Buscar

Disclaimer / Advertencia

Advertencia: la información contenida en esta web es obtenida a partir de artículos científicos publicados en revistas médicas y es presentada de la forma más objetiva posible. No obstante, tanto la posibilidad de error, como el cambiante estado de la ciencia impiden garantizar que dicha información es absolutamente fiable o completa. En cualquier caso nunca, bajo ningún motivo, el lector debe intentar ser su propio médico a partir de los conocimientos que pueda adquirir a través de esta web. Su contenido no debe usarse para diagnosticar o tratar problema alguno. Si tiene o sospecha la existencia de un problema de salud, consulte a su médico de cabecera.

 

Red Meat Mortality Relative Risk

DOES READ MEAT INTAKE REALLY INCREASES SEVERAL MORTALITY RISKS?

 

Meat intake and mortality

            Several recent epidemiological studies associated red meat and processed meat intakes to increased total mortality, heart failure, diabetes and cancer risks. Statements such as “men with the highest average intake of red meat (almost 10 servings per week) were at a 24 per cent higher risk of heart failure than men with the lowest average weekly intakes” make newspapers headlines everyday and make everyone aware of potential adverse health effects of red and/or processed meats. These apparently impressive results are often used by the media to scandalize the population, and to catch attention. However, several key points must be considered before making general population recommendations from these epidemiological studies.

            In one of these recent studies, the “Meat intake and mortality: a prospective study of over half a million people” (Sinha et al.) [1], the author’s main conclusion was that “red and processed meat intakes were associated with modest increases in total mortality, cancer mortality, and cardiovascular disease mortality”. This was a prospective study of, as referred, more than half a million people, aged 50-71 years at baseline, followed from 1995 to 2005, where meat intake was estimated with a food frequency questionnaire according to quintiles of red meat, white meat and processed meat intakes. A casual look at this study leads one to believe that red meat was analyzed separately from the processed meat, but this is not the case. The red meat group also included processed red meats, from grain-fed animals.

            Red meat intake ranged from 9.3grs/1000kcal in the first quintile to 68.1grs/ 1000kcal in the fifth quintile. The multivariate model was adjusted for several food and lifestyle covariates, such as: age, education, marital status, family history of cancer, race, body mass index, 31-level smoking history, physical activity, energy intake, alcohol intake, vitamin supplement use, fruit consumption, vegetable consumption and menopausal hormonal therapy among women. Notice the authors assumed that saturated fat (SAFA) and cholesterol content in read meat would be responsible for the development of cardiovascular disease (CVD) and cancer, hence they might be biased in their hypothesis. These preconceived ideas frequently influence the selection of adjustment parameters, leaving other potentially more important factors ignored.

Relative and absolute risks

            The results of “Meat intake and mortality: a prospective study of over half a million people” (Sinha et al.) [1] showed namely that the maximum relative risk of increased all-cause mortality for men, that between the extreme quintiles of red meat intake, apparently is 31%. Also, according to [6], “if men and women in the studied age group (50-71) would reduce their intakes of red meat to that of the group with the lowest intake, then the mortality risk is expected to be reduced by 11% in men and 16% in women over the observed period of time. The portion of cardiovascular disease mortality of total mortality could be reduced by 11% for men and 21% for women. By reducing the intake of processed meat, the cardiovascular disease mortality for women over the period of study of 10 years could be reduced to 20%.”

            At first glance, all these statistical numbers look very impressive, and thus convincing, but they are quite similar to relative risks, which don’t take into account the size of the observed population. Actually, in this Sinha et al. study these ratios are actually hazard ratios estimated by Cox proportion. These are similar to relative risks, they also ignore the population size. Although ratio measures are commonly reported in the medical

literature, the underlying absolute risks are not. In a review of ratio measures in six major medical journals, Schwartz et. al. [12] found that “the underlying absolute risks were often difficult to access or were missing altogether”. These researches explain that “the lack of accessibility of these fundamental data may well lead journal readers (doctors, policy makers, journalists, and patients) to have exaggerated perceptions of the reported effect sizes”.

            Absolute risks and the so called “numbers needed to harm/treat” provide a much better picture of the (dis)advantages of a certain food intake and/or lifestyle change, in this case the red meat intake. When it is said that reducing red meat to that of the group with the lowest intake, the mortality (relative) risks would be reduced by 11% in men and 16% in women, we estimate that corresponding absolute risk reductions of such measures would be approximately 2.1% and 1.4%. On what concerts total mortality, we estimated that the equivalent absolute risk translates into about 4% from the lowest to highest quintile of red meat intake. In a similar way, the maximum absolute risks of cancer and CVD deaths were even smaller, probably close to 1.3% and 1.0% respectively. Also notice these are just statistical/observational associations, establishing causality requires that a plausible biological/biochemical mechanism is identified in carefully designed interventional studies. These causality proving studies against red meat intake simply don’t exist until now.

