There’s a new documentary making the rounds, “The Game Changers,” which makes the case that going vegan can improve athletic performance and is the superior diet for human health. On the other side of the coin is meat, fish, fowl, eggs, and dairy—foods that not only can hamper physical fitness—they can promote erectile dysfunction, endothelial dysfunction, and lead to heart disease, cancer, and type 2 diabetes. Also, meat is destroying the environment.

I guess I’m starting to accept that such debates are never going away. Ever. Unlike science, which generally converges to truth, I don’t think nutrition (like religion or politics) is really capable of such convergence, especially when it comes to the most hotly contested vector of nutrition—dietary restriction. [If you need a primer on nutrition and its three vectors—dietary restriction, time-restriction, and caloric restriction—PLEASE pause for a moment to watch this short video explanation.] 

There are many things wrong with this film. And there will be no shortage of people—some credible and thoughtful, many deceptive and thoughtless—who will attack this film. I find that this particular topic to be more about ideology than fact and science; in fact it’s the worst kind of ideology because it cloaks itself in what looks like science. At least we recognize religion and politics for what they are, I suppose. 

If I’m going to contribute anything to this discussion, let me do so on two points related to this film that are not likely going to get a lot of attention. 

1) Switching from the Standard American Diet (SAD)—a diet of zero DR/CR/TR (see above video to understand what this means)—to a fill-in-the-blank diet is going to improve people’s health. Full stop. That includes vegan and vegetarian diets. That includes keto diets. That includes Mediterranean diets. That includes an all potato diet. But this observation tells you nothing about how healthy a “plant-based” diet is compared to one that is “meat-based.” 

2) Vegans and vegetarians are probably some of the most health-conscious people on the planet. This also means that when you compare these people to the general U.S. population, avoiding meat is but one of a myriad of differences between them. Observing differences in health outcomes is relatively easy. Determining the cause of those differences is virtually impossible. 

Let’s address these in order.

1 | The problem with using the Standard American Diet (SAD) as a control group.

What often happens, either in the volumes of observational epidemiologic studies that get cranked out every day or, less commonly, in films like this, is that a vegan or vegetarian diet is compared to the SAD, with the SAD as stand-in for a “meat-based” diet, as if the only difference between the two diets is one eschews animal foods. Take a look below at the top 10 sources of Calories in the U.S. diet circa 2010. You tell me, that even if you think meat is a problem, tell me that meat is the only problem with this list. 

1. Grain-based desserts (cakes, cookies, donuts, pies, crisps, cobblers, and granola bars)
2. Yeast breads
3. Chicken and chicken-mixed dishes
4. Soda, energy drinks, and sports drinks
5. Pizza
6. Alcoholic beverages
7. Pasta and pasta dishes
8. Mexican mixed dishes
9. Beef and beef-mixed dishes
10. Dairy desserts

I don’t think anyone would look at this list and question why Americans are so overweight and so sick. (By the way, there is not one item on this list that I do not absolutely love, except for #4. So I’m completely empathetic to the plight of anyone reading this who also craves these foods. They make up our default food environment and until that changes, we’re in for a rough ride, as a society.)

In the film, Bryant Jennings, an American heavyweight boxer, who stopped eating meat in 2013, and went vegan in 2015, talks about his conversion. “I grew up not even knowing about half of these other vegetables. Asparagus to me just came out like five years ago . . . My early years growing up in Philly, the only thing we knew was spinach in a can, collard greens and Popeye’s, KFC, everybody frying chicken,” says Jennings. “Most people say, ‘Oh, where do you get your protein?’ As if everybody that’s in KFC is looking at the back of a bucket, like, ‘Yeah, how much protein is [in here]?’ You all don’t know . . . You all don’t know what [you’re] eating. I’d never really thought about it like that before. What else was in the food I was eating?”

Jennings made the dietary switch for health reasons. “I just thought healthy eating, clean eating was much better. And it really has been great for me. . . Lots of peanut butter and jelly, oatmeal, quinoa, avocado, a lot of fruit and vegetables. I make my own burgers from scratch with chickpeas, black beans, lentils, quinoa, flax seeds, chia seeds. It’s all wholefoods,” he told The Guardian. In other words, Jennings went from fast foods to whole foods.

