It’s an annual event. Every year, without fail, the headlines are awash with claims that ‘new’ research shows red meat causes cancer. The source for this scary claim is invariably some big data-dredging project…uh, I mean, “epidemiological” study by researchers from Harvard or some other ‘prestigious’ institution (there are no actual randomized controlled clinical trials showing red meat causes cancer).
When the yearly ‘red meat causes cancer’ circus rolls into town, I tend to yawn and quickly divert my attention to other more pressing matters, like the finer points of beef spare rib preparation. Did you know the longer and gentler you cook those little suckers, the more likely they are to slip off the bone due to absurdly mouth-watering tenderness? And that just the right amounts of sea salt, cardamon and fenugreek make these delicacies even more sinfully delicious? Try it some time; just as your taste buds are about to reach the point of orgasm, spare a thought for those poor buggers who earnestly believe all this red-meat-is-bad-for-you nonsense.
Which brings me to the latest red meat-bashing paper that has everyone’s knickers in a knot. The study comes from Harvard researchers, and unlike the usual cancer focus, this one primarily addresses the risk of overall mortality allegedly imparted by eating red meat (the full text is available here). I’ve had a lot of folks asking for my ten cents worth on this study, so here it is:
I think this paper is a load of hogwash.
That’s my ten cents’ worth, hope that helps, I’m going out now. Ciao!
You wanted something a little more scientific, like a discussion of the methods, data, and possible flaws in the study?
Damn it…I thought I was about to make an early getaway!
Oh alright, give me a second to pull the paper up again, and I’ll explain just why I think it’s a complete joke.
Epidemi-hogwash versus Clinical Trials
The first thing that should be pointed out about this paper is that it reports on two epidemiological studies, sometimes referred to as prospective, population-based or follow-up studies. These are studies in which researchers recruit a bunch of people, ask them a series of questions (sometimes only at the start of the study, sometimes at intermittent periods during the study) and then follow them up over a given number of years. In the case of health- and nutrition-oriented epidemiological projects, researchers will ask questions about diet and exercise habits, then see if there is any statistical association between diet, lifestyle and subsequent illness (morbidity) and death (mortality).
If you’ve read my book The Great Cholesterol Con, you’ll know I have an extremely poor regard for epidemiological studies, or more precisely, I have an extremely poor regard for the way in which they are routinely used as alleged ‘proof’ of a causal relationship between dietary factors and disease occurrence. Let’s be perfectly clear: epidemiological studies can not be used as proof of a causal relationship between meat and cancer, or saturated fat and heart disease, or anything else. They can only detect statistical associations; that is, the observation that people who eat this type of food tend to have more, less, or similar occurrence of X disease. Whether disease X is actually caused by that type of food, or whether people who eat that type of food more often are succumbing to the disease due to other behaviours they also engage in more often can not be conclusively determined from an epidemiological study.
For that, you need a randomized controlled clinical trial, in which you recruit a group of subjects, then randomly assign each subject to one of two groups. In the case of red meat, one group would be assigned to consume a diet low in red meat, another would follow a diet identical in every aspect except for a substantially higher red meat content. Both groups would receive clear instruction as to how to best construct and adhere to their respective diets. Although randomization isn’t perfect, it maximizes the likelihood that there is an equal number of smokers/diabetics/non-exercisers/drinkers/drug users/etc in each group, so that one group isn’t disadvantaged by a higher number of subjects with unhealthy habits and/or poorer health.
How to Avoid Prospectively Chasing Your Tail
I’ll give you an example of the futility of drawing concrete conclusions from epidemiological studies. Several decades ago, when Ancel Keys was making a big splash with his absurdly biased Six and Seven Countries studies that claimed fat and cholesterol caused heart disease, another researcher devoid of cherry picking tendencies decided to look at the problem in a little more depth. His name was John Yudkin, and he analyzed a greater number of countries than Keys’ had employed in his small handpicked sample, and found no relationship between fat and CHD. He did, however, find a strong relationship between television ownership and CHD; people who owned televisions were more likely to get heart disease.
Now, if I were a Harvard researcher and got a result like that, I’d immediately conclude televisions cause CHD and issue a sensationalist press release titled “Your TV is Killing You: New Study Shows Television Causes Heart Attacks!” And because most journalists responsible for health and nutrition stories are sheep-like creatures who know absolutely nothing of value about health and nutrition, and would assume that because I’m from Harvard I must really know my stuff when it comes to health research, my claims would get worldwide headline coverage.
