But, if anything, I was too easy on Lucchesi, whose opinion is junk science.
To recap: Lucchesi issued some opinions, and then a conclusion that Vioxx was "highly likely" to have caused Plunkett's death. But Lucchesi's conclusion is an overstated one that cannot be legitimately or scientifically drawn from the evidence. It's a for-hire opinion that epitomizes what's wrong with expert testimony in litigation.
As you recall, the APPROVe study showed a 1.92 relative risk of cardiovascular event for those who had taken Vioxx for eighteen months. A "1.92" relative risk means that Vioxx patients were 1.92 times more likely to have an event over the control group taking a placebo. Before eighteen months, the relative risk of Vioxx use was not statistically significantly different than 1. Even after 18 months, the relative risk of mortality (from cardiac causes or otherwise) was not statistically significantly different than 1. (Even Dr. David Graham, whose Lancet attack on Vioxx alleging 140,000 deaths gained so much publicity, found a relative risk for low-dose Vioxx of 1.24—and that number was not statistically significantly different from 1. Graham was forced to mix-and-match a cherry-picked higher relative-risk number from another study to reach his 140,000 figure.) APPROVe says nothing of the mechanism by which low-dose Vioxx causes more heart attacks, assuming that it does.
First, Lucchesi testifies as to his hypothesis of the mechanism. The COX-2 enzyme produces prostacyclin. Prostacyclin may inhibit the production of thromboxane, and thromboxane can cause blood clots, according to the theory. Thus, a COX-2 inhibitor such as Vioxx may not just reduce pain, but it may also result in excess thromboxane production, and the additional thromboxane may cause a clot that causes a heart attack. This may or may not be true—not even all of Plunkett's experts agree with this theory, and it's inconsistent with the evidence that different COX-2 inhibitor compounds have different rates of relative risk. It's also inconsistent with the animal studies Merck did in the late 1990s. (And it's suspicious that Lucchesi has refused to produce his unpublished animal data, which is allegedly to the contrary, and which features a much higher dose of rofecoxib, for review.) The theory may be ascribing too much power to thromboxane given the wide variety of interacting enzymes at issue: a study of the effect of ibuprofen, a COX-1 inhibitor, on platelet aggregation responses shows that platelet aggregation responses return to normal after eight hours despite significant inhibition of thromboxane by the drug. An alternative theory of mechanism is that COX-2 inhibitors act to raise blood pressure, and a long-term, if not a short-term, increase in blood pressure increases risk of cardiovascular problems; thus, one would not see an impact from Vioxx usage until after several months of use. This is consistent with the evidence to date. Should Lucchesi's testimony to the contrary be admitted here? We'll assume arguendo that he can. At most, however, this theory permits Lucchesi to opine that Vioxx might have caused Irvin's heart attack.
But what Lucchesi actually opined was that Irvin took Vioxx for 23 days, and that Vioxx was "highly likely" to be responsible for Irvin's death. And it's this opinion that I object to.
What is highly likely? Two-thirds? Four-fifths? Ninety percent? These numbers would translate into relative risks of 3.0, 5.0, and 10.0 respectively. But Lucchesi's theory requires one to interpolate the long-term relative risks as equivalent to the short-term relative risks. And there's no evidence the long-term relative risk, much less the short-term relative risk, is that high. There isn't even statistically significant evidence supporting a short-term relative risk greater than 1. Lucchesi has left the field of science and entered into the realm of advocacy of a conclusory opinion.
One could arguably conclude that Vioxx caused a heart attack if there are no other risk factors. But Lucchesi didn't evaluate the risk factors. At best, he cherry-picked the conclusion of a plaintiffs' pathologist that opined that plaque rupture in Irvin's coronary blockage didn't cause his fatal heart attack. An NPR account of Dr. Colin Bloor's testimony in the trial is telling:
"Am I correct, Doctor, that there are different types of plaque, some more prone to rupture than others?" Bloor said, "Yes."
Beck asked, "Would you agree that Mr. Irvin had a more unstable kind of plaque?" Bloor agreed.
Then Bloor insisted he hadn't seen any signs of a rupture on slides of four cross-sections of Irvin's affected artery.
"You looked at four sections (of the artery). If someone wanted to, they could take thousands of these slices, right?" Beck asked.
Beck continued, "You're aware that every other pathologist who's looked at these slides disagrees with you?"
"Yes," said Bloor.
Then Beck said, "If all these pathologists agree there was a plaque rupture, then the cause of death was plaque rupture, not Vioxx."
"That is not my opinion," Bloor maintained.
What scientific basis did Lucchesi have for adopting the position of one pathologist over the consensus of all of the others? None, I suggest.
We see a similar litigation-inspired shift in Dr. Wayne Ray's testimony Thursday. When Ray was publishing papers in peer-reviewed journals on Vioxx, he found a statistically-insignificant 1.02 relative risk for low-dose Vioxx. (Lancet. 2002 Oct 5;360(9339):1071-3.) At trial, on the witness stand, the unpublished number becomes much higher. Merck will cross-examine Ray today. If Judge Fallon adheres to his two-week trial schedule, one would expect Plunkett's case to close Monday.