Relation of change or substitution of low-and no-calorie sweetened beverages with cardiometabolic outcomes: A systematic review and meta-analysis of prospective cohort studies

Background: Adverse associations of low-and no-calorie sweetened beverages (LNCSBs) with cardiometabolic outcomes in observational studies may be explained by reverse causality and residual confounding. Purpose: To address these limitations we used change analyses of repeated measures of intake and substitution analyses to synthesize the association of LNCSBs with cardiometabolic outcomes. Study Selection: MEDLINE, EMBASE, and the Cochrane Library were searched up to 10 June 2021 for prospective cohort studies ≥1-year follow-up duration in adults. Outcomes included changes in clinical measures of adiposity, risk of overweight/obesity, metabolic syndrome, diabetes, cardiovascular disease, and total mortality. Data Extraction: Two independent reviewers extracted data, assessed study quality, and certainty of evidence using GRADE. Data was pooled using random-effects model and expressed as mean difference (MD) or risk ratio (RR) and 95% CI. Data Synthesis: Fourteen cohorts (416,830 participants) met the eligibility criteria. Change in LNCSB intake was associated with lower weight (5 cohorts, 136,206 participants; MD, -0.008 [95% CI: -0.014, -0.002] kg/y). Substitution of LNCSBs for sugar-sweetened beverages (SSBs) was associated with lower weight (3 cohorts, 165,579 participants; MD, -0.12 [95% CI: -0.14, -0.01] kg/y) and lower incidence of obesity (1 cohort, 15,765 participants; RR, 0.88 [0.88, 0.89]), coronary heart disease (6 cohorts, 233,676 participants; RR, 0.89 [95% CI: 0.81, 0.98]), CVD mortality (1 cohort, 118,363 participants; RR, 0.95 [95% CI: 0.90, 0.99]), and total mortality (1 cohort, 118,363 participants; RR, 0.96 [95% CI: 0.94, 0.98]) with no adverse associations across other outcomes. Substitution of water for SSBs showed lower weight (3 cohorts, 165,579 participants; MD, -0.10 [95% CI: -0.13, -0.06] kg/y), lower waist circumference (1 cohort, 173 participants; MD, -2.71[95% CI: -4.27, -1.15] cm/y) and percent body fat (1 cohort, 173 participants; MD, -1.51 [95% CI: -2.61, -0.42] %/y), and lower incidence of obesity (1 cohort, 15,765 participants; RR, 0.85 [95% CI: 0.75, 0.97]) and diabetes (3 cohorts, 281,855 participants; RR, 0.96 [95% CI: 95% CI: 0.94, 0.98]). Substitution of LNCSBs for water showed no adverse associations. Limitations: The evidence was low to very low certainty owing downgrades for imprecision, indirectness and/or inconsistency. Conclusions: LNCSBs were not associated with cardiometabolic harm in analyses that model the exposure as change or substitutions. The available evidence provides some indication that LNCSBs in their intended substitution for SSBs may be associated with cardiometabolic benefit, comparable to the standard of care, water.