Ultra-processed food exposure and adverse health outcomes: umbrella review of epidemiological meta-analyses

Table of Contents

Overall Summary

Overview

This umbrella review of 14 meta-analyses (45 pooled analyses, ~10 million participants) found a statistically significant direct association between higher ultra-processed food (UPF) consumption and 71% (32/45) of examined health outcomes. Convincing evidence (Class I) was found for cardiovascular mortality, type 2 diabetes (dose-response), anxiety, and common mental disorders. Highly suggestive evidence (Class II) was found for all-cause mortality, heart disease mortality, type 2 diabetes (non-dose-response), depression, adverse sleep outcomes, wheezing, and obesity.

Key Points

Comprehensive Summary of Findings (written-content)
The abstract effectively summarizes key findings, including the number of analyses (45) and the overall association between UPF consumption and adverse health outcomes (71% of 32 parameters).
Section: Abstract
Define Evidence Classification and GRADE Framework (written-content)
While the abstract summarizes findings well, defining evidence classification criteria (Class I-V) and the GRADE framework would enhance clarity and context for a broader audience.
Section: Abstract
Clear Context and Relevance (written-content)
The introduction effectively establishes the context and relevance of the research by highlighting the global shift towards UPF consumption and its potential health implications.
Section: Introduction
Quantify UPF Consumption Prevalence (written-content)
While the introduction discusses increasing UPF consumption, quantifying its prevalence with specific data would strengthen the argument and underscore the research's urgency.
Section: Introduction
Comprehensive Search Strategy (written-content)
The methods section clearly outlines the search strategy, including databases (MEDLINE, PsycINFO, Embase, Cochrane), search terms, and date range (2009-June 2023), ensuring transparency and reproducibility.
Section: Methods
Clarify Disagreement Resolution Process (written-content)
While the methods mention consensus for resolving disagreements during screening, clarifying the specific process (e.g., third reviewer, specific criteria) would enhance transparency and reproducibility.
Section: Methods
Clear and Standard PRISMA Flowchart (graphical-figure)
Figure 1 (PRISMA flowchart) clearly and visually communicates the study selection process, adhering to standard terminology and structure.
Section: Results
Justify Exclusions in PRISMA Flowchart (graphical-figure)
While Figure 1 provides numbers, adding textual summaries of key exclusion reasons at each stage would enhance understanding of selection criteria.
Section: Results
Effective Synthesis and Implications (written-content)
The discussion effectively summarizes the main findings, connecting them to potential mechanisms and policy implications.
Section: Discussion
Formulate Specific Research Questions (written-content)
While the discussion mentions further research, formulating specific research questions would provide a roadmap for future studies and enhance the discussion's practical relevance.
Section: Discussion

Conclusion

This umbrella review synthesizes a substantial body of evidence, linking ultra-processed food consumption to adverse health outcomes. The review's methodological rigor, including a comprehensive search strategy, use of the PECOS framework, and dual assessment of evidence credibility and quality (using pre-specified criteria and GRADE), strengthens the reliability of its findings. The quantitative results, showing a statistically significant association between higher UPF consumption and 71% of the examined health outcomes (including convincing evidence for cardiovascular mortality, type 2 diabetes, anxiety, and common mental disorders), are compelling and underscore the potential public health implications. While the inherent limitations of umbrella reviews and observational studies are acknowledged, the review's comprehensive scope, large sample size (nearly 10 million participants), and transparent reporting contribute to its overall high quality of evidence. The findings significantly advance the field by providing a robust overview of existing evidence and highlighting critical areas for future research, particularly mechanistic studies and targeted interventions to reduce UPF consumption. The review's clear policy implications and actionable recommendations make it a valuable resource for informing public health strategies and promoting healthier dietary patterns.

Section Analysis

Abstract

Key Aspects

Strengths

Suggestions for Improvement

Introduction

Key Aspects

Strengths

Suggestions for Improvement

Methods

Key Aspects

Strengths

Suggestions for Improvement

Results

Key Aspects

Strengths

Suggestions for Improvement

Non-Text Elements

Fig. 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses...
Full Caption

Fig. 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flowchart.

