D. J. McCann; NIH/NIDA, Bethesda, MD
BACKGROUND: Professional subjects, defined here as subjects who enroll in clinical trials only for financial gain, have an unknown impact on medication efficacy trials.
METHODS: Two types of professional subjects were considered: those “destined to succeed” (DS) and those “destined to fail” (DF). An example of a DS subject is someone who feigns depression to enroll in an antidepressant trial and then answers questions truthfully after randomization. An example of a DF subject is a smoker who enrolls in a smoking cessation trial and does not try to quit. After setting success rates in legitimate subjects at 10% for placebo and 25% for active treatment, modeling studies evaluated the impact of DS and DF subjects on apparent success rates, the number of subjects required for 90% power (α 0.05, two-sided chi squared test), and apparent effect size (odds ratio).
RESULTS: With no professional subjects, 133 subjects per group yielded 90% power, and the odds ratio was 3.00. With 10% DS subjects, apparent success rates rose to 19% and 32.5%, 219 subjects per group were required for 90% power, and the apparent odds ratio fell to 2.05. With 20% DS subjects, apparent success rates rose to 28% and 40%, 326 subjects per group were required for 90% power, and the apparent odds ratio fell to 1.71. In contrast, the impact of DF subjects was modest. At the extreme, with 20% DF subjects, apparent success rates were 8% and 20%, 174 subjects per group yielded 90% power, and the apparent effect size was 2.88.
CONCLUSION: While all professional subjects negatively impact efficacy trials, DS subjects are especially problematic. A small percentage of DS subjects in a study population can greatly reduce apparent effect size, a phenomenon that cannot be overcome by increased sample size.