The possible role of sugar-sweetened beverages in obesity etiology: a review of the evidence

Transcription

1 (2006) 30, S28 S36 & 2006 Nature Publishing Group All rights reserved /06 $30.00 REVIEW The possible role of sugar-sweetened beverages in obesity etiology: a review of the evidence School of Public Health, Division of Epidemiology & Community Health, University of Minnesota, Minneapolis, MN, USA Over the second half of the last century, soft drinks and other sugar-sweetened beverages (SSB) have assumed an increasingly significant proportion of total energy intake in the US and most westernized populations. These beverages are heavily marketed to youth and adults. As such, obesity researchers and public health activists and officials have targeted SSB as one of the primary culprits in the escalating rates of obesity. The purpose of this review is to summarize the evidence from human studies on the topic of SSB and body weight regulation, including a brief review of the purported physiological mechanisms, nutritional surveillance and ecological studies, and a more extensive review of the cross-sectional, prospective and experimental studies in humans to date. The equivocal evidence on this topic makes it difficult to draw firm conclusions regarding the role of SSB in the etiology of obesity. Many of the prospective and experimental studies are of unsatisfactory methodological rigor. Some have drawn analogies between the fight against the food industry and the fight against the tobacco industry. However, the complexity of our food supply and of dietary intake behavior, and how diet relates to other behaviors, makes the acquisition of clear and consistent scientific data on the topic of specific dietary factors and obesity risk especially elusive. Only more highquality randomized trials on this topic will provide the necessary data to more completely evaluate the possible link between changes in SSB intake and obesity risk. (2006) 30, S28 S36. doi: /sj.ijo Keywords: sugar; beverage; diet; nutrition; body weight; epidemiology; review Introduction As extensively reported on and discussed in the recent literature, there are a number of lines of evidence and observations to support the hypothesis that sugar-sweetened beverages (SSB), defined as soft drinks, colas, other sweetened carbonated beverages and fruit drinks with added sugar, may play an important etiologic role in obesity risk. 1 4 This topic is of high public health importance because the exposure and the outcome are highly prevalent. Indeed, strong temporal ecological correlations between increasing consumption of SSB and rates of obesity, and possibly diabetes, have been demonstrated. 2 A striking example is the 123% increase in soft drink consumption among children and adolescents between the late 1970s and the late 1990s, with the estimate being highest, 196%, for boys of age In the late 1970s, 12- to 17-year-old boys and girls in the US consumed 2.1 and 1.4 times more milk than carbonated soft drinks, respectively, whereas by the mid- Correspondence: Dr, School of Public Health, Division of Epidemiology & Community Health, University of Minnesota, Minneapolis, MN 55454, USA. pereira@epi.umn.edu 1990s soft drink intake was nearly twice as great as milk in both genders. 3,6 An essential criterion of any causal association is that the biological mechanism(s) are known. As recently described in detail by Bachman et al., 7 there are many possible mechanisms whereby excessive SSB intake may increase the risk of obesity; yet none of the purported mechanisms have sufficient rigorous scientific evidence in their corner. Most of these mechanisms have limited research to date in support of a direct and causal link to obesity, and others have a research base with a great deal of inconsistency in findings, such as the popular hypothesis that energy consumed in liquid form may be less satiating than energy intake from solid foods, leading to passive over-consumption, with the escalating portion sizes of many SSB exacerbating this potential problem. It is possible that when SSB are increased in the diet, overall energy intake increases due to a lack of any downregulation of other sources of energy. However, the evidence on this topic remains inconclusive. 7,8 There is also accumulating evidence that high-fructose corn syrup, used to sweeten most SSB, has unique metabolic effects with respect to obesity and diabetes risks, 1,2 but any direct evidence for a unique causal role in body weight regulation is lacking to date.

