Why do RCTs of nutritional supplements often show no benefit?
As I pointed out in a previous post, many of the large intervention trials over the past few years that have looked at the effect of nutritional supplements on health conditions have not shown a benefit. Some have even shown trends toward harm. How can we understand this?
There could be straightforward reasons, taking issue with the study design or conclusions drawn. Sometimes studies are flawed. The researchers used a poor quality supplement, they didn’t use a high enough dosage, they were looking at the wrong end points, they didn’t give the supplement long enough, they didn’t control for diet or other lifestyle variables, they gave it too late, etc. etc. etc. And while all these may be valid problems with the studies, it seems a little like hand-waving arguments to me. I’ve been thinking that maybe the problem is much more basic than that. Maybe the real issue is that applying the randomized, double-blinded, placebo-controlled crossover trial is the wrong way to test for efficacy of nutritional supplements.
For starters, each person’s physiology is unique. (I mentioned Roger Williams and his concept of biochemical individuality in a previous post.) Even with medications, a person’s response to a drug is dependent upon a large number of variables, including their genetic makeup, their nutritional status and even the specific foods they eat. A simple example to illustrate: naringenin, an ingredient in grapefruit juice, can change how a drug is metabolized by affecting key enzyme pathways in the liver. Changing the activity of those enzymes changes how quickly the drug is eliminated from the body. Some drugs even carry warnings to not drink grapefruit juice when taking the drug. The whole construct that it is possible to actually reduce the variables down to just one (the purpose of the RCT) is being challenged by some scientists, as every person is going to have their own unique response. Still, for most researchers these limitations are considered acceptable since there’s no other game in town.
Secondly, and I think most importantly, this procedure, which was developed primarily to test single compounds and is used in developing pharmaceuticals drugs, does not lend itself well to complex compounds like foods. For example, a chemical analysis reveals hundreds of different chemicals that make up garlic; and furthermore, where it was grown, when it was harvested and how long it’s been stored will cause changes in the chemical composition. It’s still garlic, but it’s going to have significant differences to other garlic, so how can it be considered a single (and consistent) variable? So what that has led to is the attempt to isolate “the active,” meaning whatever is causing the desired effect. For example garlic is known to help reduce elevated cholesterol through interfering with a particular enzyme (HMGCo-A reductase) that controls a key step in the de novo (in the body) production of cholesterol. The “active” ingredient is believed to be allicin, the compound that gives garlic its characteristic pungent odor and flavor. But as I said above, garlic has hundreds of different chemicals, and it’s totally within reason that there is no single compound that accounts for all the known effects of the herb, but the combined effect of several or many of these compounds. And that’s just garlic. Consider the challenge associated in trying to pin down which specific food or constituent of the diet is contributing “the” effect that’s helping, when a diet consists of literally hundreds of different components.
Here’s another example. There is a chemical found in grapes called resveratrol. It’s a very powerful antioxidant, and epidemiological studies indicate that it’s associated with lots of health benefits. Here’s the challenge though: when isolated and studied as a single nutrient, it takes much higher levels to reach the effect seen in the epidemiological work, which measures its effect as part of a diet. That would indicate that the chemical by itself is not as effective as when it’s consumed with other things in the diet.
My contention is that we will NEVER find a single nutrient that, by itself, provides dramatic health benefits. I think we have evolved to require lots of different things from lots of different foods that, when consumed together, contribute to improved health and reduction of disease. A friend of mine talks about “Eating a rainbow,” meaning that we should choose a wide variety of colorful fruits and vegetables, as the components that provide the color (such as resveratrol) are also what carry the health benefits. So it’s no single food, but a wide variety of different foods, consumed as part of a highly varied diet.
Therefore, and what this post is leading up to, is that I think scientists will never be able to use the randomized control trial to establish which nutrients are most beneficial, specifically because they are looking at it the wrong way.
There’s an old joke about a guy walking down a street at night and sees another person looking around for something under a streetlight. He goes over to him, and after being told the other person is searching for his car keys, he decides to help. But after a few minutes of fruitless searching, he asks exactly where the other guy was standing when he dropped his keys, and his new friend points to a spot about 20 yards away. Incredulous, our friend asks why he’s searching here rather than where he dropped his keys, and is told “There’s no light over there. I couldn’t see a damn thing, so I decided to search over here under the streetlight.”
I think our scientist friends are looking where the light is, not where the keys are.
