Making sense of the “killer meat” study

Making sense of the “killer meat” study. By Rebecca Goldin Ph.D and Trevor Butterworth
Modest risk suggests meat in moderation, but cancer researchers warn that too much is being made of the link between diet and cancer at the expense of smoking and obesity.
stats.org, March 30, 2009

Hundreds of news stories last week warned people that eating red meat raised their risk for cancer and death. The headline in the Los Angeles Times health section was succinct: “Killer meat,” and the opening graph warned:

“Before you dig into another hamburger, consider this: Americans who ate the most red meat boosted their overall risk of death by 30% during a 10-year period compared to those who ate the least, according to a new study. And before you switch to cold cuts instead, keep in mind that people who consumed the most processed meat raised their overall risk of death by at least 16%.”

Actually, the study didn’t quite say this. While this large prospective study did find a modest association between dying and eating meat, the risks cited were not due to one hamburger. “Meat Intake and Mortality: A Prospective Study of Over Half a Million People” which was published in the Archives of Internal Medicine didn’t, as many other studies on diet have done, pool numerous, smaller studies to achieve a high number of participants. The study tracked over half a million Americans aged 50 – 71 from eight states over ten years and started with a common baseline evaluation of diet, which was then tracked through questionnaires. Naturally, self-reporting always raises questions as to whether the participants are capable of complete fidelity and recall, but the researchers appear to have conducted spot checks, as well as adjust for confounders like smoking.

The researchers compared high levels of red and processed meat consumption (meaning those people in the top 20 percent for meat consumption as a proportion of their calories) to those eating low levels of red and processed meat (i.e. those in the bottom 20 percent consumption level). To give a sense of the difference among the two groups, people with the highest red meat consumption ate almost seven times as much meat as those in the lowest group. For a man, that amounted to 68.1g/1000kcal of meat per day, which is almost a 1/3 lb burger a day (based on the 2116 calorie diet these men typically ate). Those in the lowest quintile of meat consumption ate on average 9.3g/1000, which comes out to approximately the same burger once a week. So before you panic, consider how your red meat intake compares to the people in the study.

On the other hand, there was some good news for meat lovers as well: high levels of white meat consumption seem to lower your chance of death. For those in the highest quintile of white meat consumption (which includes poultry and fish), the risk of death was associated with an approximately eight percent lower chance of death in the ten years of the study, for both men and women. But a curious feature that might temper the benefits of white meat to nonsmokers is that high levels of white meat consumption seems to raise rather significantly their risk of cardiovascular disease. You’re in luck if you’re a smoker, however; for this group, white meat intake seemed to have no relationship to cardiovascular disease.

These were the results driving the interest in the study, although weirdly, the strangest association was between high red meat consumption in men versus low red meat consumption and mortality due to “injuries and sudden death.“

That result – a hazard ratio of 26 percent (meaning 26 percent more likely) – was buried by the media. The category included death from unintentional injury, adverse effects, suicide, self-inflicted injury, homicide, and legal intervention. The authors note that the number of deaths was low, but the mechanism is not clear. The finding is a reminder that mining epidemiological data can produce strange relationships. In particular, since it seems difficult to argue for causality, it suggests that red meat consumption may be linked to other behaviors that were not controlled for by the study. Are male red meat eaters likelier to take risks? Are suicidal old men more likely to eat red meat?

While the study has rather convincingly linked high levels or red meat to increased mortality, the purported risk increase is much lower than it is, for example, between smoking or obesity and cancer. Inevitably, this means that the causal link is weaker. As with any observational study, there are some limitations to drawing a causal line between red meat and cancer mortality. The study attempted to control for these factors, but it is impossible to control for everything. There is also no way to discern from this study whether eating less meat would provide the direct benefit of the magnitude of the study. One can only assume that the people who reported high levels of meat consumption had been eating that amount of meat for their entire lives.

Wider problems in nutrition research

The other, wider problem is that while red meat has provided figurative red meat for nutrition researchers, there has been increased criticism of the dramatic claims being made for the nutritional basis of cancer from actual cancer researchers. Many of the news stories said the study supported the claims by the World Cancer Research Fund linking red meat and cancer. For example, Forbes noted:

“Though nutrition experts frequently recommend eating less meat, Mozaffarian says research linking red and processed meat consumption and mortality weren't consistent. But last year, when the World Cancer Research Fund International reviewed the scientific literature on red meat intake and cancer, researchers determined a link between the two.”

Reuters quoted Ian Olver, Chief Executive Officer of Cancer Council Australia, saying that:

“This large study provides further evidence to support the recommendations by groups such as the World Cancer Research Fund in demonstrating an association between a high consumption of red and processed meats and a increase risk of death from cancer.”

But as STATS previously noted, the World Cancer Research Fund only managed to do achieve this link by excluding the largest ever study examining the association, whose publication had been delayed for three years after the results were initially made known. Those results did not show a link between cancer and meat consumption. The Harvard Pooling Project, which conducted that meta-analysis, and other recent research have thrown a wrench into the conventional scientific wisdom about nutrition and health, and the exclusion of some of its key studies from the World Cancer Research Fund has left some cancer researchers troubled.

A recent editorial in the Journal of Oncology written by the director of the International Agency for Research on Cancer (Boyle et al, Oct 2008) warned that smoking and obesity as significant causes for cancer were being minimized in the face of weak evidence for diet.

"In presenting its summary and recommendations, the [World Cancer Research Fund] report implicitly downplays the key importance of tobacco smoking in cancer causation. Contrary to that stated in the press release (the best advice for cancer prevention is to avoid weight gain), avoiding tobacco smoking and use of tobacco in other forms is the single best advice to reducing cancer risk as one-third of cancer deaths in high-income countries is attributable to tobacco use. Failure to include ‘stop smoking’ and ‘avoid exposure and exposing others to second-hand smoke’ among the 10 key recommendations undermines the most important message in cancer control. The ‘best advice’ also fails to mention the importance of a variety of established cancer risk factors including sun behaviour, occupational exposures, chronic infections and use of exogenous hormones."

At the same time, the evidence presented by the WCRF for diet’s role in cancer had gotten weaker:

"‘We think we know’ or, more accurately, ‘we thought we knew’ that a high-fat diet and low consumption of fruits, vegetables and fibres were associated with increased risks of common cancers. However, faith in the cancer prevention properties of fruits and vegetables began to crack when all the available evidence was critically reviewed by an International Agency for Research on Cancer (IARC) Working Group. Subsequently, it has crumbled as major analyses of prospective studies have continued to demonstrate consistently a lack of association between intake of fruits and vegetables and risk of several cancers. This major change in classification of one the few agents classified by WCRF in the category of strongest evidence in 1997 casts doubt on the rationale to classify ‘convincing’ to the evidence linking high meat intake to colorectal cancer risk in the current report. This also raises questions about the evaluation process and about the robustness of the classification system."

But the IARC noted

"The substantial review of the evidence in the WCRF report demonstrates that there is no discernible association between many forms of cancer and specific dietary practices. There are still some very interesting hypotheses to pursue, such as the value of an approach on the basis of the food patterns (e.g. the Mediterranean diet score) rather than individual foods and nutrients, but the cupboard is remarkably bare."

The failure of science to come up with robust conclusions about diet and cancer is one of the emerging "inconvenient truths" in public health (the other is that diets don't really work), and both are at odds with giving the public clear, comprehensible guidelines for diet. This new study has been hailed for building on existing evidence that red meat consumption is linked to cancer, but good reporting would include the naysayers as well as the yaysayers; scientific consensus is never built with one study alone.