2 It is possible that high intakes of SSB may increase risk of obesity by contributing to the dietary glycemic load. However, due to the lack of definitive long-term controlled studies, the possible causal and important role of glycemic index and glycemic load in the etiology of obesity remains a topic of intense scientific debate. 7,9,10 SSB are increasingly served in very large portions, which may promote passive over-consumption and chronic energy imbalance, but the literature on the independent role of portion size in promoting weight gain is also incomplete. 11 Whereas energy density has consistent scientific support for its direct causal role in energy intake, 11,12 despite being an ample source of energy, due to their high water content and volume SSB are not high-energy density. As there appear to be a number of plausible mechanisms under study, and because their extensive review is beyond the scope of this paper, we will turn our focus to the primary goal of this review, describing the evidence on SSB and body weight change or obesity risk from observational studies and randomized trials in humans. The overarching theoretical model that puts this review into a working context is shown in Figure 1. The goal was to identify all human studies with SSB as an exposure (independent) variable and body weight or obesity as an outcome (dependent) variable. Cross-sectional studies Not unlike the ecological studies, the cross-sectional studies on dietary habits and body weight are of limited value due to the many assumptions and possible biases, especially confounding, residual confounding, the inability to evaluate temporality in the association between exposure and outcome, and the lack of within person comparisons. Therefore, only a brief summary of these studies will be provided. Harnack et al., 3 using a United States Department of Agriculture (USDA) survey of youth in the years , found children consuming X26 ounces of soft drinks to have estimated total daily energy intakes of 2605 kcal, compared to 1984 kcal for those who reported no soft drink intake. Based on analyses of the Nutrition and Health Examination (NHANES) survey data, in which diet was assessed by a 24-h recall, Troiano et al. 13 found SSB to contribute more to total energy intake than any other beverage type, with an estimate of 10 11% total energy intake coming from these drinks in the 12- to 19-year-old group. Soft drinks, in particular, contributed more energy to the diets of the overweight youth than to the diets of normal weight youth. In further analyses of NHANES, Kant et al. 14 evaluated the association of low-nutrient density food intake with body mass index (BMI) in 4852 youth aged 8 18 years. Foods were grouped into five low-nutrient density categories: (1) visible fat, (2) sweets including candy and beverages, (3) desserts, (4) salty snacks and (5) miscellaneous. The intake of low-nutrient density foods, for which sweeteners and desserts accounted for B25% of energy, were not associated with BMI. Liebman et al. 15 reported analyses of self-reported height, weight and lifestyle data from 928 male subjects and 889 female subjects, aged years, from rural Wyoming, S29 Figure 1 Theoretical framework for SSB in obesity etiology.

3 S30 Montana and Idaho. After adjustment for other dietary factors, physical activity and sociodemographics, the probability of being overweight was directly associated with habitual intake of SSB. In a cross-sectional analysis of the Bogalusa Heart Study, 16 an elevated odds of overweight per each serving of SSB was observed in Caucasian male (1.68, 95% CI ¼ ) and female populations (1.53, 95% CI ¼ ), but not in African-American male (1.02, 95% CI ¼ ) and female populations (0.92, 95% CI ¼ ). A limitation of this study was lack of adjustment for any factors beyond age and energy intake. In a separate secular trend analysis of 10-year olds in this Bogalusa population, total energy intake was significantly higher in the highest SSB consumption group compared with the lower consumption groups. 17 However, there was no statistically significant association between BMI and SSB consumption in this study, possibly due to low statistical power and measurement error. A final cross-sectional survey of 385 school children in Santa Barbara County, CA assessed BMI and body fat directly, and diet and lifestyle by questionnaire. 18 The odds of being overweight were 46% higher (95% CI ¼ 2 110%) among those students (n ¼ 49) who reported consuming X3 SSB servings/day compared to those consuming lower amounts, after adjustment for age, gender, ethnicity and television viewing, but many other lifestyle and dietary factors were not considered in this study. In summary, the cross-sectional literature includes limited studies, mostly from national dietary surveys. The findings suggest that higher SSB intake is associated with higher energy diets and possibly higher obesity risk. Prospective studies The prospective study moves us further along the line of increasing quality in study design and related evidence, as it is possible to make repeated assessments in both the exposure and the outcome, thus allowing the necessary evaluation of temporality of the exposure outcome relation. Surprisingly, it appears that the first prospective study on this topic was not published until 2001, when Ludwig et al. 19 reported their analysis of an ethnically diverse cohort study of 548 children, aged years, enrolled in the control arm of a school-based intervention project and followed for 19 months. Weight and height were directly measured, and diet and physical activity were assessed with validated questionnaires. After