Problems and pitfalls in treatment trials of multiple sclerosis

In MS only the results of a properly randomized, double-blind, placebo-controlled trial can provide adequate evidence of efficacy. When therapy cannot be blinded one can conclude only that a therapeutic regimen which includes the test agent or procedure differs from one without it, but not that the difference is necessarily due to the agent. Measure of treatment effect in MS requires a standardized schema of assessing neurologic dysfunction due to the disease as the criterion for change (better, same, worse). Quantitation of impairment as measured by such scales can be compared between regimens only by means of nonparametric statistical tests, since no scale in use is a true numerical (equal-interval, additive, multiplicative) scale. The specific scale employed is not irrelevant; in the ACTH study of acute bouts the agent was concluded only to 'hasten improvement' with one scale, but to be an effective treatment with another.