論文に対して公の場で疑念を呈された場合、煩を厭わず応答するのが研究者としての仁義であるらしい。その点から見れば、ダンマリを決め込んだりせずにリプライを返したこの論文の著者の態度は誠実なものであったと言ってよい。
但し、この場合に限っては、残念なことに、著者の認識に大きな問題があることが、かなり露骨な形で示される形になってしまった。
先に(2)(3)で引いたレターで疑問視されていたのは、いずれもデータの「質」に関わるものであって、決してデータ数の不足を問題視したものではなかった。そんなことぐらいは、素人だってわかる。しかし、この著者は、(よりにもよって)無数の専門家たちが見守る学会誌の誌上で、まず、データ数の多さと追跡期間の長さを力強くアピールすることから反論(?)を開始するのである。もう、このあたりで不安がいっぱいである〈笑〉。
その上、広く評価されているタイプの質問票を用いていることを挙げ(使い方に難があると指摘されていたんじゃ･･･)、集計結果を可能な限り多くの変数で調整したことを高らかに告げるのである(その結果を見た上で、再検討の余地アリと指摘されていたよね???)。
ハラハラしながら見守る私の気持ちを逆なでするかの如く、著者は、「未だ明確な機序は明らかになってはいないが、我々の論文は、科学者達に、人口甘味料を含む飲み物が高い糖尿病リスクと関連を持つことについて、更なるエビデンスを提供するものだ(大意)」と、華麗にフィニッシュを決めてしまうのである。
おい！ 専門家がそんなことでいいのか？〈笑〉
Reply to HC Stevens and C La Vecchia
The American Journal of Clinical Nutrition, Volume 98, Issue 1, July 2013, Pages 250–251, https://doi.org/10.3945/ajcn.113.062885
Dear Sir:
Our study “Consumption of artificially and sugar-sweetened beverages and incident type 2 diabetes in the Etude Epidemiologique aupres des femmes de la Mutuelle Generale de l’Education Nationale-European Prospective Investigation into Cancer and Nutrition cohort” showed that both sugar-sweetened and artificially sweetened beverages (SSBs and ASBs) were associated with an increased risk of type 2 diabetes (T2D) (1). The main strengths of our study were the size of the population (n = 66,118 women free of diabetes at baseline), the long follow-up (from 1993 up to 2007), the use of a validated dietary questionnaire, and the large number of covariates to control for.
Information on SSBs, ASBs, and fruit juice consumption
In 1993, women were asked to report their usual diet using a validated 208-item diet history questionnaire, structured according to the French meal pattern. Information on diet was recorded at baseline, and not updated, so that some women may have changed their consumption of SSBs and ASBs over time. Such a misclassification, however, should only bias our relative risk estimates toward unity, and “true” associations should actually be of greater magnitude.
Categories of ASB and SSB consumption have been defined among consumers of sweetened beverages (ie, women who consume ASBs only, SSBs only, or both). Groups of consumption were then compared with nonconsumers of SSBs or ASBs taken as the reference category for the calculation of HRs.
Associations between ASBs, SSBs, and T2D
Although we observed a decrease in the magnitude of HRs observed when adjusting for multiple covariates, which may be due to a decrease in statistical power, trends in risk remained significant. Most of the variation in HRs was observed when we further adjusted for BMI, which was consistent with our hypothesis that BMI is an intermediate factor in the relation between SSBs, ASBs, and T2D risk. Part of the association between soft drinks and T2D was captured by obesity. But, as stated in the article, we still observed a direct effect of SSBs and ASBs on T2D risk, independent of BMI.
It is very unlikely that the observed associations—for both SSBs and ASBs—were entirely due to residual confounding because models were adjusted for most of the well-known risk factors of T2D.
Biological implications
As mentioned in our article’s Introduction, the literature on the effect of ASBs on health is inconsistent. We deliberately considered results from cohort studies only. Previous large cohort studies that were interested in SSB, ASB, and T2D risk have concluded either a significant increased risk (2), which became nonsignificant after controlling for BMI, or no increased risk at all (3). We also clearly stated that no clear biological mechanism was proposed to explain the associations. However, all previous studies had strong limitations, which prevents interpreting their results on ASBs. The most serious limitation was the possibility of reverse causation, possibly due to the fact that ASB consumption has already been shown to be higher in individuals with T2D or prediabetic conditions, such as obesity.
The same hypothesis of reverse causation was evoked by La Vecchia and Stevens with regard to our study. We disagree with them for 2 reasons. First, in a sensitivity analysis, we excluded the 5 first years of follow-up, and results were very similar for both SSBs and ASBs. To our knowledge, our study is the first prospective study that takes into account the reverse causation issue. Second, the E3N cohort is mostly composed of teachers, a relatively health-conscious population with a low proportion of obese participants (3.2% in 1993). The associations we observed were actually driven by women with a BMI (in kg/m2) <25, who therefore developed diabetes without being obese. Consequently, we do not believe that reverse causation could explain the direct association between ASBs and T2D.
Conclusions
Even if no clear mechanisms have been proposed to explain the association between ASBs and T2D risk, our study provides additional evidence that high consumption of these beverages is associated with a strong increased risk of T2D. This conclusion will hopefully prompt scientists to conduct biological experiments on components of soft drinks and to replicate our findings in other populations.
The authors declared that they have no competing interests.
