Moderators of response to sertraline versus placebo among recently abstinent, cocaine dependent patients: A retrospective analysis of two clinical trials

Abstract

Background and objectives: Moderators of treatment response to serotonin reuptake inhibitor sertraline (SRT) for cocaine dependence were assessed in two randomized, double blind, placebo-controlled clinical trials.

Methods: Generalized estimating equation modeling was performed on data from cocaine-dependent volunteers randomized to receive SRT or placebo (N = 126) who completed >2-week drug-free residential portions of the 12-week trials, in which subsequent outpatient treatment (weeks 3-12) included weekly cognitive behavioral therapy and thrice-weekly supervised urine toxicology.

Primary outcome measure: Relapse (2 consecutive cocaine-positive or missing urines) following residential stay. Potential moderators included treatment, sex, age, race, depression measures, baseline cocaine urine result, and alcohol dependence diagnosis (ADDx).

Results: Odds ratios (OR) for relapse showed placebo-treated participants were significantly more likely to relapse than SRT participants. Regardless of treatment condition, participants more likely to relapse were male, and those with lower Hamilton depression ratings, or baseline cocaine-negative urines. Older subjects or those with current ADDx had higher relapse risk than those without ADDx; however, treating older or ADDx participants with SRT reduced cocaine relapse more than placebo.

Discussion and conclusions: Women or those with more severe cocaine use or depressive symptoms may have fewer cocaine relapses regardless of medication treatment. SRT at 200 mg reduced cocaine relapse more than placebo, especially in older participants or in those with comorbid ADDx.

Scientific significance: SRT may be efficacious to support relapse prevention among cocaine-dependent patients in the context of brief residential followed by outpatient treatment, especially in older participants or those with comorbid alcohol/cocaine dependence. (Am J Addict 2017;26:807-814).

Figure 1. Model 1 (Univariate) Results

Forest plot of the Odds Ratios (points) and 95% Confidence Intervals (bars) for all predictor variables in the univariate model for relapse. X-axis: scale for odds ratios with a dotted line at 1.0 for reference. Y-axis: predictor variables. [Note: “Age” factor appears to be plotted at/near “1.0”, while results indicated significant (p<0.005) influence of age; this is an effect of scaling of graphic elements in the figure.]

Figure 2. Model 2 (Multivariate) Results

Forest plot of the Odds Ratios (points) and 95% Confidence Intervals (bars) for all those predictor variables and interactions that were retained in the final chosen multivariate model for relapse. X-axis: scale for odds ratios with a dotted line at 1.0 for reference. Y-axis: predictor variables.