Imagine you have a friend named Justin. He is a schoolteacher. Honest, hardworking, doesn’t smoke, rarely drinks alcohol, sleeps well, doesn’t take drugs, shows up at work every day. He has also chosen to be vegetarian.
Another friend of yours, an auto mechanic named Tommy, eats fast food, loves fried chicken, drinks too much beer on the weekends, likes to drive fast cars, and sometimes gets into legal tangles. He smokes cigarettes, though has limited it to only half-a-pack per day. Late weekends, some weekday nights, sleep cut short to just two or three hours. Tommy is not a vegetarian, but likes his burgers rare with a big side of French fries and a can or two of Coca Cola.
Who will do better in the long-run and be more likely to be spared health conditions like high blood pressure, high triglycerides, fatty liver, type 2 diabetes, psoriasis, or even a high-speed motor vehicle accident?
Justin the schoolteacher will, of course. Tommy’s hard-living lifestyle is much more likely to end in disease or disaster. If we didn’t understand that such a mixed bag of observations are virtually worthless, we could then conclude that, because Justin is a vegetarian, vegetarians are therefore healthier. Laugh, but that is how most observational studies are conducted. Potential “confounding factors,” such as Tommy’s drinking and smoking, are “factored out” via statistical manipulations, but his love for fried chicken, fast driving, and sleep deprivation are not.
This is precisely how Premarin became the number one most prescribed drug in the world for many years, earning manufacturer Wyeth billions of dollars per year. Observational studies, using methods similar to comparing Justin and Tommy, suggested that women who took Premarin experienced less heart attacks, endometrial cancer, and breast cancer than women who didn’t take Premarin. While some confounding factors such as smoking were factored out, many were not—sleep habits, quantity of visceral fat present, or vitamin D status, for example.
You probably know the rest. Two properly designed studies in which several thousand women were enrolled (The Women’s Health Initiative and the HERS Trial), then provided a pill that neither investigators nor participants knew was Premarin or placebo (double-blind, placebo-controlled) and the opposite was found: Premarin increased heart attack by 50%, increased incidence of endometrial and breast cancer, even accelerated Alzheimer’s dementia. In other words, the findings of the previous observational studies, trumpeted far and wide and used to influence physicians’ prescribing habits, proved to be tragically wrong. (This makes sense: After all, how in the world do you justify giving human females estrogens unlike human estrogens? Imagine your husband was given squirrel testosterone? The company did this for patent protection, not because horse estrogens are superior to human forms—the smell of money, once again, a familiar and recurring theme in Big Pharma.)
Premarin is not the only culprit to have been held up as beneficial only to prove the opposite. This happens more often than not with clinical studies involving placebo-control and blinding demonstrating outcomes radically different from the preliminary observations. In other words, observational data is about as good as no data at all.
Observational studies are also the basis for the “eat more healthy whole grains” advice. People are administered dietary questionnaires to recall, for instance, the foods you ate over the prior 5 days, white flour vs. whole grain intake stratified. These people are then recontacted, say, 5 or 10 years later to see what became of them. Imagine Justin and Tommy were enrolled in such a study: Justin would once again come out on top. After “adjustment” for a handful of confounding variables like smoking and exercise, the greater whole grain content of Justin’s diet would be declared the reason why Justin lives to enjoy his grandchildren while Tommy ends up with a heart attack, colitis, and dementia at age 65.
Throw on top of this the issues that I have stressed over and over again—that if something bad is replaced by something less bad and there is an apparent benefit—you cannot conclude that a whole bunch of the less bad thing must therefore be good. Replace unfiltered, full-tar cigarettes with filtered, low-tar cigarettes and witness a modest reduction in lung cancer, heart disease, and death—should we then conclude that smoking low-tar cigarettes is therefore good? Of course not. But that is the line of reasoning applied in promoting “healthy whole grains” to the unwitting public. The observational Nurses’ Health Study of 47,000 women, for example, showed that women who consumed more white flour products gained substantial weight, but that women who consumed more whole grains also gained weight, but just by a couple less pounds—not weight loss, but less weight gain.
The notorious unreliability of observational data that should be used to achieve nothing less than to generate an hypothesis but virtually never establish cause-effect relationships doesn’t stop the media from reporting the results of observational studies with bold headlines delivered as fact: “Coffee consumption linked to less Alzheimer’s dementia.” Or “Red meat consumption raises risk for colon cancer.” Reporting observational findings as fact has been the modus operandi of the media for as long as I can remember. It’s confusing due to frequent conflicting findings and it is misleading. It is essentially health fake news.
But don’t you fall for it. Don’t take Premarin, don’t eat any grains at all, and eat some pork chops.