Still more ways to cook the clinical trial books

Clinical trials are concerned with efficacy and safety. There are statistical tricks and methodologic manuevers than can be used to either give the illusion of safety when that has not been proven by the data and/or likewise give the appearance of efficacy or equivalence with a comparison drug when the data does not really allow for that conclusion. Some of these methods are intuitively obvious-once they have been pointed out-others are more subtle and require a degree of sophistication not typically possessed by the average physician and at a level not likely to be taught to most medical students.

Here are some of the simple ones :

use a dose of the comparison drug that is too low(or too high) depending it you are trying to show equivalence or superiority or comparing side effects

in a comparison trials -keep the power low by have a relatively small trial

measure multiple endpoints but report only those that are favorable

use sub-group analysis to find a set of patients in whom there appears to be efficacy

arrange multi-center trials reporting those results from centers in which the right answer is found

cherry pick the subjects to minimize likelihood of side effects

report composite end point results when demonstration of efficacy is the aim and report individual safety outcomes when safety demonstration is the aim