Challenging an exercise challenge protocol

In an interesting study, published in The Journal, HABY et al. [1] compared an exercise challenge test with histamine challenge in children. Some children responded to both challenges, and some reacted to one challenge but not to the other. Most of the children reacted to neither challenge. I agree with at least one of the conclusions, namely that the study "suggests that the two challenges identify different abnormalities of the airways". On the other hand, I object to the way in which the authors presented the "relationship between dose-response ratio to histamine and the percentage fall in forced expiratory volume in one second (FEV1) to exercise, shown with the regression line" in their figure 2. The regression line gives the uncritical reader the impression that there is a significant correlation, with a slope of about 45°C. There are at least two fallacies in this presentation. Firstly, for 91 of the 94 subjects there is no obvious correlation, at least not in agreement with the regression line presented. Thus, the regression, if any, cannot be a linear one. Secondly, the three out of 94 subjects which cause the "slope" had more than a 40% fall in FEV1 on exercise. From the Method section, it is clear that in these three "the postexercise measurements were stopped (before the 10 min value) and 200 μg salbutamol aerosol administered". Thus, the recorded postexercise values for these three subjects were not true maximal values, and this will distort the "slope" still further. My own conclusion is that their figure 2 does not show any physiological correlation between the two types of tests for bronchial hyperreactivity in spite of the "statistical significance". The figure demonstrates that the computer will draw the "best possible" straight line when ordered to do so, even when there is no true linear regression.