Accrual of patients into a multihospital cancer clinical trial and its implications on planning future studies

Abstract

Little information exists on the population of cancer patients from which individual patients are selected for admission to a clinical trial. In fact, most reports of clinical trials of cancer chemotherapeutic agents begin by describing samples of treated patients but neglect to collect data and describe the population from which the samples were taken. In a multi-institutional VA [Veteran''s Administration] Cooperative Study [USA] in which 2 different cancer treatments were compared, an attempt was made to screen all lung, colorectal, prostate, and head and neck patients seen at participating hospitals prior to randomization to a therapeutic regimen. Of a total of 2687 patients screened, 437 (16.3%) were randomized and 2250 (83.7%) were excluded for 2981 reasons. Protocol reasons were the basis for 68.6% of all exclusions, 21.3% were physician refusals and 10.1% were patient refusals. The number of patients randomized did not correlate well with the number of patients screened across participating centers. Patients admitted to the study tended to be younger and in better health than excluded patients. Overestimates of randomization rates projected initially from published information point to the need for improved screening data in the planning of future studies. Factors such as screening methods, physician acceptance of the experimental approach, number of competing protocols within each center, and cooperation among medical center departments and personnel all are important ingredients in any screening effort.