Overweight among children has increased dramatically in the last few years approaching epidemic proportions(1) and various environmental, genetic and social factors have been associated with weight gain(2, Reference Dubois and Girard3). Recently, several epidemiological and experimental studies have linked the consumption of sugar-sweetened beverages (SSB), comprising soft drinks, iced tea, fruit juices and colas, to the rising rates of obesity and overweight(Reference Schulze, Manson and Ludwig4–Reference Berkey, Rockett and Field6), and reported that regular SSB consumption from 2·5 to 4·5 years of age more than doubles the odds of being overweight at the age of 4·5 years, though this association remains controversial(Reference Dubois, Farmer and Girard7) and some authors(Reference Johnson, Mander and Jones8) argue that energy compensation may not occur. It has been suggested by the WHO that the body may be unable to detect liquid energy as effectively as energy consumed in a solid form such that any subsequent energy adjustment is less precise(2). Therefore, adding energy from fluids to the diet may lead to an increase in total energy intake as a result of inaccurate regulation of other energy sources. This has been documented for coffee, alcoholic beverages, fizzy drinks, fruit juice and milk(Reference Mattes9).
The studies(Reference Ludwig, Peterson and Gortmaker10–Reference Newby, Peterson and Berkey12) that associate SSB intake and obesity are generally carried out in populations exposed to a Western dietary pattern of intake, characterized by high intake of fat, free sugars, refined grains, red meat, sugary desserts and high-sugar drinks(Reference Pereira13). None of the above-mentioned studies were carried out in a Mediterranean country.
Considering that the different socio-economic and food environments in the studied populations may be important in the association between SSB intake and obesity, these relationships may vary according to specific factors related to the studied population, such as socio-economic background(Reference Drewnowski and Bellisle14). Another of these characteristics may well be the global dietary pattern of intake, and studies on populations with diets that are distant from the Western dietary pattern, such as the Mediterranean diet, may provide valuable supplemental information. In the city of Porto, Portugal, the Mediterranean diet still represents an important pattern of food intake that may protect from chronic disease severity(Reference Barros, Moreira and Fonseca15).
The present study examined whether high SSB intake is associated with overweight in a sample of Mediterranean schoolchildren.
Methods
The data were derived from a community-based survey of children selected randomly from forty elementary schools, throughout the city of Porto, Portugal, of which thirty-five agreed to participate (thirty-five public schools with 5867 students). All schoolchildren’s parents were invited to participate in the present study, of which 3391 accepted. After parental written consent was received, 2512 children’s data sets were obtained. Each data set contained one FFQ regarding children’s food intake and one sociodemographic questionnaire, both of which were asked to be filled out by their parents or tutors (the major responsible for their daily food intake). Children who were sick (n 3) and/or with incomplete questionnaires (n 242) were not included in the study. After dietary analysis, we excluded energy intake outliers (n 241; <3347 kJ (<800 kcal) and >16 736 kJ (>4000 kcal) for boys, and <2510 (<600 kcal) and >14 644 kJ (>3500 kcal) for girls)(Reference Willett16) and/or adolescents (>10 years old; n 50) from the study. As many studies showed interesting results with SSB ingestion of >1500 ml(Reference Forshee, Anderson and Storey17), we also assumed any student with a mean daily intake of SSB higher than 1500 ml (n 301) as an outlier. The final sample included 1675 schoolchildren aged between 5 and 10 years.
Dietary intake was recorded using a semi-quantitative FFQ of the previous 12 months, comprising eighty-two food items or beverage categories and a frequency section with nine possible responses ranging from never to six or more times/d, developed by Lopes et al.(Reference Lopes18, Reference Lopes, Aro and Azevedo19). The FFQ was designed according to Willett(Reference Willett16) and adapted by inclusion of a variety of typical Portuguese food items.
Nutrient intake data were obtained by multiplying the frequency of consumption of each food item by the nutrient content of the specified serving size, with once per day being equal to one. Seasonal variation of food consumption was also considered according to indications of participants on the average number of months of consumption per year. Energy and macronutrient intakes were estimated using an adapted Portuguese version of the nutritional analysis software Food Processor Plus (ESHA Research Inc., Salem, OR, USA). The analysis considered the intake of the SSB servings (three serving categories/d, such as <1, 1–3 and >3; 1 serving size = 330 ml) derived from the FFQ.