Typically, when people make the switch from the SAD to a vegetarian or vegan diet, it doesn’t just mean they’re striking a few numbers (3 and 9 in the case of vegetarians, and in the case of vegans, 5 and 10, too) from the top 10 list and adding in more sugar-sweetened beverages, grain-based desserts, and bread. They didn’t have a list to begin with. They weren’t paying attention to what they were doing, as Jennings points out. When people make a conscious decision to go from SAD to a vegan or vegetarian diet, they’re not going from a SAD to a “plant-based SAD.” Their diet, and often other aspects of their lifestyle that may impact their health, changes in myriad ways.

They eat more vegetables, fruits, legumes, whole grains, and nuts and seeds. They eat less refined carbohydrates, sugar, and saturated fat. They exercise more and may put more of an emphasis on stress management as well as social support (as Dean Ornish’s program includes, for example). During the transition, they may quit smoking, quit or reduce the consumption of alcohol, and focus more on getting adequate sleep. While we could debate the evidence for and against each one of these interventions, the point is that there are many changes taking place, in addition to cutting out animal products, that are consistent with what many people believe to be healthy living.

At the other end of the spectrum, we have the SAD. Not only does it represent the foods in the top 10 list above, but it also tracks more closely with many behaviors that are not consistent with healthy living. Less likely to exercise, more likely to smoke, etc. What was the change, or what were the changes, that improved the health of people transitioning away from the SAD? The problem is that when you have so many variables that can and do change—often simultaneously—when leaving the SAD, it is an experimentalist’s and an epidemiologist’s nightmare.

2 | The healthy user effect.

In “The Game Changers,” there are several studies and articles [see a complete “List of studies” at the bottom of this post] that flash on the screen comprising “overwhelming scientific evidence connecting animal foods to many of the most common deadly diseases,” according to the film’s narrator. All of these were observational epidemiologic studies (or meta-analyses of observational studies) and carry many limitations that I’ve discussed before (in detail in the five-part Studying Studies series), with the healthy user effect as one of the most important.

Why are people reportedly healthier on a vegetarian diet compared to many non-vegetarians? Is it exclusively because these people eat fewer animals? Imagine a representative survey of adults in the U.S. collecting health habits of these individuals that included the following item in it for people that follow a vegan diet.

Select any of the reasons below why you are a vegan:
⃞       To reduce the impact of my food on the environment
⃞       To eat a healthier diet for my personal health
⃞       To reduce the suffering of animals

How many respondents would not check the “health” box? I suspect the proportion would be vanishingly small. In other words, they’re conscious about their health and they’re engaging in a challenging behavior—avoiding the consumption of animals—they think will improve it. Think about how health-conscious individuals compared on this survey to individuals in the U.S. who eat the Standard American Diet (SAD), a survey that also gathers information on physical activity, access to healthcare, cigarette smoking, vegetable consumption, fruit consumption, vitamin consumption, annual physical exam attendance, sleep habits, meditation, alcohol consumption, socioeconomic status, prescription drug adherence, shift work, and community ties. Think about how anyone following any particular diet that they think is healthy—Atkins, Weight Watchers, keto, Mediterranean, Jenny Craig, paleo, etc.—compare to people eating the SAD.

People who are health conscious are different from people who are not in many different ways. This is the so-called healthy user effect, in a nutshell. I would be hard-pressed to think of a situation in which there’s more potential for a healthy user bias (covered in more detail in Studying Studies: Part III) than comparing a group of people who adhere to a particular diet for health reasons and comparing them to people who unconsciously eat the SAD. When we observe people who consciously adhere to a particular diet, we also see accepted healthy behaviors increase and unhealthy behaviors decrease. It’s not just the diet in these people that is healthier. This shows up again and again (and again) in observational studies.

For example, one such study flashed on the screen was of the diet of Seventh-Day Adventists (SDAs) and their risk of colon cancer. A large proportion (about 30%, based on the reported person-years in the study) of SDAs adhere to a church-recommended vegetarian diet. According to a self-reported questionnaire, the vegetarians not only eliminated meat from their diet, less than 1% of them currently smoked cigarettes (compared to 7% in non-vegetarians), 12% of them had ever smoked in their lifetime (compared to 28% in non-vegetarians) less than 1% of them drank alcohol (compared to 9% in non-vegetarians), and they were more physically active (though not statistically significant) than SDAs that ate meat one or more times per week. There were other differences, some noted in the study, including lower aspirin use in the vegetarians (13% compared to 29%), higher age (54) at baseline (compared to 52 in non-vegetarians, though not statistically significant), and more females (61% compared to 45%, though not statistically significant) and almost assuredly others that weren’t accounted for by the investigators.

