Until recently, the Food and Drug Administration criteria for new drugs in development to treat type 2 diabetes required at least one placebo-controlled, proof-of-efficacy study for regulatory acceptance and review. Our study was initiated in 1994 in patients participating in a proof-of-efficacy study authorized by the Food and Drug Administration and approved by 24 independent ethics and institutional review boards. We now have a deeper understanding of the harmful effects of hyperglycemia, as indicated by the Diabetes Control and Complications Trial (1) and in the immediate wake of the recent United Kingdom Prospective Diabetes Study (2). With this increased understanding, the ethics of placebo-controlled study designs can be challenged.

An important issue when considering the ethics of a short-term study such as ours is the potential for harmful effects during 6 months of placebo use in patients previously treated with sulfonylureas. It clearly does not benefit the patient; however, on the basis of our current understanding, drawn from long-term prospective studies, the harmful effects of short-term withdrawal from sulfonylurea is likely to be low. Moreover, it is important to note that glycosylated hemoglobin levels tend to increase over time in patients receiving long-term sulfonylurea therapy (2), so it can be argued that only a portion of the elevation in our placebo group can be attributed to the interruption of sulfonylurea therapy. Finally, Dr. Misbin's comment about an increase in glycemia that "will predictably cause symptoms that interfere with daily activities" does not apply to our study because the study protocol allowed for patients who developed symptomatic hyperglycemia to be withdrawn from the study. Six patients were withdrawn for this reason. Their data were included in their initial treatment groups according to the intention-to-treat analysis.

The ethics of clinical research, with particular reference to the placebo-controlled trial, involves the effective resolution of a clinically important question while striking a balance between potential risks to study participants and potential benefits to them and society. We conducted our study after careful consideration of these counterbalancing factors according to information available at the time. Although different decisions might be made today, the demonstration of the effectiveness of a new oral agent for treatment of a disease that has been notoriously difficult to treat provided a substantial benefit for what we believe to be a small risk to study participants.