            Despite of these unfavorable statistical associations found for red meat intake and several unhealthy conditions that lead to increased mortality, even if direct biological causality existed, these results might still be not very relevant for most people, those in the many countries where their red meat intakes fall in the lower quintiles of this American study. For example, in the case of Germany, as discussed in a review of the Federal Institute for Risk Assessment, the average intake of red meat for men lies between the 2nd and 3rd quintile of the US data, and that of women between the 1st and 2nd quintile [6]. So the current, absolute amounts of red meat intake in each country should also be taken into account before assuming that recommending eating less red meat would provide any relevant mortality risk reductions.

Questionnaire inaccuracies

            According to Fraser [14], “the potential correlations between nutrients, and to a lesser extent foods, make it difficult to know whether the nominated variable is actually the active principle or whether there is some other dietary risk factor that is closely associated. It is not generally recognized that all traditional analyses of this sort are based on a powerful but incorrect assumption: that there are no errors in dietary assessment. If the incorrect assumption is not satisfied, relative risk estimates become distorted—reduced by one-half or more in some cases.”

C. Masterjohn text/ideas:

            According to Chris Masterjohn, a researcher affiliated with the Weston A. Price Foundation, because of important questionnaire inaccuracies, the Sinha et. al. study found “a correlation between increased mortality and a population's propensity to report eating meat, not a correlation between mortality and true meat intake”. In an extensive blog article, Masterjohn explains that the food frequency questionnaire (FFQ) contained 124 questions, each question about a particular type of food or group of food. Participants were asked how often they consumed those foods over the course of the previous year, giving them ten options. Then it asked how large of a serving size they consumed, giving them usually three or four options. Sometimes they were given additional instructions, like including sandwiches in some cases or excluding sandwiches in other cases.

 

            The problem is any individual trying to quantify his or her average intake of 124 foods over an entire year is going to have to engage in a lot of guess work. Even 24-hour recalls of what a person ate the day before are subject to a great deal of error. For this reason, researchers will commonly "validate" an FFQ or a 24-hour recall to test whether these accurately measure the intake of the foods of interest. In order to do this, they have the participants make a weighted dietary record where they meticulously weigh everything they eat with a dietetic scale and record it as they prepare each meal. Then the researchers compare the FFQ or 24-hour recall to the weighted dietary record, assuming that the weighted dietary record is the best indicator of true dietary intake.

 

            According to Masterjohn’s article, this type of validation was not applied and, instead, a 24-hour recall was used. With this simplified procedure, “the author's validation study found that the true intake of protein, carbohydrate, fat, cholesterol, fiber, vitamins, minerals, fruits, and vegetables could explain between 5 percent and 45 percent of the variation in the participants' answers on the FFQ, but they never validated the FFQ's ability to predict the true intake of meat”. When working with this type of questionnaires, other researchers found that FFQ predict true intake of some foods very well, and true intake of other foods very poorly. The ability of FFQ to predict true intake of meats is probably horrible. When some foods, in a certain cultural context, are socially and emotionally charged, participants are more likely to lie about their intake of those foods, or more likely to deceive themselves about how much of those foods they are really consuming.

            As an example of FFQ inevitable inaccuracies, consider what the researchers who validated the Nurses' Health Study FFQ had to say: “Focusing on the second questionnaire, we found that butter, whole milk, eggs, processed meat, and cold breakfast cereal were underestimated by 10 to 30% on the questionnaire. In contrast, a number of fruits and vegetables, yoghurt and fish were overestimated by at least 50%. These findings for specific foods suggest that participants over-reported consumption of foods often considered desirable or healthy, such as fruit and vegetables, and underestimated foods considered less desirable. This general tendency to over-report socially desirable foods, whether conscious or unconscious, will probably be difficult to eliminate by an alteration of questionnaire design.”

Confounding factors (and results)

            Should we consider these results meaningful enough to establish solid recommendations for the general population? Are all these epidemiological data sufficiently adjusted to the several lifestyle confounding factors that exist, and thus trustable? When dealing with large populations, it is difficult to control for all possible causing factors and, although results are adjusted for several factors, others cannot be ruled out, as the authors state. For example, the way meat is produced and cooked may affect the production of carcinogenic compounds such as heterocyclic amines, polycyclic aromatic hydrocarbons, nitrates, nitrites and N-nitroso compounds [15]. Other factors which could be involved in the potential disease promoting properties of meat, depending on the way it is produced and/or cooked [16], are the time meat stays in the intestines, fruit and vegetable consumption, hormone residues or salt.

