One problem with creating a uniform group of clinical trial subjects is finding reliable screening criteria to establish the cohort.

Cell Pathways Inc. thought that by using physician judgement and one-year patient histories, it had picked patients who would form 10-40 polyps per year for its Phase III trial of exisulind to treat adenomatous polyposis coli. But the company found that was an imperfect method, and ended up with randomization problems that prevented the trial from reaching statistical significance.