In Reply:-We agree that Mr. Olofsen has identified a fundamental flaw in the mathematical basis of our technique for analyzing binary data. We are surprised that our simulations did not reveal this flaw and can only conclude that we found local maxima to the log likelihood. We regret presenting a technique that did not have a sound mathematical basis. Although we agree with Olofsen that the technique we describe is invalid, we do not agree that all of the conclusions of this paper are invalid. We still contend that the measure of the steepness of the concentration-response curve, gamma, frequently is underestimated because of the inability to distinguish intra- and interpatient variability. We believe our manuscript does identify the potential errors associated with naive pooled data analysis of binary response data, a problem that has not been explicitly identified in the literature and that is well illustrated by figures 1 and 4. We are hopeful that the technique presented by Sheiner [1]will be a means of distinguishing the two sources of variability, but this must be confirmed by further study. However, given the flaws of our technique, we believe that our paper should be withdrawn.

James M. Bailey, M.D., Ph.D.

Department of Anesthesiology; Emory University School of Medicine; 1364 Clifton Road, N.E.; Atlanta, Georgia 30322

(Accepted for Publication September 10, 1997.)

Sheiner LB: A new approach to the analysis of analgesic drug trials, illustrated with bromfenac data. Clin Pharmacol Ther 1994; 56:309-22.