adjustment for baseline anthropometrics, demographics, physical activity, television viewing, and changes in energy intake and intake of energy from dietary fat and fruit juice, change over time in SSB intake predicted changes in BMI ( þ 0.24 kg/m 2, 95% CI ¼ , per each daily serving) as well as the odds of becoming obese (Xage- and gender-specific 85th %ile of referent population for BMI and triceps skinfold thickness) among the 398 children who were non-obese at baseline (odds ratio ¼ 1.60, 95% CI ¼ ). By comparison, changes in the intake of diet/non-ssb were inversely associated with the odds of becoming obese (OR ¼ 0.44, P ¼ 0.03). An important limitation of this study is the lack of adjustment for a variety of dietary components that may confound the associations of SSB intake and weight gain, including milk, fruits, vegetables, whole grains and fiber. The authors suggested that their findings may be explained by incomplete compensation for energy consumed in liquid form. However, adjustment for energy intake did not attenuate the associations, perhaps due to the poor precision of energy intake measurement. An ethnically diverse sample of 196 girls between the ages of 8 and 12 years were followed for an average of 7 years with objective measures of height, weight and body fat, with diet assessed by a validated food frequency questionnaire. 20 Intake of SSB was the only energy-dense snack food directly associated with changes in BMI z-score in this study, although there was no association when the dependent variable was body fat percentage. After taking into account a number of possible confounding demographic, lifestyle and dietary factors, the mean change in BMI z-score, relative to those in the lowest quartile of SSB intake (o0.74% daily energy), was 0.09 ( % energy), 0.17 ( % energy) and 0.18 (X3.2% energy) (P for trend o0.001). Berkey et al. 21 followed boys and girls, aged 9 14 in 1996, in the US Growing Up Today Study. Heights and weights were self-reported over time, and diet was assessed repeatedly over 2 years with a validated food frequency questionnaire. In both boys and girls, increased intake by 2 or more servings/day was associated with modest weight gain (boys: þ 0.14 kg/m 2, P ¼ 0.01; girls: þ 0.10 kg/m 2, P ¼ 0.046), with adjustment for age, height, Tanner stage, menarche, milk intake, and physical activity and inactivity. Unlike the study by Ludwig et al., 19 these associations were completely explained by adjustment for energy intake. Blum et al. 22 reported changes in beverage consumption in association with BMI z-score change over 2 years in a convenience sample of 164 elementary school children. A comprehensive beverage analysis, based on a single 24-h recall, included milk, 100% juice, diet sodas and SSB. In agreement with other studies, change in milk consumption was inversely correlated with SSB consumption. Compared to normal weight subjects, increases in diet soda consumption were significantly greater for overweight subjects and those who gained weight. Diet soda consumption was the only type of beverage associated with year 2 BMI z-score, as SSB intake and other dietary factors were not correlated with BMI in this prospective study. Newby et al. 23 examined beverage consumption in association with weight gain in a cohort study of 1345 preschool (aged 2 5 years) children from low-income families receiving special supplemental nutrition assistance. Height and weight were directly assessed, and diet was assessed by a food

4 frequency questionnaire administered to the mothers. No associations were observed for intake of any beverage types, including SSB, with changes in body weight (or BMI), with adjustment for height, age, gender, energy intake and sociodemographic factors. The authors noted limitations in this study, including lack of economic assistance from this program for SSB, thus resulting in relatively low and homogeneous intakes across children. Welsh et al. 24 reported on a retrospective cohort study of children of ages 2 and 3 years at baseline. Total intake of SSB (including juices, fruit drinks and sodas) was directly associated with development of overweight, although the association was not statistically significant for those children who were underweight or normal weight at baseline. For those who were at risk of overweight at baseline (BMI 85th o95th %ile), ORs, adjusted for demographics, birth weight, sweet and high-fat food intake, and energy intake, with o1 serving/day as the referent, were 2.0 (95% CI ¼ ) for 1 o2 servings/day, 2.0 (95% CI ¼ ) for 2 o3 servings/day and 1.8 (95% CI ¼ ) for X3 servings/day. Thus, no linear association was observed for overweight risk with increasing intake. Data from the National Heart Lung and Blood Institute Growth and Health Study have revealed conflicting findings for the association between SSB and weight gain among Caucasian and African-American girls who were followed from a mean age of 9.5 years to a mean age of 18.6 years. 