Ultra-processed food consumption and metabolic disease risk: an umbrella review of systematic reviews with meta-analyses of observational studies

Abstract

Background and aims: There is an ongoing debate on whether to advocate reducing ultra-processed food (UPF) in dietary guidelines to control metabolic disease (such as obesity and type 2 diabetes mellitus [T2DM]). We aimed to summarize the evidence from systematic reviews with meta-analyses between UPF consumption and metabolic diseases risk, assess the credibility, and verify the robustness of these associations.

Methods: We systematically searched PubMed, Web of Science, Embase, and Cochrane Library databases from their inception to July 15, 2023, to identify relevant systematic reviews with meta-analyses. We used the random-effects model to evaluate the summary effect size, along with 95% confidence interval and prediction interval. We also assessed heterogeneity, evidence of small-study effects and excess significance bias, and categorized the credibility of each association based on quantitative umbrella review criteria. Additionally, we conducted subgroup and sensitivity analyses to assess the robustness of associations based on continents, study design, dietary assessment methods, definition methods of UPF, population, and units of UPF consumption.

Results: Overall, 6 systematic reviews with 13 meta-analyses were included. Three (23.08%) meta-analyses were classified as highly suggestive evidence for meeting the criteria that associations were significant at p < 10^-6, had more than 1,000 cases, and presented the largest study with significance at p < 0.05. Among them, the highest UPF consumption quantile was associated with an increased risk of obesity (OR = 1.55, 95% CI: 1.36-1.77) when compared with the lowest UPF consumption quantile. The highest UPF consumption quantile was associated with an increased risk of T2DM (RR = 1.40, 95% CI: 1.23-1.59) when compared with the lowest UPF consumption quantile, and a 10% increase in UPF consumption (% g/d) was associated with an increased risk of T2DM (RR = 1.12, 95% CI: 1.10-1.13). Meanwhile, the robustness of these associations was verified by a series of subgroup and sensitivity analyses.

Conclusion: UPF consumption may be a risk factor for several metabolic diseases. However, well-designed studies are still needed to verify our findings in the future.

Keywords: meta-analysis; metabolic diseases; observational study; ultra-processed food; umbrella review.

Figure 1

Flow diagram of the study selection process.

Figure 2

Summary random effect estimate with 95% confidence interval from 13 meta-analyses evaluating the association between ultra-processed food consumption and metabolic disease risk. CI, confidence interval, MetS, metabolic syndrome; NAFLD, non-alcoholic fatty liver disease; OR, odds ratio; RR, risk ratio; T2DM, type 2 diabetes mellitus. ^a “lowest” was defined as the lowest UPF consumption quantile, “moderate” was defined as the first exposure quantile, and “highest” was defined as the highest UPF consumption quantile. ^b The unit of UPF consumption was % kcal/d. ^c The unit of UPF consumption was % g/d.

Figure 3

Credible assessment of 13 meta-analyses evaluating the association between ultra-processed food consumption and metabolic disease risk. MetS, metabolic syndrome; NAFLD, non-alcoholic fatty liver disease; T2DM, type 2 diabetes mellitus. “Lowest” was defined as the lowest UPF consumption quantile, “moderate” was defined as the first exposure quantile, and “highest” was defined as the highest UPF consumption quantile. ^a The unit of UPF consumption was % g/d. ^b The unit of UPF consumption was % kcal/d.

Figure 4

Subgroup analyses of summary random effect estimate with 95% confidence interval from meta-analyses evaluating the association between ultra-processed food consumption and metabolic disease risk. CI, confidence interval; FFQ, food frequency questionnaire; MetS, metabolic syndrome; NAFLD, non-alcoholic fatty liver disease; OR, odds ratio; RR, risk ratio; T2DM, type 2 diabetes mellitus. ^a “lowest” was defined as the lowest UPF consumption quantile, “moderate” was defined as the first exposure quantile, and “highest” was defined as the highest UPF consumption quantile. ^b The unit of UPF consumption was % kcal/d. ^c The unit of UPF consumption was % g/d.

Figure 5

Subgroup analyses of credible assessment of meta-analyses evaluating the association between ultra-processed food consumption and metabolic disease risk. FFQ, food frequency questionnaire; MetS, metabolic syndrome; NAFLD, non-alcoholic fatty liver disease; T2DM, type 2 diabetes mellitus. “Lowest” was defined as the lowest UPF consumption quantile, “moderate” was defined as the first exposure quantile, and “highest” was defined as the highest UPF consumption quantile. The color of the text reflected the change of credibility compared with the main analyses: black, unchanged; red, degraded; green, upgraded. ^a The unit of UPF consumption was % kcal/d. ^b The unit of UPF consumption was % g/d.

Figure 6

Sensitivity analyses of summary random effect estimate with 95% confidence interval from meta-analyses evaluating the association between ultra-processed food consumption and metabolic disease risk. CI, confidence interval; MetS, metabolic syndrome; NAFLD, non-alcoholic fatty liver disease; OR, odds ratio; RR, risk ratio; T2DM, type 2 diabetes mellitus. ^a “lowest” was defined as the lowest UPF consumption quantile, “moderate” was defined as the first exposure quantile, and “highest” was defined as the highest UPF consumption quantile. ^b The unit of UPF consumption was % g/d.