Merck and the Vioxx Decision
James Hamilton has a very nice post on the dubious nature of this decision. Hamilton walks us through some of the statistical arguments and concludes the following,
So how do you explain these issues to a Texas jury? Well, the defendants argued these 5 to 1 odds aren’t statistically significant, to which the plaintiff’s lawyer Mark Lanier offered this folksy rejoinder (hat tip: Newmark’s Door):
Have you got $6 on you? I’m going to give you a dollar and you give me the six. It is not statistically significant in the difference. What do you think, are you in or out?
Sure, Mr. Lanier, I’m in, though here’s how the deal actually works– you give me the dollar, but you don’t know whether I give you $6 or I give you nothing. Or, to be a little more accurate, even if there actually is an elevated risk of the magnitude the studies suggest but can’t prove, the question is whether I might want to accept a 1 in 4,000 risk of dying from a heart attack in order to get the only medication that makes my pain bearable and a mobile life livable. And if I say no to the Vioxx, I may end up taking something that is less effective for my pain but has risks of its own.
And now I have a question for Mr. Lanier. How did we arrive at a system in which 12 random Texans are assigned responsibility for evaluating the scientific merits of statistical evidence of this type, weighing the costs and benefits, and potentially sending a productive blue-chip American company into bankruptcy protection?
I strongly recommend reading the entire post. For one thing while the Lancet study ultimately relied on 10 people. The study started with 8,143 patients who had a heart attack, of which 68 were taking a nonsteroidal anti-inflammatory drug (Vioxx is a nonsteroidal anti-inflammatory drug ). Also, the study looked at 31,496 patients who had not had heart attacks, but had also taken a nonsteroidal anti-inflammatory drug (NSAID). In the end the risks of taking a NSAID was not statistically significant in general, but there was a statistically significant dose-effect relationship. However, those statistics are based on just 10 patients–2 from the group with heart attacks and 8 from the group with no heart attacks.
In short, it looks like the legal system failed to accurately evaluate the weight of the statistical evidence. This is of course not at all unsurprising.