25,26 In their report of a strong and independent inverse association between physical activity changes and weight gain, Kim et al. 23 found that energy intake, assessed by three 24-h recalls at each time point, was not a predictor of weight gain. However, Striegal-Moore et al. 24 recently reported a direct association between SSB intake and weight gain in this same cohort, but they did not adjust for physical activity or other lifestyle and dietary factors. It would appear, based on the earlier paper by Kimm et al., 25 that physical activity may be an important confounder of the SSB weight gain association in this study. From their analyses of the Nurses Health Study II, Schultz et al. 27 reported on what appears to be the first prospective study of SSB and weight gain, as well as risk of diabetes, in adults. A total of women in the diabetes analysis and in the weight change analysis were followed from 1991 to 1999, with weight, heights and diabetes being selfreported, and diet measured by a validated food frequency questionnaire over time. Most of the women in the study had consistently low intakes of SSB, with women reporting p1 serving/week over the first time interval, Weight gain estimates per category of SSB intake were thoroughly adjusted for many possible confounding factors, including demographics, baseline BMI, physical activity and other lifestyle factors and many dietary factors including energy intake from other sources, fiber, fat and a long list of food groups. The mean weight gain of these low-intake women ( kg) over this first time interval was no different than that of the 2366 women who consistently reported X7 servings/week ( kg). However, in 1007 women who increased their intake of SSB from p1 to X7 serving/week, the mean weight gain was slightly higher ( kg). Weight gain was lowest in the 1020 women who decreased their intake from X7 to p1 serving/week ( kg). Note that 8473 women did not fall into any of these categories, and had similar weight gains ( kg) as the women reporting consistent intakes. Analyses for BMI change, and for the second time interval of revealed very similar patterns, as did analyses of fruit punch and 100% fruit juices. By contrast, changes in diet beverages had inverse associations with body weight change. In the diabetes analysis, risk was increased among women reporting X7 servings/week of all SSB (relative risk of diabetes ¼ 1.8, 95% CI ¼ ) or fruit punch (2.0, ), compared to women reporting o1 serving/month of these beverages, with intermediate beverage categories at intermediate risk (P-values for trend across categories p0.001). However, 100% fruit juice intake was not associated with diabetes risk (0.97, , P for trend ¼ 0.84), despite its strong and direct relation to weight gain. In an accompanying editorial to the above paper in the Journal of the American Medical Association, Apovian 28 suggests that 1 serving of SSB per day could lead to 6.8 kg of weight gain in 1 year, implying no compensation by energy intake or expenditure. This hypothetical estimation flies in the face of the results of the very study the editorial was written about, as well as a number of controlled studies demonstrating that most individuals at least partially compensate for increased or decreased intake of liquid and solid sources of energy through a number of subtle behavioral and physiological adjustments Indeed, there was no difference in weight gain between women consistently reporting p1 serving of SSB/week and those reporting X7 servings/week. 27 Furthermore, the weight change data between those women who increased intake from p1 to X7 servings/week and those who decreased by the same amount (X7 to p1 per week) was estimated to be 0.7 kg/year, suggesting the average women in the study experienced 90% compensation for the increase of B1 serving/day of SSB. Granted, such estimations based on self-reported data need to be cast in the light of the significant measurement error in assessing diet, including possible systematic biases by weight status. Bes-Rastrollo et al. 33 examined dietary predictors of weight gain in a Mediterranean cohort study of 7194 middle-aged men and women followed for a median of 28.5 months. Diet was assessed with a food frequency questionnaire, and height and weight by self-report. A direct association between SSB intake at baseline and weight gain over the course of the study was only noted in those participants who had a recent weight gain history of X3 kg over the past 5 years, 32% of the cohort. After extensive adjustment for demographic factors, baseline weight, physical activity, television viewing, other lifestyle factors and many dietary factors including other beverages, the OR for a weight gain of S31

5 S32 at least 1 kg during following up was 1.55 (95% CI ¼ ) for those in the highest quintile of baseline SSB intake (X87.5 ml/day) compared to those in the lowest quintile (o5.5 ml/day, P-value for trend across quintiles ¼ 0.02). Fruit juice intake also had a direct, albeit weaker, association with weight gain (OR for highest quintile compared to lowest ¼ 1.16, 95% CI ¼ , P for trend ¼ 0.04). It is unclear why these authors chose to model any weight gain as a dichotomous outcome rather than incidence of obesity or incidence of large weight gains. A final prospective study from the Baltimore Longitudinal Study of Aging 34 used data from 7-day dietary records to identify five dietary patterns using cluster analysis healthy, white bread, alcohol, sweets and meat and potatoes in 459 men and women of mean age 60 years. Mean intake of SSB was highest in those individuals characterized by the sweets (1.6 servings/day) and meat and potatoes (1.8 servings/day) patterns, compared to a range of servings/day across the other three patterns. Multiple regression analysis to adjust for a number of important covariates revealed small changes in BMI and waist circumference for the sweets pattern that were not different than those of the healthy pattern (BMI difference ¼ kg/m 2, waist difference ¼ cm). By comparison, those characterized by the meat and potatoes pattern had relatively large increases in both BMI (difference ¼ , Po0.05) and waist circumference (difference ¼ , Po0.10) when compared to be healthy pattern, and those characterized by the white bread pattern had large increases in waist circumference (difference ¼ , Po0.05) when compared to the healthy pattern. The authors commented that low SSB intake was an important aspect of the healthy eating pattern likely responsible for weight gain avoidance, a statement in direct contrast to their results. Those individuals following the sweets pattern high in SSB had very similar changes in body weight and waist as those in the healthy pattern. Finally, results of a risk analysis model for the hypothetical effect of removing school vending machines that include SSB on changes in BMI was published recently. 