Sometimes when you do a hit job on someone you get caught, especially if your target is Bob Temple. Daniel Carpenter and the self-serving Jerry Avorn tried to claim that there was a higher incidence of black box warnings around drugs approved right before user fee deadlines (they used 2 weeks, an artificial number) than what they considered routine (I guess that means two weeks and one day).
They got caught when Temple and FDA economists found significant, uh, omissions in their daabase and "rounding" errors that when accounted for essentially eliminated any difference in the number of black box warnings.
According to the WSJ Carpenter and Avorn did a fix. If you read the NEJM apology for getting caught by Temple the Carpenter and Avorn fix is this: they just add new black box warnings gleaned from the "addition of data through December 2007" Ta-da! They "found more modest but still significant association between just-before-deadline approvals and safety problems."
Here's the trick: The black-boxing of a drug is a response to policy pressure as much as writing an article about black boxes is. Adding data throughout 2007, when pressure was on the FDA to ADD black box warnings from Grassley and others, data that is not included in Temple's analysis is a sure-fire way to demonstrate "significant associations." Even the sloppy Carpenter and Avon admit: most of this difference stems from the different time frames used in the two analyses.
But the media will likelly not pick up on this. After all the headline of the WSJ article regarding this issue is: Scholar Fixes Flaws, Defends Finding of FDA Errors In Too-Hasty Approvals
The WSJ article missed the sleight of hand Carpenter and Avorn use to 'fix' things...
They got caught when Temple and FDA economists found significant, uh, omissions in their daabase and "rounding" errors that when accounted for essentially eliminated any difference in the number of black box warnings.
According to the WSJ Carpenter and Avorn did a fix. If you read the NEJM apology for getting caught by Temple the Carpenter and Avorn fix is this: they just add new black box warnings gleaned from the "addition of data through December 2007" Ta-da! They "found more modest but still significant association between just-before-deadline approvals and safety problems."
Here's the trick: The black-boxing of a drug is a response to policy pressure as much as writing an article about black boxes is. Adding data throughout 2007, when pressure was on the FDA to ADD black box warnings from Grassley and others, data that is not included in Temple's analysis is a sure-fire way to demonstrate "significant associations." Even the sloppy Carpenter and Avon admit: most of this difference stems from the different time frames used in the two analyses.
But the media will likelly not pick up on this. After all the headline of the WSJ article regarding this issue is: Scholar Fixes Flaws, Defends Finding of FDA Errors In Too-Hasty Approvals
The WSJ article missed the sleight of hand Carpenter and Avorn use to 'fix' things...
