Survival analysis in drug program evaluation. Part II. Partitioning treatment effects.

Survivorship analysis is applied to both a summative process (during treatment) and a summative outcome (after discharge) evaluation comparing data obtained from 297 admissions to multiple-site methadone maintenance programs in three California counties. Measures included rates of retention, incarceration, addiction, crime, dealing, and loss of employment. In order to take full advantage of the multiple outcome measures that were used, a method for within-subject simultaneous consideration of all outcome measures is presented. This method is an adaptation of the Friedman nonparametric analysis of variance and a Bonferroni post hoc comparison of the mean ranks.