Retrospective analyses of pooled data from CREST I and CREST II trials for treatment of cocaine dependence

ArticleinAddiction 100 Suppl 1(s1):91-101 · April 2005with21 Reads
Impact Factor: 4.74 · DOI: 10.1111/j.1360-0443.2005.00986.x · Source: PubMed


    To analyze pooled data from the Cocaine Rapid Evaluation Screening Trial (CREST). Pooling data from these small pilot trials into four major drug classes permitted data exploration for treatment and covariate effects with increased sample size.
    Small pilot trials were conducted to screen fifteen medications as prospective treatments for cocaine dependence. Studies included a flexible 2-week to 4-week screening/baseline period followed by an 8-week randomized treatment condition. Participants were randomized equally to one of up to three active medications or placebo.
    Five Medications Development Research Units at the five academic centers of University of Cincinnati, New York University, University of Pennsylvania, University of California Los Angeles and Boston University.
    The pooled data set consisted of 357 total subjects. Standardized inclusion and exclusion criteria were employed in subject selection to enhance consistency of cocaine-dependent study participants across all sites (see reports on individual trials in this supplement for details). All participants provided at least two urine samples that were positive for cocaine metabolite during a two-week period prior to being randomized.
    All subjects in these trials, those randomized to placebo and active medications, received active treatment in the form of evidence-based cognitive behavioral therapy.
    Quantitative urine benzoylecgonine (BE), self-report of cocaine use, and total Brief Substance Craving Scale (BSCS) scores were compared between each class of medication and its matched-placebo group.
    Regression analysis of pooled data did not identify any statistically significant differences between treatment and matched-placebo for any of the four classes. Exploration of the effects of baseline covariates indicated that gender and African American status were associated significantly with outcome. Female gender was consistently associated with poorer outcomes for medication and placebo groups, while the direction of association between African American status and outcome differed by treatment groups. Retention was also examined: dropout rates may have been somewhat higher for placebo than treatment groups during the early active-treatment period. Classification trees were used to identify characteristics of subjects who were abstinent for at least two weeks during the eight-week trial; only 4.0% of females while 17.9% of males achieved this criterion.
    Results presented here may prove useful for planning future clinical trials for therapies targeting cocaine dependence.