lower among adults, with 114 g/day (SE 6.3) for men and 82.1 g/day (SE 4.6) for women. Consumption of alcoholic drinks for adult men and women averaged 160 g/day (SE 9.0) and 71 g/day (SE 4.8), respectively. The intake of alcoholic drinks among the oldest participants (65-75 years) averaged 143 g/day (SE 19.0) for men and 53 g/day (SE 9.8) for women.
The principal sources of total dietary water, by gender and age group, are shown in Figure 3. For both men and women, the main source was water, followed by milk. Children and adolescents consumed higher amounts of water from milk and juices than adults, for both genders; however, adolescents consumed less water from hot beverages and diet soft drinks than adults. Adults and older adults had a lower intake of caloric soft drinks and juices but consumed greater amounts of hot beverages and alcohol, for both men and women. In general (Table 3), the contribution of water intake from food increased with age, from 434 g/day (SE 13.9) among younger participants (13-17 years) to 584 g/day (SE 24.8) among older adults (65-75 years). This finding is likely owing to lower consumption of fruits and vegetables, which are rich in water, for the youngest participants. The water contribution from beverages declined from 1202 g/day (SE 21.6) among adults (18-64 years) to 1002 g/day (SE 44.0) among older adults.
For adolescents, the mean consumption of caloric soft drinks was 167 g/day (SE 19.1) for men, and 139 g/day (SE 15.9) for women (equivalent to about three cans per week for each). However, consumption was lower among adults, with 114 g/day (SE 6.3) for men and 82.1 g/day (SE 4.6) for women. Consumption of alcoholic drinks for adult men and women averaged 160 g/day (SE 9.0) and 71 g/day (SE 4.8), respectively. The intake of alcoholic drinks among the oldest participants (65-75 years) averaged 143 g/day (SE 19.0) for men and 53 g/day (SE 9.8) for women.
The principal sources of total dietary water, by gender and age group, are shown in Figure 3. For both men and women, the main source was water, followed by milk. Children and adolescents consumed higher amounts of water from milk and juices than adults, for both genders; however, adolescents consumed less water from hot beverages and diet soft drinks than adults. Adults and older adults had a lower intake of caloric soft drinks and juices but consumed greater amounts of hot beverages and alcohol, for both men and women.   TWI was highly correlated with beverage weight and water from beverages (r = 0.95), and was more weakly correlated with food intake (r = 0.54, Table 4). Caloric soft drinks and alcoholic drinks had a moderate correlation with total energy from beverages (r = 0.47 and 0.49, respectively), whereas coefficients were lowest for hot beverages and diet soft drinks.
Regarding beverage variety of eight types of beverages in our classification, the variety score was positively correlated with TWI (r = 0.39, p < 0.001) and with EI (r = 0.23, p < 0.001), suggesting that beverage variety is an indicator of a higher consumption of food and drinks. Figure 4 represents the influence of day of the week on beverage consumption. The total amount of beverage intake (g) was slightly higher on Fridays among men and on Saturdays among women than on other days of the week. This appears to be attributable to a higher consumption of alcoholic drinks on the weekends for both groups. Consumption of water, milk, fruit and vegetable juices, hot beverages, and diet soft drinks did not vary greatly by day of the week. With respect to beverage consumption according to the time of day, over a 24 h period, in general, no significant differences were found by age or time of day in either men or women (Table 5). One exception was seen in differences in the amount of beverages consumed by males during lunch when comparing children and adolescents to adults, the consumption is higher in 18-64 year-old subjects.  TWI was highly correlated with beverage weight and water from beverages (r = 0.95), and was more weakly correlated with food intake (r = 0.54, Table 4). Caloric soft drinks and alcoholic drinks had a moderate correlation with total energy from beverages (r = 0.47 and 0.49, respectively), whereas coefficients were lowest for hot beverages and diet soft drinks.
