« Reform watch: look to states | Main | Lobbying and Part D »

January 20, 2006



I just re-read both the expression of concern and the original article. While the points that NEJM makes are valid, they also indicate that the editors did a poor job of reviewing the manuscript. First off, by allowing the investigators to use Vioxx as the reference, rather than naproxen makes it more difficult to realize the magnitude of the increased risk of Vioxx relative to naproxen (relative risk curves are logarithmic, so they can be difficult to interpret). In addition, there were no tables in the original article disclosing CV risk, it was only briefly mentioned in the text. I will grant that some of this criticism is coming from hindsight--i.e. knowing now that all coxibs should be carefully vetted for CV safety. All that said, the WSJ is ridiculously out of line. Especially since the Bombardier article makes no mention of when the safety database was closed--implictly they asserted that there had been no undisclosed events from the start of the trial to the last moment before printing the journal.


I don't know of any study where adverse events that happened after the predetermined close date of a study (when the analysis begins)were included in that publication. Another paper might report on post study follow up or an extension period of a study. Otherwise, analysis begins with a closed file at the end of the study and that analysis is what gets published. It is also interesting that a publication scandal accusing P & G and researchers at Sheffield University of not including 40% of data because it would not support their pre-study hypothesis and marketing message for their osteoporosis drug (exposed Dec 5) has not been mentioned at all in the US media. It has been reported in the British papers and SLATE Magazine by a NY writer. Very odd.

The comments to this entry are closed.