OBJECTIVE: To assess whether intake of fruits and vegetables was associated with change in body mass index (BMI) among a large sample of children and adolescents in the United States.
DESIGN: Prospective cohort study of children and adolescent who were 9–14 y of age in 1996, when the study began.
SUBJECTS: The subjects included 8203 girls and 6715 boys in an ongoing cohort study who completed at least two questionnaires between 1996 and 1999.
MEASUREMENTS: Fruit and vegetable intake was assessed in 1996–1998 with a validated food frequency questionnaire designed specifically for children and adolescents. The outcome measure was change in age- and gender-specific z-score of BMI (kg/m2). Self-reported weight and height, which were used to calculate BMI, were collected annually from 1996 to 1999.
RESULTS: During 3 years of follow-up, annual changes in BMI were slightly greater among the boys than the girls. After controlling for Tanner stage of development, age, height change, activity and inactivity, which are known or suspected predictors of change in BMI, among the girls there was no relation between intake of fruits, fruit juice, or vegetables (alone or combined) and subsequent changes in BMI z-score. Among the boys, intake of fruit and fruit juice was not predictive of changes in BMI, however, vegetables intake was inversely related to changes in BMI z-score (β per serving=−0.003). However, after adjusting for caloric intake, the magnitude of the effect was diminished and no longer significant.
CONCLUSION: Our results suggest that the recommendation for consumption of fruits and vegetables may be well founded, but should not be based on a beneficial effect on weight regulation.
Diets rich in fruits and vegetables are associated with a decreased risk of many chronic diseases1,2,3 and therefore both the National Institute of Health, in the Healthy People 2010 objectives,4 and the National Cancer Institute5,6 recommend a diet high in fruits and vegetables. Despite these recommendations, few children or adults meet the five a day goal.7 According to Continuing Survey of Food Intakes by Individuals (CSFII), a study conducted by the USDA between 1994 and 1996, approximately 22% of adolescents consume at least two servings of fruit per day and 38% of adolescent girls and 55% of adolescent boys consume at least three servings of vegetables per day, thus less than 25% of adolescents meet the five a day goal.
The association of fruit and vegetable intake to weight gain is not well understood. On one hand, there has been a concern that high intake of fruit juice could promote the development of obesity; however, the results have not been consistent across studies.8,9 Despite a lack of data, several obesity-prevention interventions have included efforts to increase the consumption of fruits and vegetables, in the hope that these foods will replace more energy-dense food choices popular among children and adolescents.10,11,12,13 To assess whether fruit, fruit juice, or vegetable intake is predictive of increases in body mass index (BMI), we analyzed data from approximately 15 000 preadolescent and adolescent girls and boys in the Growing Up Today Study (GVTS), a prospective cohort study of youth throughout the United States.
The GUTS was established in 1996 by recruiting children, 9–14 y of age, of women participating in the Nurses' Health Study II (NHS II). Using the NHS II data, we identified mothers who had children ages 9–14 y. In total, we identified over 53 000 children in the eligible age range. We wrote a detailed letter to the mothers, explaining that the purpose of GUTS was to study the predictors of weight change during adolescence and sought parental consent to enroll their children. The mothers who responded to the invitational letter provided us with the names, age, and gender of more than 26 000 children in the eligible age range. In October 1996, we mailed letters and baseline questionnaires to the 13 261 girls and 13 504 boys whose mothers had granted us consent to invite them to participate in GUTS. The invitation letter to the child explained the study and asked them to complete the questionnaire if they wished to participate. The letter assured the children that no information they provided would be made available to anyone, including their parents. Approximately 68% of the girls (N=9039) and 58% of the boys (N=7843) returned completed questionnaires, thereby assenting to participate in the cohort. Since ongoing participation is crucial to the validity of cohort studies, we did not make efforts to increase baseline participation from the unwilling children (ie the nonrespondents).
Dietary intake, physical activity, inactivity, Tanner stage of development, weight, and height were assessed annually from 1996 to 1999. Height and weight were self-reported. We calculated BMI using self-reported weight and height information (wt(kg)/ht(m)2)and calculated percentiles and z-scores based on the Centers for Disease Control and Prevention and the National Center for Health Statistics growth charts,14 which are age- and gender-specific. Participants with a BMI between the national 85th and 95th percentile for age and gender were classified as overweight, those with a BMI >95th percentile were classified as obese. In addition, drawings of the five Tanner stages of development of pubic hair were used for assessing pubertal development.