The investigators tried to adjust for these differences, but the problem is that this is not only hard to do, there are also other ways that these two groups are different that were not considered. (Both of these problems fall under the heading of residual confounding.) After a six-year follow-up, they found a statistically significant positive association with meat intake and the incidence of colon cancer, with a risk ratio of 1.85.¹“[M]ost epidemiologists interviewed by Science said they would not take seriously a single study reporting a new potential cause of cancer unless it reported that exposure to the agent in question increased a person’s risk by at least a factor of 3—which is to say it carries a risk ratio of 3,” writes Gary Taubes in his 1995 article. “Even then, they say, skepticism is in order unless the study was very large and extremely well done and biological data to support the hypothesized link. Sandler Greenland, a University of California, Los Angeles, epidemiologist, says a study reporting a twofold increased risk might then be worth taking seriously—‘but not that seriously.’”

Remember, you should always ask, what are the associated absolute risks? In this case, I calculated (it was not reported in the study) the associated (and unadjusted) six-year risk of colon cancer to be approximately 0.38% and 0.54% for vegetarians and non-vegetarians, respectively. Whether this associated 0.16% difference over six years is attributable to the consumption or avoidance of meat is highly debatable.²Interestingly, the unadjusted absolute differences reveals an associated relative risk increase of 42%, while the reported 85% increase, see 1.85 above, came after adjustment. If the vegetarians were engaging in more accepted healthy behaviors, and fewer unhealthy ones, we would expect the association to get smaller after adjustment, not larger, as was reported. However, also take into account that non-vegetarians were two years younger at baseline. To give you an idea of the potential difference in risk based on age, the 5-year absolute risk difference of developing colon cancer at age 54 compared to 52 is an 0.1% increase in men [0.3% versus 0.4%], as well as an 0.1% increase in women [0.2% versus 0.3%] according to a calculator provided by the National Cancer Institute. What’s less highly debatable is how small the “effect” is, in absolute terms.

The sooner we all can move away from using observational epidemiology like this as a palette to paint our preconceptions with the better. Yes, observational epidemiology helped determine that cigarette smoking is a cause of lung cancer, for example. Contaminated water and cholera. Scrotal cancer and chimney sweeps. I can assure you that those studies were not reporting associated relative risk ratios less than two and absolute risk increases of less than two-tenths of one percent. (For example, the reported relative risk of lung cancer in current smokers of more than 1.5 packs of cigarettes a day was 111.3 and 108.6 in men and women, respectively—over a 10,000% relative risk increase.)  Perhaps most important (other than experiment) of Bradford-Hill’s “nine different viewpoints from all of which we should study association before we cry causation” is the strength of the association.

With such small associations between the consumption of meat and chronic diseases, combined with the high probability of confounding that should strengthen the association, we may be looking at a contrapositive case: If observational epidemiologic study after study either shows a small association, no association, or a negative association, it’s possible to read through the same observational studies most investigators and journalists use to implicate the consumption of red meat and increased risk for colon cancer for example, and conclude that since the observed associations are very small, the risk of confounding very high, the associated absolute differences (which are barely reported) minuscule, and these studies cannot establish cause and effect, that it’s more likely that meat does not cause colon cancer. In other words, a lack of association is evidence for a lack of causation. (For more of my thoughts on the issue of red meat causing cancer, check out my post, “Is red meat killing us?”)

My problem with this film is with its mode of inquiry, not the diet it espouses, per se. Let me state that another way. I’m not against a plant-based diet. I’m against misleading uses of subpar data and speaking with absolute certainty when a bit of humility and nuance should be the standard. The fact is that virtually any diet is better than the SAD from a health perspective. I’m sure we’d see associated improvements with any of the 47 diets listed on the U.S. News & World Report website if people consciously switched to one of these diets from the SAD. The more insidious problem is the healthy user effect. There are many variables that can and do change that could explain the small associations in these studies between meat or vegetable consumption and disease that have nothing to do with the meat or vegetable, making it virtually impossible to establish cause and effect.

– Peter

List of studies