Chatbot Use by Patients Seeking Urogynecologic Consultation: A Randomized Trial

Urogynecology patients increasingly investigate their problems using technology-based resources, yet the effect of artificial intelligence (AI) chatbots on patient decision making remains unclear.

The objective of this study was to investigate the effect of the use of an AI chatbot platform on patient decisional conflict during an initial urogynecology visit.

In this institutional review board-approved, single-center randomized trial, patients presenting with prolapse, incontinence, or lower urinary tract symptoms were assigned to 1 of 3 groups: (1) ChatGPT-4o (OpenAI) use before the visit, (2) after the visit, or (3) usual care without chatbot use. English- and Spanish-speaking participants used ChatGPT to ask questions about their primary concern. Demographics, prior information-seeking behavior, and visit details were collected. Surveys administered immediately and at 3 months assessed understanding of diagnosis and treatment, decisional conflict (Decisional Conflict Scale), visit satisfaction, and chatbot satisfaction. Chatbot transcripts were analyzed for readability.

Among 125 randomized patients (July 2024-December 2024), attrition at 3 months was 3.2%. Most participants were White (78.2%) with at least a high school education (92.8%), and 98.8% had no prior medical chatbot use. Demographics and diagnoses were similar across groups. Decisional conflict was low and did not differ immediately (P = 0.691) or at 3 months (P = 0.875). Understanding of diagnosis, visit satisfaction, and chatbot satisfaction were similar between groups. By 3 months, home chatbot use was more common in the intervention arms (P = 0.011). Chatbot responses averaged an 11th-12th grade reading level.

Patient use of a chatbot peri-visit did not affect decisional conflict. While increasingly accepted, chatbots may not yet optimally support urogynecologic patient-centered decision making.