Meaningful comparison between the four groups of interventions has several prerequisites. First, the size of each cohort must be sufficient for the comparison to have adequate statistical power. Although the authors mention a sample size calculation, they do not provide the estimated effect size. Second, the groups should be matched for all but the intervention, which is not the case. In particular, the proportion of patients having an esophagectomy (associated with the highest rates of PPC) was unmatched between the groups: control, 0 of 35; flow incentive spirometry, 2 of 33; deep breathing, 6 of 35; and volume incentive spirometry, 1 of 34. This imbalance alone could account for the observed increased rate of PPC in the deep breathing group. The population studied was extremely heterogeneous, both with regard to type of operation and route of access, with laparoscopies and laparotomies considered together. The patients were at intermediate risk of PPC (based on a 10% incidence of PPC); to maximally demonstrate a reduction in PPC, it would have been prudent to selectively study those at the highest risk.