1068 — Patient-Centered Pain Care Using Artificial Intelligence and Mobile Health Tools: A Randomized, Comparative Effectiveness Trial

Lead/Presenter: Nicolle Marinec ,  San Francisco VAMC
All Authors: Piette JD (Ann Arbor Department of Veterans Affairs Center for Clinical Management Research), Newman SC (School of Public Health, University of Michigan, Ann Arbor and Ann Arbor Department of Veterans Affairs Center for Clinical Management Research) Krein SL (Ann Arbor Department of Veterans Affairs Center for Clinical Management Research and Department of Internal Medicine, School of Medicine, University of Michigan, Ann Arbor) Marinec N (Ann Arbor Department of Veterans Affairs Center for Clinical Management Research) Chen JS (School of Public Health, University of Michigan, Ann Arbor and Ann Arbor Department of Veterans Affairs Center for Clinical Management Research) Williams DA (Chronic Pain and Fatigue Research Center, University of Michigan, Ann Arbor) Edmond SN (Pain Research, Informatics, Multimorbidities, and Education (PRIME) Center, VA Connecticut Healthcare System and Department of Psychiatry, Yale School of Medicine) Driscoll M (Pain Research, Informatics, Multimorbidities, and Education (PRIME) Center, VA Connecticut Healthcare System and Department of Psychiatry, Yale School of Medicine) LaChappelle KM (Pain Research, Informatics, Multimorbidities, and Education (PRIME) Center, VA Connecticut Healthcare System) Kerns RD (Pain Research, Informatics, Multimorbidities, and Education (PRIME) Center, VA Connecticut Healthcare System, Department of Psychiatry, Yale School of Medicine, Department of Psychology, Yale University, Department of Neurology, Yale School of Medicine) Maly M (School of Public Health, University of Michigan, Ann Arbor and Ann Arbor Department of Veterans Affairs Center for Clinical Management Research) Kim M (School of Public Health, University of Michigan, Ann Arbor and Ann Arbor Department of Veterans Affairs Center for Clinical Management Research) Farris KB (Department of Clinical Pharmacy, College of Pharmacy, University of Michigan, Ann Arbor) Higgins DM (VA Boston Healthcare System and Boston University School of Medicine) Buta E (Yale Center for Analytical Sciences, Department of Biostatistics, Yale School of Public Health) Heapy AA (Pain Research, Informatics, Multimorbidities, and Education (PRIME) Center, VA Connecticut Healthcare System and Department of Psychiatry, Yale School of Medicine)

Objectives:
Cognitive behavioral therapy for chronic pain (CBT-CP) is a safe and effective alternative to opioid analgesics. Because CBT-CP requires multiple sessions and trained therapists are scarce, many Veterans have limited access or fail to complete treatment. We evaluated whether a CBT-CP program that personalized Veterans’ treatment using reinforcement learning (a field of artificial intelligence [AI]) and interactive voice response (IVR) calls, i.e., AI-CBT-CP, was non-inferior to standard telephone CBT-CP while saving therapist time.

Methods:
This was a randomized non-inferiority, comparative effectiveness trial with Veteran participants recruited from two VA healthcare systems. 278 patients with chronic back pain were randomized between June 2017 and September 2019. 89.2% were male, 81.8% were White, and the average age was 63.9 years. All patients received up to 10 weeks of CBT-CP. For the AI-CBT-CP group, patient feedback via daily IVR calls was used by the AI engine to make weekly recommendations for: a 45-minute or 15-minute therapist-delivered telephone session, or an individualized IVR-delivered therapist message. Patients in the comparison group were offered ten 45-minute therapist-delivered telephone CBT-CP sessions. The primary outcome was the Roland Morris Disability Questionnaire (RMDQ; range 0-24), measured at 3- (primary endpoint) and 6-months post baseline. Secondary outcomes included pain intensity and pain interference. We used consensus guidelines to identify clinically meaningful improvements for responder analyses (e.g., a 30% improvement in RMDQ scores and pain intensity).

Results:
The 3-month mean RMDQ score difference between AI-CBT-CP and standard CBT-CP was -0.72 points (95% CI -2.06 to 0.62). RMDQ scores met the non-inferiority criterion at both 3- and 6-month endpoints (both non-inferiority p < .001). A greater proportion of AI-CBT-CP than comparison patients had clinically meaningful improvements at 6-months in RMDQ scores (37% versus 19%, p = .01, NNT = 6) and pain intensity scores (29% versus 17%, p = .03, NNT = 9). There were no other significant differences in secondary outcomes. AI-CBT-CP required less than half the therapist time as standard CBT-CP.

Implications:
AI-CBT-CP was non-inferior, and possibly achieved some better clinical outcomes than therapist-delivered telephone CBT-CP.

Impacts:
Multi-session programs that use AI to allocate clinician time may dramatically increase the number of Veterans who can receive care without increasing program costs.