Reader Comments
Post a new comment on this article
Post Your Discussion Comment
Please follow our guidelines for comments and review our competing interests policy. Comments that do not conform to our guidelines will be promptly removed and the user account disabled. The following must be avoided:
- Remarks that could be interpreted as allegations of misconduct
- Unsupported assertions or statements
- Inflammatory or insulting language
Thank You!
Thank you for taking the time to flag this posting; we review flagged postings on a regular basis.
closeRegression to the mean
Posted by trobc on 04 Jun 2012 at 20:16 GMT
The technical error (TE) defined as the within-subject standard deviation derived from these measurements was computed. An adverse response for a given risk factor was defined as a change that was at least two TEs away from no change but in an adverse direction.
http://plosone.org/article/info:doi/10.1371/journal.pone.0037887#article1.front1.article-meta1.abstract1.sec2.p1
I commend Bouchard and his colleagues on undertaking and reporting their study of the potential for actual harm in a subset of individuals who are induced to undertake rigorous exercise. Such a finding has implications for the universal recommendation to exercise for those whose current physical condition does not contraindicate it. They report that about 8-13% of the study population experience worsening in four biomarkers of cardiac risk after a course of exercise.
Unfortunately, there is a simple competing explanation for the observed phenomenon: regression to the mean. Assuming no beneficial effect of exercise and that technical error is equal to the standard deviation in the population (as for systolic blood pressure), a simple simulation shows that using the authors’ criterion the fraction of apparent negative changes would be approximately 8%, not far from the 12.2% observed for systolic blood pressure by the authors. However, since there is a well established beneficial effect, it should work against this and so the 8% expected by regression to the mean has to be adjusted for the average beneficial effect.
The upshot is that the authors should provide an additional analysis that takes regression to the mean into account explicitly, as it could considerably lower the estimate of the fraction with truly adverse effects. It would also be interesting to see what the percentage of “harmful changes” take place in the control groups of the randomized studies, for comparison.