Weight and height were measured in accordance with international standard methods for the collection of anthropometric data in children(20, 21). Weight was measured to the nearest 0·05 kg and height to the nearest 0·1 cm. Both measures were determined using a precision stadiometer (model 702; Seca, Hamburg, Germany) and taken in light clothing without shoes.
The weight status of the participants was also classified as recommended by the International Obesity Task Force, making a correspondence between the traditional adult cut-offs (>25 kg/m2 to define overweight and 30 kg/m2 to define obesity) and specific values for adolescents considering sex and age in complete and half years(Reference Cole, Bellizzi and Flegal22).
The sociodemographic questionnaire contained information on the age of the children and the education of the parents, evaluated in years, using the following categories: 0, 1–4, 5–9, 10–12 and >12 years. For the purpose of the data analysis, these were collapsed into: up to 9, 10–12 and >12 years.
In order to assess the level of physical activity, the questionnaire included items regarding the time spent watching television (TV)/video and playing computer/electronic games. Answers to each of these questions were divided into weekdays, Saturdays and Sundays, and for the category, these were further sub-divided into the following: <1 h, 1–2 h, 2–4 h, 4–6 h and >6 h. Later, these variables were recoded for the analysis into: up to 1 h/d and ≥1 h/d.
Statistical analysis was performed by the Statistical Package for Social Sciences statistical software package version 14·0 (SPSS Inc., Chicago, IL, USA) and the significance level was set at 5 %. OR estimates, including CI, were computed using binary logistic regression to estimate the association between SSB intake (considering three categories: <1 serving/d (reference category), 1–3 and >3 servings/d) and overweight. The adjustment considered variables that presented a statistically significant effect in univariate analysis and were considered to sustain a plausible biological and temporal relationship with the outcome (TV watching, sleep time, energy intake, parental education level, total carbohydrates, sugars, MUFA and questionnaire responder).
Throughout the paper, unless stated otherwise, it can be assumed that the overweight group also includes obese children.
Results
Girls showed higher percentages of overweight (25·8 %) than boys (24·7 %) and boys showed higher percentages of obesity (14·1 %) than girls (10·8 %).
Energy intake was not significantly different between overweight and non-overweight children. Regarding the percentage contributions to total energy intake (Table 1), no differences were found between normal and overweight girls; however, significantly higher mean intakes of total carbohydrates (52·3 v. 51·2 %, P = 0·014) and sugars (23·9 v. 22·9 %, P = 0·008) were found in overweight boys, compared with normal-weight boys. In contrast, overweight boys presented significantly lower contributions of MUFA (13·3 v. 13·6 %, P = 0·035).
Contributions to total energy intake and descriptive social factors by gender, weight and SSB intake
SSB, sugar-sweetened beverages; TV, television.
Student’s t test and Mann–Whitney U test were used to compare nutritional variables according to overweight status and SSB intake, in each gender; for other variables, χ 2 tests were used to estimate their association according to overweight status and SSB intake, in each gender.
*1 kcal = 4·184 kJ.
Parents of non-overweight girls had a higher education level than those of overweight girls, but in boys we found that the overweight group had mothers with higher education levels than those in the non-overweight group. Significant statistical differences in overweight and non-overweight children were found by the girl’s mother’s education level (P = 0·001) and the boy’s father’s education level (P = 0·009).
Overweight children spent more time watching TV, but significant results were found only on weekdays (P = 0·042 for girls and P = 0·036 for boys).
Table 1 shows different distributions between the children’s SSB intake (categories of the intake of SSB servings/d) and mother’s education (P = 0·050), father’s education (P = 0·012) and TV watching on weekdays (P = 0·010) in girls, and father’s education (P = 0·002), TV watching during Saturday (P = 0·017) and TV watching during Sunday (P = 0·031) in boys.
We did not find any association between overweight and SSB consumption (Table 2) in any consumption group even after adjustment for TV watching, sleep time, energy intake, parental education level, total carbohydrates, sugars, MUFA and questionnaire responder (1–2 servings/d v. <1 serving/d: OR = 1·67, 95 % CI 0·76, 3·66 in girls and OR = 1·63, 95 % CI 0·76, 3·47 in boys; and >2 servings/d v. <1 serving/d: OR = 0·63, 95 % CI 0·33, 1·22 in girls and OR = 0·64, 95 % CI 0·33, 1·52 in boys).
Association between overweight and SSB consumption
SSB, sugar-sweetened beverages; Ref., reference category.
Binary logistic regression analysis adjusting for energy intake, parents’ education level, time of sleep, questionnaire responder, total carbohydrates, sugars, MUFA and television watching.
