Pathological Gambling Among Cocaine-Dependent Outpatients

The subjects were 313 research volunteers who were each recruited to participate in one of three outpatient clinical trials of treatment for cocaine dependence conducted in the outpatient research clinic of the National Institute on Drug Abuse Intramural Research Program. Two of the trials were for cocaine dependence (N=113), and one was for cocaine plus opiate dependence (N=200). Of the trials for cocaine dependence only, one was an open-label evaluation of bupropion plus bromocriptine (N=34) that lasted 8 weeks (25), and the other was a double-blind evaluation of carbamazepine (N=79) that lasted 8 weeks (26). The trial involving subjects with cocaine plus opiate dependence was a double-blind evaluation of buprenorphine (N=200) that lasted 13 weeks (27). All of the subjects received weekly individual drug abuse counseling in addition to their study medication. They had to meet the DSM-III-R criteria for current cocaine (and opiate, if appropriate) dependence and to have no current dependence on another psychoactive substance (except nicotine). The exclusion criteria included 1) current severe psychiatric symptoms, based on a clinical interview by a master’s-level drug abuse counselor or a general distress score higher than 70 on the SCL-90-R (28); 2) current need for psychiatric treatment (either psychotherapy or psychotropic medication), judged by the study psychiatrist; 3) active medical disorder or current need for medical treatment; and 4) for women, nursing, pregnancy, or having child-bearing potential and not using a medically accepted birth control method. Other than current substance dependence, no specific DSM-III-R diagnosis automatically excluded an otherwise eligible subject. The three studies were approved by an institutional review board. After complete description of the study to the subjects, written informed consent was obtained.

After entry into a study, each subject was assessed with the National Institute of Mental Health Diagnostic Interview Schedule (DIS) (29) to generate a detailed psychiatric history and psychiatric diagnoses (DSM-III-R criteria) and with the Addiction Severity Index (30) to elicit sociodemographic, socioeconomic, and drug use information. Pathological gambling was diagnosed according to DSM-III-R criteria applied to information obtained from the DIS.

The subjects attended the outpatient clinic three times each week for 8 weeks (daily for 12 weeks in the buprenorphine study). At each visit urine samples for drug testing were collected under direct staff observation, and they were tested by using a screening radioimmunoassay. The two treatment outcome measures were l) percentage of urine samples positive for cocaine metabolite (benzoylecgonine) or positive for opiates and 2) length of stay in treatment (in days). Length of stay was used as a secondary outcome measure because it has been shown to correlate with treatment outcome among cocaine-dependent outpatients (for instance, see reference 31). Subjects who consented to treatment but did not appear for the first visit were assigned a length of stay of 0. Because of varying lengths of scheduled treatment in the clinical trials, length of stay was analyzed as a percentage of the possible maximum length of stay.

Univariate comparisons of characteristics at treatment admission for the subjects with and without pathological gambling were done by t tests for continuous variables and Fisher’s exact or chi-square tests for categorical variables. To further evaluate possible differences between subjects with and without pathological gambling, stepwise logistic analysis was performed to determine whether any combination of subject characteristics (antisocial personality disorder and sociodemographic and socioeconomic variables) successfully discriminated between these two groups. The analysis was conducted with the SAS LOGISTIC procedure (Cary, N.C., SAS Institute) by using the forward selection method (p=0.05 to enter) and logit link function.

Comparisons of outcome variables for the subjects with and without pathological gambling were made with the nonparametric Kruskal-Wallis test (chi-square approximation with continuity correction) because the outcome variables had non-Gaussian distributions. To further evaluate the possible influence of pathological gambling on outcome, we performed stepwise multiple linear regression analyses, including lifetime history of pathological gambling (along with antisocial personality disorder and sociodemographic and socioeconomic variables) as a subject characteristic potentially predictive of outcome. The analyses were conducted with the SAS REG procedure by using the forward selection method (p=0.10 to enter and stay). The outcome (dependent) variables were natural log transformed because of their non-Gaussian distributions.

All statistical tests used a two-tailed significance level of 0.05.

