Comparing speeds of effectiveness of two treatments

A variety of methods have been proposed for analyzing and summarizing data from clinical trials of pharmaceutical agents. Methods directed at evaluating the speed of activity of an agent are an important component of such analyses. We review several parameters that are useful for comparing speed of effectiveness and demonstrate their utility, both with didactic examples and with real data. The "survival function" is well suited to evaluating speed of cure, particularly when there is a fraction of the population that does not respond to a particular treatment. Some conditional parameters, such as the survival function among those who are curable, are shown to be inappropriate for the comparison of speeds of cure because they summarize speed of activity in different, and therefore noncomparable, populations. Comparisons based on these conditional survival curves are inappropriate and can lead to the wrong interpretation even when one treatment completely dominates another with respect to speed and efficacy. Methods appropriate for determining an optimal sequence of treatments are briefly reviewed. Examples from the published literature, as well as data from a recent study of venlafaxine, are used to demonstrate both appropriate and flawed interpretations of clinical trial data.