Sir: In the intriguing report by Templeton and Palazzo [1], the authors analyzed the impact of chest physiotherapy in the time to become ventilator-free (defined as the time between initiation of ventilation and when ventilator-free) in 172 critically ill patients under mechanical ventilation. The study was a well-designed single-blinded randomized trial, and the authors suggested that chest physiotherapy was associated with a longer time to become ventilator-free than a control group.

The choice of time to become ventilator-free as the primary endpoint requires some criticism because many factors influence it [2], confounding the effects of chest physiotherapy and consequently decreasing the power to detect clinical benefits. In an elegant study, Bernard et al. [3], using a statistical model, showed that the use of ventilator-free days (defined as the number of days between successful weaning from mechanical ventilation and day 28 after study enrollment) as primary end point has substantial advantages, improving the statistical power to detect clinical benefits in trials addressing critically ill patients.

One other important point is the Kaplan–Meier curve analysis. We agree with the authors that, if the assumption of proportional hazards is not true, the curves might be adequately compared using the Gehan–Breslow–Wilcoxon test, which does not require a constant hazard ratio.

However, the study presented an imbalance in baseline covariates, with higher APACHE II scores (49 vs. 41) and more patients with neurological damage (22 vs. 13) in the physiotherapy group. This imbalance, in itself, might lead to worse outcomes in the treatment group, confounding the results. Thus, the use of proper adjustment should produce a profound impact on the results [4] and could increase the statistical power and the ability to detect a treatment effect [5] in their study.

In conclusion, we believe that, had they used a better primary clinical end point and adequately adjusted for baseline covariates, the authors probably would have found more robust results and could have chosen a more appropriate title for their study.