In their letter, Gan and Lubarsky emphasize the use of incremental or marginal cost-effectiveness in comparing older drugs and technology. We disagree with their suggestion that there were fundamental errors in the calculations of the cost-effectiveness ratios for the antiemetic drugs we studied. In our article [1], we used a decision analysis to determine the total costs associated with strategies to prevent postoperative nausea and vomiting (PONV) and calculated the total costs to prevent PONV in one patient with each strategy. We used these values as a reflection of the cost-effectiveness of the regimen, rather than an incremental cost-effectiveness analysis, which has been used by other authors. Methods similar to the ours have been used in recent comparisons of cost-effectiveness in major journals, including: 1. the use of smoking cessation services, in which the average costs to the health plan per user who quit smoking were calculated [2] 2. shortened screening intervals for breast screening programs [3] 3. regimens for eradication of Helicobacter pylori [4] 4. routine immunization against varicella [5]. Although Gan and Lubarsky present arguments in favor of the use of incremental cost-effectiveness analysis to determine the additional costs for prevention of one additional case of PONV, this approach also has some significant limitations. For example, small changes in mean costs or efficacy can result in vast changes in the incremental cost-effectiveness ratios, even if there are no statistically significant differences between the study groups in either the costs or the efficacy. In addition, incremental cost-effectiveness ratios may not provide meaningful values in some clinical situations. Consider, for example, the cost-effectiveness of a prophylactic regimen of a hypothetical new, second-generation antiserotonin drug ("bettersetron") that prevents PONV in 85% of subjects at an overall cost of $2.25 per patient. The formula recommended by Gan and Lubarsky would result in a negative value for incremental cost-effectiveness that has no clinical meaning. These investigators would recommend the use of yet another term, "dominant therapy," for this particular situation. However, the approach used in our analyses is much simpler and is more meaningful from the clinical care perspective. Our analysis suggests that this new regimen would cost the healthcare institution $2.65 (2.25/0.85-2.65) to avoid one case of PONV, compared with $7.06 for droperidol 0.625 mg IV, and $27.65 for ondansetron 4 mg IV. Regardless of the approach used, the message from our study remains the same: droperidol 0.625 mg IV provides antiemetic prophylaxis comparable to ondansetron 4 mg IV without increasing side effects or delaying discharge and is associated with decreased costs. Of interest, our basic findings have been confirmed in a subsequent multicenter study [6] in which Dr. Gan was one of the principal investigators. Mehernoor F. Watcha, MD Department of Anesthesiology and Critical Care; University of Pennsylvania; Philadelphia, PA 19104-4399 Paul F. White, PhD, MD Department of Anesthesiology & Pain Management; University of Texas Southwestern Medical Center at Dallas; Dallas, TX 75235-8894