Introduction
The World Health Organization non sugar sweetener guideline identified two issues with non-nutritive sweetener (NNS) research potentially contributing to the low quality of evidence; possible differences in the effects of varying NNS types, and different outcomes identified between shorter term interventions and longer term observational studies (World Health Organization, 2023). The aim of this review is to (1) identify all systematic reviews investigating NNS consumption and metabolic health or cancer outcomes and (2) to extract and compare research methods from all primary studies assessed in these systematic reviews.
Methods
A literature search was conducted using Cochrane Databases of Systematic Reviews, Medline, PubMed, Scopus, Web of Science (Core Collections). Systematic reviews relating to NNS and cancer or metabolic disease were included. The primary studies from these systematic reviews were then extracted and methodologies described and compared.
Results
216 primary studies were extracted from 45 systematic reviews. One hundred and fourteen experiments were described across the 103 intervention primary studies; 63 of which investigated beverages only. Forty-seven of these studies had NNS exposures shorter than one day. The two most investigated NNSs were aspartame (n = 53) and sucralose (n = 27). One hundred and seventy-six observational studies were identified across 113 observational papers; 111 of which were cohort studies. Food frequency questionnaires (FFQs) were the dominant nutrition assessment tool used for assessing NNS intake (n = 97), 78 of which only investigated beverages. Only 26 studies investigated specific NNS types. Forty-nine of the 111 cohort studies measured NNS intake only once during the study.
Conclusion
NNS observational research relies heavily on infrequent FFQs investigating beverage consumption to determine NNS consumption, and intervention studies also prioritise beverages as the main exposure method of NNS. This body of research heavily excludes food, which may contribute to inconsistencies in NNS research.