In reply: Our economic analysis of COX-2-specific inhibitors (CSIs) was part of a research program on priority setting. The application to osteoarthritis (OA) involved cost–utility analyses of 19 interventions (written up in a 195-page research report1), and was subject to peer review by an advisory panel including senior clinicians. The research drew on over 200 references, 23 on CSI, but a limit of 50 references for articles in the Medical Journal of Australia meant that full referencing was not possible.

Evidence of efficacy in OA and adverse events (gastrointestinal and cardiac) are incorporated in our QALY estimates, the latter based primarily on the seminal CLASS trial (Celecoxib Long-term Arthritis Safety Study), and the United States Federal Drug Administration (FDA) analysis2-4 of this trial (summarised previously5). The FDA report concludes that for the primary endpoint specified in the study protocol — clinically significant upper gastrointestinal event (CSUGIE) — for the entire study period there was no significant difference in adverse events between the celecoxib arm and the combined diclofenac/ibuprofen arms (P = 0.45).2 Using the broader definition — combined CSUGIE and gastroduodenal ulcer (CSUGIE/GUD) — a significant difference is reported between combined non-specific non-steroidal anti-inflammatory drugs (NSAIDs) and celecoxib (P = 0.040), but not with diclofenac (P = 0.295).2 The FDA concluded that “celecoxib was not able to demonstrate it was statistically superior to diclofenac in terms of the clinically important UGI [upper gastrointestinal] endpoints and conditions defined in this study. The same is not true when comparisons are made to ibuprofen.”2 The dominance of non-specific NSAIDs reflects celecoxib priced at $32.13/month (for 200 mg/day) and diclofenac at $13.42/month (for 75 mg/day)6 and evidence of equivalence in management of OA and GI side effect profile. We agree this does not support a conclusion about class dominance. All NSAIDs and all CSIs are not the same.

Celecoxib is listed on the Pharmaceutical Benefits Schedule, but, as submissions by companies to the Pharmaceutical Benefits Advisory Committee (PBAC) are confidential, the research team may not have had access to all relevant evidence. We would welcome access to such information from which to prepare revised estimates. Placing all submissions to the PBAC in the public domain, as now occurs with reports of the PBAC, would allow a more informed public debate on these matters and would be most welcome.