First Reference in Text
The systematic search identified 430 de-duplicated articles (fig 1).
Description
  • Overall Process: The flowchart depicts the step-by-step process of identifying and selecting relevant studies for the umbrella review. It begins with the initial search across databases and other sources, followed by the removal of duplicates. Subsequently, studies are screened based on titles and abstracts, and then full-text articles are evaluated for eligibility. The flowchart quantifies the number of studies included and excluded at each stage, culminating in the final set of meta-analysis studies and discrete pooled analyses included in the review.
  • PRISMA Flowchart: The PRISMA flowchart is a standardized diagram used in systematic reviews and meta-analyses to visually represent the flow of information through the different phases of study selection. It provides a transparent and replicable method for documenting the search strategy, screening process, and reasons for excluding studies.
  • Numbers at Each Stage: The flowchart starts with 668 records identified from databases and registers. After removing 238 duplicates, 430 records were screened. From these, 383 were excluded during the screening phase based on titles and abstracts. Of the remaining 47, one study could not be retrieved, leading to the assessment of 46 full-text articles. Out of these, 33 were excluded for various reasons. Finally, 14 meta-analysis studies were included, encompassing 45 discrete pooled analyses.
Scientific Validity
  • Methodological Rigor: The use of the PRISMA flowchart is a standard practice in systematic reviews, enhancing the transparency and reproducibility of the study selection process.
  • Transparency: The flowchart clearly indicates the number of studies screened and excluded at each stage, providing a clear audit trail of the selection process. This level of detail allows readers to critically appraise the authors' decisions and assess the potential for selection bias.
  • Comprehensiveness: The inclusion of 45 discrete pooled analyses from 14 meta-analysis studies suggests a comprehensive search and selection process. This strengthens the review's scope and reduces the risk of publication bias by including a wide range of relevant studies.
Communication
  • Clarity and Visual Appeal: The flowchart is visually clear and easy to follow, effectively communicating the study selection process.
  • Standard Adherence: The use of standard PRISMA terminology and structure ensures that the information is presented in a recognizable and consistent manner for researchers familiar with systematic review methodology.
  • Justification of Exclusions: The flowchart could benefit from a brief, textual summary of the key reasons for exclusion at each stage. While the numbers are provided, a concise explanation of the major reasons for excluding studies would enhance the reader's understanding of the selection criteria and the characteristics of the excluded studies.
Fig. 2. Forest plot of non-dose-response relations between greater exposure to...
Full Caption

Fig. 2. Forest plot of non-dose-response relations between greater exposure to ultra-processed foods and risk of adverse health outcomes, with credibility and GRADE (Grading of Recommendations, Assessment, Development, and Evaluation) quality assessments.

First Reference in Text
Figure 2 and figure 3 show the direction and sizes of effect estimates using equivalent odds ratios for both the non-dose-response and dose-response relations between exposure to ultra-processed foods and each adverse health outcome, respectively.
Description
  • Forest Plot: This forest plot summarizes the findings of meta-analyses examining the associations between ultra-processed food consumption and various health outcomes. Each row represents a specific health outcome, such as all-cause mortality or type 2 diabetes. The plot displays the estimated effect size, represented as an equivalent odds ratio, for each outcome. The equivalent odds ratio is a standardized metric that allows for comparison across different types of original effect measures (e.g. hazard ratios, risk ratios, odds ratios). An odds ratio greater than 1 indicates that higher ultra-processed food consumption is associated with an increased risk of the outcome, while an odds ratio less than 1 indicates a decreased risk. The horizontal lines extending from each square represent the 95% confidence interval (CI) for the effect estimate. If the CI crosses 1, the association is not statistically significant. The size of the square represents the weight of each study in the meta-analysis, with larger squares indicating more influential studies.
  • Confidence Intervals: The forest plot visually represents the uncertainty associated with each effect estimate through the use of 95% confidence intervals. The width of the confidence interval reflects the precision of the estimate. Wider intervals indicate greater uncertainty, while narrower intervals suggest more precise estimates. If a confidence interval crosses the null value (1 for odds ratios), the result is not considered statistically significant, meaning there is not enough evidence to conclude that there is a true association between ultra-processed food consumption and the outcome.
  • Credibility Assessment: The color of each square corresponds to the credibility of the evidence, based on a pre-defined classification system (Class I: Convincing, Class II: Highly Suggestive, Class III: Suggestive, Class IV: Weak, Class V: No evidence). This classification system assesses the strength of the evidence based on several factors, such as the p-value, heterogeneity (I-squared statistic), and potential biases.
  • GRADE Quality Assessment: The GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach is used to assess the quality of the evidence for each outcome. GRADE considers factors such as risk of bias, inconsistency, indirectness, imprecision, and publication bias. The quality of evidence is categorized as high, moderate, low, or very low.
  • 'k' Value: The 'k' value indicates the number of original research articles included in the meta-analysis for each health outcome. This provides an indication of the amount of data supporting the effect estimate, with higher 'k' values suggesting more robust evidence.
Scientific Validity
  • Standardized Effect Measure: The use of equivalent odds ratios facilitates comparisons across different studies that may have used different effect measures, improving the overall interpretability of the findings.
  • Comprehensive Quality Assessment: The inclusion of both credibility and GRADE assessments provides a comprehensive evaluation of the strength and quality of the evidence for each outcome. This allows readers to critically appraise the findings and consider the limitations of the available data.
  • Visual Representation of Uncertainty: The visual presentation of confidence intervals allows for an immediate assessment of the statistical significance of each effect estimate. This is crucial for determining the reliability and generalizability of the findings.
  • Transparency and Potential for Bias: Providing the 'k' value enhances transparency and allows readers to understand the volume of research supporting each outcome. However, 'k' alone doesn't guarantee the quality or consistency of the included studies. The authors should have elaborated on the assessment of the methodological quality of the included meta-analyses using a validated tool like AMSTAR-2.
Communication
  • Visual Presentation of Results: The forest plot effectively visually presents the results of the meta-analyses for each outcome, enabling quick comparison of effect sizes and confidence intervals. The color-coding for credibility and the symbols for GRADE quality add another layer of information, facilitating the interpretation of the strength and quality of evidence for each outcome.
  • Clarity regarding Equivalent Odds Ratios: While the forest plot clearly displays the equivalent odds ratios, the caption should explicitly state that these are equivalent odds ratios and not the original effect measures reported in the included studies. This clarification is crucial to avoid misinterpretation of the results, especially for readers unfamiliar with the concept of equivalent odds ratios as a way to standardize different effect measures for comparative purposes.
  • Explanation of Rating Systems: Consider adding a brief explanation within the figure or caption about the meaning of the different credibility classes (I-V) and GRADE quality levels (high, moderate, low, very low). This would enhance the accessibility of the plot for readers who may not be familiar with these specific rating systems.
Fig. 3. Forest plot of dose-response relations between greater exposure to...
Full Caption

