Abstract
Background: Numerous systematic reviews (SR) and meta-analyses (MA) on low calorie sweeteners (LCS) have been published in recent years, concluding that LCS have beneficial, neutral, or detrimental effects on various health outcomes, depending on the review.
Objectives: The objective of this overview of reviews was to determine how the methodologies of SR investigating the association between LCS consumption and body weight (BW) influence their findings and whether MA results can provide a consistent estimated effect.
Methods: Systematic searches of PubMed, Scopus, and Cochrane Library were conducted in November 2022 to identify SR of randomized controlled trials (RCT) or non-randomized studies (NRS) investigating the association between LCS consumption and BW. The methods, MA results, and conclusions were extracted from each eligible SR.
Results: Of the 985 search results, 20 SR evaluated the association between LCS and BW, drawing from publications of 75 RCT, 42 prospective cohort studies, and 10 cross-sectional studies. There was a considerable lack of overlap of studies included within each SR attributed, in part, to the inclusion of studies based on design; thus, each SR synthesized results from distinctly different studies. Evidence synthesis methods were heterogeneous and often opaque, making it difficult to determine why results from certain studies were excluded or why disparate results were observed between SR.
Conclusions: SR investigating the effect of LCS on BW implement different methodologies to answer allegedly the same question, drawing from a different set of heterogeneous studies, ignoring the basic assumptions required for MA, resulting in disparate results and conclusions. Previous MA show the large effects of study design, which results in inconsistent estimates of the effect of LCS on BW between MA of RCT and NRS. Given the availability of long-term RCT, these studies should be the basis of determining causal relationships (or lack thereof) between LCS and BW. This trial was registered at PROSPERO as CRD42022351200.
Summary
This umbrella review sought to determine how the inclusion criteria, methodologies, and availability of relevant data used in systematic reviews (SRs) influence their findings and whether meta-analysis (MA) results can provide a consistent estimated effect among SRs investigating the association of low and no calorie sweeteners (LNCS) consumption with body weight related outcomes.
To pursue this aim, a systematic literature search was conducted, retrieving a total of 20 systematic reviews that investigated the association between LNCS and weight-related outcomes, and which were analysed in this umbrella review.The SRs cumulatively identified 75 publications from RCTs, 42 publications from prospective cohort studies, and 10 publications from cross-sectional studies.
SRs included in this study used different inclusion criteria, literature search methods, and evidence synthesis methods that resulted in the combination of highly heterogeneous studies each designed to address their own specific question, and the conclusions drawn by each SR depended on the type of studies included in each review.
Despite these differences in SR methodology, MAs of RCTs consistently suggested that LNCS are beneficial for weight management (particularly when compared with sugars) or have no effect on body weight. Conversely, weight status effect estimates of MA of prospective cohorts produce a range of results from significantly beneficial to significantly adverse. The certainty of evidence ranged from very low to moderate. When studies included within an MA are highly heterogeneous, inclusion or exclusion of individual studies can lead to large deviations in effect estimates. This may explain why SR and MA designed to address the same question of the effect of a dietary exposure on a health outcome often come to different conclusions, sparking debates as to whether SR and MA should be the basis to draw scientific conclusions.
The authors conclude that more SRs are not going to provide answers to whether there is a universal causal relationship between LNCS and body weight. Instead of trying to statistically combine studies to explain this complex relationship that continually yield mixed results, it is more appropriate to rely on the high-quality RCT designed to test specific hypotheses and mechanisms to explain under which conditions LNCS intake decreases, increases, or has no effect on body weight.