Author(s)
Ha Eun Kim, BS
Alex W. Yang, MD
Gregory S. Hill MD
Jason Peart, BS
David Holovac, BS
Charles A. Riley, MD
Anthony M. Tolisano, MD
Affiliation(s)
Walter Reed National Military Medical Center
Abstract:
Objective: To assess whether integrating a large language model (ChatGPT) into pre-operative counseling improves patient understanding, satisfaction, and communication in otolaryngology surgery, and to examine how pre-operative preparedness relates to post-operative satisfaction and engagement.
Study Design: Prospective randomized controlled trial.
Setting: Tertiary military academic center.
Patients: 31 adults scheduled for otolaryngology surgery.
Interventions: After standard counseling, intervention arm participants interacted with ChatGPT to ask additional surgery-related questions. Both groups completed pre- and post-operative surveys using a 5-point Likert scale.
Main Outcome Measures: Pre-operative satisfaction and confidence metrics (Q1-Q8), post-operative satisfaction measures (Q1.1-3.1), correlation between pre- and post-op understanding and confidence metrics between the control and intervention groups. