Reductionism Gone Awry

Sometimes a study leaves you gobstruck – and not in a good way. An article in Circulation: Heart Failure looks at how dietary protein intake impacts the risk of heart failure. Basically, this is a study of 2441 Finnish men age 42 to 60 studied over a mean period of 22 years looking at the incidence of heart failure based on the amount of protein in their diet. Their finding

“In middle-aged men, higher protein intake was marginally associated with increased risk of HF (heart failure).”

It requires a great deal of spin to reach that conclusion. But let us look at two glaring misuses of the concept of breaking a problem down into its parts – reductionism. Reductionism is the basis for most science since so many factors can be involved, isolated them in a lab bench experiment can yield valuable insights; for epidemiological studies, it doesn’t work as well.

Reductionism #1

The participants’ diets were used to stratify them into groups. The determinant of the diet was a four-day food survey conducted when they entered the study, twenty-two years ago. Leave aside the vagaries of dietary information from surveys (the authors spent some time making sure it was “accurate”) but can you base the development of a syndrome [1] on four days of eating? Must be pretty powerful superfoods they were eating to have that long-term effect. How can they claim that the survey done so long ago has any bearing on their eating habits during those twenty-two years?

Reductionism #2

The authors stratified the participants not only on the amount of protein consumed during that four-day period but also on the amount of animal vs. plant sources of protein. But humans rarely eat proteins by themselves, they are found in food, which has varying amounts of carbohydrates and fats. Reducing diet to simply animal or plant protein casts aside all of a foods' other components and whatever role they might play.

The authors did note that the food survey was a limitation, that “could introduce random error and thus attenuate associations.” So I am left with two questions, why did the researchers believe that one food survey was sufficient? Please let it be some sort of mistake in their reporting rather than magical thinking. More importantly, what peer-review permitted an at best weak finding, “marginally associated,” stratified on eating habits that were more than 20 years old to be published? Circulation is a respected journal, but it is printing scientism, not science. Once again, more noise, less signal.

[1] Heart failure is a constellation of symptoms, such as shortness of breath, and abnormal contractility of the heart musculature, and is rightly considered a syndrome. There are variants of heart failure that can be more precisely defined which might better be called a disease.

Source: Intake of Different Dietary Proteins and Risk of Heart Failure in Men Circulation DOI: 10.1161/circheartfailure.117.004531