Do we know anything about nutrition?

If you eat something and it makes you feel bad, stop eating it.

I was pointed to this infographic on superfoods, which aims at summarizing the scientific information we have about the health benefits of specific foods. It ranks health claims about specific foods by the strength of scientific evidence for the claim, and provides direct links to the relevant studies. The first thing you’ll notice if you look at it is that most claims fall well below promising. However, there are a few that are listed as good or strong. Since I’m somewhat of a nutrition geek, I of course went immediately and looked at the claims that are supposed to be the strongest. I was dismayed to find that of the four claims listed as very strong, at least two are rather dubious. Those two are barley and oats. (I stopped reading after that. The other two may well be dubious as well.)

Interactive superfoods infographic from

Figure 1: Interactive superfoods infographic from

There are two issues with those two claims: First, there is no credible health claim. Second, regardless of the validity of the claim itself, the claim is not well supported by evidence. Let’s consider these two pieces in turn.

Credibility of the health claim: Both oats and barley are listed for cholesterol. The problem is, cholesterol is not a disease. At best, cholesterol is a marker for a disease (cardiovascular disease, CVD). But it’s a really weak marker. It basically doesn’t work. The main reason we are obsessed with cholesterol is that in the 1950s, it was one of the first physiologic markers of CVD that could be easily measured. As a consequence, cholesterol testing became standard, and we all started to worry about cholesterol levels. But the science hasn’t held up. By now, it is pretty well established that CVD is caused primarily by inflammation, and inflammation is caused by an excess number of oxidized Low-Density Lipoprotein particles (LDL particles). A standard cholesterol test measures the amount of cholesterol carried by LDL particles. It is entirely possible to have a small number of particles carry a large amount of cholesterol (high cholesterol, low disease risk) or to have a large number of particles carry a small amount of cholesterol (low cholesterol, high disease risk). Thus, unless a food lowers the number of LDL particles, its effects on cholesterol levels are irrelevant. If this is news to you, read the nine-part series on cholesterol by Peter Attia, in particular part vi and part ix.

Strength of evidence supporting the claim: The link between oats/barley and cholesterol is supported by nutritional epidemiology, a field that has largely failed to produce any useful insight. Nutritional epidemiology can reveal associations between components of a diet and diseases/disease markers, but it cannot establish any causal relation. Further, because the number of potential confounding variables is huge, and the effect size generally small, nearly all results in nutritional epidemiology are driven by unobserved or uncontrolled confounders. Yes, you could use nutritional epidemiology to establish that eating potassium cyanide is bad, or even that eating arsenic in large quantities is bad, but you’ll generally not be able to ascribe specific health benefits to regular foods that people have eaten for generations.

The biggest confounder in nutritional epidemiology is the healthy person bias, which means that people who take good care of themselves will generally be healthier, and they will also do what is commonly believed to be healthy. At that point, nutritional epidemiology becomes a self-fulfilling prophecy. If we’re being told that oats and barley are good for us, then the most healthy-conscious people are going to eat more oats and barley, and hence we’ll see an association between consumption of oats and barley on the one hand and better health profiles on the other hand. The number of people who are like me, who are very health conscious but think that a diet of eggs, steak, and butter is healthy, is rather small. Therefore, I don’t expect any association study in the near future to support my view. But even if it did, it would be just as meaningless, for the same reasons. If you want to learn more about the problems with nutritional epidemiology, I recommend the work by Gary Taubes, in particular his book “Good Calories Bad Calories.” For a quick online read, check out this article.

Summary: I don’t think we know much about nutrition at all at this time. I’m generally sceptical of any health claim, unless it is linked to a solid, experimentally verified mechanistic explanation. (There can be little doubt, for example, that an excess of refined carbohydrates is a main driver in Type-II diabetes.) But even if a mechanistic explanation is available, we all have different genetic backgrounds and react to food differently as a consequence of that. For example, some people can eat grains and sugar all their life and be fine, while others need only look at a slice of bread and develop insulin resistance.

For myself, I’ve come up with the following list of simple guidelines that I think are well supported and reasonable:

  1. If you eat something and it makes you feel bad, stop eating it.
  2. If you’re frequently tired, have low energy, or have frequent digestive issues (heartburn, bloating, constipation, loose stools), you may be eating things your body can’t handle well. Try an elimination diet.
  3. If you suffer from allergies, you may be eating things your body can’t handle well. Try an elimination diet.
  4. If you frequently have energy supply problems, i.e., you easily get tired in the afternoon, or you frequently feel like your brain will shut off unless you eat something immediately, you lack metabolic flexibility. Play around with the macronutrient composition of your diet (more fat, more protein, fewer carbs) and see if it makes a difference.
  5. If you feel great, have good energy, generally feel awake and productive, it’s very unlikely that your diet has major issues.
Claus O. Wilke
Professor of Integrative Biology


comments powered by Disqus