Meta-analysis: were rofecoxib cardiac problems visible earlier?
A meta-analysis of all available controlled clinical trials that reported relevant data suggests that the increase in cardiovascular events associated with rofecoxib was potentially visible three or four years ago. The authors used a broad search strategy to locate as many controlled trials of rofecoxib as possible that reported data on cardiovascular events. They also attempted to locate any studies that determined whether or not naproxen had any significant protective effect, to decide whether or not this was relevant. Overall, they found 18 controlled trials of rofecoxib involving 21432 patients up to 2001. In this group, there were 64 myocardial infarctions. Patients in the rofecoxib group had an increased risk of MI - relative risk 2.24, independent of the comparator group. Studies of naproxen, all of which were epidemiological, showed it to have limited cardioprotective effect insufficient to explain the discrepancy in the VIGOR trial. The authors conclude that an increased risk of MI with rofecoxib was apparent from 2000, and therefore that the drug should have been withdrawn earlier. [Editor's comment: this appears to have been a well conducted analysis - nevertheless, it should be remembered that it is looking at secondary outcomes from trials, not those which the trials were designed to show. As a result, it is less robust than a typical meta-analysis looking at data on primary outcomes. The number of events included is also quite small, which probably reflects the generally low-risk trial population. Perhaps most important now is determining the mechanism - is it specific to rofecoxib (as the industry would maintain) or a class effect?]
Lancet, early online publication 5th November 2004 - currently available from the Lancet website here (registration required, will move after paper publication); BBC News report
|
