We thank Wang et al for their comments on our recent trial.

First, we understand that some clinicians do not have access to droperidol. However, the midazolam-olanzapine combination has an efficacy and safety profile very similar to that of midazolam-droperidol and has potential as a suitable alternative.

Of the 88 patients who received sedative medications before the first study medication, 19 were administered out-of-hospital intramuscular midazolam according to the paramedic protocol. The median dose and time before the first study medication were 10 mg (interquartile range [IQR] 6 to 16 mg) and 66 minutes (IQR 38 to 125 minutes), respectively. In the emergency department, a total of 105 doses of sedative medication were administered before the first study medication: median time before the first study medication was 48 minutes (IQR 23 to 89 minutes). The majority were oral doses of diazepam (n=45) or olanzapine (n=36), with median doses of 10 mg (IQR 10 to 10 mg) for both medications. The remaining sedative medications included other benzodiazepines, antipsychotics, and ketamine through a range of administration routes.

The use of any oral or parenteral benzodiazepine or neuroleptic (typically oral diazepam or olanzapine in our institutions) before enrollment was dichotomized into yes or no and added into the Cox regression model. This variable had no significant effect on the model. Given the small numbers and variability in medications, doses, and timing, no further modeling of pre-enrollment sedation was possible.

We can confirm that the discrepancy between the trial registry information and our report was a clerical oversight. We designed the trial to have the most clinically relevant primary outcome, ie, proportion of patients adequately sedated at 10 minutes. This primary outcome and the sample size calculation based on it were recorded in the study protocol that was approved by our ethics committee before the study began. The term “time to sedation” that we recorded in the registry was too broad because it could include both proportions sedated at various points and actual times to sedation.

The propensity of midazolam (alone or in combination) to cause respiratory events is well known. We agree that the risks and benefits of midazolam-droperidol use need to be weighed. However, although the event rate is important, the consequences of these events also need to be considered. In this regard, the majority of respiratory events are transient and easily managed and are of no clinical consequence. As Wang et al suggest, there are confounding variables that affect respiratory events. In our study, patients intoxicated with alcohol had more respiratory events than those who were not (18.7% versus 12.1%, respectively). However, patients intoxicated with recreational drugs did not have more respiratory events (14.1% versus 15.5%, respectively). In a multicenter study, we are currently investigating the effect of sedative medications, intoxication, and other confounding variables on the risk of respiratory (and other) events.