The observational NutriNet-Santé study of Debras et al. on artificial sweetener (AS) intake and cardiovascular disease (CVD)  raises logical and methodological issues.
1/ Diabetes is a strong risk factor for CVD, and subjects with abnormal glucose metabolism are generally advised to replace sugar with AS. If subjects with diabetes at the start of follow-up were excluded from analyses, subjects with incident diabetes during follow-up were part of the cohort, and there was no adjustment for incident diabetes or new use of anti-diabetic drugs during the 9-year follow-up.
2/ A soda can (250 mL) contains around 200 mg of aspartame. A daily intake of AS of about 318 mg for a European adult of 80 kg is common . In the article, mean daily intakes of AS in lower and higher consumer groups were 7.5 and 77.6 mg/day, respectively. Hence compared to general populations, daily intakes of AS of subjects included in the NutriNet-Santé cohort are trivial. If risk estimates between AS or aspartame and CVD reported by Debras et al. were true, a marked epidemic of CVD should have been noticed since long among regular soda drinkers.
3/ There is a tenfold difference in mean daily AS consumption AS consumption between the lower and higher consumers. However, hazard rations (HR) for CVD are nearly equal (HR=1.19 and 1.20 in Supplement Table 1). Results for CHD suggest a higher risk for the lower than for the higher consumer group. These results are hardly believable.
4/ Debras et al report that “Competing risks were accounted for in all analyses”. But methods used for competing risk analyses are not clear. Authors have used the Fine-Gray (FG) method. But the FG method is not appropriate in the context of this study. This study is about aetiology (is AS intake associated with raised risk of CVD?), and not about prediction (what is the probability of being diagnosed with CVD according to AS intake?). When a study is about aetiology, the cause-specific HR is to be used for causal inference thinking, not subdistribution HR obtained using the FG method [3, 4]. Moreover, if the FG method is used, articles need to report all cause-specific hazards and cumulative incidence functions .
Methodological and logical concerns should be addressed before going further in the interpretation of study results.
Philippe Autier, MD, PhD, and Patrick Mulie, PhD
International Prevention Research Institute (iPRI)
Chemin des Cuers 18
69570 Dardilly (France)
ORCID : 0000-0003-1538-5321
Ph Autier and P Mullie have no conflict of interest related to their comments.
1. Debras, C., et al., Artificial sweeteners and risk of cardiovascular diseases: results from the prospective NutriNet-Santé cohort. Bmj, 2022.
2. Huvaere, K., et al., Dietary Intake of Artificial Sweeteners by the Belgian Population. Food additives & contaminants. Part A, Chemistry, analysis, control, exposure & risk assessment, 2011. 29: p. 54-65.
3. Lau, B., S.R. Cole, and S.J. Gange, Competing risk regression models for epidemiologic data. Am J Epidemiol, 2009. 170(2): p. 244-56.
4. Austin, P.C. and J.P. Fine, Practical recommendations for reporting Fine-Gray model analyses for competing risk data. Stat Med, 2017. 36(27): p. 4391-4400.
5. Latouche, A., et al., A competing risks analysis should report results on all cause-specific hazards and cumulative incidence functions. J Clin Epidemiol, 2013. 66(6): p. 648-53.
Competing interests: No competing interests