Alternatives to randomized clinical trials.

Data bases describing the natural history of patients with multiple sclerosis or the clinical course of patients treated with placebos might serve as "historical controls" in future clinical therapeutic trials. The results of clinical trials with such controls can be misleading. There is a strong tendency for the new treatment to appear efficacious when historical controls are the comparison group. Therefore, claims of efficacy deduced from trials using such controls should be closely questioned. Thus, such comparison groups probably would be useful for preliminary and early phase II (pilot) trials rather than in more definitive phase III (full) trials.