Statistics from Altmetric.com
In their paper on bronchodilator reversibility testing in COPD, Calverley and colleagues1 come to the intuitively sensible conclusion that, in severe COPD, bronchodilator responsiveness is a continuous variable. However, this conclusion is based on an analysis in which the change in forced expiratory volume in 1 second (FEV1) effected by inhalation of a bronchodilator aerosol is related to the baseline (that is, initial) level. As a result, the reported bronchodilator responses are subject to the error that can result from regression to the mean.2 The error can be minimised by relating the change to the mean level instead of the initial level, and it would be reassuring to see the data expressed in this form.
We thank Dr Cotes for his interest in our work1 and his thoughtful comments. We did not include the suggested analysis as our purpose was to address the utility of a single bronchodilator reversibility test which is the way this is applied in routine practice. We agree that regression to the mean is a real possibility, but also believe that physiological variation in baseline airway calibre contributes to the “noise” in this particular signal. Unfortunately the number of times this test must be undertaken to minimise these confounding effects is not clear, but is likely to be too large to be possible in most clinics. Hence our caution about interpreting the significance of small changes in spirometry recorded on one visit.