A “disappointing” lack of good research
Gibson lists 11 areas of methodological weakness in the research, each of which could bias the conclusion toward positive or negative results. She notes that the “disappointing outcomes” in many studies may have resulted from the failure to assess what else study participants ate and drank, or substituted for soda. The reliance on self-reporting weight, height, activity, and soda consumption in some studies rather than independent measuring and verification may confound their reliability, as might small sample sizes; researchers do not know if these weaknesses skew the data towards, or away, from finding effects.
The Dietary Guideline Advisory Committee to the US Department of Agriculture, which included many of the leading researchers on nutrition and obesity, noted in its discussion (pp384-406) on the 2010 dietary guidelines that the weight of evidence for sugar sweetened beverages contributing to weight gain was marred by a “disappointing” lack of good research.
So what does the best available evidence tell us?
Randomized control trials, a form of intervention study, writes Gibson, “constitute the highest level of evidence for a hypothesis” because they take a group of people and randomly assign them to one of two groups, intervention or control. In one study, (Ebbeling et al. 2006, ) 103 adolescents aged 13 to 18 who regularly consumed soda were randomly assigned to either a group that received non-energy beverages for six months by way of home delivery or a control group that carried on drinking soda as normal.
The intervention group reduced their sugar sweetened beverage consumption by 82 percent. While both the intervention and control groups increased BMI [body mass index, a measure of body fat based on height and weight], the intervention group gained slightly less,
0.14kg/m2 to be exact, but the result was not statistically significant. However, when the intervention group was broken down by BMI at the start of the intervention, saw the most benefit from reduced consumption of SSBs their BMI actually went down, compared to the control group's BMI, which went up. The difference was 0.75kg/m^2, and this change in BMI was statistically significant (for an average height kid, this works out to be about a 4.6 pounds difference in weight.
Thus the strongest form of evidence only provides, “limited evidence that avoidance of sugar-containing soft drinks of substitution with other lower-energy beverages may help prevent weight gain in overweigh individuals,” Gibson writes. “By contrast, little effect was observed in normal-weight individuals.” (Ebbeling et al. is also one of the studies found to have been widely cited in a misleadingly positive manner by Allison and Cope, which is to say other researchers neglected to note the non-significant results.)
A similar study of 1140 Brazilian children (Sichieri et al., 2009) found a smaller reduction in consumption which resulted in no difference between the weight gain between the intervention and control groups.Other intervention studies offer limited corroboration given the shortness of duration (10 weeks), or were marred by significant self-reporting failures.
There are other caveats which needed to be noted about studies involving children. Bremer et al. note (Sugar-Sweetened Beverage Intake Trends in US Adolescents and Their Association with Insulin Resistance-Related Parameters, J Nutr Metab. 2010), that teens who are low and medium soda drinkers consume considerably less sugar sweetened beverages than high consumers (an overall mean consumption between .05 and 2.5 servings respectively, compared to an overall mean of 7.1 servings, or 57 oz per day).
Moreover, Bremer et al. note in their disaggregation of dietary reporting data from the National Health and Nutrition Examination Survey that it is the high consumers who have been consuming more over time, and not the low to medium consumers. These high consumers are also more likely to be male, and assuming that their self reporting was accurate, the highest adolescent male consumers of SSBs were consuming 24.2 servings per day, the equivalent of 16 cans of soda, (193.6 ounces or 2,420 calories).
Other research supports the importance of disaggregating soda consumption in order to understand who, exactly, is gaining weight. For example, one cross sectional study of Mexican-American infants (Warner et al.2006), found that those infants who drank soda everyday were more likely to be overweight at two years of age than those infants who drank no soda, although this result may be confounded by failing to control for fruit juice consumption.
A longitudinal study of low-income African American preschoolers (Lim et al., 2009) also found an association between fruit juice and soda consumption and weight gain over two years. The weight of the child’s caregiver was also a significant predictor of childhood obesity, leading the researchers to note the importance of “genetic, environmental, and parental influences on early development of obesity.”
These studies are important to note because if you look at the broader, non-subgroup specific, research on soda and fruit juice consumption and weight gain in preschool populations, you find either no association or only a non-significant positive association.
Children versus adults
A second important caveat is to distinguish between studies involving children and adolescents and those involving adults. Children are supposed to gain weight, but their manner of doing so is not always consistent. At any given moment, some children and teens will be crossing over from normal to overweight or from overweight to obese ,while others will be going in the other direction. This makes it very difficult to draw conclusions for adults from soda and weight gain in children.
Is it the soda or is it the calories?
Remarkably, there are few randomized control studies for SSB consumption and weight gain in adults. In a review of the evidence by an advisory committee of the nation’s top experts on nutrition for the 2010 US Department of Agriculture’s dietary guidelines, the conclusion was that:
This raises the question of how the evidence for higher energy intake can be weaker than the evidence for increased body weight from increased consumption (i.e., energy intake)? The transcript of the advisory committee’s discussion shows the members agonizing over the inconsistency between these two statements and the relative weakness of the connection between sugar and body weight due to limited or poorly conducted data.
One advisory member, Lawrence Appel, Professor of Medicine at Johns Hopkins, describes the evidence as “crappy” but getting better and that it is important to consider the “direction” of this better evidence (which is for an association between increased consumption of SSBs and weight gain) in making recommendations for the public. At the same time, Joanne Slavin, Professor of Food Science and Nutrition at the University of Minnesota, points out that the direction of the newer evidence still doesn’t amount to strong evidence of a link between sugar and body weight. “It is disappointing how little there is,” she says of the research, “for such an important topic.”
The Mattes et al., review and meta-analysis of intervention studies, suggests that increasing consumption of calories in a controlled environment does lead to weight gain –more so for increasing consumption through soda than for increasing consumption through other foods. One cannot discern from this whether the "natural" environment would have the same effect, since the study that examined this was structured to ask people to eat additional foods. It gives some evidence that soda might be “different” than solid foods, i.e. that solid calories more readily lend themselves to compensating by eating less of other things than soda. Decreasing consumption of soda also seems to have an impact on heavier people and those who are heavy soda consumers to begin with.
A third caveat is that the observed correlation between weight gain and soda consumption among those who are already overweight also holds for other high calorie foods too, such as chips (McCarthy et al. 2006). A study of rural communities in the US (Liebman et al. 2003) also found that overweight and obese adults were more likely to consume supersized portions and watch television.
A study of demographic and behavioral factors associated with drinking soda in New York (Rehm et al., 2008) also found that frequent consumption was associated with more television viewing and less physical activity.
A well-designed randomized control trial (RCT) is needed to determine the relationship between sugared drink consumption and weight gain. As a number of researchers have noted, observational studies are just not up to the task of determining causality, and further observational studies are not going to provide any insight.
The best that can be said of the evidence so far is that children and adults who are already overweight and who are high consumers of sugar sweetened beverages do benefit from reducing consumption. Would a soda tax reach them and alter their consumption patterns?
Back to the homepage
The Statistical Assessment Service (STATS)
933 N. Kenmore St., Suite 405
Arlington, VA 22201