We thank Xue et al . for their interest in our subgroup analysis of obese patients1randomized to 80% or 30% oxygen in the PROXI Trial.2
We agree that it would have been interesting to screen all patients for obstructive sleep apnea, but that was not done. Patients were nursed after surgery in the semirecumbent position in all departments participating in the trial. Chest physiotherapy may be beneficial, but the evidence is not very convincing, and the effect depends on the patient's ability to cooperate.3
We do not agree that every possible risk factor has to be standardized in a randomized clinical study. We sought to optimize and standardize care as much as possible,4but we realized that minor differences would still be present among centers. In our pragmatic and randomized trial, we stratified for study center, diabetes mellitus, and acute or elective surgery, and this will most likely reduce the probability of intergroup differences in known as well as unknown risk factors of postoperative complications. The aim was to enable us to detect a clinically relevant difference according to inspiratory oxygen fraction, if present. Pragmatic trials should inform decision-makers about the effectiveness of treatments in the settings in which they are to be implemented.5Tight control over every anesthetic and surgical variable may not be possible in large multicenter trials.6Actually, tightly controlled explanatory trials will have little generalizability.7,8
