Primary vs Secondary Endpoints--Are we doing more harm than good using obsolete classifications?
*Qi Jiang, Amgen 


The level of statistical rigor surrounding pharmaceutical industry-sponsored clinical trials has improved considerably over the years. For example, in past years it was common for protocols to simply list a number of endpoints without designating them as primary or secondary and without specifying how multiplicity would be controlled. Even when a single primary endpoint was specified, there was typically no multiplicity adjustment specified for the multiple secondary endpoints. The current situation is much more rigorous, and protocols are typically designed with strong control of the type 1 error rate for all primary and secondary endpoints. However, the variety of methods available to achieve strong control has increased dramatically, and some of these methods do not easily lend themselves to designating specific endpoints as “primary” or “secondary.” In some cases these designations,