Dietary intake was assessed with the Youth/Adolescent Questionnaire (YAQ), a self–administered semiquantitative food frequency questionnaire assessing intake over the past year.15 The questionnaire asks participants how often, on average, they consumed each of the 131 foods listed (such as mixed vegetables, macaroni and cheese, and peanut butter sandwich). The food portion sizes were determined by a review of US Department of Agriculture Handbook No. 8 serving sizes, NFCS Foods Commonly Eaten by Individuals (specifically for ages 9–18 y), and the ‘natural’ serving size (eg bread slice; apple 1). Foods that do not have a ‘natural’ serving size, such as spinach, do not have a portion size listed on the questionnaire. There are 11 questions on fruit and juice intake (when we use the classification scheme probably used by most children that includes tomatoes as a vegetable) and 19 questions on vegetables. Daily servings of fruit, daily servings of vegetables, and daily servings of fruits and vegetables were estimated by summing these items. Children who reported consuming less than 500 or greater than 5000 calories were outliers in terms of plausibility and therefore excluded from the analysis.
Physical activity was assessed with 18 questions on hours per week within each of the four seasons that a participant engaged in a specific activity (eg, volleyball, soccer). A summary score of average hours per week of physical activity was computed. Reports of an average of more than 40 h per week were considered implausible and therefore set to missing and not used in the analysis. Inactivity was estimated by asking participants the average number of hours per week spent watching TV, watching videos or VCR, reading/homework, or playing Nintendo/Sega/computer games. Reports of an average of more than 80 h/ week of inactivity were considered implausible and therefore set to missing and not used in the analysis.
Reliability and validity of self-reported Tanner stage, height, and weight
Pubertal development has been assessed annually since 1996 with drawings of the Tanner stages of development of pubic hair and (for girls) breasts, as well as questions about onset of menses. Self-assessment of maturation has been validated separately by three teams of investigators.16,17,18 Overall, subjects' assessments correlate highly with physicians' assessments. Duke et al16 found kappa coefficients between 0.81 and 0.88 for male and female children ages 11–18 y. Morris and Udry,17 assessing a sample of black, white and Hispanic children report moderately high correlations for females (r=0.81 for genital hair and r=0.60 for breast development) and males (r=0.63 for genital hair distribution).
The validity of self-reported weight and height among preadolescents and adolescents has been investigated by several groups of researchers. Shannon et al,19 in a sample of 806 sixth graders, report moderately high correlations for weight and height for both boys (r=0.90, r=0.74) and girls (r=0.84, r=0.62). More children tended to under-report than over-report their weight, with the taller and heavier children and girls showing the greatest tendency to do so. The authors did not find a systematic bias in reported heights. Among 1657 adolescents, aged 12–16 y, in the NHANES III study, Strauss observed high correlations between self–reported weight and actual weight (r=0.87 and 0.94, depending on gender or race) and self-reported height and actual height (ranged from r=0.82–0.91). Moreover, the use of self-reported weight and height resulted in the correct classification of weight status in 94% of children.20 In addition, in a nationally representative sample of youth from the National Longitudinal Study of Adolescent Health, Goodman et al21 observed that the correlation between BMI calculated from self-report vs measured height and weight was 0.92 and that only 3.8% of the youth were misclassified as obese from that calculated from self-reported BMI. Thus, the results suggest that preadolescents and adolescents provide information on weight and height that is as valid as the information provided by adults.
The outcome was annual (1996–1997, 1997–1998, and 1998–1999) weight change. Weight change was modeled as the differences in age- and gender-specific z-score of BMI (based on the CDC reference data) over a 1-y period, with the age- and gender-specific z-score of BMI at the beginning of the interval as the predictor and the age- and gender-specific z-score of BMI at the end of the 1-y interval as the outcome. The z-scores were used instead of raw BMI values because the former had a lower correlation with height (Spearman r=0.1 vs 0.3).
