(PS1-B29) Type of Retrospective Assessment Moderates the Effects of Cognitive-behavioral Treatment on Gambling Frequency: A Meta-analysis

Meta-analyses indicate that cognitive-behavioral treatment (CBT) significantly reduces gambling frequency and the amount of money spent on gambling relative to minimal and inactive treatment controls. Despite the promise of CBT for reducing these outcomes, the effects of treatment are highly heterogeneous.

One reason for the heterogeneity of treatment effects might be the use of different assessments of gambling behavior. For example, gambling frequency and the amount of money spent on gambling are often measured using questions that ask individuals to retrospectively report how many times they gambled and how much money they gambled in the past 30 days. The global summation of gambling behavior indicated by such retrospective assessments may introduce error that impacts estimates of treatment effects in contrast to more resource-intensive assessment measures like the timeline followback (TLFB) method.

The TLFB method promotes more detailed reporting of gambling behavior because participants are required to complete a day-by-day summary of gambling over a specified time interval. Additionally, the TLFB method encourages individuals to use memory aids (e.g., financial records) and key dates (e.g., birthdays, holidays, sporting events) to help to qualify and clarify the pattern of gambling behavior. Thus, the purpose of the current meta-analysis was to determine whether the TLFB method yields different estimates of the effect of CBT on gambling frequency and intensity compared to other retrospective assessments.

Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses, 18 studies representing 2,419 participants were identified for inclusion. Hedges’s g effect sizes were calculated to represent the effect of CBT relative to minimal and inactive controls on the number of days gambled (i.e., gambling frequency) and the amount of money spent on gambling (i.e., gambling intensity) at post-treatment. A mixed-effects subgroup analysis was conducted to examine the moderating effect of assessment method (TLFB versus other retrospective assessment) on the Hedges’s g effect sizes for days and amount of money gambled.

Results indicated that the effect of CBT relative to control on gambling frequency was significantly lower for studies using the TLFB method (g = -0.20) than studies using other retrospective assessments (g = -0.71). There was no significant difference in the effect of CBT on gambling intensity between studies using the TLFB (g = -0.22) and studies using other assessments (g = -0.38).

These results suggest that the type of retrospective assessment method accounts for some of the heterogeneity in the effect of CBT on gambling frequency. The TLFB method appears to yield a more precise and conservative estimate of CBT’s effect on this outcome, whereas the use of other more global retrospective assessments to assess gambling frequency may overestimate CBT’s effect.