Heart health, exercise, and misleading correlations

THANK YOU FOR VISITING SWEATSCIENCE.COM!

As of September 2017, new Sweat Science columns are being published at www.outsideonline.com/sweatscience. Check out my bestselling new book on the science of endurance, ENDURE: Mind, Body, and the Curiously Elastic Limits of Human Performance, published in February 2018 with a foreword by Malcolm Gladwell.

- Alex Hutchinson (@sweatscience)

***

Just noticed a great post on Amby Burfoot’s Peak Performance blog where he interviews Paul Thompson, the Hartford Hospital cardiologist who’s the go-to guy for questions about exercise and heart health. It’s a fantastic interview, very wide-ranging, with Thompson sharing his thoughts, hunches and beliefs about a bunch of the current controversies in this area.

One part that caught my attention in particular was a question about a recent German study that found a high prevalence of hardened arteries among marathon runners (I blogged about it here). Thompson was a co-author on the study, and he shared a frank assessment of some of the study’s strengths and weaknesses:

[H]is marathon group includes a number of former smokers and others who might have been quite unhealthy before they began running…

The key issue with Mohlenkamp’s runners is that their cardiac risk scores are compared using their present cholesterol, blood pressure, and other health numbers. They might have had terrible numbers before they started running, so when their coronary calcium is compared with folks who are not athletes, but had good risk numbers all their lives, it looks like the runners had more calcium, ie, more atherosclerosis than predicted by their risk factors… Many of the runners “got religion” when they turned 40 or so.

This is a classic illustration of why a single study, or even a single group of studies, can so easily point us in the wrong direction. On the surface, it looks simple: take a bunch of marathoners, compare them to controls, and presto — marathoners have worse arteries. But it’s easy to be led astray by underlying factors (not to mention, as Thompson points out, that the hard, stable arterial plaques found in the runners may actually be a good thing, as opposed to soft, unstable plaques that are easily dislodged).

Anyway, it’s an interesting read, and Burfoot does a great job “pinning Thompson to the mat” to get his (well-informed) opinions and best guesses on a bunch of topics.

2 Replies to “Heart health, exercise, and misleading correlations”

  1. The whole point of randomized control group study is to result in a control group that does not have any differences in relevant background variables from the intervention/treatment group.

    If there is self-selection among the marathoners that is influenced by important baseline variables than the study findings are simply not valid. I haven’t read the paper (did read the article). Thompson implies that alot of the marathoners were prior smokers. I would like to know what ‘alot’ is and how it differs from the control group and the general population. Did the study report on this? If there were important differences that would seem at a minimum to call the results into question.

    The hard vs soft plaque issue is intersting. Somewhat similar to the ‘pathological’ differences seen in runners knees which don’t seem to be pathological for runners.

  2. Good point, Seth — the control group is definitely supposed to match the experimental group in all relevant parameters. The fact that Thompson is commenting on the smoking-history issue means that they are at least aware of this potential confounder. The problem is that, almost by definition, we don’t actually know ALL of the relevant variables that affect any given outcome — that’s why we’re studying it. So for example, if you take 100 people recruited from running clubs, and compare them to 100 people who you’ve recruited by putting up ads offering $100 to participate in the study, who knows what systematic differences there are between those populations even after matching for age, sex, weight, etc. People who have lots of time on their hands and need $100 might share all sorts of underlying traits. Same problems with recruiting subjects from university campuses. Of course, all these issues are well known to people who care about proper study design — but I thought this was a nice clear illustration of the problem, and how results that seem straightforward and unambiguous might actually be misleading.

Comments are closed.