https://orcid.org/0000-0002-9294-1056
Abstract
Observational research suggests higher red and processed meat intakes predict greater risks of developing or dying from cardiovascular diseases (CVD) and type 2 diabetes mellitus (T2DM), but this research limits causal inference. This systematic review of reviews utilizes both observational and experimental research findings to infer causality of these relations. Reviews from four databases were screened by two researchers. Reviews included unprocessed red meat (URM), processed meat (PM), or mixed URM + PM intake, and reported CVD or T2DM outcomes. Twenty-nine reviews were included, and causality was inferred using Bradford Hill’s Criteria. Observational assessments of CVD outcomes and all meat types consistently reported weak associations while, T2DM outcomes and PM and Mixed URM + PM assessments consistently reported strong associations. Experimental assessments of Mixed URM + PM on CVD and T2DM risk factors were predominately not significant which lacked coherence with observational findings. For all meat types and outcomes, temporality and plausible mechanisms were established, but specificity and analogous relationships do not support causality. Evidence was insufficient for URM and T2DM. More experimental research is needed to strengthen these inferences. These results suggest that red and processed meat intakes are not likely causally related to CVD but there is potential for a causal relationship with T2DM.
Introduction
The 2020–2025 Dietary Guidelines for Americans recommends eating patterns that are low in “red and processed meat intake” with the caveat that “lean meats” can be part of a healthy eating pattern (as mainly fresh, frozen or canned versus further processed) when recommendations for sodium and saturated fat intakes are met (DGA Citation2020). These recommendations are consistent with others made by the American Heart Association (Lichtenstein et al. Citation2021) and American Diabetes Association (Evert et al. Citation2019). Yet, these recommendations are based largely on observational research (i.e., prospective cohorts) that show inconsistent associations and often do not include evidence from short term experimental studies (i.e., randomized controlled trials (RCT), including controlled feeding) (DGA Citation2015; O’Connor and Campbell Citation2017; DGA Citation2020). Observational studies are well suited to assess associations of self-reported dietary intake with incidence and related mortality of cardiometabolic diseases (CMD) such as cardiovascular disease (CVD) or type 2 diabetes mellitus (T2DM) over time. However, residual and unmeasured confounding, as well as documented biases in self-reported dietary data hinder capacity for determining causality (Shim, Oh, and Kim Citation2014; Subar et al. Citation2015). Alternatively, randomized controlled feeding trials can assess causality between a specific exposure and disease risk factors despite typically having smaller sample sizes and shorter durations (Weaver and Miller Citation2017). Assessing the relation between red and processed meat intake on CMD using evidence from both observational and randomized controlled feeding trials would advance our understanding of whether these relationships are likely causal or not (Beyerbach et al. Citation2022).
The Bradford Hill Causality Criteria were developed to infer the potential for causal relations of public health concern, such as smoking and lung cancer, by interpreting findings from observational research in conjunction with experimental evidence if available (Hill Citation1965). These criteria can be operationalized by researchers and public health professionals to elicit an inference, which is a conclusion reached based on evidence and critical thinking. These long-established criteria are adapted to inform nutrition-related public health recommendations (WCRF Citation2018). Independent researchers have also applied modified versions of Bradford Hill’s Causality Criteria to infer causality between several dietary factors and disease endpoints, encompassing individual nutrients, whole foods, eating patterns, or other dietary exposures (Mente et al. Citation2009; Micha et al. Citation2017; Livesey and Livesey Citation2019; Livesey et al. Citation2019). Such an approach could be applied to the body of literature about red and processed meat intakes and risk for cardiometabolic disease to provide further clarity for future recommendations. There is a plethora of long-term observational studies assessing red and processed meat intake and clinical endpoints (e.g., stroke, myocardial infarctions) as well as shorter-term randomized controlled feeding trials assessing established clinical risk factors (e.g., blood pressure, low-density lipoproteins). Yet, it is uncommon that both types of study designs are considered together in evidence reviews for dietary recommendations (DGA Citation2015; USDA Citation2015; DGA Citation2020; USDA Citation2020).
The objective of this umbrella (Aromataris et al. Citation2015) systematic review is to apply the Bradford Hill Causality Criteria (Hill Citation1965) to infer the potential for causal relations between intakes of red and processed meat and CVD or T2DM using systematic reviews and meta-analyses of both observational and experimental research, including randomized controlled feeding trials, on the topic. In addition, AMSTAR 2, a systematic review appraisal tool (Shea et al. Citation2017), is used to assess the quality of evidence.
Methods
Search strategy
A preliminary search strategy and database selection (Supplemental Table S1) were developed prior to conducting the search. The databases searched were: (1) PubMed, (2) Scopus, (3) Cochrane Database of Systematic Reviews, and (4) Cumulative Index of Nursing and Allied Health Literature (CINAHL). Searches were conducted without filters or restrictions and the reference lists of the included articles were reviewed for additional eligible articles. Grey literature was included in our search, but only peer-reviewed systematic reviews or meta-analyses were of interest. A health sciences research librarian (BM) coordinated the development of the search strategy. The protocol and methods were designed following the PRISMA reporting checklist (Moher et al. Citation2009) and registered at PROSPERO (PROSPERO Citation2011), the International Prospective Registrar of Systematic Reviews (the review protocol can be accessed at https://www.crd.york.ac.uk/prospero/ using ID number: CRD42020127976). Before registration, preliminary searches and pilot testing was conducted, but the formal screening process did not begin until after registration. The search was conducted up to July 2021 and there were no significant deviations from our approved PROSPERO protocol.
Table 1. Description of PICOS criteria for an umbrella review assessing causal relations between red and processed meat intake and cardiometabolic disease.
Search criteria
Our research question aimed to address whether intakes of red and processed meat are causally related to CVD or T2DM in generally healthy adults. The PICOS (Population, Intervention/Exposure, Comparator, Outcome, and Study Design) (Aslam and Emmanuel Citation2010) criteria applied to the research question are outlined in . We conducted an “umbrella systematic review,” i.e., a systematic review of systematic reviews or meta-analyses of observational and/or experimental literature (Aromataris et al. Citation2015). Throughout this manuscript we use the abbreviation “OBS” to identify outcomes from observational studies, which are mostly prospective longitudinal cohort studies reporting relative risks (RRs). We use the abbreviation “RCT” to identify outcomes from experimental studies, which are mostly randomized controlled feeding trials reporting mean differences or change values.
Inclusion criteria were determined a priori and were: (1) systematic reviews or systematically searched meta-analyses of cohort, experimental, and/or cross-sectional design; (2) red meat consumption as an a priori independent exposure or variable as listed in the search terms, objectives, or methods sections of the reviews; (3) report of CVD and/or T2DM risk factors (RF), incidence, or mortality outcomes, as described in the next section; (4) populations without a diagnosis of cardiometabolic disease of interest at enrollment; and (5) a participant group mean age ≥19 years old. Exclusion criteria were: (1) narrative reviews or individual primary research studies; (2) systematic reviews that did not include red meat consumption as an a priori independent exposure or variable; (3) systematic reviews of specific clinical or diseased populations; (4) population with a mean age < 19 years old; (5) animal or in vitro models; (6) reviews not published in English; and (7) meta-analytical approaches that lacked systematic assessment of included studies. The references and rationales for excluding articles after full text review are presented in Supplemental Table S2.
Table 2. The Bradford Hill causality criteria descriptions (Hill Citation1965) for an umbrella review assessing red and processed meat intake and cardiometabolic disease.
Screening and data abstraction
Two researchers (ERH and YW) independently screened and crosschecked each article from the literature search, and a third researcher (CMC) was consulted to settle any discrepancies. Two researchers (ERH and YW) independently abstracted and crosschecked the following data from each included review: (1) author names, (2) publication year, (3) detailed outcome, (4) type of meat reported/consumed, (5) definition of meat type, (6) summary estimates, (7) 95% confidence intervals (CI), (8) heterogeneity measures (I2 value), (9) details of a dose-response analyses when available, (10) number and type of primary research studies in the systematic review or meta-analysis articles, (11) participant characteristics, and (12) funding source. Authors were not contacted because additional information was not needed.
Definitions for the meat type categories
Due to inconsistent use and description of meat terminology in chronic disease research (Gifford, O'Connor et al. Citation2017; O'Connor et al. Citation2020), we re-categorized meat exposures (i.e., consumption of skeletal muscle and associated tissue of mammal, avian, or aquatic species)(Seman et al. Citation2018) that were reported in the systematic reviews and meta-analyses. We used the following categories: Unprocessed Red Meat (URM), Processed Meat (PM), and Mixed Unprocessed Red Meat and Processed Meat (Mixed URM + PM) based on the 2015–2020 Dietary Guidelines for Americans (DGA) definitions (DGA Citation2015), as this was the current DGA version at study inception. These descriptions are consistent with those used for the 2020–2025 DGA as well (DGA Citation2020). The 2015–2020 DGA defined red meat as “all forms of beef, pork, lamb, veal, goat, and non-bird game (e.g., venison, bison, elk)” (DGA Citation2015). Processed meat was defined as red meat or poultry that is “preserved by smoking, curing, salting, and/or the addition of chemical preservatives” (DGA Citation2015). By default, we defined unprocessed red meat as all forms of beef, pork, lamb, veal, goat, and non-bird game (e.g., venison, bison, elk) not preserved by smoking, curing, salting, and/or the addition of chemical preservatives. Most of the included reviews did not disaggregate processed meat into independent processed poultry and processed red meat intake categories. About 30% of the processed meat consumed in the US is processed poultry (O'Connor, Wambogo, et al. Citation2021) therefore, we assume that the PM category in our review includes both processed red meat and processed poultry. Unprocessed poultry fell outside of the scope of our umbrella review. If included reviews used terms such as “red meat,” “red and processed meat,” or “total red meat” without providing detailed descriptions to determine degree of processing, then these reviews were categorized into the Mixed URM + PM category due to ambiguity. Therefore, the Mixed URM + PM category is mutually exclusive from the URM and PM categories and is not a combination of the two. See Supplemental Table S3 which includes the meat terminology and accompanied descriptions provided in the included reviews and how those were then re-categorized for our analysis into URM, PM, or Mixed URM + PM.
Table 3. Bradford Hill causality criteria summary for all meat type categories and CVD and T2DM outcomes.
Exposure and outcome stratification
In this review, exposures and outcomes were stratified by meat type: Mixed URM + PM, PM, and URM, and then further categorized by outcome (see Supplemental Table S3): CVD RF (OBS), CVD RF (RCT), CVD Incidence, CVD Mortality, T2DM RF (OBS), T2DM RF (RCT), T2DM Incidence and, T2DM Mortality. We are mindful that CVD encompasses more specific diagnoses (i.e., myocardial infarction, stroke, ischemic heart disease, etc.), however there were insufficient reviews to meaningfully assess theses outcomes individually.
