From afar I have not had a chance to absorb the FDA panel review of rivaroxaban, but the news reports indicate that the panel (by a slim majority) felt that there was too much missing data to draw a firm conclusion about benefit risk balance in the setting of acute coronary syndromes. Thus, after an enormous investment with many thousands of patients volunteering for a human experiment, a declaration of a positive trial in a New England Journal publication and review by the international community there is a chance that the treatment won’t be labeled for use for this indication in the US.  Several key points:
1. The FDA is not bound to agree with the panel–it is advisory and the final decision will reside with the FDA itself. For those who criticize the FDA, it’s worth a second to pause and think about the pressure that is generated by situations like this in which enormous money and human talent has been invested and in the end we have strong opinions. The rest of us can have opinions, but those at the FDA essentially have a binary decision to make, and regardless of the decision there will be major criticism.
2. Missing data is a frustrating problem that is not appreciated enough by biomedical scientists. A recent IOM report reviews this in detail, but essentially there is no truly satisfactory way to deal with missing data.
3. In clinical trials those who are lost to follow up, withdraw consent, stop taking the drug or coming for the intervention, or those who miss an important test are almost always the participants most likely to have relevant outcome information.
4. In an individualistic society there is a primacy of the individual right to participate. In a society with a primary focus on the good of all, there is a primacy of continuing participation even if you cannot adhere to the treatment. This difference is profound in terms of the proportion of people who withdraw consent or become lost to follow-up. The contrast of the US with other countries in this regard is increasingly concerning. The US was the model for clinical trial conduct in the past–this is no longer the case.
One thing for sure: the riva panel will focus attention on this issue.
more...