Internet-based Versus Synchronous Cognitive Behavioural Therapy for Patients With Gambling Disorder

PROJECT GROUP: Anna Westh Stenbro (Psychologist, MSc.); Lisbeth Frostholm (Professor, Psychologist, MSc., Ph.D.); David Hodgins (Professor, Psychologist, MSc., Ph.D.)

The trial will be conducted by the research group at the Research Clinic for Gambling Disorder, based at Aarhus University Hospital, Denmark. The clinic was established in 2007 and offers assessment and specialized treatment to patients with GD.

MEASUREMENTS:

Self-reported data from the participants are collected at seven time points: At baseline (T1), randomisation (T2), three time points mid-treatment (T3, T4, T5), end-of-treatment (T6) and at 3-month follow-up (T7, primary endpoint). Due to different treatment durations in the two arms, an additional measurement point (T7b) is added to control for the effect of time for participants in the SF group, at the time point when participants in the sCBT group complete the 3-month follow-up (week 30). At T6, all participants will be invited to participate in a 12-month follow-up study (T8).

During treatment, therapists register the length, and themes, of phone conversations and face-to-face sessions in SpilleFri and equally the length and overall themes of sessions in the sCBT arm. In both arms, therapists register involvement of concerned significant others in treatment sessions. In the sCBT arm, therapist register manual fidelity and covered intervention topics after each session. Deviations are described. At the end of the treatment course, the therapists evaluate participant engagement and treatment effect (CGI-I).

For a detailed list of the measures and measurement time points included, see the section "Outcome measures".

RANDOMIZATION:

Included patients will be randomized to SF or sCBT in a 1:1 ratio. Randomization takes place immediately following the inclusion conference. A member of the research group will perform the randomization using the randomization module in the online data collection program, REDCap. When the randomization module is activated, REDCap automatically and randomly assigns participants to one of the predefined study groups in randomization blocks randomly ranging between 8 and 12. All research group members and clinicians are blinded to the current randomization block size.

POWER CALCULATION:

Sample size estimation was done using simulation of mixed linear models in Stata version 17. The relative effects of SF and sCBT will be analysed using a pre-set non-inferiority margin, determining the margin within which SF can be less effective than sCBT without being declared clinically inferior to the standard treatment. Within this margin, any difference in effect will represent a clinically unimportant loss of effect.

Simulation parameters were informed by 1) previous clinical data from the RCGD on reduction in NODS for patients given sCBT and 2) reduction in NODS for patients in the SF feasibility study. Assuming baseline-EOT-FU trajectory in NODS-score of 7-1-1 in individual CBT condition and 7-2-2 in SF condition, power of .80 was reached for 100 patients (50 in each condition). To account for dropout and attrition, a sample size of 150 is ideal.

ANALYSIS:

The hypothesized non-inferiority of SF compared to individual CBT as measured by NODS will be analysed using 2-level mixed linear models with random intercept - i.e., repeated measures nested within patients. REML estimation will be employed. Model-predicted mean change in NODS from baseline to FU is estimated in both CBT and SF condition, and a 95% confidence interval for the difference in change, i.e., ΔCBT-ΔSF, is calculated. The null-hypothesis of inferiority is rejected if the upper bound of this confidence interval is greater than the inferiority bound defined to be -2 NODS-scale points, i.e., SF leads to a change in NODS not more than 2 points less than CBT. The inferiority bound was decided based on clinical experience from RCGD, judging how large a difference in NODS that is clinically visible.

Results will be analysed and reported according to the CONSORT statement. A statistician is continuously involved in the project.

• Change in Gambling Symptoms past week as Assessed by Gambling Symptom Assessment Scale (G-SAS) [Time frame: Baseline, during treatment (weeks 4, 7, 10), post (week 12/18), 3 month FU (week 24/30), 12 month FU (week 64/70)]
• Change in Self-reported Gambling Net Loss past week as assessed by a single item [Time frame: Baseline, during treatment (weeks 4, 7, 10), post (week 12/18), 3 month FU (week 24/30), 12 month FU (week 64/70)]
• Overall health improvement [Time frame: Post (week 12/18, clinician rating), 3 month FU (week 24/30, participant and clinician rating)]
• Change in Symptoms of anxiety, depression, and psychological distress as Assessed by th SCL-90 [Time frame: Baseline, post (week 12/18), 3 month FU (week 24/30), 12 month FU (week 64/70)]
• Change in Overall Quality of Life and General Health Satisfaction as Assessed by the WHOQoL [Time frame: Baseline, post (week 12/18), 3 month FU (week 24/30), 12 month FU (week 64/70)]
• Change in Quality of Current Romantic Relationship as Assessed by the DAS-4 [Time frame: Baseline, post (week 12/18), 3 month FU (week 24/30), 12 month FU (week 64/70)]
• Participant treatment expectancy and satisfaction as Assessed by the Experienced Treatment Credibility and Expectancy [Time frame: Baseline and post (week 12/18)]