The overall occurrence rate of lifetime pathological gambling in these 313 cocaine-dependent outpatients was 8.0% (N=25): 6.2% (N=7) among those with cocaine dependence only and 9.0% (N=18) among those with both cocaine and opiate dependence. One-half (48.0%) of the pathological gamblers reported current (within past month) gambling behavior (N=12), while 40.0% (N=10) reported no such behavior in more than a year.

The patients with pathological gambling differed in sociodemographic and socioeconomic characteristics from those without pathological gambling only in that they were less likely to be currently employed, were more likely to have recently engaged in illegal activity for profit, and had a longer duration of lifetime incarceration (Table 1 and Table 2). These differences derived primarily from the patients with dual cocaine and opiate dependence. These pathological gamblers also differed from those with only cocaine dependence in being more likely to be male (Table 1) and to spend their free time with friends (Table 2). The stepwise logistic analysis yielded results similar to those from the univariate comparisons. Only history of antisocial personality disorder and lifetime duration of incarceration contributed significantly to the correct classification of subjects by history of pathological gambling. These two variables correctly classified 68.4% of the subjects, incorrectly classified 15.3%, and left 16.4% indeterminate.

The patients with pathological gambling did not differ significantly from the other subjects in age at onset of cocaine or opiate dependence or duration of dependence (Table 3). However, the pathological gamblers with opiate dependence did show a nonsignificantly longer duration of cocaine dependence than those without opiate dependence (Table 3). The onset of pathological gambling preceded the onset of cocaine dependence in 18 (72.0%) of the subjects, came afterwards in six (24.0%), and was within the same year for one (4.0%). Age, sex, marital status, and employment status did not differ significantly by order of onsets (data not shown). Among the 18 opiate-dependent pathological gamblers, the onset of pathological gambling preceded the onset of opiate dependency in eight (44.4%), came afterward in seven (38.9%), and was within the same year for three (16.7%). For only one patient did the onset of gambling come between the onsets of cocaine and opiate dependence.

The rates of most other DSM-III-R axis I psychiatric disorders were not significantly different in the patients with and without pathological gambling. The respective rates of depression were 8.0% (N=2) and 4.9% (N=14) (χ²=0.45, df=1, p=0.38), the rates of posttraumatic stress disorder were 12.0% (N=3) and 4.9% (N=14) (χ²=2.28, df=1, p=0.15), and the rates of phobias were 24.0% (N=6) and 22.9% (N=66) (χ²=0.02, df=1, p=0.89). However, tobacco dependence was more common among the pathological gamblers: 84.0% (N=21) versus 61.1% (N=176) (χ²=5.17, df=1, p=0.02). Also, there was a significantly higher occurrence of antisocial personality disorder among the pathological gamblers (56.0%, N=14) than among the other subjects (19.8%, N=57) (χ²=16.87, df=1, p<0.0001).

The patients with and without pathological gambling did not differ significantly in either mean length of stay (60.4% versus 59.8% of possible maximum length of stay) (χ²=0.00, df=1, p=0.94) or mean proportion of urine samples positive for cocaine (44.0% versus 36.8%) (χ²=0.91, df=1, p=0.59). This was also the case when we considered only the 12 outpatients who had current (within the past month) pathological gambling: the mean lengths of stay for the patients with and without pathological gambling were 64.1% and 59.6% (χ²=0.08, df=1, p=0.78), and the mean rates of cocaine-positive urine tests were 42.6% and 37.2% (χ²=0.18, df=1, p=0.67). Among the 200 opiate-dependent outpatients, those with and without pathological gambling did not differ significantly in mean proportion of urine samples positive for opiates (45.5% versus 31.8%) (χ²=1.14, df=1, p=0.27). As in the univariate comparisons, lifetime history of pathological gambling did not contribute significantly to the prediction of either outcome variable. In the final stepwise multiple linear regression models, its F-to-enter was 1.86 (p=0.17) for predicting length of stay and 0.03 (p=0.86) for predicting percentage of cocaine-positive urine samples.