            Individuals consuming red and processed meats also have lifestyle behaviors that may significantly affect mortality types, like education, physical activity, smoking, alcohol use, adiposity and fruit/vegetable intake. According to Mozaffarian [2], the model of Sinha et al. did not adjust for parameters like income, air pollution exposure, intake of high-glycemic index starches, sugars, and processed foods, and lower intake of dietary fiber, whole grains, and nuts, seeds, and legumes. Also there was no record about personal history of CVD, or related conditions like hypertension, diabetes, dyslipidemia, nor the use of medications, and such these factors were also not included in the multivariate analysis [4]. As referred above, the authors assumed that SAFA and cholesterol content in read meat would be related to CVD but, despite of this, they don’t inform of differences in saturated fat intake between quintiles or if this was included at all in the multivariate analysis.

            Given the known relationship between glycated hemoglobin (HbA1c), diabetes and heart disease, without the carbohydrate intake information (modern high-glycemic index starches and sugars tend to raise HbA1c) this study could not evaluate if it was the eventually higher amounts of carbohydrates, consumed along with the higher portions of meat, that raised the mortalities. Also, since high blood sugars tend to suppress the immune system and, at the same time, feed glucose to cancer cells, this might help explain the link found between increased cancer deaths and higher red meat intakes. Other similar observational studies, where meat intake was more accurately estimated, didn’t found such association.

            As Mozaffarian [2] refers in his comments, in such a large population, with broad social/economical, geographic, ethnic/cultural, and lifestyle diversity, adequate control for confounding factors assumer even higher importance. In fact, the small observed risk differences (eg, relative risks of 1.1-1.3), that are rendered statistically significant because of the large cohort, are those most susceptible to being due to bias. In these cases, the use of a “negative control”, for which an outcome for which the exposure would have no plausible mechanism, would be recommended. The multivariate model should be adjusted/calibrated until the association of observed variables with the negative control shows no effect. In the Sinha et al. study, such residual confounding actually exists, the “all other deaths” item. Since the association found of meat intake with this item was the strongest found in this study, this strongly suggests that insufficient adjustment for residual confounding variables is present.

Healthy cohort effect

            Regarding other factors involved in mortality associated to meat intake, it is noteworthy to mention that, in this Sinha et al. study, the participants in the top quintile of red meat consumption were 3 times more likely to smoke, half as likely to do regular exercise, much less likely to have a college degree, were substantially heavier and also had higher caloric intakes than low red-meat consumers [3].

             Furthermore, it is common that people who believe that red meat is not healthy tend to adopt other lifestyle factors that favor their health like doing more exercise, not smoking, not drinking alcohol, not frying and taking supplements. In summary, cause-effect cannot be deduced from observational studies as cancer or CVD are multifactorial diseases, in which several genetic and environmental factors are involved. If some relevant (unknown?) confounding factors are not adjusted in the multivariate model, then the risk factors are often overestimated. Large samples, such as this one, with different ethnic, socioeconomic, geographic, cultural and lifestyle need to be carefully controlled for many confounding factors.

            Also, when health authorities decide that a certain food and/or lifestyle behavior are healthy, from that point it becomes increasingly difficult to measure their impact on public health with epidemiological studies. This happens because health conscious people tend to gravitate toward the official, mainstream recommendations. As Dr. Stephan Guyenet, a researcher from University of Washington, explains in his blog Whole Health Source, “if a theory manages to become implanted early on, it will become a self-fulfilling prophecy as healthy, conscientious people adopt the behavior and are detected by subsequent observational studies. People who don't care about their health or aren't motivated enough to make a change will keep living how they used to, and that will also be detected”.

 

            Dr. Guyenet then adds that “you can't measure all the little things that accompany a health-conscious lifestyle. Do the participants take the stairs or the elevator? Do they take supplements, and if so, which ones? How much sunlight do they get? Do they have positive relationships with their friends and family? (…) What is the quality of the foods they buy? How often do they visit the doctor, and how often do they follow her advice? I believe there are too many confounds to measure and correct for. In my opinion, this means that observational data gathered from populations that already have opinions about the factor you're trying to study are unreliable and will tend to reinforce prevailing notions.

Evolutionary evidence above epidemiological uncertainties

            As a whole, epidemiological/observational studies can only provide statistical associations, but an epidemiological association does not equal causation and as we well know, very often, epidemiological studies show something and then randomized controlled trials show no association whatsoever, or even the opposite. They should not be completely worthless, but only by themselves we believe they are not sufficient to establish solid recommendations and dietary guidelines for the general population.

            Given all these epidemiological limitations, it is then worthwhile not forgetting the most powerful paradigm in human health: evolution. It is weird that a food which has been part of the human diet for 2.6 [17], or even 3.4 [18], million years is now responsible for the cause of CVD or cancer, among other diseases associated to meat intake [19]. Our genome was shaped during the long period of the paleolithic era with little change (0,005%) since the agricultural revolution, 10,000 years ago, despite an enormous change in human diet and other lifestyle factors.

