Dietary sugar consumption and health: umbrella review

Abstract

Objective: To evaluate the quality of evidence, potential biases, and validity of all available studies on dietary sugar consumption and health outcomes.

Design: Umbrella review of existing meta-analyses.

Data sources: PubMed, Embase, Web of Science, Cochrane Database of Systematic Reviews, and hand searching of reference lists.

Inclusion criteria: Systematic reviews and meta-analyses of randomised controlled trials, cohort studies, case-control studies, or cross sectional studies that evaluated the effect of dietary sugar consumption on any health outcomes in humans free from acute or chronic diseases.

Results: The search identified 73 meta-analyses and 83 health outcomes from 8601 unique articles, including 74 unique outcomes in meta-analyses of observational studies and nine unique outcomes in meta-analyses of randomised controlled trials. Significant harmful associations between dietary sugar consumption and 18 endocrine/metabolic outcomes, 10 cardiovascular outcomes, seven cancer outcomes, and 10 other outcomes (neuropsychiatric, dental, hepatic, osteal, and allergic) were detected. Moderate quality evidence suggested that the highest versus lowest dietary sugar consumption was associated with increased body weight (sugar sweetened beverages) (class IV evidence) and ectopic fatty accumulation (added sugars) (class IV evidence). Low quality evidence indicated that each serving/week increment of sugar sweetened beverage consumption was associated with a 4% higher risk of gout (class III evidence) and each 250 mL/day increment of sugar sweetened beverage consumption was associated with a 17% and 4% higher risk of coronary heart disease (class II evidence) and all cause mortality (class III evidence), respectively. In addition, low quality evidence suggested that every 25 g/day increment of fructose consumption was associated with a 22% higher risk of pancreatic cancer (class III evidence).

Conclusions: High dietary sugar consumption is generally more harmful than beneficial for health, especially in cardiometabolic disease. Reducing the consumption of free sugars or added sugars to below 25 g/day (approximately 6 teaspoons/day) and limiting the consumption of sugar sweetened beverages to less than one serving/week (approximately 200-355 mL/week) are recommended to reduce the adverse effect of sugars on health.

Systematic review registration: PROSPERO CRD42022300982.

Fig 1

Flowchart of systematic search and selection process

Fig 2

Significant dose-response relations between dietary sugar consumption and multiple health outcomes. Estimates are relative risks, summary mean difference is weighted mean difference, and effect models are random unless noted otherwise. Δ=final value – baseline value; AMSTAR=a measurement tool to assess systematic reviews; C=cohort studies; CHD=coronary heart disease; CI=confidence interval; CVD=cardiovascular disease; GRADE=Grading of Recommendations Assessment, Development and Evaluation; NA=not available; P=population based case-control and/or cross sectional studies; SSB=sugar sweetened beverage; T=total No of studies; T2DM₌type 2 diabetes mellitus. *1 serving/week increment. †355 mL/d increment. ‡250 mL/d increment. §1 serving/d increment. ¶25 g/d increment. **Hazard ratio. †Children

Fig 3

Significant non-dose-response relations between dietary sugar consumption and endocrine and metabolic outcomes. Comparisons are highest versus lowest, estimates are relative risks, summary mean difference is weighted mean difference, and effect models are random unless noted otherwise. Complete associations between dietary sugar consumption and endocrine and metabolic outcomes are shown in supplementary table A. Δ=final value – baseline value; AMSTAR=a measurement tool to assess systematic reviews; C=cohort studies; CI=confidence interval; GRADE=Grading of Recommendations Assessment, Development and Evaluation; HDL-C=high density lipoprotein cholesterol; LADA=latent autoimmune diabetes in adults; LDL-C=low density lipoprotein cholesterol; NA=not available; P=population based case-control and/or cross sectional studies; R=randomised controlled trials; SSB=sugar sweetened beverage; T=total No of studies. *Odds ratio. †Children. ‡Any versus none. §Fixed effects model. ¶Standardised mean difference

Fig 4

Significant non-dose-response relations between dietary sugar consumption and cardiovascular outcomes. Comparisons are highest versus lowest, estimates are relative risks, summary mean difference is weighted mean difference, and effect models are random unless noted otherwise. Complete associations between dietary sugar consumption and cardiovascular outcomes are shown in supplementary table B. Δ=final value – baseline value; AMSTAR=a measurement tool to assess systematic reviews; C=cohort studies; CI=confidence interval; CVD=cardiovascular disease; GRADE=Grading of Recommendations Assessment, Development and Evaluation; NA=not available; P=population based case-control and/or cross sectional studies; SBP=systolic blood pressure; SSB=sugar sweetened beverage; T=total No of studies. *Children and adolescents. †Odds ratio

Fig 5

Significant non-dose-response relations between dietary sugar consumption and cancer outcomes. Comparisons are highest versus lowest, estimates are relative risks, and effect models are random unless noted otherwise. Complete associations between dietary sugar consumption and cancer outcomes are shown in supplementary table C. AMSTAR=a measurement tool to assess systematic reviews; GRADE=Grading of Recommendations Assessment, Development and Evaluation; C=cohort studies; CI=confidence interval; NA=not available; P=population based case-control and/or cross sectional studies; SSB=sugar-sweetened beverage; T=total No of studies

Fig 6

Significant non-dose-response relations between dietary sugar consumption and other outcomes. Comparisons are highest versus lowest, estimates are relative risks, summary mean difference is weighted mean difference, and effect models are random unless noted otherwise. Complete associations between dietary sugar consumption and other outcomes are shown in supplementary table D. ADHD=attention deficit/hyperactivity disorder; AMSTAR=a measurement tool to assess systematic reviews; BMD=bone mineral density; C=cohort studies; CI=confidence interval; GRADE=Grading of Recommendations Assessment, Development and Evaluation; IHCL=intrahepatocellular lipids; NA=not available; NAFLD=non-alcoholic fatty liver disease; P=population based case-control and/or cross sectional studies; R=randomised controlled trials; SSB=sugar-sweetened beverage; T=total No of studies. *Children. †Odds ratio. ‡Fixed effects model. §Never/low versus moderate/high consumption. ¶Standardised mean difference. **Any versus none

Fig 7

Map of outcomes associated with dietary sugar consumption