Fig. 3. Forest plot of dose-response relations between greater exposure to ultra-processed foods and risk of adverse health outcomes, with credibility and GRADE (Grading of Recommendations, Assessment, Development, and Evaluation) quality assessments.

First Reference in Text
Figure 2 and figure 3 show the direction and sizes of effect estimates using equivalent odds ratios for both the non-dose-response and dose-response relations between exposure to ultra-processed foods and each adverse health outcome, respectively.
Description
  • Dose-response Forest Plot: This forest plot summarizes the dose-response relationships between ultra-processed food consumption and various health outcomes. Each row corresponds to a specific health outcome, and the plot displays the equivalent odds ratio, a standardized effect measure allowing for comparison across different types of original effect measures. An equivalent odds ratio greater than 1 signifies that each increment of ultra-processed food intake is associated with an increased risk of the outcome, while a value less than 1 indicates a decreased risk with each increment. The horizontal lines represent the 95% confidence intervals (CIs) for the effect estimates. If the CI crosses 1, the dose-response relationship is not statistically significant.
  • Credibility, GRADE, and 'k' value: Similar to Figure 2, the color of each square represents the credibility of the evidence (Class I to V), while different symbols indicate the GRADE quality of the evidence (high, moderate, low, or very low). 'k' denotes the number of original research articles included in each meta-analysis.
  • Confidence Intervals: The confidence intervals (CIs) shown on the plot visually represent the precision of the effect estimates. A wider CI indicates more uncertainty, while a narrower CI suggests a more precise estimate. If a CI crosses the null value (1 for equivalent odds ratios), it signifies that the dose-response relationship is not statistically significant.
Scientific Validity
  • Dose-response Analysis: The focus on dose-response relationships strengthens the analysis by exploring the association across different levels of ultra-processed food consumption, providing a more nuanced understanding of the potential health impacts compared to simple categorical comparisons.
  • Standardized Effect Measure Considerations: The use of equivalent odds ratios, while facilitating comparison, requires careful consideration of the underlying assumptions and limitations of converting different effect measures to a common scale. The authors should have clearly outlined the method used for this conversion and discussed any potential implications for the interpretation of the results.
  • Assessment of Bias and Confounding: The inclusion of credibility and GRADE assessments adds valuable information about the strength and quality of the evidence. However, the authors should provide more details on the criteria used for these assessments and how potential biases, such as publication bias and confounding, were addressed in the included meta-analyses.
Communication
  • Visual clarity and accessibility: The forest plot effectively communicates the dose-response relationships by visually presenting the equivalent odds ratios and their confidence intervals. The color-coding and symbols for credibility and GRADE assessments, respectively, enhance the plot's interpretability by providing a clear visual representation of the strength and quality of the evidence for each outcome. The inclusion of an interactive version enhances accessibility and allows for detailed exploration of the data.
  • Clarity and precision of language: While the use of equivalent odds ratios facilitates comparison across studies, the caption should explicitly state that these are equivalent odds ratios and not the original effect measures. This is essential to avoid misinterpretation, especially for readers unfamiliar with the concept of equivalent odds ratios. Furthermore, a brief explanation of how the dose-response was modeled (e.g., per serving, per 10% increase) would improve clarity.
  • Accessibility for a broader audience: The plot would benefit from a brief explanation of the credibility classes (I-V) and GRADE levels directly within the figure or caption. This would make the plot more self-explanatory and accessible to a broader audience, including those not familiar with these specific assessment tools.