However, having graduated not from Harvard but the University of Commonsense, and having majored in Bullshit Detection 101, my first reaction would be to ask, “Is there anything inherently toxic about televisions themselves, or is there some other associated factor at play here?”
Well, we know that television does not emit some toxic poison that seeps into the arteries of viewers, causing them to clutch their chests in terror and inconveniently die just as Lieutenant Horatio is about to nail the bad guys in CSI Miami.
What we do know is that, if you’re watching television, you’re sitting on your butt. Which means you’re not exercising. Instead, you’re probably eating high-calorie junk. If you watch a lot of TV, then that means you’re probably eating even more junk and even less likely to be exercising at all. In which case, you’re more likely to be overweight, unfit, and have the arterial elasticity of a steel girder.
Using the powers of deduction endowed to me by my advanced degree in BS Detection 101, I would promptly suspect that the actual causal factors for increased CHD risk are your poor physical condition and intemperate nutritional habits. This would then prompt me to make recommendations that would seek to directly improve your fitness and nutritional habits. Concluding that TV was the causal factor and telling you to get rid of your TV would do you a fat lot of good if you stayed on the couch and kept snacking your way to an early grave.
This, ladies and gentlemen, is the very real danger inherent in the modern obsession with epidemiological studies. They are vastly inferior to randomized clinical trials, but when treated as if they are essentially the same thing – as is the widespread practise nowadays – they can lead to wildly inaccurate and misleading conclusions.
There are circumstances in which epidemiological research is indeed useful. If there is an outbreak of a new viral contagion, for example, researchers don’t have time to sit around discussing the finer nuances of a clinical trial. Epidemiological analysis allows them to determine what geographic locations/age groups/occupations/habits/ethnicities/etc seem to be most susceptible, and then take the appropriate action to contain the outbreak.
The other thing epidemiological research can be valuable for is as a prelude to clinical research, i.e. to determine associations worthy of confirmation or refutation in a far more controlled setting. Concluding from an epidemiological study that meat causes cancer or that driving Holden Barinas causes men to develop effeminate habits is plain stupid. However, noting that these relationships seem to be consistently observed in prospective studies and then commissioning a clinical trial to examine whether the relationship still remains evident in a far more controlled setting is not stupid – it’s good science. You know, the kind of science that would be practised in a world where drug companies and food conglomerates didn’t virtually own the medical and health fields and where researchers were a noble group of independent thinking mavericks who placed more value on relentlessly seeking out the truth than simply milking the same epidemiological study for one headline-grabbing paper after another. But I digress…
The bottom line is that concluding red meat increases cancer risk from an epidemiological study is very sloppy science. And that’s exactly what’s happening with this latest Harvard paper by Pan et al.
It Wasn’t the Inactivity, Smoking and Diabetes that Killed Them – it Was the Red Meat, Damnit!
One of the first things I look at when reading a paper like this is the baseline characteristics of the subjects. Doing so often gets alarm bells ringing, and right now, as I peruse Table 1 of the paper, my head is ringing like an incoming call centre. There are so many discrepancies it actually boggles my mind that anyone would take a paper like this seriously.
The researchers divided meat intake into five quintiles, from lowest to highest. And as you peruse the table, something quickly becomes blindingly obvious: The more red meat people ate in these studies, the more likely they were to live unhealthier lives overall.
Remember how I said this paper covered two studies? The first was the Health Professionals Follow-Up Study (HPFS), the baseline data of which shows that as red meat intake rose, so too did the prevalence of all the following:
Those who ate the most red meat also drank more alcohol, were less likely to use multivitamin supplements, and had a self-reported daily caloric intake over 800 calories more than those in the lowest quartile (2396 vs 1659 cals/day, respectively…more on that later).
So in HPFS, those who ate the most red meat also tended to live lifestyles that were unhealthier all-round.
And it’s much the same in the second project, the Nurses’ Health Study (NHS). The baseline data again show that those who ate the most red meat did less exercise, were more likely to smoke and have diabetes, were less likely to take a multivitamin, and also self-reported eating an extra 800+ calories daily.
