Chatbot-assisted therapy for patients with methamphetamine use disorder: a preliminary randomized controlled trial

Abstract

Background Methamphetamine (MA) use disorder is associated with a large public health burden. Despite the therapeutic effects of psychosocial interventions based on current evidence, finding an approach to retain patients in treatment remains a real-world challenge. The rapid development of mobile health (mHealth) systems suggests the potential to provide real-time personalized care at any time and from any location, minimize barriers to treatment, maximize use, and promote the dissemination of accessible therapeutic tools in at-risk populations. Our study aimed to investigate the feasibility and effectiveness of chatbots for the treatment of MA use disorder. Method The inclusion criteria were (a) a diagnosis of MA use disorder as defined by the DSM-5, (b) age between 18 and 65 years, (c) no acute exacerbation of severe mental illness during the initial assessment, such as schizophrenia or bipolar I disorder, (d) willingness to participate in standard outpatient treatment for ≥ 6 months, and (e) an Android phone. Participants were randomly allocated to either a chatbot-assisted therapy via smartphone (CAT) group or a control group following simple randomization procedures (computerized random numbers) without blinding. All participants were followed up for 6 months. Treatment retention and monthly urine test results were analyzed as outcome measures. Participants' satisfaction with CAT was also assessed. Results In total, 50 and 49 participants were allocated to the CAT and control groups, respectively. There were no significant differences in retention time between the two treatment groups ( df = 1, p = 0.099). The CAT group had fewer MA-positive urine samples than the control group (19.5% vs. 29.6%, F = 9.116, p = 0.003). The proportion of MA-positive urine samples was positively correlated with the frequency of MA use ( r = 0.323, p = 0.001), severity of MA use disorder ( r = 0.364, p < 0.001), and polysubstance use ( r = 0.212, p = 0.035), and negatively correlated with readiness to change ( r = −0.330, p = 0.001). Totally 55 participants completed the study at the 6-month follow-up and 60% reported relative satisfaction. Conclusion Participants in this study had favorable acceptance and generally positive outcomes, which indicates that chatbot is feasible for treating people who use MA.