Sirs. IN their paper in PLOS ONE (1), Bouchard et al concludes that there are subjects that responds adversely to regular exercise in cardiovascular risk factors, specifically Systolic blood pressure, HDL-cholesterol, triglycerides and fasting insulin Adverse responders (AR).
However, even if their study demonstrates some subjects with a change in adverse direction for one or more of these factors, the question is whether the number is large enough to be significant.
In their paper, they have defined the significance of the response in terms of variability of the actual measurements. Each parameter was measured three times over a presumably stable period of three weeks, in a sub study. The within-subject standard deviation was calculated for each measurement, and termed the technical error (TE). This definition of TE takes both measurement variability and day-to-day biological variability. Defining an adverse measurement value as 2xTE is consistent with a definition of normalcy as the range between the 2.5 and 97.5 per cent percentiles of a reference population.
However, in their paper, Bouchard et al defines a new variable, the change in measurement from baseline to post training intervention, and the adverse response as a change in adverse direction of more than 2xTE. But a compound variable from two measurements, has a wider standard deviation. Each measurement has a variability of 2xTE, and the variability of the compound variable is the variability of each measurement x √2. Thus from their paper, and adverse response should be Tex2x√2. From the curves, given in the paper, this would seem roughly to cut the number of AR roughly by about 50%, but it would still mean that a normal response is the range between 2.5 and 97.5 percentiles of a reference population.
Secondly, given this definition of adverse response, the question remains whether the number of adverse responders really is significant. For this the number of adverse responders should be significantly higher than expected, the expected number being 2.5% from measurement variation alone.
From table 5 in the paper, using the corrected value for the variability parameter, the number of adverse responders would be roughly around 65 for ∆insulin, 210 for ∆HDL-C, 85 for ∆ triglycerides and 85 for ∆ systolic BP. The expected number would be about 42 for each measurement. The presence of adverse response to exercise would thus be dependent on whether the number of AR is significantly higher than the expected.
Until this is shown, I cannot see that the presence of adverse response to exercise in risk factors is conclusively evident.
Although the present database could give some supportive evidence by statistical means, it should preferably be shown in a randomised controlled study.

1. Bouchard C, Blair SN, Church TS, Earnest CP, Hagberg JM, Häkkinen K, Jenkins NT, Karavirta L, Kraus WE, Leon AS, Rao DC, Sarzynski MA, Skinner JS, Slentz CA, Rankinen T. Adverse metabolic response to regular exercise: is it a rare or common occurrence? PLoS One. 2012;7(5):e37887.