On March 12, a new study was published in the Archives of Internal Medicine and it’s created a low-fat, whole grain feeding frenzy by mainstream media. Headlines such as “Eating All Red Meat Increases Death!” and “Too Much Red Meat May Shorten Lifespan!” or my favorite: “Red Meat: a Ticket to Early Grave!” flooded the airwaves and internet the past couple of days.
It’s all quite dramatic and conjures up images of a beef-induced apocalypse worthy of a Hollywood blockbuster. But is all of this attention much ado about nothing? Let’s take a closer look, shall we?
What we have here is a basic observational study. That is, the researchers studied the diets of a group of about 121,000 people. They then drew a line from a particular component of their diet, in this case, red meat, to all causes of early death including Cardiovascular Disease (CVD) and cancer. They found that one serving of red meat per day was associated with a 12 percent increase in total mortality.
Of the 23,926 deaths documented, 5,910 were from CVD and 9,464 were from cancer. I admit I have limited math skills but so far that adds up to 15,374 deaths. That leaves 8,552 other deaths that we can assume were from anything from other diseases to automobile accidents to being bitten by a snake or other venomous critter.
For the record, you have a 1 in 100,000 chance of being killed by such a bite or sting. If my math is correct that means eating red meat increases your chances of dying by something like a spider bite from .0001% to a earth shattering .00112%. Better watch your step. Just saying. Oh... wait... correlation is not causation. In other words, this study can’t conclude that red meat causes increased deaths from cancer or heart disease any more than it can conclude that it causes increased death by something like a bee sting.
On the other hand, if this were a real experiment it would look something like a randomized crossover clinical trial. While the observational study, at best, can generate hypothesis, a clinical trial may actually be able to show proof of something. This type of study takes groups of people, puts them in a controlled environment and changes ONLY the aspect of the diet they are studying. In that situation the researches can observe changes that can clearly by attributed to the specific item being investigated. This is the gold standard in scientific research.
The Confounding Confounders Conundrum
Confound: verb. To cause to become confused or perplexed.
In the Red Meat Consumption and Mortality study there are so many confounders it actually becomes amusing to try and spot them all.
As it turns out, the folks that ate the least amount of red meat, also were the most physically active. Perhaps being more physically active might account for some of the longevity seen in the group eating less red meat as opposed to the folks who ate the most red meat who, lo-and-behold, were the least physically active.
Being physical activity is generally a good thing and sure is healthier than being physically sloth-like.
Moving on we also see that the group with the highest red meat intake also happened to be the heaviest drinkers of alcohol out of the bunch. What can I say, apparently meat-eaters like to party. Well, I like a good party as much as the next person but yet again, we have a confounding factor that could easily be attributable to a person’s early departure from this world.
What else do we know about the highest red meat intake group? Hmm ... not sure if this is relevant but, what-the-heck, I’ll throw it out there anyway. That group also happens to have the highest percentage of current smokers. This study is starting to have more holes in it than ... something with a lot of holes in it. Sorry I ran out of metaphors.
There are other confounders and imbalances in the report as well, like multivitamin usage and body mass index. The closer we look the harder it becomes to narrow down a causal role for mortality yet the researchers insist that it must be red meat.
The value of a well controlled and randomized trial and why it’s the gold standard in research is that it removes these types of confounders. When done correctly, for example, no one group would have an overwhelmingly higher percentage of smokers or drinkers in it.
Explain in detail everything you ate last week. Not so easy is it? Now tell me what you ate in the last 12 months. Impossible you might say yet that’s exactly what was expected of the subjects in the study. The data were collected using something called a Food Frequency Questionnaire (FFQ). This quote comes directly from the study:
“In 1980, a 61-item FFQ was administered to the NHS participants to collect information about their usual intake of foods and beverages in the previous year. In 1984, 1986, 1990, 1994, 1998, 2002, and 2006, similar but expanded FFQs with 131 to 166 items were sent to these participants to update their diet.”
As you can see, participants were asked to recall details about their diet as far back as 4 years! A far better representation is something called a Dietary Record. By contrast this quote is from The International Journal of Epidemiology, Vol. 18, No. 4, 1998.
Between June 1980 and June 1981, participants completed a seven-day dietary record four times at intervals of approximately three months. Women were provided with a Pelouze dietetic scale for weighing food servings. Alternative methods employing common household measures were taught for use when away from home. All food and beverages consumed for a one-week period were recorded in a specially designed booklet. Participants provided detailed descriptions of each food, including brand and method of preparation and recipes whenever possible. The records were then coded by trained dietitians and reviewed by LS to minimize variability of interpretation. Participants were telephoned to resolve any discrepancies or to obtain further information when necessary.
That sure seems a lot more reliable. In fact, when comparing the FFQ to the Dietary Record that same article finds the following discrepancies (brackets in bold are mine):
As a second measure of validity we compared the absolute mean amounts of each food estimated by the two different methods. Although a ranking of subjects by intake is sufficient for many epidemiological analyses, an accurate estimate of the absolute amount consumed is desirable to formulate recommendations, [like a recommendation to avoid red meat like the plague?] and compare different studies. Focusing on the second questionnaire, we found that butter, whole milk, eggs, processed meat and cold breakfast cereal were under-estimated by 10 to 30% on the questionnaire. In contrast, a number of fruits and vegetables, skimmed or low fat milk, 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. A similar bias was found in an Australian study. Hauser and Bebb found similar over- and underestimations in a dietary history compared with dietary records; citrus fruits, 'other cooked vegetables' and tea were overestimated by the dietary history and luncheon meat was underestimated.
I need to highlight this part one more time:
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.
Ummm.... really?! Shocker!
Despite these glaring over- and underestimations, the Red Meat and Mortality Study concludes that “Red meat consumption is associated with an increased risk of total, CVD, and cancer mortality.” Of course I can’t say it enough, correlation does not equal causation and folks, even the correlation here is about as flimsy and overestimated as the cooked spinach reported in FFQ. We also can’t be sure what else these folks were eating. Since hamburgers were included in the report we can safely assume that at least some portion of those burgers came from fast foods joints. Do you want fries with that? Of course you do and don’t forget to supersize me.
This is nothing more than another attempt to blame some old foods on some very new diseases. The same thing we’ve seen with things like eggs and butter. We’ve been eating these items for thousands of years but somehow they are causing the sudden rise in disease we’ve seen in the past 50 years. Of course some of these foods have changed dramatically in recent years and have a multitude of chemical and become more processed than ever before.
Certainly we should avoid food like pepperoni pizza, Slim Jim’s and other processed meats that contain nitrates, nitrites, antibiotics, hormones and other growth stimulants. Likewise we should limit our intake of meats that are fed an unnatural diet that’s comprised of grains, corn and soy. Cows and other ruminants are meant to eat grass. Feeding them anything else makes them sick and in turn, makes the meat from them unhealthy. So, should we really conclude that ALL red meat is bad based on one very poorly designed study? Of course not. Does it mean we should do our best to choose healthy meats from humanely raised, grass-fed animals? Indeed this food is some of the healthiest and nutrient-dense foods available to us and should be enjoyed regularly. As an added bonus, this type of animal husbandry is essential to a healthy environment.
So fear not my fellow omnivores and get your burger on. Just be sure to lose the bun and no, you don’t want fries with that.