Regarding beverage variety of eight types of beverages in our classification, the variety score was positively correlated with TWI (r = 0.39, p < 0.001) and with EI (r = 0.23, p < 0.001), suggesting that beverage variety is an indicator of a higher consumption of food and drinks. Figure 4 represents the influence of day of the week on beverage consumption. The total amount of beverage intake (g) was slightly higher on Fridays among men and on Saturdays among women than on other days of the week. This appears to be attributable to a higher consumption of alcoholic drinks on the weekends for both groups. Consumption of water, milk, fruit and vegetable juices, hot beverages, and diet soft drinks did not vary greatly by day of the week. With respect to beverage consumption according to the time of day, over a 24 h period, in general, no significant differences were found by age or time of day in either men or women (Table 5). One exception was seen in differences in the amount of beverages consumed by males during lunch when comparing children and adolescents to adults, the consumption is higher in 18-64 year-old subjects. Table 4. Partial correlations between water intake, energy intake and beverage consumption. (3 days mean data adjusted for age, gender, body weight and physical activity).   juices, hot beverages, and diet soft drinks did not vary greatly by day of the week. With respect to beverage consumption according to the time of day, over a 24 h period, in general, no significant differences were found by age or time of day in either men or women (Table 5). One exception was seen in differences in the amount of beverages consumed by males during lunch when comparing children and adolescents to adults, the consumption is higher in 18-64 year-old subjects.  Appendix Table A1 shows a multiple regression model including four covariates (gender, age, weight, and physical activity) as well as total beverage intake, which is mathematically equivalent to substituting caloric beverages (sum of hot beverages, milk, fruit and vegetable juices, caloric soft drinks, and alcoholic drinks) with an equal weight of non-caloric beverages (diet soft drinks and water). The predicted effect of replacing 100 g of caloric beverages with 100 g of non-caloric drinks was associated with a reduction in EI of 50 kcal (Model 1). When food EI was constrained as a constant (i.e., disallowing compensation), the net impact of 100 g of caloric beverages was estimated at 40 kcal (Model 2). Further adjustment for level of education, economic level, employment, etc. did not change the reported results.
The second regression model analysis (Appendix Table A2) used a within-person change model to address whether a change in a participants' beverage consumption habits on any day could be associated with a change in their total EI (compared with their three-day mean). Modeling each beverage separately, with total beverage weight held constant, each of the above two non-caloric beverages (diet soft drink and water) was negatively associated with energy (as were hot beverages), whereas the five caloric beverages mentioned above were positively associated with energy (Model 3). When combining caloric beverages, the final models (Models 4 and 5) gave an estimated effect of 43 kcal per 100 g of caloric drinks substituted, or 34 kcal if EI from food was held constant.
Participants who fulfilled the EFSA AI recommendations of TWI for men and women (2.5 L and 2.0 L, respectively) were classified as Criterion 1. Participants with ratios of water/energy intake >1.0 were included as Criterion 2 (considering a value of 1 g of water per 1 kcal of energy intake). Finally, participants who met both definitions (Criteria 1 and 2) were classified as Criterion 3. For children 9-13 years old, the EFSA AI is 2.1 L for boys and 1.9 L for girls. Following this analysis, Table 6 shows that for both genders, more than 75% of women, about 80% of men, and nearly 90% of children did not meet the AI recommendation for water consumption. Table 6. Combined classification for the Total water intake (TWI) following established criteria.

Discussion
This study provides analyses of total water intakes from all sources among a nationally representative sample of the Spanish population aged 9-75 years, included in the 2013 ANIBES study database. To our knowledge, the present analyses represent one of the few explorations of the consumption of water and beverages in Spain, as well as the association with energy intake, consumption according to time of day and day of the week, the association between beverage variety and increased fluid intake, and compliance with current AI recommendations, by gender and age.
The main findings of this study indicate that for the entire sample, TWI was 1625 g/day (SE 14.2). In general, and as expected, male individuals had statistically higher intakes than female individuals for both food and beverages. However, neither men nor women consumed sufficient amounts of water (1664 g/day for men, 1585 g/day for women), according to EFSA AI reference values [15]. Men consumed approximately 33% less than the AI and women nearly 21% less.
Most of the data analysis in this work was based on an earlier survey by Gibson and Shirreffs [16], who analyzed the weighed dietary records from the National Diet and Nutrition Survey (2000/2001) of 1724 British adults. Contrary to our findings, mean TWI in that population was nearly identical to the EFSA reference AI for both genders.
