Substantial data have shown that both celecoxib and rofecoxib cause sodium retention similar to that caused by nonselective NSAIDs. We appreciate the reminder from Pfister and colleagues that other manifestations of sodium retention (such as pleural effusions) can also occur. The primary purpose of our study was to directly compare the peak effects of rofecoxib, indomethacin, and placebo on glomerular filtration rate and other variables of renal function in patients predisposed to such effects. The study that most similarly addresses the same question with celecoxib showed comparable effects (1). In that study, a single 400-mg dose of celecoxib (200 mg twice daily is the highest approved daily dose) caused a significant peak decrease of approximately 13 mL/min in glomerular filtration rate. In our study, a single 250-mg dose of rofecoxib (10 times the highest clinical daily dose) resulted in a significant decrease of 14 mL/min in glomerular filtration rate. No published studies directly compare the two COX-2 inhibitors; those cited by Dr. Whelton are not comparisons. The only trial suggesting a lack of effect of celecoxib on renal function examined a patient sample different from that of our study and obtained glomerular filtration rates at time points that probably missed peak effect (2). Nonetheless, rofecoxib did not affect glomerular filtration rate in that sample.

It is difficult to comment on Dr. Whelton's unpublished study comparing once-daily doses of rofecoxib, 25 mg, and celecoxib, 200 mg. However, a prospective study presented at the same EULAR meeting showed that 25 mg of rofecoxib was superior to 200 mg of celecoxib in relieving ostearthritis pain and that adverse experiences did not differ among patients taking each drug (3). Interpretation of the unpublished trial cited by Dr. Whelton is limited in light of this information, which suggests that equally effective doses were not used.

On the basis of preclinical and clinical studies (including recent case reports [4]), readers should assume that renal effects are likely to be class related (and, therefore, mechanism based) and that clinically relevant differences are unlikely to exist between individual COX-2 inhibitors or between COX-2 inhibitors and nonselective NSAIDs. In the absence of randomized, dose–response, comparative studies with rigorous and direct assessments of renal function, it is not appropriate to suggest differences. Such a suggestion could lead to insufficient monitoring of at-risk patients and may be detrimental. We therefore concur with Dr. Dunn's comments that selective COX-2 inhibitors as a class must be used cautiously in patients with predisposing diseases, including chronic renal failure, severe cardiac disease, and hepatic failure (5).