35 Data from nutrition surveillance systems were used to estimate baseline SSB intake from school vending machines, and data from the prospective study by Ludwig et al. 19 were used to estimate the association between SSB change and BMI change. The overall conclusion from these authors was that removal of the vending machines is not a worthwhile endeavor because it would have no impact on changes in BMI. This appeared to be a forgone conclusion primarily because the estimated baseline intake of SSB coming from vending machines is relatively low compared to other sources. 5 However, these authors did not address a number of important possibilities, most notably the potential for vending removal to have an important impact on subpopulations of high BMI and/or high habitual intake of SSB. Finally, what of the possible impact of messages children are getting by the presence of vending machines stocked with nutrient-poor foods in the schools, possibly affecting dietary attitudes and behaviors outside of school? In summary, the prospective studies of SSB in youth and adults provide some support for the hypothesis that high or increasing intakes of these drinks over time may increase the risk for weight gain and obesity. The studies are not without limitations, especially lack of adequate control for important confounding factors in several studies, and questionable internal validity due to discrepancies of within-study analyses. In some cases, the findings are misinterpreted by the authors themselves or by others. Fortunately, randomized trials, of which the few to date are discussed next, are logistically feasible on this topic, and should ultimately answer these questions. Randomized controlled trials Randomized controlled trials are looked upon as the definitive test of causality, as they utilize the true experimental design and are the least susceptible to bias. Five published randomized trials of SSB intake and body weight change were found in the literature, and these are summarized in Table 1 and described in the following text. In a small within-person masked cross over study, Tordoff and Alleva 36 provided nine normal weight women and 21 men 1135 g of soda beverage sweetened with artificial sweetener (aspartame, 3 kcal) for one 3-week period and high-fructose corn syrup (530 kcal/day) in the other 3-week period, separated by a 3-week washout period. Interestingly, even though the aspartame treatment supplied essentially no energy, the same degree of dietary compensation occurred for both treatments, due primarily to a decrease in sugar elsewhere in the diet of nearly 200 kcal/day. As such, slight weight loss over the 3 weeks was observed during the aspartame treatment, whereas weight gain was observed during the high-fructose corn syrup treatment. The mean weight gain for women during the high-fructose corn syrup treatment was kg (Po0.01) and for men was kg (NS), suggesting a partial compensation for the increased SSB intake and that this compensation was greater for men than for women. Using a similar design as the study just discussed, DiMeglio and Mattes 37 enrolled 15 men and women into 4-week interventions separated by 4-week washout in order to compare the effects of SSB to isoenergetic (450 kcal) portions of solid jelly-bean candies, thereby testing the hypothesis that the degree of dietary compensation for the excess energy would be less for the liquid vs solid food form. Indeed, according to analysis of diet records, compensation appeared to be complete (mean ¼ 118%, 95% CI ¼ ) during the jelly-bean treatment, whereas there was essentially no compensation during the SSB treatment (mean- 17%, 95% CI ¼ ). No significant difference in

6 S33 Table 1 Randomized trials of SSB and body weight 1st author Sample Design Duration Intervention Results Comments Dietary intake Body weight Tordoff and NW adults 9 ~, Alleva # Two treatment crossover with washout 3 weeks SSB (530 kal/day) vs artificial sweetener beverages Energy intake 4 for SSB 4For SSB Artificial sweetener caused some reduction in overall energy intake and weight loss Partial dietary compensation during SSB resulting in less weight loss than predicted DiMeglio and Mattes 37 NW adults 7 ~, 9 # Two treatment crossover with washout 4 weeks Jelly beans vs SSB, isocaloric (450 kcal) Energy intake 4 for SSB No differences between groups Discrepant findings between energy intake and body weight Raben et al. 31 OW adults 35 ~, 6# Two-group parallel 10 weeks Sucrose (150 g/day, 600 kcal) mostly as SSB vs artificial sweetener Energy intake 4 for sucrose 4For sucrose Weight gain during SSB, 1.6 kg, was much less than predicted by diet records James et al ~ and # school children Two-group parallel grouprandomized 1 year Four 1-h education sessions over 1 year, or control Carbonated beverages o for intervention (0.6 servings/3 days), no difference in SSB %OWo for intervention, Mean BMI change null Low overall intakes of SSB No information on total dietary intakes Discrepant findings for % OW vs mean BMI Ebbeling et al. 40 OW and NW youth, 56~, 47# with high SSB intake Two-group parallel 6 months Diet