Discussion

Key Aspects

Strengths

Suggestions for Improvement

Non-Text Elements

Fig. 4. Credibility and GRADE (Grading of Recommendations, Assessment,...
Full Caption

Fig. 4. Credibility and GRADE (Grading of Recommendations, Assessment, Development, and Evaluation) ratings for associations between greater exposure to ultra-processed foods and risks of each adverse health outcome

First Reference in Text
Considering the evidence classification criteria assessments, we graded 9% of the pooled analyses as providing convincing evidence (class I), including those measuring risks of cardiovascular disease related mortality, common mental disorder outcomes, and type 2 diabetes (dose-response) (fig 4).
Description
  • Overall Structure and Purpose: This figure summarizes the credibility of evidence and GRADE quality assessments for associations between ultra-processed food consumption and various adverse health outcomes. It uses a stylized human figure as a visual aid to organize the outcomes by body system (e.g., mortality, cancer, cardiovascular). Each listed outcome has two associated ratings: one for credibility and one for GRADE quality.
  • Credibility Ratings: The credibility assessment categorizes the evidence into five classes (I to V), representing the strength of the association. Class I is 'Convincing' evidence, while Class V signifies 'No evidence'. The classes between represent varying degrees of suggestive or weak evidence.
  • GRADE Quality Ratings: The GRADE (Grading of Recommendations Assessment, Development and Evaluation) system assesses the quality of the evidence based on multiple factors, including risk of bias, inconsistency, indirectness, imprecision, and publication bias. It classifies the quality into four levels: high, moderate, low, and very low.
  • Visual Representation of Credibility: The figure visually represents the credibility assessments using a color-coded system, with darker shades of red corresponding to more convincing evidence (Class I and II) and lighter shades for weaker evidence (Class III and IV). Grey represents no evidence (Class V).
  • Visual Representation of GRADE: The GRADE quality assessments are shown using symbols: a filled circle for high quality, a partially filled circle for moderate, an empty circle for low, and an empty square with a cross inside for very low quality.
  • Dose-Response vs. Non-Dose-Response: The figure also distinguishes between dose-response and non-dose-response relationships for certain outcomes, indicating whether the analysis examined the effect of varying levels of consumption (dose-response) or just the presence/absence or high/low categories of consumption (non-dose-response). A "/" separates the two ratings where applicable, with the dose-response rating appearing first.
Scientific Validity
  • Dual Assessment Framework: Presenting both credibility and GRADE assessments provides a comprehensive evaluation of the evidence, allowing readers to understand both the strength and quality of the associations. This dual assessment framework enhances the rigor of the review.
  • Organization and Overview: The visual representation of the ratings by body system provides a clear overview of the evidence base across different health areas. This facilitates the identification of areas where strong evidence exists and those requiring further research.
  • Transparency and Clarity of Criteria: The specific criteria used for the credibility assessment (Classes I-V) are not clearly defined in the figure, reducing the transparency of the evaluation process. A brief explanation of the criteria used for assigning each class would enhance the scientific rigor.
  • Dose-Response Differentiation: The distinction between dose-response and non-dose-response relationships is a valuable addition, highlighting the importance of considering the level of consumption when evaluating the health impacts of ultra-processed foods.
Communication
  • Visual presentation and clarity: Visually summarizing the credibility and GRADE assessments for each health outcome alongside a human figure adds an intuitive layer of interpretation and facilitates quick identification of areas with stronger or weaker evidence. The color-coding and symbols for credibility and GRADE enhance visual distinction and rapid understanding.
  • Standalone readability and context: While the figure is generally clear, providing a brief explanation of the credibility and GRADE rating systems within the figure itself would improve its standalone readability. A legend defining the color codes and symbols, or a short description of the rating criteria, would be beneficial.
  • Potential for misinterpretation: The human figure enhances the visual appeal and thematic connection to health outcomes, but it could be misinterpreted as suggesting causal links or specific physiological pathways, which may not be fully supported by the evidence. Consider revising the figure to emphasize that the associations are observational and not necessarily causal.

Conclusions

Key Aspects

Strengths

Suggestions for Improvement

↑ Back to Top