3.1. Baseline demographic and drug use
On the whole, the two medication groups, acamprosate and placebo, were very similar in demographics and baseline drug use characteristics (see ). The average age of the subjects was about 45. Most were African American men and most smoked crack cocaine. On average, subjects had used cocaine about 16-18 days in the month prior to treatment (range 2-30 days) and spent around $700 on drugs in the month prior to treatment. There were no significant differences between the two groups in any baseline demographic or drug use variable.
Subject characteristics, expressed as percents or means (standard deviation).
3.2. Treatment retention
Sixty percent of the subjects completed the trial, 18/26 (69%) in the placebo group and 18/34 (53%) in the acamprosate group. Survival analysis showed no significant differences between groups (Log rank test = 2.0, p =.158). Treatment retention, measured by the number of visits attended favored the placebo group. On average, acamprosate-treated subjects attended 27 visits and placebo-treated subjects attended 34 visits (Man-Whitney U = 338, p = .12).
There were a number of missed visits resulting in missing data. The UBT data were most affected. 363/960 of the urine samples were missing. For the Urine Test results, there was no significant difference between the two groups during the study period of the trial, neither time by medication group interaction effects were significantly different from zero. But the subjects were more likely to drop out (F=16.04, P<.001) significantly over time. The Acamprosate -treated subjects were more likely to skip the later visits, but there was no significant difference compared with placebo group. The time to last visit was not significantly different between groups (F=2.46, p=0.11).
3.3. Urine Benzoylecgonine Test Results
Overall, out of 17 UBT collected while the patients were receiving study medication, 23% of the UBT submitted by the placebo treated subjects were benzoylecgonine negative vs. 22% for the acamprosate treated subjects. There was no significant difference between the two groups (GEE χ2=0.61, df=1, p=0.44) during the study period of the trial. Neither time effect nor time by medication group interaction effects was significantly different from zero. Analyses using mixed effects models (F(1,589)=0.58, p=0.53) and pattern mixture models (GEE χ2=0.20, df=1, p=0.53) gave very similar results for the test of medication effect, suggesting that missing data has very little influence on the test for medication effect.
3.4. Results from the Addiction Severity Index
Days of alcohol use, days of cocaine use, dollars spent on alcohol, dollars spent on drugs, the ASI composite alcohol scores, drug scores, legal scores, family/social scores, and psychiatric scores tended to decline in both groups in the trial, but we found no medication effects. Days of cocaine use (F=13.02, P=.003), dollars spent on drugs (F=8.04, P=0.05), ASI composite drug scores (F=12.69, P=.004), and ASI composite Psychiatric scores (F=7.14, P=.0075) declined significantly during the trial.
3.5. Results from BSCS
The BSCS composite measure combines three cocaine craving domains: intensity, duration and frequency. Cocaine craving showed a significant decline over the trial in both groups (F=19.89, P<.001). And there was significant interaction between baseline scores and time, subjects with higher craving at baseline showed a greater reduction in craving during the trial (F=9.51, P=.002). But no significant difference was found between the two groups for BSCS composite scores.
3.6 Results for the Clinical Global Impression Scale
Improvement was rated by the nurse practitioner weekly using the CGI. Improvement was rated on a scale from very much improved (0) to much worse (6). The improvement scores from CGI were evaluated as continuous variables and as dichotomized variables (very much improved and much improved vs. all other scores). There was no group difference for these two different analyses. When treated as continuous variables, CGI improvement scores (F=13.41, P<.001) declined significantly in both groups over time, showing an overall improvement in patients’ clinical state. However, there was no significant time by treatment interaction. If we dichotomize improvement scores there is a significant trend toward improved / very much improved over time (F=7.60, P=.006), but no significant time by treatment group interaction.
3.7. Results from CSSA
The CSSA composite measures overall cocaine withdrawal symptoms. CSSA composite score showed a significant decline over the trial in both groups (F=21.45, P<.001). There was no significant difference found between the two groups for CSSA composite scores.
3.8. Medication adherence
Medication adherence was measured by pill count. The percentage of pills taken was calculated by subtracting the number of pills returned each week from the number of pills dispensed. Both groups showed good adherence. On average, patients in the placebo group took their pills as prescribed on 90% of trial days, and patients in the acamprosate group took their pills on 86% of trial days (t = .898, df =58, p = .373).
3.9. Safety analyses
Adverse events were assessed at each visit. Acamprosate was well tolerated. Adverse events were mainly mild and evenly distributed between the acamprosate and placebo groups. The most commonly reported adverse events included aches and pains, upper respiratory tract infections, headache (n=6), diarrhea, insomnia, flatus, nausea, and skin infections. There were no significant differences noted in the occurrence of any adverse event between the two medication groups.