drinks delivered to homes, or control condition 82% reduction in SSB intake for intervention, no change in control (Po0.0001) Significant intervention effect on weight loss for those in the highest third of BMI It is biologically plausible that the intervention would be most efficacious in those with highest baseline body mass index, perhaps due to gene environment interactions. Most studies have not recognized the importance of this concept. Abbreviations: NW¼ normal weight, OW ¼ overweight, SSB¼ sugar-sweetened beverages. hunger ratings was noted. Whereas one would predict at least moderate differences in body weight or body fat over time based on this important difference in energy intake, the changes in body weight and body fat between the two conditions were negligible. Based on 0% compensation for the 450 kcal/day of SSB, one would predict a weight gain of at least 1 kg, assuming some compensatory increase in energy expenditure. However, the mean body weight data presented for baseline and post-treatment suggests a 0.2 kg difference in weight change between the two treatments, which was not statistically significant. Body composition findings were similar. Compelling as the findings are for energy intake compensation, enthusiasm is tempered by the small size of the study, the limited body weight ranges and the null findings for body weight and fat. Larger and longer studies are needed to fully test the hypothesis that differential compensation for energy consumed in liquid vs solid form will be persistent and important for body weight over the longer term. Raben et al. 31 conducted a parallel randomized trial to compare daily food and beverage supplements high in either sucrose (152 g/day), of which 70% was provided in the form of SSB, or artificial sweeteners in 41 overweight men (n ¼ 6) and women (n ¼ 35). Over the 10 weeks, changes in energy intake, sugar intake, and body weight and body fat were higher for the sucrose group compared to the artificial sweetener group. Dietary compensation was estimated to be approximately 56% in the sucrose group, in contrast to the complete lack of compensation when SSB was increased in the above study by DiMeglio and Mattes, 37 and interestingly, there was a mean reduction in energy intake from total carbohydrate in the artificial sweetener group of 18%. Mean weight change for the sucrose groups was þ kg, compared to kg for the artificial sweetener group

7 S34 (Po for the diet time interaction). Changes in fat mass were smaller, but the pattern was similar. These authors noted that weight gain in the sucrose group was approximately half of what was predicted based on actual intake, suggesting increases in metabolic rate over time. James et al. 38 conducted a school-based randomized group trial to examine the effect of a brief classroom-based intervention focused on reducing SSB intake on body weight changes over 1 year. The study involved six primary schools in southwest England, including 644 students between the ages of 7 and 11 years. The intervention was administered in only four one-hour sessions (one per term of the academic year) that included education about avoiding high-sugar drinks and the importance of a healthy balanced diet. Weight and height were directly measured every 6 months and 3-day records for beverage intake were completed at baseline and follow-up. The main findings of the study demonstrated a mean decrease in carbonated drinks (both SSB and diet) of 0.6 servings per 3-day period in the intervention group, compared to a mean increase of 0.2 in the control classrooms (Po0.05 between groups). The percentage of overweight and obese children in the intervention group decreased over 12 months by 0.2%, compared to a 7.5% increase in the control classrooms (mean difference ¼ 7.7%, 95% CI ¼ %). However, mean changes in BMI or z-scores were similar between the two groups (difference in BMI change ¼ 0.1 kg/m 2, 95% CI ¼ ). A letter to the editor of the British Medical Journal by French et al. 39 pointed out a variety of important methodological problems with this study, including the low response rate (B50%) for completing the beverage diaries, lack of information on changes in other beverages, lack of mediation analyses specifically examining whether changes in SSB intake predicted changes in obesity status and failure to explore why the control classrooms had such large increase in obesity rates. The most recently published randomized trial on this topic was conducted from the laboratory of Dr David Ludwig, Children s Hospital, Boston, MA, USA. 40 In this 6-month two-arm parallel controlled trial, Ebbeling et al. 40 randomized an ethnically and anthropometrically diverse group of 113 adolescent boys and girls in order to compare effects of replacing SSB in the habitual diet with water and other lowenergy beverages. An important feature of this study was the inclusion criteria of consuming at least 1 SSB/day in the habitual diet. The beverages were delivered weekly to the homes of those randomly assigned to the intervention arm, along with some educational guidelines and follow-up monitoring around the possible benefits of decreasing SSB intake and behavioral strategies for doing so. By way of example, each time the intervention youth went to the refrigerator, they were faced with a magnet on the door that read Think Before You Drink. Two unannounced 24-h dietary and activity recalls were administered at baseline and follow-up to assess total energy intake and intake of specific food groups and nutrients. Heights and weights were directly assessed at baseline and follow-up. Quality control and adherence were excellent, with an 