            Nowadays, more than 70% (cereal grains, dairy products, refined vegetable oils and refined sugars) of the calories of the typical western diet come from foods unavailable for our ancestors during the paleolithic era. This discordance between our ancient, genetically determined biology and the nutritional characteristics of the actual diet may be the real cause of the so called diseases of civilization, including CVD and cancer [17].


Ricardo Carvalho and Maelán Fontes1


1. Center for Primary Health Care Research, Faculty of Medicine, Lund University


References

[1] Sinha R, Cross AJ, Graubard BI, Leitzmann MF, Schatzkin A. Meat intake and mortality: A prospective study of over half a million people. Arch Intern Med 2009;169:562-71.

 

[2] Meat intake and mortality: evidence for harm, no effect, or benefit? Mozaffarian D. Arch Intern Med. 2009 Sep 14;169(16):1537-8; author reply 1539.

 

[3] ACP Journal Club. High consumption of red meat and processed meat was associated with increased risk for mortality. Schectman J. Ann Intern Med. 2009 Jul 21;151(2):JC1-15.

 

[4] Higher red meat intake may be a marker of risk, not a risk factor itself. de Abreu Silva EO, Marcadenti A. Arch Intern Med. 2009 Sep 14;169(16):1538-9; author reply 1539.

 

[6] - Study on meat intake and mortality: BFR opinion. BfR - Federal Institute for Risk Assessment, Germany

 

[7] Stewart FH, Shields WC, Hwang AC. Presenting health risks honestly: Mifepristone, a case in point. Contraception 2004;69:177-8.

 

[8] Montori VM, Jaeschke R, Schunemann HJ, Bhandari M, Brozek JL, Devereaux PJ, Guyatt GH. Users' guide to detecting misleading claims in clinical research reports. BMJ 2004;329:1093-6.

 

[9] Felicia H. Stewart, Wayne C. Shields, Ann C. Hwang, MD. Presenting health risks honestly, mifepristone a case in point. Contraception 69 (2004) 177–178.

 

[10] Victor M Montori, Roman Jaeschke, Holger J Schünemann, Mohit Bhandari, Jan L Brozek,

P J Devereaux, Gordon H Guyatt. User’s guide to detecting misleading claims in clinical research reports. BMJ 2004;329:1093–6.

 

[11] Ratio measures in leading medical journals: structured review of accessibility of underlying absolute risks (BMJ)

Simple tools for understanding risks: from innumeracy to insight (BMJ)

 

[12] Edward F. Vonesh. Relatives risks can be risky. Peritoneal Dialysis International, Vol. 13, pp 59.

 

[13] Schwartz LM, Woloshin S, Dvorin EL, Welch HG. Ratio measures in leading medical journals: Structured review of accessibility of underlying absolute risks. BMJ 2006;333:1248.

 

[14] Gary E Fraser . A search for truth in dietary epidemiology. American Journal of Clinical Nutrition, Vol. 78, No. 3, 521S-525S, September 2003.

 

[15] Sugimura T. Nutrition and Dietary Carcinogens. Carcinogenesis. Vol. 1 Nº 3 pg 387-395. 2000

 

[16] Sinha, R., Rothman, N., Brown, E., Salmon, C., Knize, M., Swanson, C., Rossi, S., Mark, S., Levander, O., and Felton, J. 1995. High concentrations of the carcinogen 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) occur in chicken but are dependent on the cooking method. Cancer Research 55(20): 4516-9.

 

[17] Cordain L. Et al. Origins and evolution of the Western diet: health implications for the 21st century. Am J Clin Nutr 2005;81:341–54

 

[18] McPherron SP. Evidence for stone-tool-assisted consumption of animal tissues before 3.39 million years ago at Dikika, Ethiopia. Nature.  Vol. 466. 12 August. 2010

 

[19] Ashaye A. Red meat consumption and risk of heart failure in male physicians. Nutrition, Metabolism & Cardiovascular Diseases (2010).

 

 

Maelan Fontes Villalba, MS

Maelan Fontes Villalba, MS

Algunas recomendaciones de alimentación están basadas en estudios observacionales y no siempre los datos de los estudios epidemiológicos se corroboran con estudios de intervención. Es decir, los datos de estudios observacionales establecen asociaciones, pero no causa-efecto, que permiten formular hipótesis para posteriormente testarlas con estudios de intervención.

Leer más

Universidad de Lund

Anuncio

NutriScience

Miembro del equipo de:
Anuncio

Conf. Dr. Staffan Lindeberg

Ver video
Video de la conferencia aquí

Canal Youtube

Visita mi Canal en Youtube
Anuncio

Acceso usuarios

El registro permite acceder a más contenidos de la web.