Abstract
There is a current debate on the effects of artificial sweeteners on health-related outcomes among different populations. Even when meta-analytic evidence has revealed that the use of non-nutritive sweeteners reduces body mass when substituting sugar intake, concerns about how these additives could impact glucose homeostasis in people with diabetes persist.
A recent Cochrane systematic review and meta-analysis of randomised trials conducted by Lohner and colleagues was generated after a request from the Nutrition Guidance Expert Advisory Group of the World Health Organization to provide evidence on the use of non-nutritive sweeteners in people with diabetes. Considering the transcendency of these guides, we are concerned that critical methodological considerations were overlooked, yielding imprecisions on the results’ interpretation.
Summary
In this Letter, Laviada-Molina and colleagues present scientific comments and concerns on the methodology and interpretation of the outcomes of the systematic review by Lohner et al (2020) on “Non-nutritive sweeteners for diabetes mellitus”, that was supported by the WHO. The authors comment that meta-analysing studies with such high methodological and clinical heterogeneity could lead to severe bias. Laviada-Molina et al are concerned that critical methodological considerations may have been overlooked, yielding imprecisions on the results’ interpretation.
The authors discuss limitations of the Lohner et al systematic review in their Letter. For example, they note that the meta-analysis comparing sugar with low/no calorie sweeteners included only 36 people derived from 3 cross-over trials assessing glycated heamoglobin (HbA1c). None of these trial designs included a washout period between the interventions, introducing overlapping effects between exposures. Furthermore, the inclusion criteria were inconsistent across studies, including in the same analysis both type 1 and type 2 diabetes patients despite the entirely different pathophysiology and treatments for these conditions. The authors note that meta-analysing studies with such critical methodological heterogeneity is inappropriate, as supported by the 74% statistical heterogeneity found in the analysis. On the other hand, Lohner and colleagues’ conclusion that the effects of low/no calorie sweeteners on HbA1c remain unclear seems unjustified based on the study’s results from the second comparison between sweeteners vs placebo, which indicates that low/no calorie sweeteners do not promote any glycaemic variation. Lastly, Laviada-Molina et al point out that the extensive evidence from clinical trials exploring acute effects of low/no calorie sweeteners on glycaemia, and which consistently show that low/no calorie sweeteners cause a lower spike in blood glucose compared to sugars, was not discussed at all by Lohner et al.