The following is an example of how we categorized and stratified exposures and outcomes. Bechthold et al conducted independent analyses for “red meat” and “processed meat” intake and CVD Incidence (specific outcomes included: CHD, heart failure (HF), and stroke) totaling six different analyses (Supplemental Table S4) (Bechthold et al. Citation2019). We categorized the “red meat” exposure into the Mixed URM + PM category because no explicit definition was provided for red meat (i.e., no mention of degree of processing) and we categorized the “processed meat” exposure into the PM category. Each analysis was treated as a separate point estimate and stratified by type of meat exposure and type of outcome. We refer to each of these point estimates as independent “assessments” throughout this manuscript. We refer to the entirety of each systematic review or meta-analysis included in our analysis as a “review.” See for a full breakdown of the number of reviews and assessments by meat type and outcome category.
Figure 1. Article screening process flow chart for an umbrella review assessing red and processed meat intake and cardiometabolic disease.
Table 4. Mixed unprocessed red meat + processed meat and CVD outcomes: Bradford Hill causality criteria and AMSTAR 2 quality findings.
Causal inference assessment
The Bradford Hill Causality Criteria were described in 1965 to help infer the likely causal association between smoking and lung cancer (Hill Citation1965). These criteria have since been adapted by researchers and policymakers to develop guidelines of causal inference for public health purposes (Mente et al. Citation2009; Micha et al. Citation2017; WCRF Citation2018; Livesey and Livesey Citation2019; Livesey et al. Citation2019). The criteria are used to develop an inference, which is a conclusion reached based on evidence and reasoning (Rothman and Greenland Citation2005). Long term randomized controlled feeding trials assessing effects of red and processed meat intake on chronic disease outcomes (e.g., stroke, myocardial infarction) would be preferred to infer causality but this study design is infeasible and often unethical (Todt and Luján Citation2017). Therefore, observational studies are relied upon for dietary guidance purposes (DGA Citation2015; DGA Citation2020). It is practical to assess effects of red and processed meat intake via a randomized controlled feeding trial with shorter-term risk factors as outcomes. However, due to the short-term nature of the approach, this study design is not often included in evidence reviews for dietary guidelines development (DGA Citation2015; DGA Citation2020). Thus, the Bradford Hill Causality Criteria are well suited to infer causal relations among both longer-term observational studies and shorter-term randomized controlled feeding trials. The Causality Criteria include (1) Strength, (2) Consistency, (3) Specificity, (4) Temporality, (5) Biological Gradient, (6) Plausibility, (7) Experiment, (8) Coherence, and (9) Analogy () (Hill Citation1965). The Causality Criteria were dichotomized as “Supported” or “Not supported” based on how the evidence met each of the criteria. However, all nine Causality Criteria were considered collectively to infer potential causal relations between each meat type and outcome category.
In this umbrella review, we are including systematic reviews and meta-analyses of observational and experimental studies. Observational studies typically assess risk by conducting analyses comparing the highest percentile vs. lowest percentile of intake or via dose-response continuous intake analyses. To consider each type of evidence under the appropriate criterion, the point estimates for high vs. low assessments are summarized in the Strength Criterion and the point estimates for dose-response assessments are summarized in the Biological Gradient Criterion. Reviews of experimental studies, i.e., randomized controlled feeding trials, reporting on risk factor outcomes are summarized in the Experiment Criterion.
Strength
As previously stated, this criterion includes point estimates that are comparing the high vs. low percentiles of intake from reviews of observational studies. In the included reviews, statistical significance (95% confidence interval does not cross 1.0) is reported for these assessments. However, with regards to inferring causality, statistical significance is not necessarily indicative of a strong association (Sorkin et al. Citation2021). Therefore, we used an a priori cut off: RR ≥1.2 to indicate a strong association. A 20% increased risk is biologically meaningful at the population level and considers potential attenuation of associations due to biases or poor control of confounding factors (Byers and Lyle Citation1999; NCEP Citation2001).
Consistency
We included high vs. low assessments from reviews of observational studies which included different prospective cohort populations from varying geographic locations and time periods. We considered associations to be consistent if ≥ 67% of reported high vs. low assessments had a RR ≥ 1.2 (Mente et al. Citation2009).
Biological gradient
This criterion includes dose-response or trend assessments from reviews of observational studies. We documented the percentage of assessments that reported a significant dose-response curve and if the shape of the relationship (i.e., linear, non-linear, J-shape, etc.) was reported then it was described in Supplemental Material.
Experiment
This criterion includes assessments on risk factor outcomes from reviews of experimental studies, i.e., randomized controlled feeding trials. We reported the percentage of assessments that the 95% CI of the mean differences did not include 0.0 between higher and lower meat intakes.
Coherence
We compared the results from the Strength Criterion and Experiment Criterion to infer the presence of coherence across study designs.
Plausibility
We qualitatively describe various proposed biological mechanisms between the three meat categories and CVD or T2DM outcomes.
Temporality
Only reviews of prospective observational and experimental studies were included to meet this criterion.
Specificity
We qualitatively describe notable relationships between red meat and processed meat intake and other health outcomes.
Analogy
We qualitatively described associations between other animal-based protein sources (white meat, dairy, seafood, and eggs) and CVD and T2DM outcomes, then cross-compared those associations with our results from red and processed meat.
Risk of bias and quality analysis
AMSTAR 2 (A MeaSurement Tool to Assess systematic Reviews) was applied to assess study quality and risk of bias of the included systematic reviews and meta-analyses as well as our own umbrella review. AMSTAR 2 is used to assess systematic reviews and meta-analyses of both randomized and non-randomized trials through a series of 16 questions (Shea et al. Citation2007; Shea et al. Citation2017). Per AMSTAR 2 protocol, reviews were given quality ratings of either High, Moderate, Low, or Critically Low based on the presence of critical or non-critical biases. This tool assesses the quality of the reviews but not the quality of the included primary research studies. Two researchers (ERH and YW) independently entered and crosschecked data for the AMSTAR 2 assessment and a third researcher (CMC) was consulted to settle disputes. The appropriateness of using AMSTAR 2 as a tool to assess quality of evidence in nutrition research is supported by its use by the 2015 Dietary Guidelines Advisory Committee (USDA Citation2015).
Publication bias assessment
Publication bias of the included reviews was assessed for each meat type and outcome category. Natural log-transformed 95% CI for pooled Relative Risk (RR) and 95% CI for mean differences from each review were used to determine the standard error of summary risk estimates using the delta method (Feiveson, University Citation2017). Funnel plots and Egger’s tests were computed to assess publication bias using STATA SE software (version 17.0) with statistical significance level set at 5%. This is a commonly used method to assess publication bias among meta-analyses but these funnel plots are subject to certain limitations and results may be explained by other types of biases other than publication bias (Terrin, Schmid, and Lau Citation2005). Publication bias was assessed for all quantifiable assessments but was done separately for the highest vs. lowest, dose-response, and risk factor assessments. Assessment of publication bias is a criterion of the AMSTAR 2 assessment to be considered a High quality review.
Results
Of the 605 articles screened, 133 full text articles were assessed for eligibility and 29 systematic reviews and meta-analyses (each referred to as a “review”) were included in this umbrella systematic review. The references and rationales for excluding articles after full text review are presented in Supplemental Table S2. presents a flow chart of the number of reviews and assessments. “Assessments” are referred to as individual point estimates between a meat exposure and type of outcome. Twenty-two reviews included only observational studies (Aune, Ursin, and Veierød Citation2009; Micha, Wallace, and Mozaffarian Citation2010; Kaluza, Wolk, and Larsson Citation2012; Akesson et al. Citation2013; Chen et al. Citation2013; O'Sullivan et al. Citation2013; Abete et al. Citation2014; Lippi, Mattiuzzi, and Sanchis-Gomar Citation2015; Wang et al. Citation2016; Yang et al. Citation2016; Kim et al. Citation2017; Schwingshackl, Hoffmann, et al. Citation2017; Schwingshackl, Schwedhelm, et al. Citation2017; Tian et al. Citation2017; Kim and Je Citation2018; Bechthold et al. Citation2019; Fan et al. Citation2019; Schlesinger et al. Citation2019; Vernooij et al. Citation2019; Zeraatkar, Han, et al. Citation2019; Yang et al. Citation2020; Papier et al. Citation2021) and seven reviews included only experimental studies (Maki et al. Citation2012; O'Connor, Kim et al. Citation2017; Schwingshackl et al. Citation2018; Guasch-Ferré et al. Citation2019; Zeraatkar, Johnston et al. Citation2019; An, Liu, and Liu Citation2020; O'Connor, Kim, et al. Citation2021). Notably, no review assessed observational and experimental studies together. Of the 29 included reviews, the following outcomes were examined: 15 with only CVD (Kaluza, Wolk, and Larsson Citation2012; Maki et al. Citation2012; Akesson et al. Citation2013; Chen et al. Citation2013; O'Sullivan et al. Citation2013; Abete et al. Citation2014; Lippi, Mattiuzzi, and Sanchis-Gomar Citation2015; Wang et al. Citation2016; Yang et al. Citation2016; Kim et al. Citation2017; O'Connor, Kim et al. Citation2017; Schwingshackl, Schwedhelm, et al. Citation2017; Bechthold et al. Citation2019; Guasch-Ferré et al. Citation2019; Papier et al. Citation2021), six with only T2DM (Aune, Ursin, and Veierød Citation2009; Schwingshackl, Hoffmann, et al. Citation2017; Tian et al. Citation2017; Fan et al. Citation2019; Yang et al. Citation2020; O'Connor, Kim, et al. Citation2021), and eight with both CVD and T2DM (Micha, Wallace, and Mozaffarian Citation2010; Kim and Je Citation2018; Schwingshackl et al. Citation2018; Schlesinger et al. Citation2019; Vernooij et al. Citation2019; Zeraatkar, Han, et al. Citation2019; Zeraatkar, Johnston et al. Citation2019; An, Liu, and Liu Citation2020). Three reviews were strictly systematic reviews without a meta-analysis (Akesson et al. Citation2013; Lippi, Mattiuzzi, and Sanchis-Gomar Citation2015; Schwingshackl et al. Citation2018).
Presentation of results and synthesis of quality assessment
We first present Strength, Consistency, Biological Gradient, Experiment, Coherence, and AMSTAR 2 results for each meat type and outcome stratum for CVD. We then describe post- hoc how inferences change when including only reviews that had AMSTAR 2 ratings of High and Moderate quality to see if inferences would change based on the quality of the reviews included. Plausibility is then described globally for all meat types and CVD. This same format is followed for T2DM. Lastly, we present results for the Temporality, Specificity, and Analogy Criteria for CVD and T2DM combined because qualitative operationalization of these is not specific to either disease or type of meat. We summarize whether each Bradford Hill Causality criterion was supported or not supported for each meat/outcome stratum in , and compile conclusions of causal inference in . show the findings from the Strength, Consistency, Biological Gradient, Experiment, Coherence and AMSTAR 2 quality analyses. Supplemental Table S4 presents population characteristics, meat terminology and description, and outcomes of each review. Supplemental Table S5 presents the AMSTAR 2 assessment and ratings for each included review. Additionally, our review meets the AMSTAR 2 criteria for a High quality review (Shea et al. Citation2017; Supplemental Table S5.1). Supplemental Tables S6–S20 present the individual assessments used for the Strength, Biological Gradient, and Experiment Criteria and include indication of statistical significance, I2 heterogeneity metric, AMSTAR 2 quality of each assessment, and the shape of the dose-response assessments, if available.