82% reduction of SSB intake in the intervention arm vs no change in the control arm (Po0.0001), resulting in a mean total intake difference between groups of approximately 250 kcal/day. Whereas this difference in energy intake would equate to a difference in weight change between groups of approximately 5 kg, the net change between groups was negligible and not statistically significant ( kg/m 2 ), suggesting the dietary intake data were invalid and/or there were declines in resting energy expenditure rendering the intervention meaningless for overall energy balance. However, there was a significant interaction between SSB change and baseline BMI (P ¼ 0.02), such that a stronger and statistically significant intervention effect was observed only among those youth in the top tertile (425.6 kg/m 2 ) of baseline BMI, although still considerably less than would have been expected based on the dietary intake records (intervention effect ¼ kg/m 2, P ¼ 0.03). Based on the few randomized trials of SSB and body weight, there appears to be growing evidence to date for the hypothesis that manipulating intake of SSB may cause important changes in habitual energy intake that could lead to increase obesity risk over an extended period of time. Of the three trials in adults, two were of short duration and small sample size, and observed little or no evidence of effects on body weight. The trial by Raben and co-workers demonstrates a potentially important effect of a very high sucrose diet, primarily through SSB consumption, on changes in body weight over 10 weeks. Two studies in youth provide further support of the SSB and obesity hypothesis, but the school-based trial by James et al. 38 has a number of important methodological shortcomings already discussed. The final study by Ebbeling et al., 40 which was of high methodological quality, reported a potentially important interaction between baseline BMI and decreased SSB intake on weight loss over time, pointing to a number of possibilities that have not been addressed by other studies in this area, including potential gene environment interactions. The subgroup characterized by the phenotypes of high BMI and high habitual SSB intake likely has a significant prevalence in the population. Summary There are many plausible mechanisms that may support a causal link between SSB intake and obesity risk, but the definitive pathways that may set SSB apart from its myriad solid-food counterparts in this regard remain to be determined. The surveillance studies demonstrate clear temporal associations between per capita intake of SSB and increasing rates of obesity. The applied human studies on this topic, reviewed in this paper, provide some support for this hypothesis, but a variety of inconsistencies and methodolo-

8 gical difficulties preclude definitive conclusions at this time. The 2003 report of the World Health Organization and the Food and Agricultural Organization of the United Nations, Diet, Nutrition and the Prevention of Chronic Diseases 41 classified the scientific evidence on the question of SSB and increased obesity risk as probable, whereas the evidence on high intakes of energy-dense micronutrient-poor food as a promoter of obesity risk was classified as convincing. Based on the contributions of the newer literature on SSB and obesity, we are still a ways from seeing the probable change to convincing. Only additional high-quality randomized controlled trials, building upon what has been done, will provide the necessary data to more fully evaluate this question. Recommendations and policy decisions to limit the intake of nutrient-poor foods or beverages served up at low cost in excessive portions, such as those by the Academy of Pediatrics 42 and the American Heart Association, 43 would appear to be important for the promotion of improved nutrition habits. However, the literature reviewed in this paper suggests that the potential impact of recommendations to reduce intake of SSB on obesity rates is currently difficult to predict. References 1 Bray GA, Nielsen SJ, Popkin BM. Consumption of high-fructose corn syrup in beverages may play a role in the epidemic of obesity. Am J Clin Nutr 2004; 79: Gross LS, Li L, Ford ES, Liu S. Increased consumption of refined carbohydrates and the epidemic of type 2 diabetes in the United States: an ecologic assessment. Am J Clin Nutr 2004; 79: Harnack L, Stang J, Story M. Soft drink consumption among US children and adolescents: nutritional consequences. J Am Diet Assoc 1999; 99: Popkin BM, Nielsen SJ. The sweetening of the world s diet. Obes Res 2003; 11: French SA, Lin BH, Guthrie JF. National trends in soft drink consumption among children and adolescents age 6 to 17 years: prevalence, amounts, and sources, 1977/1978 to 1994/1998. JAm Diet Assoc 2003; 103: Yen ST, Lin B-H. Beverage consumption among US children and adolescents: full-information and quasi maximum-likelihood estimation of a censored system. Eur Rev Agric Econ 2002; 29: Bachman CM, Baranowski T, Nicklas TA. Is there an association between sweetened beverages and adiposity? Nutr Rev 2006; 64: Almiron-Roig E, Chen Y, Drewnowski A. Liquid calories and the failure of satiety: how good is the evidence? Obes Rev 2003; 4: Pawlak DB, Ebbeling CB, Ludwig DS. Should obese patients be counselled to follow a low-glycaemic index diet? Yes. Obes Rev 2002; 3: Raben A. Should obese patients be counselled to follow a lowglycaemic index diet? No. Obes Rev 2002; 3: Rolls BJ. The supersizing of America: portion size and the obesity epidemic. Nutr Today 2003; 38: Rolls BJ, Bell EA. Intake of fat and carbohydrate: role of energy density. Eur J Clin Nutr 1999; 53 (Suppl 1): S166 