Neuroepidemiology. Part II: Assessment of therapeutic trials

The gold standard for therapeutic trials is the randomized, double-blind, placebo-controlled study design. Lack of blindness and placebo makes tenuous the attribution of results to the specific agent; lack of randomization between concurrent and comparable groups makes it impossible. The chi-square test of the null hypothesis of no difference between treatment groups is the best method of assessment, as no overall rating system used in neurology is a true numerical scale but at best a rank-order scale. If quantitation of results is desired, a nonparametric rank-order test is necessary. Results over time can be assessed with a logrank (life-table) test for single events as end point, and by chi-square or rank-order tests for multiple events among individual patients.