Discussion
Beverages can play an important role in energy intake due to the evidence that high intake of liquid carbohydrates may lead to weight gain because of a lack of dietary compensation compared to that observed in similar amounts of solid carbohydrates(Reference DiMeglio and Mattes23). This SSB association with overweight in children found in some studies from other countries(Reference Ludwig, Peterson and Gortmaker10, Reference Sanigorski, Bell and Swinburn24) was not supported in our study.
We did not find any association between SSB and overweight, for both genders, even after statistical adjustment for potential confounders. Although findings from some cross-sectional(Reference Ariza, Chen and Binns25, Reference Andersen, Lillegaard and Overby26) studies suggest a positive trend in the relationship between SSB intake and obesity, meta-analysis(Reference Forshee, Anderson and Storey17) failed to find such an association. Controversy between studies may arise from different instruments to measure SSB consumption, different measures of weight gain, different statistical models to estimate the effect sizes and different units of time. Indeed, Drewnowski et al.(Reference Drewnowski and Bellisle14) suggest that the socio-economic environment should be taken into account as a factor of SSB intake, and not only with its isolated ingestion. Nevertheless, even after adjustment for these potential confounders in our study, no significant association was found between SSB and overweight.
In our study, parents were interviewed and reported on the consumption patterns of their children, and although this data collection method does not rely on children’s memory, it may be associated with bias due to lack of parental knowledge of children’s eating habits (particularly on food eaten away from home), inability to estimate exact serving sizes and frequency of consumption and the possibility of a differential recall bias by parents across children’s weight status categories(Reference Livingstone, Robson and Wallace27).
One discussion point in the present study could be the inclusion of all SSB (colas, iced teas and fruit juices) in the same analysis group. This decision was based on our previous data(Reference Bessa, Valente and Cordeiro28) in which no effect of isolated SSB consumption (colas, iced teas and fruit juices) on weight was found. The intake of each SSB per se was probably not enough to show any effect on children’s weight.
Another perspective could be that regular overweight SSB consumers may unwittingly be consuming more sweetened beverages than reported. Likewise, parents of overweight children may be under-reporting their children’s intake of SSB(Reference Livingstone, Robson and Wallace27), and these factors may also explain the lack of evidence in our study.
Our findings indicate a higher prevalence of overweight in children from both genders than do recent Portuguese data(Reference Padez, Fernandes and Mourao29). Our findings also show a higher prevalence of overweight in children from both genders, especially in boys (38·8 %). This high value supports the urgency of a nutritional intervention for Portuguese schoolchildren, namely using education programmes in school, parents’ education and control of the contents of school vending machines. At the macronutrient level, high intakes of fat and protein, and a low intake of carbohydrates, regarding WHO recommendations(2), seem to be one of the causes of the growing levels of overweight.
Our results also suggest that overweight children spend more time watching TV than do normal-weight children, for both genders. According to the American Academy of Pediatrics, children’s total media time (with entertainment media) should not be higher than 1–2 h/d(Reference Walsh and Gentile30), and children exposed to TV viewing for more than 1 h/d may already be at increased risk for overweight(Reference Bessa, Valente and Cordeiro28). WHO(2) classifies high dietary fibre intake and regular physical activity as convincing factors to protect against weight gain, and therefore these two factors should be the major focus base of any planned intervention in children.
To the best of our knowledge, this is the first study to describe that high SSB intake is not associated with overweight in a sample of Mediterranean schoolchildren. Our findings suggest that reducing or eliminating SSB consumption would not have a large effect on the BMI distribution of children. Therefore, dietary advice and education should clearly communicate that SSB should only be consumed in moderation as part of a balanced diet, especially for those children who are overweight or at risk of becoming overweight, because SSB still are a source of energy.
Future research should also try to evaluate whether high SSB intake is related to other food intake patterns so that we could have better knowledge of the role of SSB and children’s obesity.
Acknowledgements
The present study was supported by a scholarship from Nestle Portugal SA. The authors declare that they have no conflict of interest. H.V. supervised all aspects of the study implementation, conducted statistical analyses and wrote the paper. V.T., P.P., A.M., V.M., C.L. and J.M. helped to interpret the findings and review the drafts of the manuscript. M.B. and T.C. helped to review the drafts of the manuscript. P.M. designed the study, supervised all aspects of its implementation, conducted the statistical analyses and helped to interpret the findings and review the drafts of the manuscript.
