Another interesting fact regarding the AIR2 study was that missing data for secondary outcomes, such as safety, were imputed using the last observation carried forward method. In other words, the authors assumed interval stability from the most recently recorded data point, which could certainly underestimate negative outcomes. In addition, patients were required to complete only their symptom diaries within the 4 weeks prior to their next scheduled study visit. During the treatment phase, the visits were close, and therefore, adverse events were easy to recall, whereas follow-up patients relied more heavily on their memory of events. Patients in the treatment arm were much more likely to have experienced a respiratory treatment-related adverse event (mild, 43.6% vs 58.7%; moderate, 53.2% vs 39.8%; severe, 3.1% vs 1.5%).3 It may be that these early respiratory complications at least partially explained concern regarding unmasking of treatment assignment in this group.