Table 5. Processed meat and CVD outcomes: Bradford Hill causality criteria and AMSTAR 2 quality findings.
Table 6. Unprocessed meat and CVD outcomes: Bradford Hill causality criteria and AMSTAR 2 quality findings.
Table 7. Mixed unprocessed red meat + processed meat and T2DM outcomes: Bradford Hill causality criteria and AMSTAR 2 quality findings.
Table 8. Processed meat and T2DM outcomes: quantitative Bradford Hill causality criteria and AMSTAR 2 quality findings.
Table 9. Unprocessed red meat and T2DM outcomes: quantitative Bradford Hill causality criteria and AMSTAR 2 quality findings.
Table 10. Bradford Hill causality inference summary table for an umbrella review assessing red and processed meat intake and cardiometabolic disease.
Cardiovascular disease: strength, consistency, biological gradient, experiment, coherence, and quality
Mixed unprocessed red meat + processed meat
shows the Mixed URM + PM and CVD findings from the Strength, Consistency, Biological Gradient, Experiment, Coherence and AMSTAR 2 quality analyses. Strength and Consistency: Twenty high vs. low assessments from 10 reviews of observational studies were reported. Eleven of the 20 assessments (55%) were statistically significant, yet only two of the 20 assessments (10%) had an effect size ≥1.2 (Supplemental Table S6). Biological Gradient: Seventeen dose-response assessments from 11 reviews of observational studies were reported. Fourteen of the 17 assessments (∼82%) were statistically significant but the shape of this relationship was not consistently reported (Supplemental Table S7). Four of the 11 reviews reported a visual representation of the relationship shape with two reviews reporting a linear and two reviews reporting a nonlinear relationship (Supplemental Table S7). Experiment: Thirty-one assessments from four reviews of experimental studies were reported. Only eight of the 31 assessments (∼26%) were statistically significant, suggesting experimental evidence does not support a relationship between intakes of Mixed URM + PM and CVD risk factors (Supplemental Table S8). Coherence: There is a lack of coherence between the observational findings and experimental evidence, i.e., observational findings reported statistically significant and positive associations between Mixed URM + PM and CVD outcomes, whereas experimental evidence showed mostly non-significant effects on CVD risk factors. These findings are supported by two additional systematic reviews of observational studies identified in our search that lacked quantifiable analyses. These reviews reported that inconsistent evidence, or an overall lack of evidence contributed to an unclear relationship between mixed unprocessed red meat and processed meat intake and CVD incidence and mortality (Akesson et al. Citation2013; Lippi, Mattiuzzi, and Sanchis-Gomar Citation2015).
AMSTAR 2 quality
Among 17 total reviews assessing observational CVD outcomes, four reviews (∼24%) were rated as High or Moderate quality and contributed five assessments, while 13 reviews (∼76%) were rated as Low or Critically Low quality and contributed 22 assessments (Supplemental Tables S6, S7). Among four reviews that assessed experimental CVD risk factors, two (50%) were rated as High or Moderate quality and had 13 assessments, while the other two reviews (50%) were rated as Critically Low quality and contributed 18 assessments (Supplemental Table S8). Restricting to only Moderate and High quality reviews further supported the Strength and Consistency Criteria but the Biological Gradient Criterion was not supported (Supplemental Tables S6–S8).
Processed meat
shows the PM and CVD findings from the Strength, Consistency, Biological Gradient, Experiment, Coherence and AMSTAR 2 quality analyses. Strength and Consistency: Eleven high vs. low assessments from eight reviews of observational studies were reported. Nine of the 11 total assessments (∼82%) reported a statistically significant association but only two of the 11 total assessments (∼18%) had an effect size ≥1.2 (Supplemental Table S9). Biological Gradient: Eighteen dose-response assessments from 11 reviews of observational studies were reported. Among those 18 assessments, 15 (∼83%) were statistically significant (Supplemental Table S10). Three of the 11 reviews reported a visual representation of the relationship shape with one review reporting a linear and two reviews reporting a nonlinear relationship (Supplemental Table S10). Experiment and Coherence: No review reported findings from experimental studies on processed meat and CVD risk factors so the Experimental and Coherence Criteria could not be assessed.
AMSTAR 2 quality
Among 13 reviews with observational CVD outcomes, four (∼30%) were rated as High or Moderate quality and contributed five assessments. The remaining nine reviews (∼70%) were rated as Low or Critically Low quality and contributed 14 assessments (Supplemental Tables S9, S10). Restricting to only High and Moderate quality reviews did not change results (Supplemental Tables S9, S10).
Unprocessed red meat
shows the URM and CVD findings from the Strength, Consistency, Biological Gradient, Experiment, Coherence and AMSTAR 2 quality analyses. Strength and Consistency: Six high vs. low assessments from four reviews of observational studies were reported. Four of the six assessments (∼67%) reported statistically significant associations but of the total assessments, none had an effect size ≥1.2 (Supplemental Table S11). Biological Gradient: Eleven dose-response assessments from seven reviews of observational studies were statistically significant. Only one review reported a visual representation of the relationship shape, and it was nonlinear (Supplemental Tables S12). Experiment: One review of experimental studies reported four assessments, but none were statistically significant, suggesting that the available experimental evidence does not support a relationship between intakes of URM and CVD (Supplemental Table S13). Coherence: There is a lack of coherence between the observational findings and experimental evidence, i.e., observational findings reported statistically significant positive associations between URM and CVD outcomes, whereas experimental evidence showed non-significant effects on CVD risk factors.
AMSTAR 2 quality
Among 10 reviews with observational CVD outcomes, four (40%) were rated as High or Moderate quality and contributed five assessments, while six (60%) were rated as Low or Critically Low quality and contributed eight assessments (Supplemental Tables S11, S12). One review assessed experimental CVD risk factors and was rated as Critically Low quality and contributed four assessments (Supplemental Table S13). Restricting to only Moderate and High quality reviews further supported the Strength and Consistency Criteria but the Biological Gradient Criterion was not supported and there were no High or Moderate quality reviews from experimental studies, so Experiment and Coherence could not be assessed (Supplemental Tables S11–S13).
Plausibility
Several plausible biological mechanisms have been proposed to explain associations between red meat intake and CVD outcomes. Dietary saturated fat and cholesterol (which is in relatively high concentrations in some red and processed meat products) may increase the risk of cardiovascular disease by raising serum low-density lipoprotein cholesterol concentrations (Bergeron et al. Citation2019), which in turn could contribute to the development of atherscerosis (Wolmarans et al. Citation1991; Gascon et al. Citation1996; Hodson, Skeaff, and Chisholm Citation2001). Despite red meat containing these nutrients of concern for overconsumption, a systematically searched meta-analysis of RCTs included in our results, did not show significant effects of unprocessed red meat with any blood lipids or lipoprotein markers (O'Connor, Kim et al. Citation2017). Heme iron in red meat may contribute to oxidative stress, and DNA damage, thought to increase the risk of stroke (Cross, Pollock, and Bingham Citation2003; Kaluza, Wolk, and Larsson Citation2013), however this proposed mechanism has yet to be thoroughly investigated or confirmed by experimental research (O'Connor, Kim, et al. Citation2021). Increased consumption of sodium, which is commonly found in processed meats, was associated with risk of hypertension (Sacks et al. Citation2001; He and MacGregor Citation2002), vascular stiffness (Sanders Citation2009), and stroke (Strazzullo et al. Citation2009). Additionally, processed meats likely contain greater amounts of nitrates and nitrites than unprocessed meats, which may contribute to vascular and endothelial dysfunction (Kleinbongard et al. Citation2006; Förstermann Citation2008). Higher red meat intake is also associated with elevated circulating trimethylamine-N-oxide through gut bacterial and hepatic flavin monooxygenase activities on components such as choline and carnitine (Naghipour et al. Citation2021). While observational evidence suggests inconsistent associations between trimethylamine-N-oxide and CVD risk (Naghipour et al. Citation2021), limited experimental research with randomized-controlled feeding indicates a link between red meat intake, trimethylamine-N-oxide, and homeostatic model assessment of insulin resistance, but not other lipids or lipoproteins (e.g., triglycerides, total cholesterol, HDL-C, or LDL-C) (Krishnan et al. Citation2021). Even though there is a lack of experimental confirmation for some of these mechanisms, all are biologically plausible to support the Plausibility Criterion.
Type 2 diabetes mellitus: strength, consistency, biological gradient, experiment, coherence, and quality
Mixed unprocessed red meat + processed meat
shows the Mixed URM + PM and T2DM findings from the Strength, Consistency, Biological Gradient, Experiment, Coherence and AMSTAR 2 quality analyses. Strength and Consistency: Eight high vs. low assessments from eight reviews of observational studies were reported. Most assessments (∼88%) reported statistically significant associations and seven of the eight total assessments (∼88%) had an effect size ≥1.2 (Supplemental Table S14). No review reported an outcome for T2DM-related mortality likely because T2DM is not often the immediate cause of death, but generally an underlying condition that may contribute to the primary cause of death. Biological Gradient: Seven dose-response assessments from seven reviews of observational studies were reported. Six of the seven total assessments (∼86%) were statistically significant, but the shape of this relationship was not consistently reported (Supplemental Table S15). Experiment: Twelve assessments from three reviews of experimental studies were reported. Half of the assessments were statistically significant (50%), thus precluding a conclusion regarding the relationship between Mixed URM + PM intake and T2DM risk factors (Supplemental Table S16). Coherence: There was a lack of coherence between the observational findings and experimental evidence, i.e., observational findings reported a majority of statistically significant and positive associations between Mixed URM + PM and T2DM outcomes, but reviews of experimental evidence were too inconsistent to draw conclusions. An additional systematic review of experimental studies (Schwingshackl et al. Citation2018) was identified in our search but lacked quantifiable analyses. This review reported that when compared to other food groups, Mixed URM + PM was one of the least effective at reducing HbA1c and fasting glucose, which may suggest coherence with the observational findings (Schwingshackl et al. Citation2018).
AMSTAR 2 quality
Among 10 reviews assessed with observational T2DM outcomes, two reviews (20%) were rated High or Moderate quality and contributed two assessments, while the remaining eight reviews (80%) were rated Low or Critically Low quality and contributed eight assessments (Supplemental Tables S14, S15). All three reviews of experimental T2DM risk factor outcomes were rated High or Moderate quality and contributed 12 assessments (Supplemental Table S16). Restricting to only Moderate and High quality reviews further supported the Strength, Consistency, Experiment, and Coherence Criteria but the Biological Gradient Criterion was not supported (Supplemental Tables S14–S16).
Processed meat
shows the PM and T2DM findings from the Strength, Consistency, Biological Gradient, Experiment, Coherence and AMSTAR 2 quality analyses. Strength and Consistency: Five high vs. low assessments from five reviews of observational studies were reported. All the PM and T2DM incidence assessments reported were statistically significant and all five assessments (100%) had effect sizes ≥1.2 (Supplemental Table S17). Biological Gradient: Six dose-response assessments from six reviews of observational studies were reported. All the assessments (100%) were statistically significant, but the shape of the relationship was not consistently reported (Supplemental Table S18). Three of the six reviews reported a visual representation of the relationship shape with two reviews reporting a linear and one review reporting a nonlinear relationship (Supplemental Table S18). Experiment and Coherence: No review reported experimental findings between processed meat and T2DM risk factors, so the Experimental and Coherence Criteria could not be assessed due to lack of evidence.
AMSTAR 2 quality
Seven reviews assessed observational T2DM outcome assessments and two reviews (∼29%) were rated as High or Moderate quality and contributed two assessments. The remaining five reviews (∼71%) were rated as Low or Critically Low quality and contributed five assessments (Supplemental Tables S17, S18). Restricting to only High and Moderate quality reviews did not change results (Supplemental Tables S17, S18).
Unprocessed red meat
shows the URM and T2DM findings from the Strength, Consistency, Biological Gradient, Experiment, Coherence and AMSTAR 2 quality analyses. Strength and Consistency: No review of observational studies provided high vs. low intake assessments, so these criteria could not be assessed. Biological Gradient: Two dose-response assessments from two reviews of observational studies were included. Only one assessment (50%) was statistically significant; neither review provided a visual representation of the shape of the relationship (Supplemental Table S19). Experiment: One review of experimental studies reported three assessments, and none were statistically significant. (Supplemental Table S20). Coherence: This criterion could not be assessed due to lack of evidence.
AMSTAR 2 quality
Among two reviews of observational T2DM outcomes, one (50%) was rated as High quality and contributed one assessment and the other review was rated as Low quality and contributed one assessment. Only one review assessed experimental T2DM risk factors and was rated as High quality and contributed three assessments (Supplemental Tables S19, S20). Restricting to only High and Moderate quality reviews did not change results (Supplemental Tables S19, S20).
Plausibility
Similar to the plausible biological mechanisms as CVD, dietary saturated fat and cholesterol from red meat or processed meat may increase the risk of type 2 diabetes mellitus through increased serum low-density lipoprotein cholesterol concentration (Bergeron et al. Citation2019), which could contribute to insulin resistance (Bendinelli et al. Citation2013). However, a meta-analysis of RCTs included in our results did not show that higher compared to lower meat intakes affected T2DM risk factors, including fasting glucose, insulin, and HOMA-IR (O'Connor, Kim, et al. Citation2021). Heme iron in red meat may contribute to oxidative stress, and DNA damage which may increase risk for T2DM (Cross, Pollock, and Bingham Citation2003; Zhao et al. Citation2012). Processed meats likely contain greater amounts of nitrates and nitrites, which may contribute to a decrease in insulin secretion (Portha et al. Citation1980; McGrowder, Ragoobirsingh, and Dasgupta Citation2001). Cooking preparation methods such as open flame or high-temperature cooking has also been associated with an increased risk of T2DM (Liu et al. Citation2018). Even though there is a lack of experimental confirmation for some of these mechanisms, all are biologically plausible to support the Plausibility Criterion.
Cardiovascular disease and type 2 diabetes mellitus: temporality, specificity, and analogy
Temporality
After the search was completed, only one review used a cross-sectional design (Rouhani et al. Citation2014) which violated the Temporality Criterion; therefore, this review was excluded post hoc. Thus, all systematic reviews and meta-analyses included only prospective observational and experimental studies within populations who were free of the cardiometabolic disease of interest at baseline (Thiese Citation2014). Including only prospective study designs ensures documentation of red and processed meat intakes prior to ascertainment of outcomes of interest, thus supporting the Temporality Criterion for all meat type and outcome categories.
Specificity
“Total red meat” intake was reported to increase risk for all-cause mortality, accidental deaths, and all other causes of death which have no plausible biological mechanism related to meat intake (Sinha et al. Citation2009; Klurfeld Citation2015). Intake of “red and processed meat” is associated with several different types of cancers such as gastric, colorectal, and breast (Chan et al. Citation2011; Farvid et al. Citation2018; Kim, Kim et al. Citation2019) as well as end stage renal disease (Lew et al. Citation2017). While it is plausible that there may be a common underlying mechanism relating red and processed meat intake to CVD, T2DM, and some cancers, there is no plausible explanation for the association with all-cause mortality and accidental deaths. Intakes of red and processed meats are linked to a wide range of adverse outcomes; therefore, this evidence does not support the Specificity Criterion for any meat type and outcome categories.
Analogy
We describe parallels between red and processed meat and other animal-based protein sources (white meat, fish/seafood, dairy, and eggs) and their relations to CVD and T2DM in Supplemental Table S21. Briefly, white meat intake was not associated with an increased risk in CVD and T2DM incidence or mortality; fish and seafood intake had non-significant associations with risk of CVD and T2DM with the caveat that fish be lean; dairy intake had inverse or non-significant associations with CVD and T2DM risk factors, incidence, and mortality outcomes, particularly with low or reduced fat products; eggs were inconsistently associated with an increased risk of CVD and T2DM in observational studies but experimental studies reported non-significant findings with risk factors. The patterns that we observed in white meat, fish/seafood, dairy, and egg literature were analogous to the literature on the red and processed meat intake, but since the findings did not suggest causality, the Analogy Criterion was not supported.
Publication bias
Potential publication bias was detected for four comparisons: PM and CVD Mortality (using the high vs. low RRs; p = 0.016; Supplemental Figure S5), Mixed URM + PM and T2DM Incidence (using the dose-response RRs; p = 0.017; Supplemental Figure S12), PM and CVD Mortality (using the dose-response RRs; p = 0.0378; Supplemental Figure S14), PM and T2DM Incidence (using the dose-response RRs; p < 0.001; Supplemental Figure S15). The Cochrane Handbook recommends inclusion of at least 10 assessments to adequately power test for publication bias (Sterne, Egger, and Moher Citation2011), but none of these significant comparisons met this criterion, so these findings should be interpreted with this in mind. See Supplemental Figures S1–24 for visuals of all funnel plots and Egger’s test results.
Discussion
Observational research is the predominant evidence underlying dietary recommendations regarding red and processed meat intakes and cardiometabolic disease-related outcomes in the US (DGA Citation2015; DGA Citation2020). Our umbrella systematic review builds upon this evidence by using long-established criteria (Hill Citation1965) for causal inference in which we additionally incorporated experimental evidence from short-term randomized controlled feeding trials of relevant clinical risk factors. We infer that a causal relationship is not likely present between red and processed meat intakes and CVD risk. This is largely because reviews with observational studies report consistently weak associations (four of the 26 (∼15%) total high vs. low intake assessments from all meat type categories and CVD outcomes reported RR’s ≥1.2) which was coherent with reviews of experimental studies reporting a lack of significant effects on CVD risk factors. However, we infer a potential for a causal relationship between PM and Mixed URM + PM and T2DM. This statement is largely supported by consistently strong associations (12 of the 13 (∼90%) total high vs. low intake assessments from all meat type categories and T2DM outcomes reported RR’s ≥1.2), but additional experimental evidence is needed to corroborate coherence across study designs. Conducting more observational studies will not further advance our understanding of this topic, but rigorous randomized controlled feeding trials may help fill causal gaps (Brown, Bohan Brown, and Allison Citation2013; Brown et al. Citation2021). Overall, our results suggest the inclusion of experimental evidence and consideration of how meat variables are operationalized are prudent in conclusions regarding red and processed meat intake and cardiometabolic health.
Our results exemplify how the original nine Bradford Hill Causality Criteria (Hill Citation1965) may be adapted to review a broad scope of human nutrition research instead of relying solely on observational evidence to infer diet-disease causal relationships. This is important because long-term direct causal evidence is often not ethically, financially, or otherwise feasible to obtain in nutrition research (Weaver and Miller Citation2017). We modeled our approach based on previous Bradford Hill assessments of other nutrition topics (Mente et al. Citation2009; Micha et al. Citation2017; Livesey and Livesey Citation2019; Livesey et al. Citation2019). Consistent with those assessments, we considered an a priori RR ≥1.2 to indicate a strong association. Some consider a 10% increased risk (RR cutoff of 1.10) to indicate higher risk of disease (Neuhouser Citation2020), while a 50% increased risk (RR cutoff of 1.5 )(Willett Citation2012; Klurfeld Citation2015) is considered appropriate by others. A 20% increased risk is biologically meaningful at the population level and considers potential attenuation of associations due to biases or poor control of confounding factors (Byers and Lyle Citation1999; NCEP Citation2001). Some prior Bradford Hill assessments developed scoring systems to tally the criteria, or provide a grade based on weighting each criterion equally (Dyda et al. Citation2018; Fundora et al. Citation2020). We avoided this approach as it was advised against by Bradford Hill himself, who encouraged critical inquisition to interpret findings and infer causation rather than depend on a numbered score (Hill Citation1965; Potischman and Weed Citation1999). Instead, our review collectively considered all nine Bradford Hill Criteria to infer causation, with emphasis on the Strength, Consistency, Coherence, and Experimental Criteria, as well as the quality of evidence and the potential for publication bias. A more standardized approach for future nutrition applications would be beneficial, akin to the standardized methodology developed by the American Institute for Cancer Research’s Continuous Update Project (WCRF Citation2018).
Our review focused on meat as a whole food (Mixed URM + PM, URM, or PM) rather than specific nutritive components of meat, such as sodium or saturated fat. Reducing or limiting these nutrients is recommended for prevention of CVD (Arnett et al. Citation2019). Educating omnivorous consumers to choose lean red meats and/or unprocessed red meats could help meet daily sodium and saturated fat intake recommendations. However, our findings suggest that high intakes of red and processed meats as independent whole foods likely do not directly increase CVD risk. Simply reducing red and processed meat intake may not reduce risk for CVD, particularly depending on what is being consumed in place of it. It is also important to inform meat consumers about overall diet quality. High intakes of red and processed meats are a hallmark of a Western-style eating pattern, as are high intakes of sugar sweetened beverages, refined grains, and non-meat foods high in sodium, added sugar, and saturated fat (Shan et al. Citation2019). Often, red and processed meat intakes are associated with an increased CVD risk in Westernized countries (Iqbal et al. Citation2021). This pattern of eating is also associated with other unhealthy lifestyle behaviors such as low physical activity, high alcohol intake, and tobacco use which may also increase CVD risk (Pérez-Martínez et al. Citation2017; Chudasama et al. Citation2020). Bradford Hill stated that “one-to-one relationships are not frequent [and] multi-causation is generally more likely than single causation”(Hill Citation1965). Our results suggest that it is unlikely that consuming higher amounts of red and processed meats are one-to-one causes of CVD, but instead likely depend on the entirety of the dietary and behavioral patterns of individuals.
Our results suggest a potential for a causal relationship between PM and Mixed URM + PM intakes and T2DM. Most T2DM assessments had RR’s ≥1.2, some as large as 1.41 (95%CI: 1.25, 1.60) (Aune, Ursin, and Veierød Citation2009) for processed meat and T2DM incidence. These consistently large effect sizes that reach statistical significance are not commonly seen in nutrition epidemiology (Potischman and Weed Citation1999). In fact, this is more than double the risk associated between processed meat intake and colorectal cancer (WCRF Citation2018). Further, 13 of the 15 total (∼87%) dose-response analyses of T2DM assessments were statistically significant. High quality experimental evidence, particularly of processed meat intake, on T2DM risk factors is lacking to corroborate these notably strong and consistent associations. Publication bias also could not be ruled out for some T2DM assessments (Supplemental Figures S12, S14, and S15), meaning that there may be non-significant or weak associations that were not published. An inference for URM and T2DM was not made due to a lack of reviews with observational studies. More experimental research is needed between PM and Mixed URM + PM on T2DM risk factors to reach stronger conclusions of causality.
We used AMSTAR 2 to assess the quality of included systematic reviews or meta-analyses (Shea et al. Citation2017) as well as a guide to ensure that our review was of high quality (Supplement ). Our findings using AMSTAR 2 indicated that most of the reviews rated as High or Moderate quality were reviews of experimental evidence while most of the reviews rated Low or Critically Low quality were reviews of observational evidence. The Low or Critically Low quality reviews were frequently rated as such for a lack of one or more of the following critical items: (1) explicit statement of a priori methods, (2) assessment of risk of bias, (3) discussion of risk of bias when interpreting findings, and (4) assessment of publication bias. A post-hoc assessment of the High and Moderate quality reviews only largely supported our original findings from when all reviews were combined. A limitation with our approach and the use of AMSTAR 2 is that one review may have multiple assessments. Therefore, one review could contribute a high number of assessments and affect the total number of High, Moderate, Low, or Critically Low quality assessments, depending on the quality of the review. Further, AMSTAR 2 rates the quality of the systematic review methods, not the quality or internal validity of the included empirical studies. Our results should be interpreted with this in mind, because a High quality systematic review included in our review could have potentially included low quality empirical studies.
A noted limitation of synthesizing meat-related human research is variability in meat terminology, a lack of standardized composition of meat-containing foods, and a lack of consensus of meat definitions (Gifford, O'Connor et al. Citation2017; O'Connor et al. Citation2020). Research shows that variability among the terms used to describe meat-containing foods can result in meaningful differences in population-level intake estimates. For example, the decision to include vs. exclude processed meat as part of an exposure assessment (akin to the difference between Mixed URM + PM vs. URM in our analysis) can result in intake estimates that differ by up to 9 oz-eq/week (O'Connor, Herrick, et al. Citation2021). A strength of our approach is that we re-categorized included reviews into three mutually exclusive and explicit meat categories (URM, PM and Mixed URM + PM) based on the 2015–2020 Dietary Guidelines for Americans definitions (DGA Citation2015). We relied on the descriptions provided in the reviews, and if the descriptions were unclear the reviews were categorized into the Mixed URM + PM category. This helped ensure that reviews were being compared appropriately based on meat category and allowed for a more granular assessment of meat intake and CVD and T2DM. That said, our review focused largely on relative consumption (e.g., high vs. low intake) of red and processed meats. The review did not take into consideration the leanness or absolute quantities of meat consumed, which are other important factors to consider for public health recommendations. Further, the common practice of using food frequency questionnaires or dietary recalls to assess dietary intakes with observational studies are prone to recall error, measurement error, and have a potential for inaccurate nutrient and food group estimation (Satija et al. Citation2015). Collectively, we consider that assessing observational studies in conjunction with randomized controlled feeding trials, as we did in this review, improves understanding of diet-disease associations (Beyerbach et al. Citation2022).
Our objective for this review was to systematically assess reviews of both observational and experimental research and infer causality between intakes of red and processed meats and risk for CVD and T2DM. This approach is different from the series of systematic reviews released by the NutriRECS Consortium panel in 2019, which had the aim to provide dietary recommendations regarding red and processed meat intake (Johnston et al. Citation2018; Johnston et al. Citation2019; Valli et al. Citation2019; Vernooij et al. Citation2019; Zeraatkar, Han, et al. Citation2019). We are not making recommendations for public consumption, but rather hope that our inferences can aid in interpretation of future evidence reviews for public guidance on this topic.
We are mindful that multiple reviews may have included the same empirical research studies which may disproportionately influence our collective findings. A majority of assessments within the included reviews were deemed to have high heterogeneity, indicating significant variability among the empirical research. Some reviews had multiple summary estimates, which contributed greater weight to our findings. We only used reviews written in English, eliciting potential language or cultural biases. The majority of included reviews were primarily of Caucasian cohorts (Stern and Kleijnen Citation2020). Our inferences between CVD and T2DM are discordant despite commonalities among some of their risk factors. We recognize that T2DM is considered a risk factor for CVD (Einarson, Acs et al. Citation2018), however our a priori objective was to assess these two diseases independently. Among the strengths of this review, our search strategy was inclusive in that we did not restrict our search based on populations or eating patterns, increasing external validity of our findings (Higgins and Green Citation2011). As widely used in public health research (Mente et al. Citation2009; Micha et al. Citation2017; WCRF Citation2018; Livesey and Livesey Citation2019; Livesey et al. Citation2019), this review operationalized classic causality criteria (Hill Citation1965) to reviews that included both observational and experimental studies to infer causal relations.
Conclusion
We infer red and processed meat intakes are not causally related to cardiovascular disease outcomes due to consistently weak associations and a lack of coherence with experimental evidence. However, we infer processed meat and mixed red and processed meat intakes are potentially causally related to type 2 diabetes mellitus due to consistently strong associations. More randomized controlled feeding trials are needed, specifically for processed meat, to support or refute these inferences. These results emphasize the importance of integrating longer-term observational studies of clinical endpoints and shorter-term randomized controlled feeding trials of clinical risk factors when developing future public dietary guidance.
Supplemental Material
Download MS Word (1.2 MB)Acknowledgments
We thank Andrew W Brown, PhD for his thoughtful feedback on the manuscript draft. Hill and Drs O’Connor and Campbell designed the research. McGowan, Hill, Wang, and Clark conducted the systematic review; Hill, Wang, and Dr. O’Connor conducted the research, analyzed the data, and drafted the manuscript. Hill and Dr. Campbell had full access to all the data in the review and take primary responsibility for final content. Hill, Wang, and Drs O’Connor, Campbell and Forman made critical revisions and provided intellectual content. All authors read and approved the final manuscript. This work was supported by The Beef Checkoff.
Disclosure statement
During the time this research was conducted, WWC received funding for research grants, travel or honoraria for scientific presentations, or consulting services from the following organizations: The Beef Checkoff, Pork Checkoff, North Dakota Beef Commission, Barilla Group, Mushroom Council, National Chicken Council, Foundation for Meat and Poultry Research and Education, and American Egg Board’s Egg Nutrition Center. ERH, LEO, YW, CMC, BSM, and MRF report no conflict of interest.
Funding
This work was supported, in part, by The Beef Checkoff. The funder had no role in designing, conducting, or reporting the research.
Data availability statement
All results described in the manuscript are provided in the Supplemental Material.
References
- Abete, I., D. Romaguera, A. R. Vieira, A. Lopez de Munain, and T. Norat. 2014. Association between total, processed, red and white meat consumption and all-cause, CVD and IHD mortality: a meta-analysis of cohort studies. The British Journal of Nutrition 112 (5):762–75. doi: 10.1017/S000711451400124X.
- Akesson, A., L. F. Andersen, A. G. Kristjánsdóttir, E. Roos, E. Trolle, E. Voutilainen, and E. Wirfält. 2013. Health effects associated with foods characteristic of the Nordic diet: a systematic literature review. Food Nutr Res 57:1–21.
- An, R., J. Liu, and R. Liu. 2020. Pork consumption in relation to body weight and composition: a systematic review and meta-analysis. American Journal of Health Behavior 44 (4):513–25. doi: 10.5993/AJHB.44.4.12.
- Arnett, D. K., R. S. Blumenthal, M. A. Albert, A. B. Buroker, Z. D. Goldberger, E. J. Hahn, C. D. Himmelfarb, A. Khera, D. Lloyd-Jones, J. W. McEvoy, Jr., Sr., et al. 2019. 2019 ACC/AHA guideline on the primary prevention of cardiovascular disease: a report of the american college of cardiology/american heart association task force on clinical practice guidelines. Circulation 140 (11):e596–e646.
- Aromataris, E., R. Fernandez, C. M. Godfrey, C. Holly, H. Khalil, and P. Tungpunkom. 2015. Summarizing systematic reviews: methodological development, conduct and reporting of an umbrella review approach. International Journal of Evidence-Based Healthcare 13 (3):132–40. doi: 10.1097/XEB.0000000000000055.
- Aslam, S, and P. Emmanuel. 2010. Formulating a researchable question: a critical step for facilitating good clinical research. Indian Journal of Sexually Transmitted Diseases and AIDS 31 (1):47–50. doi: 10.4103/0253-7184.69003.
- Aune, D., G. Ursin, and M. B. Veierød. 2009. Meat consumption and the risk of type 2 diabetes: a systematic review and meta-analysis of cohort studies. Diabetologia 52 (11):2277–87. doi: 10.1007/s00125-009-1481-x.
- Bechthold, A., H. Boeing, C. Schwedhelm, G. Hoffmann, S. Knüppel, K. Iqbal, S. De Henauw, N. Michels, B. Devleesschauwer, S. Schlesinger, et al. 2019. Food groups and risk of coronary heart disease, stroke and heart failure: a systematic review and dose-response meta-analysis of prospective studies. Critical Reviews in Food Science and Nutrition 59 (7):1071–90. doi: 10.1080/10408398.2017.1392288.
- Bendinelli, B., D. Palli, G. Masala, S. J. Sharp, M. B. Schulze, M. Guevara, A. D. van der, F. Sera, P. Amiano, B. Balkau, et al. 2013. Association between dietary meat consumption and incident type 2 diabetes: the EPIC-InterAct study. Diabetologia 56 (1):47–59.
- Bergeron, N., S. Chiu, P. T. Williams, S. M. King, and R. M. Krauss. 2019. Effects of red meat, white meat, and nonmeat protein sources on atherogenic lipoprotein measures in the context of low compared with high saturated fat intake: a randomized controlled trial. The American Journal of Clinical Nutrition 110 (1):24–33. doi: 10.1093/ajcn/nqz035.
- Beyerbach, J., J. Stadelmaier, G. Hoffmann, S. Balduzzi, N. Bröckelmann, and L. Schwingshackl. 2022. Evaluating concordance of bodies of evidence from randomized controlled trials, dietary intake, and biomarkers of intake in cohort studies: a meta-epidemiological study. Advances in Nutrition (Bethesda, Md.) 13 (1):48–65. doi: 10.1093/advances/nmab095.
- Brown, A. W., S. Aslibekyan, D. Bier, R. Ferreira da Silva, A. Hoover, D. M. Klurfeld, E. Loken, E. Mayo-Wilson, N. Menachemi, G. Pavela, et al. 2021. Toward more rigorous and informative nutritional epidemiology: the rational space between dismissal and defense of the status quo. Critical Reviews in Food Science and Nutrition 1–18. doi: 10.1080/10408398.2021.1985427.
- Brown, A. W., M. M. Bohan Brown, and D. B. Allison. 2013. Belief beyond the evidence: using the proposed effect of breakfast on obesity to show 2 practices that distort scientific evidence. The American Journal of Clinical Nutrition 98 (5):1298–308. doi: 10.3945/ajcn.113.064410.
- Byers, T, and B. Lyle. 1999. The role of epidemiology in determining when evidence is sufficient to support nutrition recommendations. Summary statement. The American Journal of Clinical Nutrition 69 (6):1365s–7s. doi: 10.1093/ajcn/69.6.1365S.
- Chan, D. S., R. Lau, D. Aune, R. Vieira, D. C. Greenwood, E. Kampman, and T. Norat. 2011. Red and processed meat and colorectal cancer incidence: meta-analysis of prospective studies. PLoS One 6 (6):e20456. doi: 10.1371/journal.pone.0020456.
- Chen, G. C., D. B. Lv, Z. Pang, and Q. F. Liu. 2013. Red and processed meat consumption and risk of stroke: a meta-analysis of prospective cohort studies. European Journal of Clinical Nutrition 67 (1):91–5. doi: 10.1038/ejcn.2012.180.
- Chudasama, Y. V., K. Khunti, C. L. Gillies, N. N. Dhalwani, M. J. Davies, T. Yates, and F. Zaccardi. 2020. Healthy lifestyle and life expectancy in people with multimorbidity in the UK Biobank: a longitudinal cohort study. PLoS Medicine 17 (9):e1003332. doi: 10.1371/journal.pmed.1003332.
- Cross, A. J., J. R. Pollock, and S. A. Bingham. 2003. Haem, not protein or inorganic iron, is responsible for endogenous intestinal N-nitrosation arising from red meat. Cancer Research 63 (10):2358–60.
- DGA 2015. U.S. Department of Health and Human Services and U.S. Department of Agriculture. 2015 – 2020 Dietary Guidelines for Americans. 8th ed. https://health.gov/our-work/food-nutrition/previous-dietary-guidelines/2015.
- DGA 2020. U.S. Department of Agriculture and U.S. Department of Health and Human Services. Dietary guidelines for Americans, 2020–2025. 9th ed. DietaryGuidelines.gov.
- Dyda, A., S. Stelzer-Braid, D. Adam, A. A. Chughtai, and C. R. MacIntyre. 2018. The association between acute flaccid myelitis (AFM) and Enterovirus D68 (EV-D68) - what is the evidence for causation? Eurosurveillance 23 (3):17–00310. doi: 10.2807/1560-7917.ES.2018.23.3.17-00310
- Evert, A. B., M. Dennison, C. D. Gardner, W. T. Garvey, K. H. K. Lau, J. MacLeod, J. Mitri, R. F. Pereira, K. Rawlings, S. Robinson, Jr, et al. 2019. Nutrition therapy for adults with diabetes or prediabetes: a consensus report. Diabetes Care 42 (5):731–54. doi: 10.2337/dci19-0014.
- Einarson, T. R., A. Acs, C. Ludwig, and U. H. Panton. 2018. Prevalence of cardiovascular disease in type 2 diabetes: A systematic literature review of scientific evidence from across the world in 2007-2017. Cardiovasc Diabetol 17 (1):83.
- Fan, M., Y. Li, C. Wang, Z. Mao, W. Zhou, L. Zhang, X. Yang, S. Cui, and L. Li. 2019. Dietary protein consumption and the risk of type 2 diabetes: a dose-response meta-analysis of prospective studies. Nutrients 11 (11):2783. doi: 10.3390/nu11112783.
- Farvid, M. S., M. C. Stern, T. Norat, S. Sasazuki, P. Vineis, M. P. Weijenberg, A. Wolk, K. Wu, B. W. Stewart, and E. Cho. 2018. Consumption of red and processed meat and breast cancer incidence: a systematic review and meta-analysis of prospective studies. International Journal of Cancer 143 (11):2787–99. doi: 10.1002/ijc.31848.
- Feiveson, A. 2017. Explanation of the delta method. https://www.stata.com/support/faqs/statistics/delta-method/#.
- Fundora, J. B., P. Guha, D. R. Shores, M. Pammi, and A. Maheshwari. 2020. Intestinal dysbiosis and necrotizing enterocolitis: assessment for causality using Bradford Hill criteria. Pediatric Research 87 (2):235–48. doi: 10.1038/s41390-019-0482-9.
- Förstermann, U. 2008. Oxidative stress in vascular disease: causes, defense mechanisms and potential therapies. Nature Clinical Practice. Cardiovascular Medicine 5 (6):338–49. doi: 10.1038/ncpcardio1211.
- Gascon, A., H. Jacques, S. Moorjani, Y. Deshaies, L. D. Brun, and P. Julien. 1996. Plasma lipoprotein profile and lipolytic activities in response to the substitution of lean white fish for other animal protein sources in premenopausal women. The American Journal of Clinical Nutrition 63 (3):315–21. doi: 10.1093/ajcn/63.3.315.
- Gifford, C., L. O’Connor, W. Campbell, D. Woerner, and K. Belk. 2017. Broad and inconsistent muscle food classification is problematic for dietary guidance in the U.S. Nutrients 9 (9):1027. doi: 10.3390/nu9091027.
- Guasch-Ferré, M., A. Satija, S. A. Blondin, M. Janiszewski, E. Emlen, L. E. O’Connor, W. W. Campbell, F. B. Hu, W. C. Willett, and M. J. Stampfer. 2019. Meta-analysis of randomized controlled trials of red meat consumption in comparison with various comparison diets on cardiovascular risk factors. Circulation 139 (15):1828–45. doi: 10.1161/CIRCULATIONAHA.118.035225.
- He, F. J, and G. A. MacGregor. 2002. Effect of modest salt reduction on blood pressure: a meta-analysis of randomized trials. Implications for public health. Journal of Human Hypertension 16 (11):761–70. doi: 10.1038/sj.jhh.1001459.
- Higgins, J., and S. Green 2011. Cochrane handbook for systematic reviews of intervention Version 5.1.0. Table 5.6a: Some advantages and disadvantages of broad versus narrow review questions. [updated March 2011]. The Cochrane Collaboration, 2011. Available from www.handbook.cochrane.org.
- Hill, A. B. 1965. The environment and disease: association or causation? Proceedings of the Royal Society of Medicine 58 (5):295–300. doi: 10.1177/003591576505800503.
- Hodson, L., C. M. Skeaff, and W. A. Chisholm. 2001. The effect of replacing dietary saturated fat with polyunsaturated or monounsaturated fat on plasma lipids in free-living young adults. European Journal of Clinical Nutrition 55 (10):908–15. doi: 10.1038/sj.ejcn.1601234.
- Iqbal, R., M. Dehghan, A. Mente, S. Rangarajan, A. Wielgosz, A. Avezum, P. Seron, K. F. AlHabib, P. Lopez-Jaramillo, S. Swaminathan, et al. 2021. Associations of unprocessed and processed meat intake with mortality and cardiovascular disease in 21 countries [Prospective Urban Rural Epidemiology (PURE) Study]: a prospective cohort study. The American Journal of Clinical Nutrition 114 (3):1049–58. doi: 10.1093/ajcn/nqaa448.
- Johnston, B. C., P. Alonso-Coello, M. M. Bala, D. Zeraatkar, M. Rabassa, C. Valli, C. Marshall, R. E. Dib, R. W. M. Vernooij, P. O. Vandvik, et al. 2018. Methods for trustworthy nutritional recommendations NutriRECS (Nutritional Recommendations and accessible Evidence summaries Composed of Systematic reviews): a protocol. BMC Medical Research Methodology 18 (1):162. doi: 10.1186/s12874-018-0621-8.
- Johnston, B. C., D. Zeraatkar, M. A. Han, R. W. M. Vernooij, C. Valli, R. E. Dib, C. Marshall, P. J. Stover, S. Fairweather-Taitt, G. Wójcik, et al. 2019. Unprocessed red meat and processed meat consumption: dietary guideline recommendations from the nutritional recommendations (NutriRECS) consortium. Annals of Internal Medicine 171 (10):756–64. doi: 10.7326/M19-1621.
- Kaluza, J., A. Wolk, and S. C. Larsson. 2012. Red meat consumption and risk of stroke: a meta-analysis of prospective studies. Stroke 43 (10):2556–60. doi: 10.1161/STROKEAHA.112.663286.
- Kaluza, J., A. Wolk, and S. C. Larsson. 2013. Heme iron intake and risk of stroke: a prospective study of men. Stroke 44 (2):334–9. doi: 10.1161/STROKEAHA.112.679662.
- Kim, K., J. Hyeon, S. A. Lee, S. O. Kwon, H. Lee, N. Keum, J. K. Lee, and S. M. Park. 2017. Role of total, red, processed, and white meat consumption in stroke incidence and mortality: a systematic review and meta-analysis of prospective cohort studies. Journal of the American Heart Association. 6 (9):1–16.
- Kim, S. R., K. Kim, S. A. Lee, S. O. Kwon, J. K. Lee, N. Keum, and S. M. Park. 2019. Effect of red, processed, and white meat consumption on the risk of gastric cancer: an overall and dose-response meta-analysis. Nutrients 11 (4):826. doi: 10.3390/nu11040826.
- Kim, Y, and Y. Je. 2018. Meat consumption and risk of metabolic syndrome: results from the korean population and a meta-analysis of observational studies. Nutrients 10 (4):390. doi: 10.3390/nu10040390.
- Kleinbongard, P., A. Dejam, T. Lauer, T. Jax, S. Kerber, P. Gharini, J. Balzer, R. B. Zotz, R. E. Scharf, R. Willers, et al. 2006. Plasma nitrite concentrations reflect the degree of endothelial dysfunction in humans. Free Radical Biology and Medicine 40 (2):295–302. doi: 10.1016/j.freeradbiomed.2005.08.025.
- Klurfeld, D. M. 2015. Research gaps in evaluating the relationship of meat and health. Meat Science 109:86–95. doi: 10.1016/j.meatsci.2015.05.022.
- Krishnan, S., L. E. O’Connor, Y. Wang, E. R. Gertz, W. W. Campbell, and B. J. Bennett. 2021. Adopting a Mediterranean-style eating pattern with low, but not moderate, unprocessed, lean red meat intake reduces fasting serum trimethylamine N-oxide (TMAO) in adults who are overweight or obese. British Journal of Nutrition 1–9. doi: 10.1017/S0007114521004694.
- Lew, Q. J., T. H. Jafar, H. W. Koh, A. Jin, K. Y. Chow, J. M. Yuan, and W. P. Koh. 2017. Red meat intake and risk of ESRD. Journal of the American Society of Nephrology 28 (1):304–12. doi: 10.1681/ASN.2016030248.
- Lichtenstein, A. H., L. J. Appel, M. Vadiveloo, F. B. Hu, P. M. Kris-Etherton, C. M. Rebholz, F. M. Sacks, A. N. Thorndike, L. Van Horn, and J. Wylie-Rosett. 2021. 2021 dietary guidance to improve cardiovascular health: a scientific statement from the American Heart Association. Circulation 144 (23):472–487. doi: 10.1161/CIR.0000000000001031.
- Lippi, G., C. Mattiuzzi, and F. Sanchis-Gomar. 2015. Red meat consumption and ischemic heart disease. A systematic literature review. Meat Science 108:32–6. doi: 10.1016/j.meatsci.2015.05.019.
- Liu, G., G. Zong, K. Wu, Y. Hu, Y. Li, W. C. Willett, D. M. Eisenberg, F. B. Hu, and Q. Sun. 2018. Meat cooking methods and risk of type 2 diabetes: results from three prospective cohort studies. Diabetes Care 41 (5):1049–60. doi: 10.2337/dc17-1992.
- Livesey, G, and H. Livesey. 2019. Coronary heart disease and dietary carbohydrate, glycemic index, and glycemic load: dose-response meta-analyses of prospective cohort studies. Mayo Clinic Proceedings. Innovations, Quality & Outcomes 3 (1):52–69. doi: 10.1016/j.mayocpiqo.2018.12.007.
- Livesey, G., R. Taylor, H. F. Livesey, A. E. Buyken, D. J. A. Jenkins, L. S. A. Augustin, J. L. Sievenpiper, A. W. Barclay, S. Liu, T. M. S. Wolever, et al. 2019. Dietary glycemic index and load and the risk of type 2 diabetes: assessment of causal relations. Nutrients 11 (6):1436. doi: 10.3390/nu11061436.
- Maki, K. C., M. E. Van Elswyk, D. D. Alexander, T. M. Rains, E. L. Sohn, and S. McNeill. 2012. A meta-analysis of randomized controlled trials that compare the lipid effects of beef versus poultry and/or fish consumption. Journal of Clinical Lipidology 6 (4):352–61. doi: 10.1016/j.jacl.2012.01.001.
- McGrowder, D., D. Ragoobirsingh, and T. Dasgupta. 2001. Effects of S-nitroso-N-acetyl-penicillamine administration on glucose tolerance and plasma levels of insulin and glucagon in the dog. Nitric Oxide. 5 (4):402–12. doi: 10.1006/niox.2001.0360.
- Mente, A., L. de Koning, H. S. Shannon, and S. S. Anand. 2009. A systematic review of the evidence supporting a causal link between dietary factors and coronary heart disease. Archives of Internal Medicine 169 (7):659–69. doi: 10.1001/archinternmed.2009.38.
- Micha, R., M. L. Shulkin, J. L. Peñalvo, S. Khatibzadeh, G. M. Singh, M. Rao, S. Fahimi, J. Powles, and D. Mozaffarian. 2017. Etiologic effects and optimal intakes of foods and nutrients for risk of cardiovascular diseases and diabetes: systematic reviews and meta-analyses from the Nutrition and Chronic Diseases Expert Group (NutriCoDE). PLoS One 12 (4):e0175149. doi: 10.1371/journal.pone.0175149.
- Micha, R., S. K. Wallace, and D. Mozaffarian. 2010. Red and processed meat consumption and risk of incident coronary heart disease, stroke, and diabetes mellitus: a systematic review and meta-analysis. Circulation 121 (21):2271–83. doi: 10.1161/CIRCULATIONAHA.109.924977.
- Moher, D., A. Liberati, J. Tetzlaff, and D. G. Altman. 2009. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ (Clinical Research ed.) 339:b2535. doi: 10.1136/bmj.b2535.
- Naghipour, S., A. J. Cox, J. N. Peart, E. F. Du Toit, and J. P. Headrick. 2021. Trimethylamine N-oxide: heart of the microbiota-CVD nexus? Nutrition Research Reviews 34 (1):125–46. doi: 10.1017/S0954422420000177.
- NCEP 2001. Executive Summary of the third report of the national cholesterol education program (NCEP) expert panel on detection, evaluation, and treatment of high blood cholesterol in adults (adult treatment panel III). JAMA 285 (19):2486–97. doi: 10.1001/jama.285.19.2486.
- Neuhouser, M. L. 2020. Red and processed meat: more with less? The American Journal of Clinical Nutrition 111 (2):252–5. doi: 10.1093/ajcn/nqz294.
- O'Connor, L. E., K. A. Herrick, R. Parsons, and J. Reedy. 2021. Heterogeneity in Meat Food Groups Can Meaningfully Alter Population-Level Intake Estimates of Red Meat and Poultry. Frontier in Nutrition 8:778369. doi: 10.3389/fnut.2021.778369. PMC: 34977122
- O'Connor, L. E., C. L. Gifford, D. R. Woerner, J. L. Sharp, K. E. Belk, and W. W. Campbell. 2020. Dietary meat categories and descriptions in chronic disease research are substantively different within and between experimental and observational studies: a systematic review and landscape analysis. Advances in Nutrition. 11 (1):41–51.
- O’Connor, L. E., J. E. Kim, and W. W. Campbell. 2017. Total red meat intake of ≥0.5 servings/d does not negatively influence cardiovascular disease risk factors: a systemically searched meta-analysis of randomized controlled trials. The American Journal of Clinical Nutrition 105 (1):57–69. doi: 10.3945/ajcn.116.142521.
- O'Connor, L. E., J. E. Kim, C. M. Clark, W. Zhu, and W. W. Campbell. 2021. Effects of Total red meat intake on glycemic control and inflammatory biomarkers: a meta-analysis of randomized controlled trials. Advances in Nutrition 12 (1):115–27. doi: 10.1093/advances/nmaa096.
- O'Connor, L. E., E. A. Wambogo, K. A. Herrick, R. Parsons, and J. Reedy. 2021. A Standardized Assessment of Processed Red Meat and Processed Poultry Intake in the US Population Aged ≥2 Years Using NHANES. The Journal of Nutrition 152 (1):190–9. doi: 10.1093/jn/nxab316.
- O'Sullivan, T. A., K. Hafekost, F. Mitrou, and D. Lawrence. 2013. Food sources of saturated fat and the association with mortality: a meta-analysis. American Journal of Public Health 103 (9):e31–e42. doi: 10.2105/AJPH.2013.301492.
- O’Connor, L. E, and W. W. Campbell. 2017. Red meat and health: getting to the heart of the matter. Nutrition Today 52 (4):167–73. doi: 10.1097/NT.0000000000000225.
- Papier, K., A. Knuppel, N. Syam, S. A. Jebb, and T. J. Key. 2021. Meat consumption and risk of ischemic heart disease: a systematic review and meta-analysis. Critical Reviews in Food Science and Nutrition 1–12. doi: 10.1080/10408398.2021.1949575.
- Portha, B., M. H. Giroix, J. C. Cros, and L. Picon. 1980. Diabetogenic effect of N-nitrosomethylurea and N-nitrosomethylurethane in the adult rat. Annales de la Nutrition et de L'alimentation 34 (5-6):1143–51.
- Potischman, N, and D. L. Weed. 1999. Causal criteria in nutritional epidemiology. The American Journal of Clinical Nutrition 69 (6):1309s–14s. doi: 10.1093/ajcn/69.6.1309S.
- PROSPERO 2011. International prospective register of systematic reviews. National Institute of Health Research. https://www.crd.york.ac.uk/prospero/.
- Pérez-Martínez, P., D. P. Mikhailidis, V. G. Athyros, M. Bullo, P. Couture, M. I. Covas, L. de Koning, J. Delgado-Lista, A. Díaz-López, C. A. Drevon, et al. 2017. Lifestyle recommendations for the prevention and management of metabolic syndrome: an international panel recommendation. Nutrition Reviews 75 (5):307–26. doi: 10.1093/nutrit/nux014.
- Rothman, K. J, and S. Greenland. 2005. Causation and causal inference in epidemiology. American Journal of Public Health 95 (Suppl 1):S144–S150. doi: 10.2105/AJPH.2004.059204.
- Rouhani, M. H., A. Salehi-Abargouei, P. J. Surkan, and L. Azadbakht. 2014. Is there a relationship between red or processed meat intake and obesity? A systematic review and meta-analysis of observational studies. Obesity Reviews: An Official Journal of the International Association for the Study of Obesity 15 (9):740–8. doi: 10.1111/obr.12172.
- Sacks, F. M., L. P. Svetkey, W. M. Vollmer, L. J. Appel, G. A. Bray, D. Harsha, E. Obarzanek, P. R. Conlin, E. R. Miller, D. G. Simons-Morton, et al. 2001. Effects on blood pressure of reduced dietary sodium and the Dietary Approaches to Stop Hypertension (DASH) diet. DASH-Sodium Collaborative Research Group. New England Journal of Medicine 344 (1):3–10. doi: 10.1056/NEJM200101043440101.
- Sanders, P. W. 2009. Vascular consequences of dietary salt intake. American Journal of Physiology. Renal Physiology 297 (2):F237–243. doi: 10.1152/ajprenal.00027.2009.
- Satija, A., E. Yu, W. C. Willett, and F. B. Hu. 2015. Understanding nutritional epidemiology and its role in policy. Advances in Nutrition (Bethesda, Md.) 6 (1):5–18. doi: 10.3945/an.114.007492.
- Schlesinger, S., M. Neuenschwander, C. Schwedhelm, G. Hoffmann, A. Bechthold, H. Boeing, and L. Schwingshackl. 2019. Food groups and risk of overweight, obesity, and weight gain: a systematic review and dose-response meta-analysis of prospective studies. Advances in Nutrition (Bethesda, Md.) 10 (2):205–18. doi: 10.1093/advances/nmy092.
- Schwingshackl, L., G. Hoffmann, K. Iqbal, C. Schwedhelm, and H. Boeing. 2018. Food groups and intermediate disease markers: a systematic review and network meta-analysis of randomized trials. The American Journal of Clinical Nutrition 108 (3):576–86. doi: 10.1093/ajcn/nqy151.
- Schwingshackl, L., G. Hoffmann, A. M. Lampousi, S. Knüppel, K. Iqbal, C. Schwedhelm, A. Bechthold, S. Schlesinger, and H. Boeing. 2017. Food groups and risk of type 2 diabetes mellitus: a systematic review and meta-analysis of prospective studies. European Journal of Epidemiology 32 (5):363–75. doi: 10.1007/s10654-017-0246-y.
- Schwingshackl, L., C. Schwedhelm, G. Hoffmann, S. Knüppel, K. Iqbal, V. Andriolo, A. Bechthold, S. Schlesinger, and H. Boeing. 2017. Food groups and risk of hypertension: a systematic review and dose-response meta-analysis of prospective studies. Advances in Nutrition (Bethesda, Md.) 8 (6):793–803. doi: 10.3945/an.117.017178.
- Seman, D. L., D. D. Boler, C. C. Carr, M. E. Dikeman, C. M. Owens, J. T. Keeton, T. D. Pringle, J. J. Sindelar, D. R. Woerner, A. S. de Mello, et al. 2018. Meat science lexicon. Meat and Muscle Biology 2 (3). doi: 10.22175/mmb2017.12.0059.
- Shan, Z., C. D. Rehm, G. Rogers, M. Ruan, D. D. Wang, F. B. Hu, D. Mozaffarian, F. F. Zhang, and S. N. Bhupathiraju. 2019. Trends in dietary carbohydrate, protein, and fat intake and diet quality among US adults, 1999-2016. JAMA 322 (12):1178–87. doi: 10.1001/jama.2019.13771.
- Shea, B. J., J. M. Grimshaw, G. A. Wells, M. Boers, N. Andersson, C. Hamel, A. C. Porter, P. Tugwell, D. Moher, and L. M. Bouter. 2007. Development of AMSTAR: a measurement tool to assess the methodological quality of systematic reviews. BMC Medical Research Methodology 7:10. doi: 10.1186/1471-2288-7-10.
- Shea, B. J., B. C. Reeves, G. Wells, M. Thuku, C. Hamel, J. Moran, D. Moher, P. Tugwell, V. Welch, E. Kristjansson, et al. 2017. AMSTAR 2: a critical appraisal tool for systematic reviews that include randomised or non-randomised studies of healthcare interventions, or both. BMJ (Clinical Research ed.) 358:j4008. doi: 10.1136/bmj.j4008.
- Shim, J. S., K. Oh, and H. C. Kim. 2014. Dietary assessment methods in epidemiologic studies. Epidemiol Health 36:e2014009.
- Sinha, R., A. J. Cross, B. I. Graubard, M. F. Leitzmann, and A. Schatzkin. 2009. Meat intake and mortality: a prospective study of over half a million people. Archives of Internal Medicine 169 (6):562–71. doi: 10.1001/archinternmed.2009.6.
- Sorkin, J. D., M. Manary, P. A. M. Smeets, A. J. MacFarlane, A. Astrup, R. L. Prigeon, B. B. Hogans, J. Odle, T. A. Davis, K. L. Tucker, et al. 2021. A guide for authors and readers of the American Society for Nutrition Journals on the proper use of P values and strategies that promote transparency and improve research reproducibility. The American Journal of Clinical Nutrition 114 (4):1280–5. doi: 10.1093/ajcn/nqab223.
- Stern, C, and J. Kleijnen. 2020. Language bias in systematic reviews: you only get out what you put in. JBI Evidence Synthesis 18 (9):1818–9. doi: 10.11124/JBIES-20-00361.
- Sterne, J., M. Egger, and D. Moher. 2011. Cochrane Handbook for Systematic Reviews of Intervention. Chapter 10: addressing reporting biases. In: Higgins J., and Green S. (editors). Version 5.1.0 (updated March 2011). The Cochrane Collaboration, 2011. Available from www.handbook.cochrane.org.
- Strazzullo, P., L. D'Elia, N. B. Kandala, and F. P. Cappuccio. 2009. Salt intake, stroke, and cardiovascular disease: meta-analysis of prospective studies. BMJ (Clinical Research ed.) 339:b4567. doi: 10.1136/bmj.b4567.
- Subar, A. F., L. S. Freedman, J. A. Tooze, S. I. Kirkpatrick, C. Boushey, M. L. Neuhouser, F. E. Thompson, N. Potischman, P. M. Guenther, V. Tarasuk, et al. 2015. Addressing current criticism regarding the value of self-report dietary data. The Journal of Nutrition 145 (12):2639–45. doi: 10.3945/jn.115.219634.
- Terrin, N., C. H. Schmid, and J. Lau. 2005. In an empirical evaluation of the funnel plot, researchers could not visually identify publication bias. Journal of Clinical Epidemiology 58 (9):894–901. doi: 10.1016/j.jclinepi.2005.01.006.
- Thiese, M. S. 2014. Observational and interventional study design types; an overview. Biochemia Medica 24 (2):199–210. doi: 10.11613/BM.2014.022.
- Tian, S., Q. Xu, R. Jiang, T. Han, C. Sun, and L. Na. 2017. Dietary protein consumption and the risk of type 2 diabetes: a systematic review and meta-analysis of cohort studies. Nutrients 9 (9):982. doi: 10.3390/nu9090982.
- Todt, O, and J. L. Luján. 2017. Health claims and methodological controversy in nutrition science. Risk Analysis: An Official Publication of the Society for Risk Analysis 37 (5):958–68. doi: 10.1111/risa.12665.
- University, B 2017. Boston University School of Public Health. Confidence Intervals for the Risk Ratio (Relative Risk). https://sphweb.bumc.bu.edu/otlt/MPH-Modules/BS/BS704_Confidence_Intervals/BS704_Confidence_Intervals8.html.
- USDA 2015. Dietary Guidelines Advisory Committee. 2015. Scientific Report of the 2015 Dietary Guidelines Advisory Committee: Advisory Report to the Secretary of Health and Human Services and the Secretary of Agriculture. U.S. Department of Agriculture, Agricultural Research Service, Washington, DC.
- USDA 2020. Dietary Guidelines Advisory Committee. 2020. Scientific Report of the 2020 Dietary Guidelines Advisory Committee: Advisory Report to the Secretary of Agriculture and the Secretary of Health and Human Services. U.S. Department of Agriculture, Agricultural Research Service, Washington, DC.
- Valli, C., M. Rabassa, B. C. Johnston, R. Kuijpers, A. Prokop-Dorner, J. Zajac, D. Storman, M. Storman, M. M. Bala, I. Solà et al. 2019. Health-related values and preferences regarding meat consumption: a mixed-methods systematic review. Annals of Internal Medicine 171 (10):742–55. doi: 10.7326/M19-1326.
- Vernooij, R. W., M. D. Zeraatkar, M. A. Han, R. E. Dib, M. Zworth, K. Milio, D. Sit, Y. Lee, H. Gomaa, C. Valli, et al. 2019. Patterns of red and processed meat consumption and risk for cardiometabolic and cancer outcomes: a systematic review and meta-analysis of cohort studies. Annals of Internal Medicine 171 (10):732–41. doi: 10.7326/M19-1583.
- Wang, X., X. Lin, Y. Y. Ouyang, J. Liu, G. Zhao, A. Pan, and F. B. Hu. 2016. Red and processed meat consumption and mortality: dose-response meta-analysis of prospective cohort studies. Public Health Nutrition 19 (5):893–905. doi: 10.1017/S1368980015002062.
- World Cancer Research Fund/American Institute for Cancer Research. Continuous Update Project Expert Report 2018. Judging the Evidence. Available at https://www.dietandcancerreport.org.
- WCRF 2018. American Institute for Cancer Research Continuous Update Project. Meat, fish and dairy products and the risk of cancer. World Cancer Research Fund. www.wcrf.org/wp-content/uploads/2021/02/Meat-fish-and-dairy-products.pdf.
- Weaver, C. M, and J. W. Miller. 2017. Challenges in conducting clinical nutrition research. Nutrition Reviews 75 (7):491–9. doi: 10.1093/nutrit/nux026.
- Willett, W. 2012. Nutritional Epidemiology. Chapter 1: overview of nutritional epidemiology. Oxford University. doi: 10.1093/acprof:oso/9780199754038.003.0001.
- Wolmarans, P., A. J. Benadé, T. J. Kotze, A. K. Daubitzer, M. P. Marais, and R. Laubscher. 1991. Plasma lipoprotein response to substituting fish for red meat in the diet. The American Journal of Clinical Nutrition 53 (5):1171–6. doi: 10.1093/ajcn/53.5.1171.
- Yang, C., L. Pan, C. Sun, Y. Xi, L. Wang, and D. Li. 2016. Red meat consumption and the risk of stroke: a dose-response meta-analysis of prospective cohort studies. Journal of Stroke and Cerebrovascular Diseases: The Official Journal of National Stroke Association 25 (5):1177–86. doi: 10.1016/j.jstrokecerebrovasdis.2016.01.040.
- Yang, X., Y. Li, C. Wang, Z. Mao, W. Zhou, L. Zhang, M. Fan, S. Cui, and L. Li. 2020. Meat and fish intake and type 2 diabetes: dose-response meta-analysis of prospective cohort studies. Diabetes & Metabolism 46 (5):345–52. doi: 10.1016/j.diabet.2020.03.004.
- Zeraatkar, D., M. A. Han, G. H. Guyatt, R. W. M. Vernooij, R. E. Dib, K. Cheung, K. Milio, M. Zworth, J. J. Bartoszko, C. Valli, et al. 2019. Red and processed meat consumption and risk for all-cause mortality and cardiometabolic outcomes: a systematic review and meta-analysis of cohort studies. Annals of Internal Medicine 171 (10):703–10. doi: 10.7326/M19-0655.
- Zeraatkar, D., B. C. Johnston, J. Bartoszko, K. Cheung, M. M. Bala, C. Valli, M. Rabassa, D. Sit, K. Milio, B. Sadeghirad, et al. 2019. Effect of lower versus higher red meat intake on cardiometabolic and cancer outcomes: a systematic review of randomized trials. Annals of Internal Medicine 171 (10):721–31. doi: 10.7326/M19-0622.
- Zhao, Z., S. Li, G. Liu, F. Yan, X. Ma, Z. Huang, and H. Tian. 2012. Body iron stores and heme-iron intake in relation to risk of type 2 diabetes: a systematic review and meta-analysis. PLoS One 7 (7):e41641. doi: 10.1371/journal.pone.0041641.