Again, the more prolific red meat eaters smoked more, exercised less, and had a higher incidence of diabetes – a disease of disordered glycemic control with no plausible connection to red meat but strong links to excess caloric and refined carbohydrate consumption and physical inactivity.
I think it’s incredible that I even need to point this out in 2012, especially to Harvard researchers, but it’s pretty well established by now that smoking, diabetes and physical inactivity all greatly increase the risk of cancer, heart disease and premature mortality; the very outcomes that the researchers are ascribing in this paper to poor old red meat, everybody’s favourite whipping boy.
Some readers may object that this was countered by the multivariate analysis performed by the researchers, which adjusted for age, body mass index, physical activity level, smoking status, race, menopausal status and hormone use in women, family history of diabetes, myocardial infarction, or cancer; history of diabetes, hypertension, or hypercholesterolemia; and intakes of total energy, whole grains, fruits, and vegetables.
My response to that would be: Yeah, right.
Assuming that multivariate analysis can honestly remove all confounding variables and provide a truly accurate risk assessment is to basically claim that researchers have magical powers. Running a statistical model after the fact is a far cry from conducting a clinical trial scenario in which subjects are randomized to two groups and given explicit instructions on what they should and should not eat. Furthermore, think about all the factors that weren’t even considered; illicit drug use, which is highly prevalent in society (and a very real problem in the medical field), is rarely probed in nutrition-slanted prospective studies. Nor are many other risk-taking behaviours, or such factors as poor sleep habits or greater psychosocial stress exposure. These factors certainly didn’t feature in Pan et al’s multivariate analysis, even though there is a very real possibility that, due to their generally unhealthier lifestyles, the more frequent red meat eaters were more likely to display these characteristics.
Just How Accurate Was this Data, Anyway?
The very high prevalence of inaccurate dietary reporting in prospective studies and even free-living clinical trials is well established. Keeping that in mind, remember how in the Pan et al paper, as red meat consumption went up so too did self-reported daily calorie intake? And that as red meat intake went up, physical activity levels went down? Unless you’re one of the dopey low-carb sods who still believes in MAD, we should by all rights expect a far greater prevalence of overweight and obesity with increasing red meat consumption – but we see no such thing. Mean BMI is inexplicably similar across all five quintiles of red meat intake.
According to this study, unlike the rest of us mere mortals, doctors and nurses are metabolic freaks who can eat more, exercise less, yet still not gain extra weight.
You’ll recall that in long-running epidemiological studies, sometimes a single dietary questionnaire is filled by the subjects, in others the questionnaires are filled periodically. In the HPFS and NHS studies, “periodically” was once every four years(!)
To quote the researchers themselves: “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. Using the expanded FFQ used in the NHS, dietary data were collected in 1986, 1990, 1994, 1998, 2002, and 2006 from the HPFS participants. In each FFQ, we asked the participants how often, on average, they consumed each food of a standard portion size.”
Most people struggle to accurately remember what they ate three days ago, but this method makes the very generous (some would say absurd) assumption that the subjects not only gave highly accurate portion sizes, and highly accurate frequency estimates, but were able to do so for a period covering twelve months prior to the questionnaire being filled out.
That odour you’re smelling? It’s the not-so-sweet smell of bovine poop.
Cholesterol Down! Mortality Up!
There’s something else in Table 1 that is very worthy of close attention, but was completely ignored in all the shrill press releases. Namely, as people ate more red meat, their likelihood of having high cholesterol decreased in linear fashion. This relationship was especially pronounced in HPFS, where those in the highest red meat quartile were almost half as likely to have high cholesterol.
Stop the presses!
For years, I’ve been banging on about how the war on cholesterol is an absolute absurdity, how cholesterol is a perfectly natural substance that is absolutely essential to the continued well being of each and every one of us. And while the occasional sickeningly smug Janet Brill-type attempts to portray me as a lunatic for stating this plain fact, the indisputable truth is that cholesterol is an essential cellular substrate which humans ought to grant far more respect. Pull all the cholesterol (and saturated fat) from your cell membranes, and you’d promptly collapse to the ground in a lifeless pile of mush and bones.
And here we have a study showing that as levels of this essential substrate went down, mortality went up. But once again, this is ignored in favour of far more fashionable and politically correct red meat-bashing.
