The federal jury in Plunkett went home today without reaching a verdict. While the first two state-court juries reached a verdict in a matter of hours, neither was unanimous, while a federal jury must have unanimity. The Wall Street Journal reports that the plaintiffs have sought a mistrial over the New England Journal of Medicine allegations.
Derek Lowe has a must-read post about the New England Journal of Medicine controversy; some of his commenters have more incisive analysis than has appeared in any mainstream-media publication. Commenter "TFox" writes:
The real question is whether conclusions are changed. Considering all patients, there was a significant difference in MI even without the new data, as correctly reported in the paper. For the nonaspirin group, there was no difference reported in the paper, and when you throw in the new events, there's still no difference (as Derek pointed out). Again, the paper remains correct. Finally, the most important question, where the risk comes from. The paper ascribes the added risk to the high risk aspirin group. The risk differential here is so large that even 3 new nonaspirin events don't change it; high risk patients on rofecoxib are keeling over at a rate 20x that of their low-risk cohorts (8+% vs 0.5%). Curfman calls this conclusion "misleading", but the new data gives you the same conclusion as the old!
So are calculations made "incorrect"? Only in a way that an innumerate reader would care about. Are the conclusions "incorrect" and/or "misleading"? No and no. This is basic stuff -- I doubt this "Editorial" would have made it through peer review without changes. I wonder what process NEJM used to assess it before going ahead and publishing? And will they be reconsidering that process going forward?
The reason why, to me, the most innocent explanation makes the most sense, is that I can't imagine what motivation a dozen doctors would have to risk their professional careers over minor adjustments in data or presentation that doesn't materially change the conclusions of the paper, especially when the full data set will become public in a few months anyway. Earlier discussion: Dec. 8.