Participants included 8203 girls and 6715 boys who were 9–14 y of age in 1996 and completed at least two GUTS questionnaires between 1996 and 1999. There were 926 girls and 1235 boys who completed two, 1759 girls and 1971 boys who completed three, and 5518 girls and 3509 boys who completed all the four questionnaires.
All analyses were stratified by gender and conducted with SAS software.22 Conditional linear models, which allow for variation in the time between exposure assessments, were used for all multivariate analyses (SAS proc nlin).23 All models assessing the association between intake of fruit (with and without fruit juice), fruit juice, and vegetables (separately and in combination) and weight change over a 1-y period controlled for age, age squared, Tanner stage of pubic hair development, activity, inactivity, age- and gender-specific z-score of BMI at the beginning of the 1-y interval, and height change over the 1-y interval. To assess whether the effect of fruit and vegetable intake was because of differences in intake of total calories, we reran the linear regression models controlling for caloric intake. Three 1-y periods were assessed: 1996–1997, 1997–1998, and 1998–1999. Fruit and vegetable intake was modeled as continuous variables in the analyses and both fruit and vegetable intake at the beginning of the 1-y period and change in intake over the year were included as covariates in the models. All P-values are two-sided, with P<0.05 considered statistically significant.
In 1996, when the study began, the mean BMI of the girls was slightly lower than that of the boys (19.0 vs 19.1 kg/m2, respectively, Table 1). The prevalence of overweight and obesity was higher among the boys (Table 2). During 3 y of follow-up, annual changes in BMI were slightly greater among the boys than among the girls (0.7–0.8 units per year among the boys vs 0.6–0.7 units per year among the girls).
In 1996, amount of fruit and vegetables consumed was similar among males and females (Table 1). On average, girls and boys consumed slightly fewer than two servings of fruit per day, of which almost 50% was in the form of juice. Participants consumed fewer servings of vegetables than that of fruit per day; however, the differences were small. The decision whether or not to include french fries or other potato dishes as vegetables did not materially alter the estimates of vegetable intake (Table 1). The mean number of servings of vegetables per day was approximately 0.3 servings less when potatoes were not included (Table 1). The mean servings per day of fruits and vegetables (not including french fries) combined dropped slightly from 1996 to 1998 among both girls and boys. Fewer than 25% of the participants were meeting the recommendation to consume at least five servings of fruits and vegetables per day (Table 2).
After controlling for Tanner stage of development, age, height change, activity and inactivity, which are known or suspected predictors of change in weight and BMI, we found there was no relation between intake of fruits, fruit juice, or vegetables (alone or combined) and subsequent changes in BMI z-score among the girls (Table 3). Among the boys, intake of fruit and fruit juice was not predictive of changes in BMI; however, vegetable intake was inversely related to changes in BMI z-score (β=−0.003, Models 4, 5, and 7, Table 2). Among the boys, the magnitude of the effect was diminished and no longer significant after adjustment for caloric intake (Table 3). Thus, among the boys, consuming the same number of calories per day, vegetable intake offered no benefits in terms of change in BMI. However, after adjusting for total calories, fruit intake (β=0.003 for girls and β=0.002 for boys, Models 7 and 8, Table 3) was predictive of having a slightly larger BMI z-score at the end of the follow-up period.
In a large cohort of preadolescents and adolescents living throughout the United States, weight and height changes were relatively consistent during the 3 y of follow-up. At the beginning of the study slightly more males than females were overweight, and during the follow-up, change in BMI was slightly larger among the boys than in the girls. Among both genders, approximately 75% of the adolescents were not meeting the public health recommendation to consume at least five servings of fruits and vegetables per day.
Neither fruit or juice intake predicted changes in BMI, but among the males vegetable intake was inversely related to changes in BMI z-score. The benefit of vegetables, however, was diminished and no longer significant once total calories were included in the statistical model. The fact that the protective effect of vegetable intake among the males was diminished when calories were included in the model suggests that the benefit was acting through caloric intake. In other words, diets high in vegetables may have been lower in calories and it was the calories, not vegetables per se that were inversely predicting changes in BMI z-scores. With adjustment for total calories, there was a suggestion that consuming a diet rich in fruits might lead to larger gains in relative weight; however, the effect was very modest. With or without control for total calories, there was no evidence that vegetables intake was inversely related to gains in relative weight among the girls.
