We agree that a limitation of our study was the open-label design. However, a double-blind study was not feasible because 1) it was not possible to manufacture a suitable placebo for inhaled human insulin, 2) it seemed inappropriate to blind treatment when individualized flexible-dose titration is needed, and 3) patients or physicians would have very easily unblinded the placebo or inhaled insulin because of the latter's immediate effect on blood glucose levels.

Ours was a proof-of-concept study that was designed in 1998 according to good clinical practices at the time and was based on a previous study of similar design (1). We do not believe there were any ethical issues for patients who were randomly assigned to the control group because microvascular complications are associated with long-term hyperglycemia, and there is no evidence that a relatively short period of inadequate glucose control will have a deleterious effect. The reality is that many patients who have type 2 diabetes with poor glycemic control continue to take oral agents instead of initiating insulin therapy (2). We continued the study for 3 months to provide sufficient time to show an effect (1) while minimizing the control group's exposure to hyperglycemia. Furthermore, all participants had the option of receiving inhaled human insulin by enrolling in an open-label extension of the trial.

In response to concerns regarding the timing of the publication, we would like to reaffirm the editor's reply and definitively state that the publication of this paper was not engineered to coincide with the FDA Advisory Committee's report on inhaled insulin. It is essential that data such as ours are subject to rigorous scientific scrutiny in a peer-reviewed journal. Indeed, the paper was subjected to 2 rounds of review by the Annals editors over several months, which also required the submission of additional analyses; therefore, it would have been impossible to predict if publication would coincide with the FDA Advisory Committee's report.

As highlighted in Dr. Comi's editorial in the same issue (3), we were also pleasantly surprised at the robust response to inhaled human insulin despite what might seem to be a nontraditional approach. Furthermore, the effects observed with inhaled human insulin extended beyond the predicted pharmacokinetic activity by substantially improving fasting plasma glucose levels, which probably contributes to the robust reductions of hemoglobin A_1c levels.

Inhaled human insulin could really have an impact if health care professionals can convince adult patients with type 2 diabetes to begin using insulin much earlier and more aggressively. The availability of the product as a treatment option has already been shown to substantially increase the proportion of patients who would theoretically choose to begin insulin therapy if they cannot achieve glycemic control with a modified diet or oral antidiabetic agents (4). We agree with Dr. Comi's viewpoint that patient acceptance and preference will ultimately determine the future use of inhaled insulin.