S Troiano RP, Briefel RR, Carroll MD, Bialostosky K. Energy and fat intakes of children and adolescents in the United States: data from the national health and nutrition examination surveys. Am J Clin Nutr 2000; 72 (5 Suppl): 1343S 1353S. 14 Kant AK, Graubard BI. Predictors of reported consumption of low-nutrient-density foods in a 24-h recall by 8 16 year old US children and adolescents. Appetite 2003; 41: Liebman M, Pelican S, Moore SA, Holmes B, Wardlaw MK, Melcher LM et al. Dietary intake, eating behavior, and physical activity-related determinants of high body mass index in rural communities in Wyoming, Montana, and Idaho. Int J Obes Relat Metab Disord 2003; 27: Nicklas TA, Yang SJ, Baranowski T, Zakeri I, Berenson G. Eating patterns and obesity in children. The Bogalusa Heart Study. Am J Prev Med 2003; 25: Rajeshwari R, Yang SJ, Nicklas TA, Berenson GS. Secular trends in children s sweetened-beverage consumption (1973 to 1994): the Bogalusa Heart Study. J Am Diet Assoc 2005; 105: Giammattei J, Blix G, Marshak HH, Wollitzer AO, Pettitt DJ. Television watching and soft drink consumption: associations with obesity in 11- to 13-year-old schoolchildren. Arch Pediatr Adolesc Med 2003; 157: Ludwig DS, Peterson KE, Gortmaker SL. Relation between consumption of sugar-sweetened drinks and childhood obesity: a prospective, observational analysis. Lancet 2001; 357: Phillips SM, Bandini LG, Naumova EN, Cyr H, Colclough S, Dietz WH et al. Energy-dense snack food intake in adolescence: longitudinal relationship to weight and fatness. Obes Res 2004; 12: Berkey CS, Rockett HR, Field AE, Gillman MW, Colditz GA. Sugaradded beverages and adolescent weight change. Obes Res 2004; 12: Blum JW, Jacobsen DJ, Donnelly JE. Beverage consumption patterns in elementary school aged children across a two-year period. J Am Coll Nutr 2005; 24: Newby PK, Peterson KE, Berkey CS, Leppert J, Willett WC, Colditz GA. Beverage consumption is not associated with changes in weight and body mass index among low-income preschool children in North Dakota. J Am Diet Assoc 2004; 104: Welsh JA, Cogswell ME, Rogers S, Rockett H, Mei Z, Grummer- Strawn LM. Overweight among low-income preschool children associated with the consumption of sweet drinks: Missouri, Pediatrics 2005; 115: e223 e Kimm SY, Glynn NW, Obarzanek E, Kriska AM, Daniels SR, Barton BA et al. Relation between the changes in physical activity and body-mass index during adolescence: a multicentre longitudinal study. Lancet 2005; 366: Striegel-Moore RH, Thompson D, Affenito SG, Franko DL, Obarzanek E, Barton BA et al. Correlates of beverage intake in adolescent girls: the National Heart, Lung, and Blood Institute Growth and Health Study. J Pediatr 2006; 148: Schulze MB, Manson JE, Ludwig DS, Colditz GA, Stampfer MJ, Willett WC et al. Sugar-sweetened beverages, weight gain, and incidence of type 2 diabetes in young and middle-aged women. JAMA 2004; 292: Apovian CM. Sugar-sweetened soft drinks, obesity, and type 2 diabetes. JAMA 2004; 292: Leibel RL, Rosenbaum M, Hirsch J. Changes in energy expenditure resulting from altered body weight. N Engl J Med 1995; 332: Pereira MA, Swain J, Goldfine AB, Rifai N, Ludwig DS. Effects of a low-glycemic load diet on resting energy expenditure and heart disease risk factors during weight loss. JAMA 2004; 292: Raben A, Vasilaras TH, Moller AC, Astrup A. Sucrose compared with artificial sweeteners: different effects on ad libitum food intake and body weight after 10 wk of supplementation in overweight subjects. Am J Clin Nutr 2002; 76: Agus MS, Swain JF, Larson CL, Eckert EA, Ludwig DS. Dietary composition and physiologic adaptations to energy restriction. Am J Clin Nutr 2000; 71: S35

9 S36 33 Bes-Rastrollo M, Sanchez-Villegas A, Gomez-Gracia E, Martinez JA, Pajares RM, Martinez-Gonzalez MA. Predictors of weight gain in a Mediterranean cohort: the Seguimiento Universidad de Navarra Study 1. Am J Clin Nutr. 2006; 83: ; quiz Newby PK, Muller D, Hallfrisch J, Qiao N, Andres R, Tucker KL. Dietary patterns and changes in body mass index and waist circumference in adults. Am J Clin Nutr 2003; 77: Forshee RA, Storey ML, Ginevan ME. A risk analysis model of the relationship between beverage consumption from school vending machines and risk of adolescent overweight. Risk Anal 2005; 25: Tordoff MG, Alleva AM. Effect of drinking soda sweetened with aspartame or high-fructose corn syrup on food intake and body weight. Am J Clin Nutr 1990; 51: DiMeglio DP, Mattes RD. Liquid versus solid carbohydrate: effects on food intake and body weight. Int J Obes Relat Metab Disord 2000; 24: James J, Thomas P, Cavan D, Kerr D. Preventing childhood obesity by reducing consumption of carbonated drinks: cluster randomised controlled trial. BMJ 2004; 328: French SA, Hannan PJ, Story M. School soft drink intervention study. BMJ. 2004; 329: E315 E Ebbeling CB, Feldman HA, Osganian SK, Chomitz VR, Ellenbogen SJ, Ludwig DS. Effects of decreasing sugar-sweetened beverage consumption on body weight in adolescents: a randomized, controlled pilot study. Pediatrics. Mar 2006; 117: World Health Organization. Diet, Nutrition, and the Prevention of Chronic Disease. Geneva: World Health Organization and the Food and Agriculture Organization of the United Nations, Gidding SS, Dennison BA, Birch LL, Daniels SR, Gillman MW, Lichtenstein AH et al. Dietary recommendations for children and adolescents: a guide for practitioners. Pediatrics 2006; 117: Gidding SS, Dennison BA, Birch LL, Daniels SR, Gilman MW, Lichtenstein AH et al. Dietary recommendations for children and adolescents: a guide for practitioners: consensus statement from the American Heart Association. Circulation 2005; 112: