Non-communicable disease risk associated with red and processed meat consumption—magnitude, certainty, and contextuality of risk?

Mean global intakes per person of red and processed meats are 51 and 17 g/day respectively. Consumption is lowest in South Asia (7 and 3 g/d), and highest in Central Europe/Asia (114 and 54 g/d). While some researchers claim that red meat consumption is intrinsically harmful, the evidence does not support this being the case where intakes are below 75 and 20 g/d, respectively. Even beyond these intake levels, only small increases in relative risks are reported (<25%), there is little to no effect on absolute risk, and the certainty of evidence remains low to very low based on the best available summary evidence. Importantly the relationship is not necessarily causal - when meat consumption is part of healthy dietary patterns, harmful associations tend to disappear, suggesting that risk is more likely to be contingent on the dietary context rather than meat itself. Despite being a foundational part of human evolutionary diets and a source of high-quality protein and bioavailable micronutrients in a global context of nutrient insecurity (see elsewhere in this Special Issue; Leroy et al., 2023), the consumption of red and processed meats is nowadays increasingly discouraged by a vocal group of scientists and organizations. The rationale for this is based on a purported association of their intake with an increased risk of obesity and non-communicable diseases, such as myocardial infarction, stroke, diabetes, and particular cancers. As for most foods, one could assume that there may be compelling evidence for optimal intake levels, balancing the potential benefits and harms of meat as a human food. Such an optimum, however, is difficult to estimate, as the evidence is highly contextual and complex, or even conceptually incorrect to begin with. A first factor complicating the reliable and universal estimation of optimal intake levels has to do with interpersonal variability, based on differences in genetics, sex, age group, health status, socio-economic background, etc. For instance, whereas some population groups with relatively higher iron requirements may benefit from more meat, others may be prone to iron accumulation and overload. A second problem is that optimal intake levels are defined by the background diet and the lifestyle of an individual. Assuming that meat would indeed have to be considered as a health hazard, as proclaimed by the IARC (Bouvard et al., 2015), despite controversy (Klurfeld, 2018), this still leaves the question of its role in disease risk. Appropriate risk assessment requires considering the frequency and amount of meat intake, preparation method, and interactions with other compounds in the diet. There is considerable scientific debate over the certainty of the evidence associating intake with morbidity and mortality based on what has been provided by many nutritional epidemiologists vis-à-vis the absence of long-term randomized trials of red and processed meat intake and clinical health outcomes. All this makes it difficult to establish an upper limit for safe consumption. Whereas some researchers go as far as claiming that no amount is harmless, others oppose this assumption (Klurfeld, 2018; Johnston et al., 2019; Leroy and Cofnas, 2020; Stanton et al., 2022). Taken together, the following issues need to be addressed: 1) what are the methods and limitations of epidemiological research, 2) how can we evaluate the certainty of the underpinning evidence, 3) what can be inferred from the current data with respect to safe or optimal intake levels, 4) what is the role of exposure in the framework of risk assessment, and 5) how can we arrive at a trustworthy message, by contextualizing these findings and relating them to the various benefits of meat (products), including contributions to food culture and nutrient security? Patients, members of the public, and health care professionals should rely on the best available evidence to guide their lifestyle and health care decisions. Typically, the standard for making causal connections between an exposure and the risk of desirable (benefits) and undesirable (harms) health outcomes is to conduct robust randomized controlled trials of the putative causal factor and to measure clinical outcome events. For optimal decision-making, clinical trial evidence should then be summarized using high-quality, up-to-date systematic review, and meta-analysis methodology. In the absence of long-term trials of red and processed meats as isolated dietary interventions for health outcomes, dietary guidelines and policies can be informed by high-quality systematic reviews with meta-analysis of the evidence originating from observational studies. Informed decision-making requires knowledge of the magnitude of the potential benefits and adverse health outcomes—ranging from trivial to large—and the corresponding certainty of evidence for all important health outcomes. These outcomes include quality of life, mortality, and major morbidity (e.g., stroke, cancer incidence). Investigators may express the impact of an intervention or exposure for dichotomous outcomes in either relative terms (i.e., odds ratio, relative risk, or hazard ratio) or in absolute terms (risk difference, also known as absolute risk reduction or increase, or as the number needed to treat or harm). There are upsides, as well as downsides, to the presentation of exposure or treatment effects using either approach (Alonso-Coello et al., 2016). Exclusive use of relative risk estimates can be highly misleading, since the relative risk typically yields larger, often much larger, treatment/exposure effects than if absolute risk is used. For example, a relative risk of 0.50 equivalent to 50% relative risk reduction can, when based on a low baseline risk, mean an absolute risk reduction of a mere 1%; i.e., from 2% to 1%. This difference not only influences the judgment of lay persons and leads to hyperbolic public discourse (cf. Leroy et al., 2018), but also affects clinicians and policymakers. Relative effect estimates, however, are usually—though not always—similar across populations and subgroups, whereas absolute effect estimates typically vary with the baseline risk. Therefore, expressing a treatment/exposure effect estimate as only an absolute risk is also misleading, because it will under- or overestimate the effect for patients at high or low baseline risk, respectively. As a result, in the context of conducting and using meta-analysis for decision-making, one may need to apply the relative effect estimate to a range of baseline risks typically seen in the population of interest (Guyatt et al., 2013). This may require ascertaining clinically identifiable risk groups and clarifying the period over which the associated baseline risk applies, ideally based on the largest available cohort study or a summary of cohort or controlled studies that best represent one’s population of interest, as was done in the recent NutriRECS guideline on red and processed meats (Johnston et al., 2019). To optimize data interpretation, systematic reviews and meta-analyses should always present the estimates of absolute risks in intervention/exposure and control groups, alongside the corresponding relative risks, together with the 95% confidence intervals for all important desirable and undesirable outcomes. Cochrane, the Joanna Briggs Institute, and the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) system all endorse and require this approach to data presentation. The above-mentioned treatment or exposure effects, which lead to risk estimates, can be obtained from a set of different study designs. The latter are usually categorized as observational studies, e.g., cross-sectional, case-control, and cohort studies, and interventional studies with either humans or animals. It is important to emphasize that observational studies on their own very rarely allow the inference of causal relationships. Causal relationships from observational data should be supported by large exposure effects, e.g., relative risk > 2.0, and additional evidence based on intervention studies, plausible mechanisms, clear dose–response relationships, etc., as also echoed in the Bradford Hill criteria (Hill et al., 2022). For good practice, the certainty of estimates for each target outcome using the GRADE system should be considered when systematically evaluating a body of evidence (see below). When the certainty is low, authors should avoid causal inferences or strong public health recommendations (Schünemann et al., 2011). The limitations that are inherent to nutritional epidemiology can be illustrated by the case of the IARC Monograph Working Group evaluation of the carcinogenicity of red and processed meat (Bouvard et al., 2015), and the criticism related to this procedure (Klurfeld, 2018). According to IARC methodology, most weight was given to prospective cohort studies, supported by case-control studies and information from mechanistic studies. It was concluded that red and processed meat consumption was not a hazard for almost all cancers, except for colorectal cancer. The hazard classification of the latter was primarily based on positive associations in 7 of 14 cohort studies on red meat and 12 of 18 cohort studies on processed meat. Importantly, it was also concluded that chance, bias, and confounding could not be ruled out with the same degree of confidence for the data on red meat compared to processed meats. Red and processed meats were nonetheless classified as hazards for colorectal cancer from a precautionary principle, although this was done without further assessment of risk (cf. “Contextuality and risk assessment”). The focus on observational research by IARC and similar organizations relates to the fact that human intervention studies of a size and follow-up period sufficient to measure people-important health outcomes of noncommunicable diseases are incredibly challenging and costly. Therefore, if data from intervention studies is available, this typically comes from short-term human intervention studies, in which biomarkers are measured as proxies, or from animal studies. In human intervention studies, the effects on biomarkers are small or neutral, and sometimes even benign, for example, in the case of red meat and glycemic control and inflammatory biomarkers (O’Connor et al., 2021) or cardiovascular risk factors (O’Connor et al., 2017; Zeraatkar et al., 2019). Animal studies suffer from an often nonrepresentative dietary context and from problems of extrapolation (issues of indirectness according to GRADE). Because of such limitations, the evidence to confirm a mechanistic link between the (moderate) intake of unprocessed red meat as part of a healthy dietary pattern and colorectal cancer risk should be considered insufficient (Turner and Loyd, 2017; Kruger and Zhou, 2018; Johnston et al., 2019; Lescinsky et al., 2022), and similarly so for cardiovascular diseases (Johnston et al., 2019; Mente et al., 2020; Delgado et al., 2021; Lescinsky et al., 2022). In the domain of life sciences, the reliability of evidence varies widely across studies. Making health recommendations, either strong or weak, needs to be based on a common, rigorous, and transparent evaluation of the certainty of the evidence for all health outcomes. The current standard in guideline development is the GRADE system. The latter is used by >110 organizations worldwide (e.g., Cochrane, the World Health Organization, and the Centers for Disease Control) and comprehensively and transparently allows for rating the certainty of evidence based on systematic review(s) of the As a common, transparent GRADE of the evidence to the evidence and or with the For a given research GRADE certainty using low, or very low for each target outcome A systematic of the evidence as certainty evidence when it is based on randomized clinical trials and certainty evidence when based on observational studies, e.g., cohort and case-control studies. of the certainty as in the Grading of Assessment, and Evaluation (GRADE) system of the certainty as in the Grading of Assessment, and Evaluation (GRADE) system reviews of randomized trials at high certainty evidence because a high of control for confounding that may the reviews of such trials can be for issues of risk of bias, or risk of In systematic reviews of observational studies are considered low certainty evidence because are not in a controlled rely on and are at high risk of even for certainty can be if there is a large effect of treatment or exposure that confounding is to on the of this with a relative risk > or a risk compared to et al., or if there is evidence of a dose–response and Johnston 2019). high or certainty evidence may in strong do recommendations if the benefits harms and (e.g., or evidence leaves the impact of in The latter for decision-making, potential and are between such as and health care so that can their own and decisions. While GRADE is based on over of the GRADE system may in their and of the of the system is that it requires with are when making on the certainty of evidence for each and when the of the to be transparent with all different researchers or groups of researchers may arrive at different When use the same by transparent on the certainty of evidence and of such as and are to informed decisions. an at cardiovascular risk to the evidence for meat and the risk of stroke, a standard most vocal nutritional et al., however, claim that an should be for the nutritional sciences, with a higher for evidence in the or et al., The rationale for this claim relates to the fact that randomized trials are difficult and to in the that should be more robust across all health et al., To for trustworthy and population dietary recommendations, a common, and reliable framework is needed for transparent evaluation of 1) the magnitude of relative and absolute estimates of and the certainty of evidence for each important health outcome et al., 2021) and GRADE the NutriRECS reported on systematic reviews of all randomized trials and cohort studies (Johnston et al., 2019). The reported only or very evidence that diets in either red meat or processed meats could have impact on the risks of important outcomes infarction, stroke, and and for cancer and the a that most should their current red and processed meat Despite these findings and recommendations, some recent of red and processed meats from the human diet. The for instance, a intake of red and processed meats of 14 et al., one per The of Disease and the on Health and reported that were by consumption of unprocessed red meat et al., 2020; et al., et have that each of a in of life and et that from a which and of unprocessed red meat and processed to a diet which these foods, would life by for and by for and estimates by et et and are on 1) that the risk exposure levels or optimal intakes of red and processed meats are and 2) that risk even with consumption of red and processed meats. these do not to be with the from the large epidemiology studies that have the relationships of red and processed meats with mortality and Relative risk estimates for mortality from all cohort studies of populations at by levels of unprocessed red meat Relative risks and 95% confidence intervals are on the and red meat intake on the and studies are in and respectively. first and of are for each cohort are provided in the Relative risk estimates for mortality from all cohort studies of populations at by levels of processed red meat Relative risks and 95% confidence intervals are on the and processed meat intake on the and studies are in and respectively. first and of are for each cohort are provided in the For unprocessed red meat, an intake range of to is not associated with risk in of the large cohort studies of populations higher intake levels, and being contingent on the studies, from reported small increases in relative risk at to 75 g/d, the global study from studies, studies, and one study either no increased risk at intake or only increased risk at intakes 75 For processed meats increased relative risks were reported in most studies where intake intakes to 20 g/d), only studies from reported relative risks The to that there is different the risk estimates in study Typically, studies that the that more meat were also more likely to be current and to have and levels, a higher body and intake of and and confounding by these factors is very of the studies the unprocessed red meat for processed meat, no study to has for food intake, despite for of the in the diets of et al., 2016). are of of and typically with using a of little if are often and increasingly associated with noncommunicable diseases et al., 2018). it that related to meat intake need to be using methods (Hill et al., and that there is no robust to assume a for unprocessed red meat of or even below the range of to 75 g/d, equivalent to to For processed the be at to 20 even g/d), equivalent to to 3 Even if such low could be then these risks would still need to be not these foods, in of increased nutrient elsewhere in this Special Issue; Leroy et al., It is that on of the in to Stanton et has that the of the red meat to in the was not and that the of for red meat will be in all 2022). The from the of and the World et al., for further or of the of for unprocessed red meat by in their estimates very likely that of the in the will impact many other studies et al., 2019; et al., 2021; et al., 2021; et al., 2022), if do assume red and processed meats of the their systematic of the health effects associated with consumption of unprocessed red meat, that the evidence for increased risk in disease or mortality is weak, and insufficient to strong or recommendations et al., 2022). their that the 95% for the for unprocessed red meat is very from to the of red meat and processed meats from to more still estimates as (e.g., to 75 and to 20 g/d, to be by further a different on the current consumption levels of these The reported that the mean global intake per person of unprocessed red meat and processed meats in was 51 and 17 g/d, et al., 2022). there is considerable in and processed meat consumption are lowest in South Asia (7 and 3 g/d, in including the and and g/d), and highest in and Central Asia (114 and 54 g/d). estimates of optimal intakes of meat with estimates of consumption are of in the of populations with and higher than optimal the of and to human and it should be that the certainty in may low given the of diet and lifestyle patterns, and the limitations of current methods of as Red and processed meats are of many diets across the and are typically together with other in the of meats and the effects on are by the and diet as well as by factors and the et al., This is very and interactions may of and for instance, may be associated with cancer risk estimates increased meat to the of or (e.g., et al., confounding lifestyle factors are usually for in observational studies by including these in the that the risk estimates will be however, is likely in nutritional epidemiology given the of diets and and the limitations of control for these relationships. may if disease risk assessment of food groups such as red processed meat makes to begin and nutritional observational studies should not be to dietary and is also Because relative risks for meat tend to when for known confounding for the Health study et al., 2022), the higher disease risks associated with a relatively high meat consumption are related to the dietary not the intake of red or processed meat as such (Johnston et al., 2019; Zeraatkar 2019). To further the association of iron intake with colorectal risk is likely by the dietary et al., 2016). This may to the fact that one of the for the red meat and colorectal cancer association is the effect of e.g., by using and in may this as in of studies et al., an effect of has been et al., 2013). In present in and may food preparation and et al., The of harmful compounds the of a and between and the of and as it is to a food or food as or in this This is when the of iron in for human consumption are compared to the much higher and used in studies (Turner and the degree of of a in meat for is likely contextual and its impact on health remains a of to and in the evidence et al., and et al., 2019). the same it has clear that the for potential adverse effects of red meat consumption on cardiovascular diseases, i.e., support from studies and with the of some or et al., This that either on associations based on observational studies be or that these if to be are a of interactions that be by a The latter further that one should be in a food or food for a health the scientific on the potential associations between meat and noncommunicable diseases is often no a transparent assessment of the evidence, but is by including and It is nonetheless important to high of evidence a to the of dietary recommendations and To this the GRADE approach a for and of evidence, making use of that are across health A for from such is to be from the evidence in this a reduction of meat intake below the current levels of consumption is not supported to public for health This to be the case when study quality (e.g., case-control cohort studies randomized clinical trials robust systematic reviews with meta-analysis of all available when confounding is to a higher when are as absolute estimates of effect and the certainty of evidence for the estimates is based on the GRADE red and processed meats are in a healthy background diet. on the of the dietary recommendations should focus on healthy to rather than on foods, that red meat food is a good source of that are not always available from other dietary in from (see elsewhere in this Special Issue; Leroy et al., health or dietary guideline for a reduction of meat intake should that this may also lead to additional and from and and A more robust of potential harms and benefits of red and processed meat consumption is This relates to differences in interpersonal and needs as well as to differences in dietary and which in the case of processed if or are used. Johnston is an in the of and at obtained in from the of and at in Health and in and As the and of NutriRECS and leads an of researchers and research high nutritional guideline recommendations on major public health is a for dietary guidelines and is a in Animal at as a and obtained a in research with quality of foods, related to factors and with a strong interest in the impact on human and is of the of and as a of the Working Group at IARC on the evaluation of the carcinogenicity of red and processed meat. Leroy as a and obtained a in at the where a in food and research with food human and animal and food studies. is a of various i.e., the of and for and a also on various (e.g., the World and the on Mente degree in from the of is an in Health at the Health Institute, the largest study of intake and cardiovascular disease risk, and mortality in the a study of from on Mente has an on the role of and in cardiovascular disease in also has large systematic reviews of diet and cardiovascular Stanton is a is a in at the of in and of Health at on the to and as an for the is a of the and Health and of the World and Health has many healthy diets from food including the at the

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.