Subject numbers and placebo outcome variability in clinical trials of new CNS medications

Few investigations hove compared placebo group outcomes across therapeutic areas. We examined placebo group outcomes from 171 clinical trials reported in the Food and Drug Administration (FDA) Summary Basis of Approvals describing 33 medications for six neurologic or neuropsychiatric conditions: migraine headache, advanced Parkinson's disease, depression, postoperative (nondental) pain, postdental surgery pain, and schizophrenia. We used funnel plots to create a visual estimate of expected placebo outcome, to characterize the variability around the expected value, and to establish an empirically derived number of subjects to minimize the risk that sampling variability will lead to a placebo outcome that prevents a positive trial result for an effective drug. The results confirmed our hypothesis that in trials from the FDA's preapproval database placebo outcomes are distributed around an expected value. Variability diminished with increasing placebo group enrollment, allowing empirical determination of a minimum placebo group size associated with low risk that sampling variability will cause a placebo outcome that results in a negative trial.