There has been a concern that high intake of fruit juice by young children could promote the development of obesity; however, the results of studies on the topic have not been consistent.8,9,24 Unless we controlled for total caloric intake, we did not observe an association between consumption of fruit juice and change in BMI during adolescence or find that fruit intake predicted change in BMI.
There are numerous benefits to consuming a diet rich in fruits and vegetables, but it is not entirely clear why this type of diet would prevent obesity or excessive weight gain. One possible mechanism would be that fruits and vegetables might serve as healthy substitutes for more calorie-dense foods. The effect of substituting fruits and vegetables for calorie dense foods could be to lower caloric intake. However, it is possible to consume a diet rich in calorie dense foods, as well as fruits and vegetables. For example, a vegetarian burrito could contain several types of vegetables cooked in oil, cheese, sour cream, refried beans, and a large tortilla, thus it would be high in fat and calories, as well as provide one or more vegetable servings. Moreover, if a child occasionally substituted a piece of fruit for cookies, cake, or other sweets, it would be unlikely to have a material impact on total caloric intake over an extended period of time unless the child otherwise would be eating a very large serving of the sweets. Therefore, the lack of a strong relation between intake of fruits and vegetables and change in BMI should not be surprising.
Although there has been support for promoting fruit and vegetable intake as a means for preventing weight gain, there is one mechanism through which certain vegetable intake could promote obesity. Glycemic index is a property of carbohydrate-containing food, which describes the increase in blood glucose after a meal. Foods that are digested and absorbed rapidly, such as refined grains and potatoes, have a high glycemic index.25,26 The physiologic responses to oral glucose suggest a mechanism that links glycemic index to weight gain. Oral administration of glucose or foods with a high glycemic index produces rapid elevations in blood glucose and insulin levels, which are followed in many individuals by a period of reactive hypoglycemia with continued modest elevation in insulin levels. This situation results in hunger and increased food intake, possibly leading to cycles of hypoglycemia and hyperphagia. The relative hyperinsulinemia would also promote storage of fat.27 A recent feeding study among humans provides support for this hypothesis. In the study, which was a crossover design, 12 obese teenage boys consumed meals with a high-, low-, and medium-glycemic index. Voluntary energy intake after the high-glycemic-index meal was 81% greater than after the low-glycemic-index meal, with the medium-glycemic-index meal showing intermediate values.28 However, in our study we did not observe any consistent evidence that vegetables, with or without potato products, promote weight gain relative to height gain. Although there was a suggestion that increases in vegetable intake were predictive of gains in BMI z-score among the females, the lack of a similar association among the males and the fact that the effect was diminished after control for caloric intake makes it unlikely that high glycemic index is the explanation for the finding about vegetable intake among the girls.
Our study has several limitations: it does not represent a random sample of all US adolescent males and females. Since the participants are children of nurses, we are unlikely to have children of low socioeconomic status in the sample. Therefore, the results of the sample may not be readily generalizable to economically disadvantaged populations. Another limitation is that we analyzed the impact of fruit and vegetable intake without taking into account other dietary patterns. The protective effect of vegetables among the males was quite modest, therefore if females consume vegetables with calorie-dense foods, such as cheese or salad dressing, it is possible that the protective effective of vegetables might be overwhelmed by the potentially obesegenic effect of other foods consumed with vegetables. Unfortunately, dietary patterns, such as combinations of foods or composition of salads or vegetable dishes, are extremely complex and are best studied with diet records or diet recalls, rather than with food frequency questionnaires.
In conclusion, our data suggest that the recommendation for consumption of fruits and vegetables may be well founded, but should not be based on a beneficial effect on regulation of BMI. Randomized trials are needed to assess whether it is possible to convince children and adolescents to replace unhealthy snack foods with fruits and vegetables and whether such a change would have a beneficial impact on weight change.
We thank Catherine Berkey and Nan Laird for their comments and suggestions on analytic issues. The analysis was supported by a Special Interest Project grant (U48-CCU115807) from the Centers for Disease Control and Prevention, the Boston Obesity Nutrition Research Center (DK 46200), a research grant (DK-46834) from the National Institutes of Health, and the Kellogg Company.
About this article
World Journal of Pediatrics (2015)