Unguided Chatbot-Delivered Cognitive Behavioural Intervention for Problem Gamblers Through Messaging App: A Randomised Controlled Trial

Internet-delivered intervention may be an acceptable alternative for the more than 90% of problem gamblers who are reluctant to seek face-to-face support. Thus, we aimed to (1) develop a low-dropout unguided intervention named GAMBOT integrated with a messaging app; and (2) investigate its effect. The present study was a randomised, quadruple-blind, controlled trial. We set pre-to-post change in the Problem Gambling Severity Index (PGSI) as the primary outcome and pre-to-post change in the Gambling Symptom Assessment Scale (G-SAS) as a secondary outcome. Daily monitoring, personalised feedback, and private messages based on cognitive behavioural theory were offered to participants in the intervention group through a messaging app for 28 days (GAMBOT). Participants in the control group received biweekly messages only for assessments for 28 days (assessments only). A total of 197 problem gamblers were included in the primary analysis. We failed to demonstrate a significant between-group difference in the primary outcome (PGSI - 1.14, 95% CI - 2.75 to 0.47,p = 0.162) but in the secondary outcome (G-SAS - 3.14, 95% CI - 0.24 to - 6.04,p = 0.03). Only 6.7% of the participants dropped out during follow-up and 77% of the GAMBOT group participants (74/96) continued to participate in the intervention throughout the 28-day period. Integrating intervention into a chatbot feature on a frequently used messaging app shows promise in helping to overcome the high dropout rate of unguided internet-delivered interventions. More effective and sophisticated contents delivered by a chatbot should be sought to engage over 90% of problem gamblers who are reluctant to seek face-to-face support.