The trials are often run on small numbers of unrepresentative patients, and the statistical analyses are massaged to give as rosy a picture as possible.
Nissen points out that based on other Merck analyses made public last month, it's possible to argue that the risk emerged after only three months of exposure.