The comparison is difficult to do when we want to compare American and Spanish populations. The NHANES study, developed in the US between 2005 and 2010, used the proposal by the Institute of Medicine of the United States of America (IOM) as reference value, set as 3.7 L/day for men and 2.7 L/day for women [19]. This leads us to consider the need to investigate why the recommendations of the EFSA and the IOM are so different if both recommendations include the water from both food and beverage sources. The clarification of this issue deserves further study.
Mean daily EI was 1809 kcal/day (SE 11.1), of which 12% was provided by the beverages, close to the 10% proposed by some international authorities (EFSA, WHO) [15,20] who recommended that no more than 10% of the daily calorie intake should come from beverages. In the NHANES study [21], the proportion of energy from beverages was 21%, while it was 16% in the British study [16]. Regarding this issue, in recent years, the impact of caloric soft drink consumption on obesity and metabolic disorders has come under intense scrutiny and debate worldwide [22][23][24][25], and large differences between countries have been observed. The present study showed that for the entire Spanish population, caloric soft drinks contributed only 2% of the total EI, or 41.4 kcal/day (SE 1.5) out of a total EI of 1809 kcal/day (SE 11.1). Lower consumption was seen in older adults, at 26.5 g/day (SE 6.6). Higher consumption was found among adolescents, at 167 g/day (SE 19.1), followed by adults and children, with similar mean consumptions of 114.2 g/day (SE 6.3) and 113 g/day (SE 14.8), respectively). This finding is perhaps one of the most interesting findings of our study. The relatively low contribution of caloric soft drinks to the EI could be attributable to the Mediterranean pattern of consumption that this society keeps. By contrast, the NHANES study of the United States [21] has the highest contribution to EI from sweetened beverages. For adults, soda accounted for near 6% of energy intake.
On the other hand, beverage consumption is uneven throughout the day; in this study, it tended to be concentrated at lunchtime and was slightly higher on Fridays in men and on Saturdays in women. Although we did not find large variation, these findings reflect certain cultural trends of consumption that appear to be attributable to a higher intake of alcoholic beverages on weekends among both genders. The contribution of alcoholic drinks to the diet in the ANIBES study (135.7 g/day, SE 7.5) was slightly higher than that in the ENIDE dietary survey of Spanish adults in 2011 (117 g/day) [7]. It is important to note that beverages with lower alcohol content (beer, wine, and cider) represented over 90% of the energy contribution in adult and older populations. Furthermore, although these data are of the highest quality obtainable, alcohol intake is one of the dietary components for which underreporting may occur, especially among women and participants with higher education and socioeconomic levels [26,27]. In fact, the reported amount of alcohol consumed by men was twice that reported by women in this study. Anyway, this relatively high consumption of alcohol shown could also be attributable to the Mediterranean dietary pattern that certainly remains rooted in Spanish society. The Spanish Society of Community Nutrition (SENC) recommends the maximum consumption of 1 to 1.5 servings/day of alcoholic beverages in women and 2 to 2.5 servings/day for adult males in the context of a Mediterranean balanced diet. Consumption of alcoholic beverages is associated with a more or less healthy dietary pattern. The types of alcoholic beverages that are closer to the Mediterranean environment are [28] fermented drinks, wine, beer, and cider consumed during the principal meals.
The strengths of this study include the careful design, protocol, and methodology used in the ANIBES study. The present analysis can be used to inform approaches to improving the overall quality of diet and hydration status of the Spanish population.
Two important limitations of this study must be noted. This study is a cross-sectional design, which provides evidence for association but not causal relationships. The second limitation is that, for logistical reasons, there was no inclusion of a hydration biomarker, which would allow assessment of dietary beverage intake and hydration status without the bias of self-reported dietary intake and also intra-individual variability. This biomarker should be included in further studies.

Conclusions
The present study shows clearly that neither men nor women consumed adequate TWI when compared with EFSA reference values. The ANIBES study demonstrated that well-conducted national surveys have the potential to yield rich contextual value data that can emphasize the need to undertake appropriate health and nutrition policies to increase the total water intake at the population level, promoting a healthy Mediterranean hydration pattern.