Participants

The study participants were adults, aged 18–65 years, who visited the gambling clinic in Kangbuk Samsung Hospital between May 2002 and December 2011. From February 2012 to March 2013, we retrospectively reviewed the medical charts of these patients with GD.

Using the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV) for Axis I disorders (SCID-I), a tool for diagnosing GD and identifying psychiatric comorbidities [14], a psychiatrist (Y.C. Shin, one of the co-first authors) diagnosed all of the patients with GD. We enrolled subjects who scored at least 5 on the Korean version of the South Oaks Gambling Screen (K-SOGS) [15].

There were 23, of the initial 824, cases that were excluded from the study because the family members sought counseling in lieu of the person who gambled. According to the exclusion criteria, additionally 11 participants were excluded due to lifetime diagnosis of a psychotic disorder, mental retardation, and substance use disorder, except for alcohol and nicotine dependence. None of the participants were provided with a therapist’s recommendation to discontinue treatment due to improvement, given that “improvement” is regarded as temporary in GD, and all of the patients were advised to continue treatment throughout their lifetime. However, an additional 32 patients were excluded from the current study because they terminated treatment in this outpatient GD clinic upon agreement with a therapist. Of these patients, 20 were referred to gamblers anonymous (GA) or other outpatient clinics due to the far travel distance to the study clinic. The remaining 12 of the 32 patients who terminated treatment at the clinic were referred to inpatient clinics, or other institutes, because of severe recurrence including the increase of debt or gambling severity. Finally, a total of 758 participants were included in this study. The study protocol was approved by the Ethics Committee of the Kangbuk Samsung Hospital. All of the study subjects provided informed consent prior to participation.

Intervention

The GD clinic provided individual psychotherapy (PT), pharmacotherapy, and group CBT. The patients simultaneously received distinct therapeutic alternatives, such as GA, if they chose to participate.

Individual PT was based on CBT and motivational enhancement therapy, and lasted for 15–20 min in the outpatient clinic. Regular follow-up was recommended, regardless of whether other types of treatment were utilized [16]. The duration of an individual PT session cannot be longer because the Korean National Health Insurance System pays one CBT fee irrespective of the experts’ work experience (approximately 25 USD for 30 min of CBT with a psychiatrist) [17].

The patients received medication based on their symptoms and clinical presentation. The patients received anticraving drugs for cravings to gamble, and antidepressants to cope with anxiety or depressed mood. To compare the effect of each pharmacotherapy, the psychotropic drugs were classified as anticraving agents (naltrexone, acamprosate) and antidepressants (selective serotonin reuptake inhibitors—escitalopram, paroxetine, sertraline; serotonin–norepinephrine reuptake inhibitors—venlafaxine, milnacipran; norepinephrine–dopamine reuptake inhibitors—bupropion). The ‘no-pharmacotherapy’ group was not administered any major psychotropic prescription drug during the treatment period. Augmented benzodiazepines (as needed) were permitted and not considered in the data analysis. The patients whose main therapy was antipsychotics or mood stabilizers were excluded from the analyses because we assumed that they had an additional major diagnosis.

The group CBT for GD was composed of weekly 2-h sessions, for 8 weeks [18, 19]. All of the group CBT sessions were conducted by a single psychiatrist (Y.C. Shin) three or four times per year from 2004 to 2012, except for the two groups in 2006 (by S.W. Choi). The group CBT included conventional therapeutic elements from CBT, such as psychoeducation, cognitive restructuring, and decisional balance. Moreover, therapeutic skills from motivational enhancement and logotherapy, including problem solving in other areas of life and the meaning of gambling in life, were combined in the treatment. Additional details on the CBT are shown in Fig. 1.

Measurements

Treatment maintenance duration was evaluated by chart review. If a patient did not visit the GD clinic (unless according to an agreement with a therapist), the patient was regarded as a dropout. For the participants who had not dropped out, the treatment maintenance duration was calculated from the first visit to the day of the last visit before the chart review.

The characteristics of the participants’ gambling behavior were also evaluated at the first visit. The GD patients had a tendency to distort or minimize their debt, due to their gambling behavior [20]. If there was a lack of consensus between a patient and an accompanied family member’s report, then we chose the higher value. The classification of main gambling type was based on the following question: “What type of gambling has most disturbed your life and resulted in the greatest loss of money?” We classified gambling according to structural game characteristics [21]. “Strategic gambling” was defined as gambling activities for which the outcome is believed to be the result of the players’ skill or analysis, such as Poker or Go-stop. “Nonstrategic gambling” was defined as the gambling activities for which the outcome is generally believed to be a result of random probability, such as slot machines, roulette, or lotteries [16].

