Total fluid intake and its determinants: cross-sectional surveys among adults in 13 countries worldwide

The present study is a cross-sectional analysis of original or published data collected in adults and elderly (≥18 years) by 13 different surveys conducted in Latin America (Mexico [11], Brazil and Argentina), Europe (Spain [12], France, UK [13], Germany, Poland and Turkey) and Asia (Iran [14], China [15], Indonesia and Japan) by public (Iranian National Nutrition and Food Technology Research Institute, NNFTRI, and Chinese Centre for Disease Control, CDC) and private organizations. The primary objective of these surveys was to assess the sources of fluid consumption, including drinking water and different types of beverages. The individual surveys took place between 2008 and 2014 and were referred to as Liq.In⁷. The participants in each country were randomly recruited either from a database of volunteers for population surveys or via systematic door-to-door recruitment until the quotas for age, gender, region, habitat and/or socioeconomic characteristics in relation to the total country population were met.

The following individuals were excluded from participating: those working in company advertising; marketing; market research; the media; the manufacture, distribution and sale of water; and all kinds of beverages. Likewise, people not able to read and write in the language of the questionnaire were not eligible to participate in the survey. Having a specific diagnosed disease and/or following a medically prescribed diet was additional exclusion criteria in UK, Iran and China. The surveys in Argentina, Poland and Japan also excluded participants who had taken part in a survey about non-alcoholic drinks in the previous 6 months. Participants who did not complete the full fluid intake record, participants reporting a mean total daily fluid intake below 0.4 L/day or higher than 6 L/day or those who had participated in a market research study in the previous 6 months were excluded from the analysis. Pregnancy or lactation was not a specific exclusion in the most countries, except in Iran and China. The effective sample size for the present study was 16,276 participants. Individuals who agreed to be part of the survey received detailed information about the survey’s objectives, what was expected of them, and information about the study’s provisions to preserve confidentiality, risks and benefits, and a clear explanation about their option to participate voluntarily or not in the study. After being given a fully informed description of the study, following the principles of informed consent, participants were asked for their oral approval to participate. No monetary incentive was offered for taking part in the study. All data were recorded anonymously. Therefore, subjects included in the dataset cannot be identified, either directly or through identifiers. The survey protocol of the unpublished surveys was reviewed and approved by the University of Arkansas Review Board (ref. 14-12-376).

All participants were provided with a 24-h fluid-specific record so that they could collect information on their fluid intake over seven consecutive days. The 7-day fluid-specific record was always presented in the official language of the country. In all countries except France, Germany and Japan, a paper version of this 7-day fluid-specific record was delivered and explained to the participants during an initial interview at home. After a period of 7 days, the fluid record was collected from the participant’s home by the researcher and checked with the participant. In France, Germany and Japan, participants completed 7-day fluid-specific record online. On the morning of the first day, these participants received an electronic reminder with written instructions on how to fill in the fluid record. Paper memory cards were made available to the participants so that they could make notes during the day and subsequently complete the fluid record online. Both the paper and online records had the same structure: The participants were asked the type and temperature of the beverage, the volume of the intake, the reason for the intake, and where and when it was consumed. The questionnaire also asked whether the fluid was consumed by itself or with some food, but did not record the food consumed. To assist the participants in estimating how much fluid was consumed, the records were supported by a photographic booklet of standard fluid containers. The questionnaire items on different types of fluids included: water (tap water, still bottled water and sparkling bottled water); hot beverages (coffee, tea and other hot beverages); milk and milk derivatives; regular sweet beverages (carbonated soft drinks, non-carbonated soft drinks, juices, energy drinks, sports drinks, other sugared soft drinks); diet sweetened beverages (diet carbonated soft drinks, diet non-carbonated soft drinks, other diet soft drinks); and alcoholic drinks. Total fluid intake was defined as the sum of all these categories.

The percentage of individuals following the EFSA adequate intake of water from fluids was calculated. The EFSA sets an AI of total water intake for men and women at 2.5 and 2 L, respectively. The sources of this water can be food moisture, drinking water and different types of beverage. The EFSA assumes that foods usually contribute about 20 % of total water intake [3]. Therefore, we set the AI of water from fluids, preferably water, to be at 2 L/day for men and 1.6 L/day for women. Within this manuscript, we will as of now on refer to these values as AI of fluids.

Socioeconomic level was assessed using a self-administered questionnaire in most of the countries and was categorized using the Market Research Society classification [16, 17].

Height in meters (m) and weight in kilograms (kg) were self-reported by participants, except in Poland, Iran and China where these variables were measured. The body mass index (BMI) was calculated (kg/m²). In Mexico, Brazil, Argentina, Indonesia and Japan, no anthropometric data were available.

Data are presented either as means and 95 % confidence intervals (95 % CI), medians and 25th and 75th percentiles for continuous variables, or numbers and percentages for dichotomous variables. We compared the distribution of the selected characteristics between groups using χ ² tests for categorical variables or Student’s t tests or analysis of variance (ANOVA), as appropriate, for continuous variables. Logistic regression models were fitted to assess the associations between compliance with the AI of fluids for total fluid intake (dependent variable) and gender (two categories) or age categories (four categories) as exposure. The models were adjusted for gender (except when gender was the independent variable), age in years (except when age categories were the independent variable), BMI in kg/m² and socioeconomic characteristics (lower and middle-low, middle, upper-middle and high). All statistical tests were two-tailed, and the significance level was set at p < 0.05. All analyses were performed using the SPSS software version 22.0 (SPSS Inc, Chicago, IL).