We conducted a secondary analysis of data from a randomized, double-blind clinical trial testing the efficacy of sublingual buprenorphine (2 mg, 8 mg, or 16 mg daily, or 16 mg every other day) for the treatment of comorbid cocaine and opioid dependence (Montoya et al., 2004
). The study was approved by the Johns Hopkins Bayview Institutional Review Board, and was conducted in a sample of 200 opioid- and cocaine-dependent subjects (DSM-IIIR criteria) in the outpatient clinic of the National Institute on Drug Abuse Intramural Research Program (NIDA IRP; Baltimore, MD).
Subject inclusion criteria were age 21-50 years, comorbid current cocaine and opioid dependence (by DSM-IIIR criteria), self-reported use of cocaine and opioids within the past 14 days, use of at least $50 per day of heroin and $100 per week of cocaine at some time over the past month, two urine samples positive for opioids and for cocaine during the screening process, and not currently in drug abuse treatment elsewhere. Excluded were individuals unable to understand or fill out questionnaires, with a current unstable medical or psychiatric disorder, and pregnant or nursing women. HIV-infected individuals with CD4 T cell count < 200/mL were also excluded due to the risk that their impaired immune status might interfere with study participation.
Of the 200 subjects enrolled in the trial, 179 were considered evaluable, having completed the 4-day buprenorphine induction period and having achieved their target buprenorphine dose. For this report, we analyzed data from the 90 patients who completed the scheduled 70-day maintenance buprenorphine treatment period. This avoided any confounding of treatment outcome by differences in study retention.
Applicants received a thorough medical and behavioral/psychological evaluation. After qualification and consent, subjects had their first clinic visit within one week. Subjects were scheduled to participate in a 91-day treatment program that required daily clinic visits. At each visit they ingested a medication dose, had physiologic measures taken, and provided self-report data on outcome measures. At three visits each week (usually Monday, Wednesday, Friday), they gave a urine sample (for drug assay) under staff observation. Subjects were discharged for missing three consecutive medication doses or six psychotherapy sessions. Cancelled or rescheduled sessions did not count as missed sessions. Urine samples were assayed semi-quantitatively using the Abuscreen OnLine DAT immunoassay (Roche Diagnostics Corporation, Indianapolis, IN) for morphine and benzoylecgonine (BZE), both with a lower limit of quantification of 100 ng/mL.
Sublingual buprenorphine doses escalated to the targeted dose by day 5 (dose escalation phase) then remained at the target until day 70 (maintenance phase). Over the last 20 days, doses were decreased to 0 (withdrawal phase). Matching buprenorphine placebo was given on days of 0-mg buprenorphine dosing.
The therapy manual had two parts. Part I provided an overview of the course of treatment and discussed general issues of conducting treatment at each session. Part II provided a more detailed description of the 21 therapeutic techniques that could be used during the treatment based on the specific needs of the patient. The therapy manual offered a wide range of techniques and options for use by counselors. These included the identification of drug craving cues, recognition of interpersonal relationships, reframing and improvement of the decision-making process, overcoming helplessness, managing symptoms of depression and anxiety, increasing ability to relax without drugs, and increasing structure in the patient's life.
The therapy itself had four phases: (1) review of personal history, formulation of problems and goals, and development of a therapeutic alliance; (2) development of strategies to achieve treatment goals and control drug use; (3) strengthening of strategies and skills that prevent drug use, learning to use available support resources; and (4) resolution of separation and termination issues (Covi & Lipman, 1987
; Covi, Hess, Kreiter, & Haertzen, 1995
The urine toxicology test results were available to counselors and were part of the evaluation of patients' treatment progress. In addition, these results were discussed with members of the treatment team in the weekly clinical case conferences in order to formulate treatment goals for the following week. There were no negative contingencies for having drug-positive urine tests and no positive contingencies (payment or vouchers to exchange for goods or services) for attending psychotherapy sessions or having drug-free urine tests.
Individuals with a master's degree in a psychotherapy-related discipline provided the psychotherapy. One of the authors (LC) developed the therapy manual (Covi et al., 2002
), trained the therapists and provided therapy supervision throughout the study. Fidelity to the therapy manual and study protocol, as well as patients' treatment progress, were reviewed at weekly clinical case conferences attended by all therapists, the therapist supervisor, the study nurse, and at least one of the investigators.
2.5. Data analysis
Because psychotherapy attendance was mandatory only during weeks 2-10 of the study (the buprenorphine maintenance phase) and because we were interested in drug use outcomes while subjects were being actively maintained on buprenorphine, rather than when medication was being withdrawn, the psychotherapy attendance and urine toxicology data used for this report were limited to this 9-week period. Psychotherapy attendance data were analyzed as percent visits kept; i.e., the number of psychotherapy visits attended, whether as scheduled or rescheduled, divided by the total number of scheduled or rescheduled visits.
Thrice-weekly quantitative urine benzoylecgonine and morphine levels were used as outcome measures. Quantitative urinalysis (drug metabolite concentration) was used rather than qualitative urinalysis (metabolite present/absent) because drug metabolite levels have been shown to have good discriminative validity, correlate well with self-reported drug use (Delucchi, Jones, & Batki, 1997
), and confer additional statistical power for detecting a treatment effect (Batki, Manfredi, Jacob, & Jones, 1993
). Urine test results were log-transformed to normalize their right-skewed distributions. Histograms of the log-transformed concentrations looked approximately Gaussian and did not reveal any potential outliers.
Sociodemographic and psychiatric correlates of psychotherapy attendance were identified using t-tests and Spearman correlation coefficients. To identify those variables that could potentially confound the relationship between psychotherapy attendance and urine metabolite levels, any subject characteristic showing a statistically significant (p < 0.05) association with overall percent psychotherapy visits kept (weeks 2-10) was then tested for a significant association with urine metabolite levels by including each separately as an independent variable in repeated-measures linear regression models having log(BZE) and log(morphine) for weeks 2 through 10 as outcome measures.
The relationship between psychotherapy attendance and urine metabolite levels was first evaluated using scatterplots of the cumulative percent of visits attended through week 9 versus urine metabolite levels during week 10 ( and ). Relationships observed in the graphs were further evaluated using repeated-measures linear regression. The outcome measures were the log-transformed concentrations of urine BZE and morphine collected thrice weekly during weeks 3 to 10, with psychotherapy attendance as the time-varying covariate (percent visits attended through the previous week). For example, percent visits attended through week 2 was the covariate for urine drug metabolite levels during week 3, percent visits attended through week 3 was the covariate for urine drug metabolite levels during week 4, and so on through week 9 for psychotherapy and week 10 for urine drug metabolite levels. For each outcome measure, an initial regression model was fit that included psychotherapy attendance, treatment group, study week, the interaction terms (psychotherapy attendance × treatment group, psychotherapy attendance × study week, study week × treatment group, and psychotherapy attendance × study week × treatment group), and any potential confounding variables. Statistical significance of model terms was evaluated using F tests; whether regression coefficients differed from zero was determined using t tests. If the three-way interaction term was not significant (p > 0.05), it was dropped and the model re-fit. If two-way interaction terms in the re-fit model were not significant, they were dropped and the model again re-fit. The final model retained only significant interaction terms and all main effects. All analyses were conducted using SAS v. 8.2 (SAS Institute, Cary, NC).
Scatterplot of relationship between psychotherapy attendance and urine benzoylecgonine (BZE) levels for 90 cocaine- and opioid-dependent outpatients treated with buprenorphine.
Scatterplot of relationship between psychotherapy attendance and urine morphine levels for 90 cocaine- and opioid-dependent outpatients treated with buprenorphine.