Phil B.FontanarosaMD, Deputy EditorIndividualAuthorStephen J.LurieMD, PhD, Fishbein FellowIndividualAuthor
In Reply: Although rofecoxib is designated
an NSAID, it differs biochemically from the comparator NSAIDs used in our
analysis in that it specifically inhibits cyclooxygenase 2 (COX-2) and does
not inhibit cyclooxygenase 1 (COX-1) at the doses tested.1- 3
Thus, as stated in our article, "The primary hypothesis involved a biologically
meaningful comparison of GI [gastrointestinal] safety with COX-2 specific
inhibition versus nonspecific COX-1/COX-2 inhibition." Dr Freston is correct
that the test for treatment by protocol interaction suffers from low power.
We noted that there were generally too few events within any single protocol
to provide precise study-specific results, and we outlined the analyses conducted
to evaluate the appropriateness of assuming a common RR. We acknowledged that
no PUBs occurred with nabumetone in the 1 small study that included it. It
is worth noting that no PUBs occurred with rofecoxib in that study either,
making that study uninformative in the combined analysis.
Langman MJ. Rofecoxib and the Risk of Adverse Upper Gastrointestinal Effects—Reply. JAMA. 2000;283(15):1960-1961. doi:10.1001/jama.283.15.1957