We read with the interest the report by Li et al.1 of their single-center retrospective registry analysis on reversal of neuromuscular blockade and postoperative pulmonary complications. No difference in the odds of postoperative pulmonary complications was observed between patients receiving sugammadex (4.2%) or neostigmine (5.9%) (adjusted odds ratio, 0.89; 95% CI, 0.65 to 1.22). This result is consistent with the largest prospective cohort study2 and the two small randomized controlled trials3,4 to date, but contrasts with a much larger multicenter retrospective registry analysis5 where sugammadex was reported to be superior to neostigmine. A large randomized controlled trial is clearly required to resolve this issue.One reason for differences between the abovementioned studies may be inconsistency in the definition of the composite primary endpoint (table 1). Well-constructed composite endpoints summarize the important consequences of a condition or intervention.6 Each component should be mechanistically linked to the condition or intervention, should meet a threshold for impact on health, and, ideally, should be of a similar type (i.e., diagnoses, clinical events, or management strategies). This was not the case for Li et al.’s retrospective registry analysis1 or the other studies mentioned in this letter.2–5 Although there is a clear pathophysiologic relationship between inadequate reversal of neuromuscular blockade and pneumonia, mechanical ventilation for 48 h or longer and unplanned intubation may be required to treat conditions that are not associated with residual paralysis, including cardiac and neurologic events and surgical complications such as bleeding.1 The question is how this affected the apparent incidence of postoperative pulmonary complications in the study of Li et al. and the validity of the conclusions they drew about reversal.In 2018, Abbott et al.6 conducted a systematic review and consensus-building Delphi process to identify a recommended definition of postoperative pulmonary complications. They identified 27 different definitions but could not reach consensus about the best existing definition. They therefore developed a new definition consisting of diagnoses that are mechanistically related to anesthesia and that exceed a threshold for impact on health: atelectasis, pneumonia, acute respiratory distress syndrome, and aspiration pneumonitis. We propose that future registry analyses extract data in line with this definition rather than relying on clinical events that may arise from nonpulmonary conditions.Dr. Peyton has received consultant fees from Getinge (Solna, Sweden). The other authors declare no competing interests.