Do Low-Calorie Sweetners Affect Body Weight?Friday, May 18, 2012
This is certainly one of those topics that attracts vehement and passionate views and arguments on both side – arguments, for which, it turns out, there is little more than ‘circumstantial’ evidence (on both sides).
Readers interested in this issue are directed to a rather entertaining and thought-provoking review by the University of Toronto’s Harvey Anderson, co-authored by my good friend and colleague David Allison from the University of Birmingham, Alabama, published in the Journal of Nutrition.
The article appears in a supplement reporting the proceedings of a conference on Low-Calorie Sweeteners, Appetite and Weight Control: What the Science Tells Us, held in Washington, DC, April 7–8, 2011. The conference was sponsored by the Committee on Low-calorie Sweeteners of the International Life Sciences Institute North America.
The short answer to the question at hand, in the words of the authors, following their review of the scientific evidence, is simply:
“We conclude that there is no evidence that LCS can be claimed to be a cause of higher body weights in adults. Similarly, evidence supporting a role for LCS in weight management is lacking. Due to the confounders in most observational studies, randomized controlled trials are needed to advance understanding.”
Because much of the speculations about whether or not LCS is part of the problem (or the solution) is based on observational studies (rather than well-designed definitive randomised-controlled trials), the authors provide a rather succinct and insightful primer into what we can (or cannot) learn from this type of studies.
Here is a brief summary of their arguments:
“Observational studies only tell us about the associations among variables. Relative to true experiments, observational studies offer only weak information and they may suggest the plausibility of causation but cannot demonstrate causation.”
“…the fundamental limitation is the possibility of confounding. …. Confounding cannot be unequivocally eliminated, because we can only control for those variables that we think to measure, can measure well, and know how to statistically model appropriately. Unfortunately, knowledge is limited (if it were not, research would not be needed) and measurement capabilities for many potentially confounding factors such as diet, physical activity patterns, drug intake, acculturation, and social status are often quite limited. Therefore, there is no certainty that all confounders have been controlled for in observational studies.”
Unfortunately, despite these well know limitations, observational studies are usually misinterpreted or misrepresented.
“Three common examples of misleading statements are as follows. The first example arises from extrapolation beyond the range of the observed data. For instance, if it is found that people who run 5 mi/d (8.1 km/d) live 5 y longer on average than people who run 0 mi/d, the data cannot be extrapolated to suggest that if people run 50 mi/d, they would live 50 y longer.
Second, misleading statements can arise fromusing language that indicates causation when only association has been demonstrated. As recently documented in an empirical analysis, this is a far too common practice in the fields of nutrition and obesity.
Finally, another common, misleading approach involves reporting estimates of the strength of associations (so-called effect sizes) in a way that exaggerates their magnitude.”
Given these limitations of observational studies, the authors raise another important issue with regard to the rather common practice of adding new observational studies when other studies have more or less already looked at this issue. Thus,
“When contemplating conducting the (M+1)th study, one has to ask: “Is there reason to believe that that study will move the collective results from one category to another or settle the question of causation if that is under debate?” Often the answer is no, but the (M+1)th study is conducted nonetheless, perhaps more as a tool for rhetoric than to enhance understanding. Doing so is arguably an unethical waste of resources.”
Perhaps, if the question to be answered is indeed deemed important enough, the authors argue, such money should better be used on conducting randomised controlled studies (RCTs).
“Ultimately, in many cases, it is only the conduct of large, well designed RCT that will settle a controversy and yield satisfactory evidence about causation. Yet in some domains, particularly in the area of health-related policy, it is often expressed that RCT are impossible. Although this is assuredly true in some cases, RCT may be far more feasible than many advocates of public health policies acknowledge. RCT have been conducted to include many thousands of people and to study issues as challenging as the effects of moving poor families to less poor neighborhoods and the effects of supplemental income on long-term retention of guardianship for displaced children.
If RCT can be conducted to address such challenging, expensive, and sensitive topics, then surely RCT of many nutrition-related practices and policies can be conducted. Long-term RCTs in free-living adult populations to separately address the role of LCS in weight loss or weight maintenance will ultimately help to clarify the role of LCS in assisting motivated individuals to achieve body weight goals and to inform policy makers.”
Till then, I am sure the debate will continue – as passionate as the arguments put forward may be and as entertaining as such controversies may seem, in the absence of conclusive data, we must remember that we are listening to opinions rather than evidence.
In this respect, the debate on the pros and cons of LCS is certainly not the only one in the field of obesity that would clearly benefit from actual evidence.
Anderson GH, Foreyt J, Sigman-Grant M, & Allison DB (2012). The Use of Low-Calorie Sweeteners by Adults: Impact on Weight Management. The Journal of nutrition PMID: 22573781