Intake of water and beverages of children and adolescents in 13 countries

The World Health Organization (WHO) has raised concern regarding an excessive intake of sugar-sweetened beverages (SSB) for children [1]. This concern is based on a meta-analysis of long-term prospective cohort studies concluding that children consuming the largest intakes of SSB had greater likelihood of being overweight or obese than children with the lowest intakes [2]. Moreover, the sugars present in SSB have been associated with dental carries prevalence both in children and adults [2, 3]. Consequently, the WHO has set recommendations for intake of free sugars to <10 % of total energy intake and suggests a further reduction to <5 % of total energy intake [1]. Evidence has suggested that a reduction in energy intake facilitating weight management can be achieved among regular SSB consumers if they replace their SSB with drinking water [4, 5].

Now that this recommendation has been made, surveying populations is globally needed to assess the intake patterns of different fluid types (water and all other beverages). Several large cohort or cross-sectional studies have already reported the intake of different fluid types in children [6‐9]. In 2014, Özen et al. [10] published a systematic review of studies assessing beverage consumption across age groups. In children, plain water contributed up to 58 % of total beverage intake, with great variability from 21 to 58 % between countries [10]. This number was increased to 51–75 % in adolescents. Surprisingly, even though the studies included in this review assessed beverage consumption, they did not all report the intake of water [10]. Furthermore, other inconsistencies were noted in the study design, dietary assessment methods, classification of beverages and age categories which limit the comparison of results between countries. To make an inter-country comparison, a review containing surveys which assessed fluid intake with the same methodology and reported intake of all fluids would be needed. Ideally, the sample of such surveys would be representative of the national sample and would cover the same and wide age range. The review by Özen et al. [10] showed also that for one country, such as the USA, several studies reporting on fluid intake of children are available. However, fluid intake of children and adolescents remains to be assessed in numerous countries worldwide. The aim of the present pooled analysis was therefore to describe the intake pattern of water and all other beverages in children and adolescents, aged 4 up to 17 years in 13 countries of three continents. Differences in intakes between sex and age groups are also reported.

Table 1 describes the sample size, age and BMI of both age groups by country. The proportion of children and adolescents in the sample of each country was balanced, except for Belgium, Iran and China. In the latter three samples, 68–80 % of the participants were adolescents. Consequently, the mean age of the children in these three samples was higher than the mean age of the children in the other samples.

In general, across the total samples of the countries, the highest daily intakes were observed for water (738 ± 567 mL/day), followed by milk (212 ± 209 mL/day), RSB (168 ± 290 mL/day) and juices (128 ± 228 mL/day). There was a large inter-country variation in the intake of a given fluid type. Daily water intake ranged from 296 mL/day in Poland to 1516 mL/day in Indonesia, whereas daily milk intake ranged from 123 mL/day in Indonesia to 530 mL/day in Uruguay. The intake of RSB and juices ranged from 64 mL/day in China to 625 mL/day in Argentina and from 21 mL/day in Indonesia to 555 mL/day in Brazil, respectively. The age-, sex- and country-specific means of intake of the different fluid types are presented in Tables 2 and 3.

Despite these large differences in volumes of intake of the different fluid types, some samples had comparable patterns of contribution of fluid types to TFI in the children (Fig. 1). The intakes of the Chinese and Indonesian sample were characterized by the largest contribution of water to TFI (respectively, 67 and 73 % in total children sample) of all samples. In the samples of Spain, France, Belgium, Iran and Turkey, half of the TFI came from water (42–53 % in total children sample). In these five samples, the other fluids contributed for a similar amount to TFI, except for the hot beverages. In the total children sample of Iran and Turkey, hot beverages contributed, respectively, for 10 and 13 % to TFI, whereas in the French and Spanish sample they contributed only for 0–3 %. Large contributions of hot beverages to TFI were also reported in the total children sample of Poland (34 %) and Argentina (13 %). Besides the large contribution of hot beverages, these two samples were also characterized by a contribution of RSB to TFI (22–35 %) that was larger than the contribution of water to TFI (19–21 %). Similar results were observed in the total children sample of Brazil, Mexico and UK: in Brazil the contribution of juices to TFI (29 %) and in Mexico and UK the contribution of RSB to TFI (respectively 29–32 %) were as large as the contribution of water to TFI (29–36 %).

Similar patterns in the contribution of the different fluid types to TFI were identified among adolescents (Fig. 2). However, a comparison between the intake of children and adolescents indicated significant age effects (all with p value <0.001). The most consistently observed age effect was regarding the contribution of milk to TFI: adolescents in all samples except in Belgium and Mexico had a significantly lower milk intake than children. Moreover, adolescents had a significantly higher contribution of RSB to TFI than children in Brazil, Uruguay, Spain, Turkey and Iran. In the sample of Iran, children had a higher contribution of juices to TFI than adolescents, whereas in the Chinese sample the opposite was observed. The contribution of hot beverages to TFI was significantly higher among adolescent than among children in the sample of Brazil, Uruguay, Argentina, France Iran and China. The contribution of water to TFI was comparable between children and adolescents, except in the sample of Indonesia.

Significant gender differences in the contribution of the fluid types to TFI were observed in the individual samples, yet they were inconsistent. The contribution of water to TFI was significantly higher for females in the Belgian sample (p = 0.0057), whereas it was lower in the Chinese samples compared with males (p = 0.0001). The milk contribution to TFI was higher for females in the Chinese samples, but lower in the Indonesian sample (p = 0.004) than for males. The contribution of other beverages to TFI was significantly higher for females in the Chinese and Belgian sample than for males (p < 0.0001 and p = 0.0026, respectively). In the Chinese sample, females also had a significantly higher contribution of hot beverages compared with males (p < 0.0001). The only gender difference that was consistent across several samples was observed for the contribution of RSB to TFI: males had a significantly higher RSB contribution than females in the samples of Belgium, UK, Iran and China (p < 0.01 for all).

When analysing the gender difference within the two age categories, significant gender differences were also observed. Among children (Fig. 1), the contribution of milk to TFI was significantly higher among men than among women in the Brazilian sample (p = 0.01); however, in the Iranian sample, the effect was the opposite direction (p = 0.005). The Chinese females drank more hot beverages than males (p = 0.01). Male children in UK and China had a significantly higher RSB contribution to TFI than females (p = 0.01 and p = 0.0001, respectively). In the Belgian samples, the females had a higher contribution of other beverages than males (p = 0.01). Among adolescents (Fig. 2), most gender differences were observed in the Chinese sample: males had a significantly higher contribution of water and RSB to TFI than females (p < 0.0001 for both), whereas females had a significantly higher contribution of milk, hot beverages and other beverages (all p < 0.0001) to TFI than males. In the Iranian sample, adolescent males had a higher contribution of RSB to TFI than adolescent females (p = 0.0002).

This unique pooled analysis of individual data of 13 cross-sectional surveys provides novel insights on fluid intake for countries that, to the best of our knowledge, had no internationally published data so far. Since all 13 surveys used the same method for data collection (a 7-day fluid-specific record), this pooled analysis gave the opportunity to observe differences in intake patterns between samples of different countries. While the range in mean milk and RSB intake between samples with the lowest and largest intake volumes was 407 and 561 mL/day, respectively, the range in mean water intake reached 1220 mL/day. These large differences in mean intake are not unusual as shown in a review from Özen et al. [10] reporting an inter-country range in TFI of 1.2 L/day (0.6–1.8 L/day), with the water contribution ranging from 21 to 58 %. Moreover, the intakes estimated for the total adolescent sample in this study are very much in line with those reported by Duffey et al. [18] in adolescents of eight European countries. It is worth noting that the intake of water was higher and the intake of RSB and alcoholic beverages lower in the current analysis compared with that of Duffey et al. [18]. Discrepancies in methodologies of recording and classification of the sugar-sweetened beverages may explain this difference.

The large differences in intakes of the different fluid types between the samples may be partly explained by differences in climate. Indeed, temperature, humidity and seasonality influence both volume consumed and the preference for certain fluid types [19, 20]. Since the surveys reported here were not designed to explore inter-country variability, no data on temperature or humidity were gathered during the period of data collection nor were seasonality taken into account. Other possible explanations for the large inter-country differences observed are cultural habits and geographical location. A certain fluid type might be consumed more out of tradition in a country (e.g. tea in UK), but also certain foods or the amount of a nutrient consumed in a certain country may influence fluid intake [21]. Samples of countries with a similar geographical location indeed showed similarities in the contribution of the different fluid types to TFI. All samples of Latin America, Mexico, Brazil, Uruguay and Argentina were characterized by a high contribution of RSB and juices to TFI. Argentina, however, differed from the other three Latin American countries by a larger contribution of hot beverages to TFI, more specifically the traditional Mate. The samples of the Asian countries included in this analysis (Indonesia and China) and also those from countries relatively closely located around the Mediterranean Sea (France, Spain, Iran and Turkey) had a comparable pattern: at least half of the fluid intake in these countries came from water. The contributions of the fluid types to TFI observed in the Belgian sample seemed comparable to the pattern observed in the samples of the Latin American countries. However, hot beverages were not recorded in the Belgian survey, and therefore, a comparison with other countries should not be made. Data in adult samples also showed similar contributions of the different fluids types to TFI in countries from the same geographical area [16]. This is not surprising since a number of studies have also shown that parental food preferences and nutrient intake including SSB are adopted by children and adolescents [22‐24]. This observation suggests that there is a risk of relaying detrimental food and beverage intake habits between generations. This remains to be confirmed for fluid intake in the future.

Differences in the contribution of fluid types to TFI were observed between the two age groups, which have been reported by others [10, 18]. Among children, the intake of milk and juice was higher than among adolescents, whereas adolescents consumed more water, hot beverages, RSB and alcoholic beverages. In the total sample, both the volume and the contribution to TFI of RSB were significantly higher among children than among adolescents; however, when each sample was considered individually, adolescents always had a higher RSB intake than children. However, due to the significant differences in intake patterns between the samples and due to the unbalance in sample sizes of the countries, interpretation of the pooled data of the total sample should be done with caution. Nevertheless, among European adolescents aged 12.5–17.5 years, similar age effects on fluid intake were observed [18]. Özen et al. [10] also drew similar conclusions in their systematic review: milk intake was higher among children and was replaced by regular fluid/soft drinks among adolescents. They also reported that with age the intake of hot beverages and diet beverages increased.

The effect of gender on the intake of the different fluids was neither consistent nor very pronounced in the samples, except for the contribution of RSB to TFI. In four samples, males consumed more RSB than females. This observation suggests that females start adopting healthier hydration habits than males during adolescence, potentially due to an increased health consciousness or attention to their body image [25]. This lack of consistent gender effect on the intake of the different fluids was not in line with what has been reported previously. In a large European adolescent sample, males clearly had a higher contribution of high fat milk, SSB and alcoholic beverages and lower contribution of water than females [18]. The review by Özen et al. [10] reported also that males had a higher milk consumption than females. Until the gender effect on the intake of the different fluids has been analysed again by future surveys, it is recommended to interpret the gender and age effects country-by-country.

In eight out of the 13 samples included in this analysis, the combined mean of juices and RSB of both consumers and non-consumers was higher than 335 g/day. Intervention studies and cohort studies have shown that children and adolescents consuming SSB on a daily basis are at increased risk of becoming overweight or obese compared with non-regular consumers [26, 27]. Adolescent females who consumed more than 335 g/day had a greater overall cardio-metabolic risk, independent of their weight status (OR 3.2; 95 % CI 1.6, 6.2) (all p-trend ≤0.001) [27]. In light of the current prevalence of obesity and diabetes, health promotion strategies should focus, among others, on reducing the intake of SSB and increasing the intake of water. The large differences in intake patterns across samples indicate that in some countries the concern about an excessive SSB intake is higher than in other countries. Though only in three samples water was the majority of fluid intake, the concern is global.

All fluid intake data used in this analysis were self-reported. For children younger than 12 years, the primary care giver was responsible for filling the 7-day fluid record. Therefore, the risk of over- or underestimation of intake and a possible reporting error by the primary care giver cannot be excluded. In future studies, combining the recording of the intake with the collection of urine biomarkers may give an indication of the accuracy of the intake reporting. Also a validation of accuracy and reliability of the 7-day fluid record would be useful. It would also allow an estimation of the hydration status of children and adolescents. An estimate on how fluid intake contributed to the whole diet also cannot be made due to lack of food data. However, evidence suggested that a fluid-specific record might more accurately estimate fluid intake compared with a food and fluid record [28]. Since the primary aim of all 13 surveys was to assess fluid intake, the preference was given to record fluids only. Another limitation to acknowledge is that not all samples were necessarily representative of the national target sample of the country. Nevertheless, the methods of recruitment used in the surveys are recognized as valuable methods to provide enough sample by age of participants, regions of the country and different socio-economic groups for meaningful analysis. In the future, it would also be recommended to avoid the minor differences in the fluid classification that were currently present across countries.

This pooled reanalysis of individual data has several strengths. All surveys used a fluid-specific record over 7 consecutive days and are therefore assumed to provide data highly representative of habitual daily intakes. Moreover, all records were supported by a photographic booklet to increase accuracy of the reported volumes. An additional strength is that this compilation of 13 samples of different countries allowed to highlight the large diversity in fluid intake patterns across countries.