当然のことながら、この論文に対しては疑問が投げかけられた。内容を子細に検討すると、看過できない瑕疵がいくつかあるとの指摘である。
要約すると、
＊ データ採集の際に用いられた食事に関する質問票の設定が荒い上、実際とはズレが生じ易い記入法で被験者に依頼している。しかも、調査開始時に一回しか実施されていない。
＊ 人口甘味料について調査した他の研究結果をきちんと参照した形跡がない。
＊ だから、生理学的に有り得ない主張をしてしまっている。
･･･というものだ。いずれも穏当なものと言えるだろう。
更に重要なのは、観察研究に於けるデータ解釈の際に陥りがちな「論理の逆転(reverse causation)」にも言及していることである。
身も蓋もない言い方でいえば、「人口甘味料をたくさん摂るから糖尿病になるんじゃなくて、既に太っている人がこのままだと糖尿病になってしまうと思って人口甘味料を多く摂るようになるんじゃない？」という指摘だ。そして、この指摘は、おそらく正しい。
Diabetes and diet beverage study has serious limitations
The American Journal of Clinical Nutrition, Volume 98, Issue 1, July 2013, Pages 248–249, https://doi.org/10.3945/ajcn.113.062588
Dear Sir:
In the article “Consumption of artificially and sugar-sweetened beverages and incident type 2 diabetes in the Etude Epidémiologique auprès des femmes de la Mutuelle Générale de l’Education Nationale–European Prospective Investigation into Cancer and Nutrition cohort” (1), the authors examined the associations between sugar-sweetened beverages, artificially sweetened beverages, and type 2 diabetes. We argue against the conclusion in this study that consumption of artificially sweetened beverages increases the risk of developing type 2 diabetes due to the following study limitations.
Poor assessment of beverage consumption. The assessment of sugar-sweetened beverages and artificially sweetened beverages was inadequate. Although the study ran from 1993 to 2007, participants were asked only once in 1993 about their diet beverage consumption patterns. What and how much the subjects drank likely changed over the course of the study period, but the authors did not account for this. Furthermore, participants were asked to recall their diet beverage consumption patterns for the past year. One year is a very long recall period, which introduces significant recall bias into the study.
Results are inconsistent with other research on aspartame. The authors suggested that aspartame might increase glucose or insulin concentrations, but numerous human studies have shown that aspartame does not increase postprandial glucose or insulin concentrations (2–9). The researchers did not discuss their findings in relation to these studies, which included robust clinical trials. It was also suggested that low-calorie sweeteners may lead to increased preference for sweets and/or enhance appetite; however, neither of these have been shown to be true in scientific studies. Rather, a recent review article found that low-calorie sweeteners do not enhance appetite (10).
Biologically implausible findings. It does not seem physiologically possible for low-calorie sweeteners to cause diabetes, and the authors offered no explanations as to why such an association might exist. Middle-aged people are more likely to develop type 2 diabetes regardless of what they drink. Type 2 diabetes is strongly associated with being overweight. When the researchers controlled for BMI, a measure of body weight, the association between diet beverages and diabetes decreased. This finding points to the possibility that the results reported may be related to factors unrelated to diet beverage use, such as total body fat. Individuals seeking to lose weight or to manage their blood sugars often switch to diet beverages. Low-calorie sweetener use might therefore simply be a marker for individuals already on weight-gain or diabetes trajectories, which continued despite their switching to diet beverages. This is the most plausible explanation of these findings, not that the diet beverages caused the overweight or diabetes. Although the authors attempted to control for reverse causality, because it was not a randomized controlled trial reverse causation is still a possible explanation for the findings.
Other factors may have influenced the findings. Although researchers tried to control for many variables, it is possible that the findings were due to other factors for which the study was not controlled. This includes, for example, the actual diets of the participants during the years since 1993. The authors also failed to control for risk factors associated with type 2 diabetes, such as metabolic syndrome, hypertriglyceridemia, waist circumference, and coronary artery disease. External factors such as these may likely have influenced the results, which the authors acknowledged.
Other study limitations. The study included only women. There was also a large variation in how many diet beverages the women consumed, which makes the data less accurate. The authors admitted they had “limited statistical power in some subcategories,” so the data may be unreliable. Furthermore, ∼1 in 4 women in the cohort failed to participate, which has the potential to introduce self-selection bias. Finally, the questionnaire used to determine diabetes diagnosis in the study was administered only every 2–3 y, which may have introduced recall bias into the study.
The author is the President of the Calorie Control Council, an international association of manufacturers of low- and reduced-calorie foods and beverages, including companies that make alternative sweeteners (eg, intense sweeteners, polyols), fibers, and fat replacers used in those products.
