This week, Harvard published the results of a study which concluded:
Red meat consumption is associated with an increased risk of total, CVD, and cancer mortality. Substitution of other healthy protein sources for red meat is associated with a lower mortality risk.
I wasn’t even going to bother writing about it due to its utter silliness but I continue to get emails from multiple people asking me if they should give up meat because they read about a study on the front page of the LA Times called “All Red Meat is Bad For You” or on Yahoo called “Consuming Red Meat Increases Risk of Premature Death” or on any of the other major news sites picking up on it from around the web. Now I realize that sensationalized headlines drive readers to your site, but I truly feel there is a point where we cross the lines of journalistic responsibility.
Before we tackle the issues related to the Harvard researcher’s stated conclusions, let us talk about the study design itself. First a little background on epidemiology, and why it shouldn’t be used to form causal relationships.
Epidemiology is the observation of patterns occurring in well defined populations. It is often used by researchers in the public health arena to develop hypotheses which are subsequently studied further in more controlled studies. Most of the time epidemiological research will be categorized into one of a few types such as ecological data studies, case-controlled studies, and follow-up studies.
Ecological data studies take data from a large populations of people and compare the data with data from other large populations. As you can imagine, this often amounts to an apples-to-oranges comparison scenario and famous studies such as Ancel Keys’ “Seven Countries Study” would fall into this category. As any readers familiar with the severe failures of that study can attest, this is not a preferred method for developing cause and effect. I will leave that to a future post to elaborate on.
Case-controlled studies amount to comparisons between individuals who have experienced something (let’s say a heart attack or stroke for this example) and then compared to others in their geographic area who have not suffered from a heart attack or stroke. Researchers interview individuals from both groups to attempt and theorize what may be different between the groups as it relates to the cause of the heart attack. Often things like recall bias can render the results of these studies less than beneficial
The last of the study types is the prospective study, which the Harvard meat study falls under. It is also sometimes called a follow-up or cohort study. Essentially, a group of individuals is recruited to participate in the study and then interviewed about their diet and lifestyle choices for a given period of time. The group of individuals is then followed for a set length of time to see who develops diseases or whatever the researchers are interested in analyzing. As you can imagine, this is far from ideal as you are relying on participants giving accurate data and remembering facts from many years prior while researchers are busy attempting not to use any form of selection bias when designing the study. In addition, there are well documented flaws related to study designs where data is provided by test subjects as it relates to the hypothesis being tested. With that said, of the tree types of epidemiological study designs, this is normally seen as the most reliable.
The Harvard research team analyzed data from two prospective studies and included both men and woman (approximately 170,000 subjects total). These subjects were provided with questionnaires every few years so that they could update things related to their lifestyle. According to the research team, they had follow-up rates that exceeded 90%. Now, as I will reiterate here, relying on questionnaires is not the most accurate method for tracking diet. If I were to ask you what you had for dinner last night, could you tell me? I’m assuming you probably could. Now, what if I asked you what you had last month on a Tuesday? Gets a little bit tougher, no? How about last year? Now you can see why this data must be used cautiously when forming causal conclusions.
As we get deeper into the study we see that all subjects were health professionals, who aren’t exactly known for their stress free lifestyles; our first potential confounding variable. If we dig just a little bit deeper though, we find what might be a serious of confounding variables that really makes the conclusion seem a bit far fetched.
Men and women with higher intake of red meat were less likely to be physically active and were more likely to be current smokers, to drink alcohol, and to have higher body mass indexes. In addition, a higher red meat intake was associated with a higher intake of total energy (caloric intake) but lower intakes of whole grains, fruits, and vegetables.
I would love to know how the research team disregarded the smoking, alcohol consumption, obesity, low intakes of fruits/vegetables, and concluded it was the red meat that led to the higher mortality rates?
Referring back to Keys’ famous study boner and I’m reminded of Professor John Yudkin’s take on it which was published back in 1957. Yudkin took the same epidemiological data that Keys’ team was working with and theorized that Keys was only quoting the bits that supported his view (for those unfamiliar, this study was famous for incorrectly stating that high fat diets and cholesterol led to CHD). To prove his point, Yudkin examined not just the participating country’s fat intakes but many other potential factors. My favorite part of Yudkin’s article was that he successfully showed that the strongest predictor of future coronary mortality in countries where data was collected by Keys was the percentage of subjects who owned a TV or radio in that country. So, if we follow the logic of Harvard here, it is plainly obvious that TVs and radios cause CHD. I think I finally am starting to understand how this epidemiology thing works…
In any event, large leaps of faith like the one made by Harvard’s research team are very irresponsible because large amounts of the population do not dig into the innards of the study. The only thing these folks are seeing and hearing are what’s being published on news sites along with the sensationalized headlines and thus assuming that red meat will kill you. It has taken the average person decades to get over the fear of cholesterol, fat, and the like – and large percentages are still not there yet – caused by similar irresponsible research so let’s not go down this road again.
Links to other articles on this topic.
AddingPins' Blog
My apologies. Good website and very thoughtful comments both here and at MDA