Excuse me while I step outside for some fresh air…the smell from the Pan paper is killing me.
Unprocessed My Rump!
Like a lot of the meat-and-cancer studies, the researchers examined both unprocessed and processed meats, the latter known to often contain a number of questionable ingredients suspected of contributing to chronic diseases. In addition, processed meats are often subject to some rather dodgey manufacturing and preparation processes; in the US some of the largest ever food poisoning outbreaks were caused by cold cuts, hot dogs, and inadequately cooked hamburger patties.
So obviously, the researchers had to group different types of red meats into the “processed” and “unprocessed” categories. In the “processed” category we see “bacon”, “hot dogs”,and “sausage, salami, bologna, and other processed red meats”.
No problem there.
Now let’s check out the “unprocessed” category. All you afficionados of free-ranged, pasture-fed beef might want to sit down before I continue the discussion. You might want to keep some oxygen handy, as well.
All set? Good.
According to the researchers, ‘unprocessed red meat consumption included “beef, pork, or lamb as main dish” (pork was queried separately beginning in 1990), “hamburger,” and “beef, pork, or lamb as a sandwich or mixed dish.”‘
Since when is a hamburger “unprocessed”? Sure, if you buy fresh ground meat and make the patties yourself with no added artificial ingredients – but what tiny percentage of America’s hamburgers are actually consumed in that fashion?
And you don’t have to be a brain surgeon to know that sandwich meats are often of the processed “cold cut” variety…so just how did they sneak into the “unprocessed” category?
Folks, welcome to Bizarro World, where a fresh T-Bone and an additive-laden meat patty from MickeyD’s are considered the same thing!
While you don’t need to be a brain surgeon, it appears you must not be from the epidemiology department at Harvard to be alert to such blindingly obvious flaws.
Relative Risk Shenanigans
OK, let’s talk about the actual risk increases the researchers claimed for red meat consumption. The HPFS study included up to 22 years of follow-up, while the NHS follow-up included up to 28 years’ worth. The total number of participants in both studies was 121,342; total number of documented deaths was 23,926 deaths (including 5910 CVD deaths and 9464 cancer deaths) during 2.96 million person-years of follow-up. So around 20 percent of the cohort died during the study.
After performing their multivariate analyses, the researchers claimed a relative risk increase in overall mortality of 12% for total red meat consumption, a 13% increase for “unprocessed” red meat, and a 20% increase for “processed” meat consumption. Even in a clinical trial, increases of 12-13% are not exactly earth-shattering. But in a prospective study, hopelessly prone to confounding factors and featuring questionable dietary retrieval methodology occurring about as frequently as the Olympic games, it’s an absolute non-event. Relative risk percentage in the triple digits, despite the flaws of epidemiological studies, certainly warrant attention. But to earnestly claim from such an inherently flawed study that red meat consumption imparts a 12% increase in overall mortality is a very, very bad joke.
Only a hopelessly biased vegan could love a study like this…
To all you journalists who insist on sensationalizing pseudo-scientific rot like this, shame on you. Seriously, stick to quoting lying politicians or writing up gossip columns or whatever other inane pap you are truly qualified for; people who have no clue about scientific research and merely accept press releases from research groups at face value have absolutely no ethical business being in a position where they can influence other people’s health knowledge and behaviours.
For everyone else, be very wary of sensationalist diet and health stories appearing in the popular media, and immediately ask yourself, “Was this finding derived from a randomized clinical trial or a prospective epidemiological study?” If the latter, take the results with a wee pinch of salt.
Anthony Colpo is an independent researcher, physical conditioning specialist, and author of The Fat Loss Bible and The Great Cholesterol Con. For more information, visit TheFatLossBible.net or TheGreatCholesterolCon.com
Copyright © Anthony Colpo.
Disclaimer: All content on this web site is provided for information and education purposes only. Individuals wishing to make changes to their dietary, lifestyle, exercise or medication regimens should do so in conjunction with a competent, knowledgeable and empathetic medical professional. Anyone who chooses to apply the information on this web site does so of their own volition and their own risk. The owner and contributors to this site accept no responsibility or liability whatsoever for any harm, real or imagined, from the use or dissemination of information contained on this site. If these conditions are not agreeable to the reader, he/she is advised to leave this site immediately.