To evaluate the severity of gambling, the Korean version of the Gambling Symptom Assessment Scale (GSAS) was used [22]. Each item of the GSAS is scored on a 5-point scale from 0 (no symptoms) to 4 (extreme symptoms). The total score ranges from 0 to 48. The Beck Depression Inventory (BDI) and Beck Anxiety Inventory (BAI) were administered to assess comorbid depressive and anxiety symptoms, respectively [23].

Statistical analysis

To compare the distribution and frequency of the categorical variables, Pearson’s Chi-square test was used. The differences of scale scores and other continuous variables according to intervention were analyzed with independent t tests and one-way analysis of variance (ANOVA). Analysis of covariance (ANCOVA) was conducted with BDI, GSAS, and age as the covariates to analyze the total treatment maintenance duration among the three groups and examine the main and interactive effects of other demographic variables. To analyze the independent factors that were associated with the follow-up duration, a generalized linear model was used to adjust each confounding factor. All the analyses were performed using PASW Statistics 17.0 software (SPSS Inc., Chicago, IL, USA), with the cut-off for statistical significance set at p < 0.05.

Demographic and clinical characteristics of the participants

Of the 758 subjects, 717 participants (94.59 %) were male. The participants had a mean age of 39.60 ± 9.94 years. The mean age at the onset of GD was 29.84 ± 8.39 years. Of the subjects, 391 (51.58 %) had graduated college or university. More than half of the participants (61.6 %) had a full-time job for more than half of their adult life, and 157 participants (20.8 %) never had a job during their lifetime. Of the patients, 469 (61.87 %) were married.

All of the 758 patients had debt due to GD. Gambling debt at the first visit was 119,897 ± 360,259 US dollars (exchanged from Korean Won). The mean GSAS and K-SOGS scores were 15.56 and 16.27, respectively. The greatest amount of money that each participant had ever won at one time by gambling, a so-called ‘big win’ [24], was 166,471 ± 962,448 US dollars, on average. A total of 150 subjects (19.7 %) gambled online. Of the gambling types, strategic gambling was most preferred (51.6 %), followed by nonstrategic gambling (17.4 %) and mixed (31.0 %). A considerable portion of the subjects (n = 555, 72.9 %) had a psychiatric comorbidity. Depression (9.7 %) was the most prevalent comorbidity, followed by alcohol-related disorders (8.5 %) and bipolar disorder (2.1 %).

Treatment maintenance duration and pharmacotherapy

The mean treatment maintenance duration was 8.32 ± 16.21 months. Forty-eight patients (6.3 %) did not drop out, with a mean treatment length of 36.43 ± 31.28 months. Of the subjects, 129 (17.01 %) continued to visit the GD clinic for longer than a year, with a follow-up duration of 26.47 ± 24.48 months.

Combined effect of pharmacotherapy and group CBT

We also compared the treatment maintenance duration according to the interventions. The mean number of months of follow-up differed across the interventions (F = 35.79, p < 0.001). The patients who received individual PT (n = 409) displayed a significantly smaller number of follow-up months than did all the other intervention groups (3.39 ± 7.51 months), even according to the post hoc analysis (p = 0.03 for the comparison with the anticraving agent group; p < 0.001 for the comparison with the other groups). However, those who received combined group CBT and antidepressant treatment for GD demonstrated the longest follow-up period (38.86 ± 41.77 months) compared with all the other intervention groups (p < 0.001) (Fig. 2).

Additionally, we classified the patients into the following four groups regardless of the type of medication they were prescribed: (1) individual PT (n = 409), (2) group CBT (n = 68), (3) pharmacotherapy (n = 228), and (4) combined group CBT and pharmacotherapy (n = 53). There was a significant difference in follow-up duration among the four groups (F = 48.41, p < 0.001). This result was preserved after adjusting for confounding factors such as age, BDI, and GSAS (F = 27.41, p < 0.001). However, combined group CBT and pharmacotherapy (24.70 ± 28.37 months) was superior to pharmacotherapy only (10.32 ± 16.95 months, F = 26.16, p < 0.001), but not superior to group CBT only in terms of the number of follow-up months (18.53 ± 23.39 months, F = 0.514, p = 0.416).