There was an increase in cardiac events on Vioxx in the AD trial 078.
The published article misrepresented the cardiovascular outcomes of the trial. Although the study protocol stated that the primary endpoint would be an ITT analysis, the published paper did not report any such analysis. Merck presented the on-drug and off-drug data separately and disaggregated fatal and nonfatal myocardial infarctions and cardiac arrests. When combined, the results indicated that Vioxx treated-volunteers had suffered a statistically significant increased rate of MI/SD (sudden death)* of 2.16 (CI 1.12, 4.35). In addition, there was a statistically significant increase in adjudicated CV/T deaths RR = 4.72 (CI 1.59 - 14.00). These volunteers used prophylactic anti-thrombotic medications, including Plavix and aspirin. (Thal LJ 2005). Aspirin's anti-thrombotic capabilities may have concealed the negative cardiovascular effect of Vioxx, thus acting as a significant confounder.
The problem with this conglomeration of numbers is that it assumes that Vioxx has a latent effect, and it is thus appropriate to ascribe a heart attack that occurred over a year after a patient stopped taking Vioxx to Vioxx, rather than to random chance. In e-mail, Dr. Egilman indicates that he does believe COX-2 inhibitors do have a latent effect on the arterial walls.
This seems inconsistent with the circumstantial evidence that plaintiffs' attorneys have refused to even propound this as a theory of recovery, though it would double or triple the number of potential plaintiffs; it's not like the trial bar is shy about cutting-edge scientific claims. (One could construct a conspiracy theory that the trial bar unanimously agreed to a cartel that would refuse to bring claims of latent injury for fear of cratering a settlement: Supreme Court precedent and the "Field of Dreams" problem make it infeasible for defendants to agree to a settlement of latent claims, and without a global settlement, Merck would be unlikely to settle non-latent claims. But Occam's Razor and the incentives of individual trial lawyers to defect from the cartel would suggest a more basic reason for refusing to bring claims of latent injury.) If Vioxx does not have a latent cardiovascular effect, then Dr. Egilman's numbers include irrelevant data and prove nothing. I further note that Psaty & Kronmal were not reticent about making claims of Merck perfidy in the presentation of the Study 078 data, but never make anywhere near as aggressive a claim in their peer-reviewed article.