Manipulating a Journal Article - New York Times
I submitted the following to the editors of the NY Times in response to the above editorial, which itself was a reaction to the NEJM’s call for a correction from the authors of the VIGOR trial. They didn’t publish it, so I’m posting it.
The Times should have mentioned that it is atypical to include in the body of a clinical research article data derived past a pre-specified cutoff date. This editorial practice deters authors from “cherry-picking” data to inflate an article’s importance. Notably, the statistical and clinical inferences regarding Vioxx’ safety don’t change with inclusion of the additional data in question. The relative risk estimates in the overall study population (4.25 and 5.0) are not meaningfully different, given the wide 95% confidence intervals around them. Likewise, confidence intervals around the similar estimates of relative risk in the low-risk population (2.25 and 3.0) both include 1.0, precluding reliable inferences of increased risk. The above notwithstanding, clinicians should view all journal-published clinical trial findings skeptically, regardless of funding source, as all journals limit the quantity and types of data published, and all authors and editors hold biases that influence what readers see.
