LAUSR

To the Editor: We read with interest the paper by Comes et al. (1) titled “Changes in Cycle-Ergometer Performance during Pulmonary Rehabilitation Predict COPD Exacerbation”. This retrospective study included COPD patients who participated in a pulmonary rehabilitation and showed that a reduction in the performance of the cycle ergometer can be used to predict future exacerbations. The study findings were the first during rehabilitation sessions to show a promising objective outcome, other than regularly monitored variables used in tele-health in COPD to identify those at risk of future COPD exacerbations (2). In addition to the limitations reported by the authors, there are some additional points that need to be discussed further. This study yielded interesting results; however, the study provided limited information on the methods of measuring the outcome. Only moderate and severe exacerbations were defined but not mild (3). This may indicate that some exacerbations were missed, which makes their results less likely to be generalizable. The exclusion of patients who did not experience any exacerbation was not justified, which might be a selection bias. Furthermore, the way to confirm exacerbations was not explicitly described. Authors indicated that comorbidities were collected for those patients; however, this was not presented along with the other patients’ characteristics. Exacerbations must be carefully differentiated from other comorbidity-related events, such as worsening cardiovascular comorbidities and pneumonia (4, 5). It could be argued that such a decrease in the cycle ergometer performance was a result of other comorbidities that may negatively impact their stable COPD status. Such possibilities should be investigated further. The evidence has shown the positive influence of pulmonary rehabilitation on reducing exacerbations (6), which would be counted for those who were regularly attending. We are not clear if the patients attended the rehabilitation sessions consecutively. If not, how possible a non-consecutive attendance may be part of the cause of the recorded exacerbations? Although a clear statistical plan was presented using post-hoc analysis with a Bonferroni multiple comparison test, it would be interesting to use a multivariate regression to identify the independent predictors of exacerbations. This would therefore allow statistical adjustments for other confounding variables leading to more rigorous outcomes. To translate the findings into a clinical decision-making tool, it would be most useful to have a test that was sensitive to predict the exacerbations. This made us interested to know what would be the cut off value in the performed ROC analysis at 80% sensitivity and how specific it would be. In its current format, with special thanks to all the authors, this paper leaves us with four important unanswered questions. 1) Would a drop in the performance of the cycle ergometer can predict exacerbations if prospectively monitored? 2) If frequent ( 2/year) exacerbators are compared with infrequent exacerbators, what would be the difference? 3) How would be the results if the performance of the cycle ergometer, physical activity and changes in physiological parameters (heart rate, respiratory rate, peak expiratory flow) are compared with a larger representative sample? 4) Which one of the aforementioned variables would have the highest predictive capacity for COPD exacerbations?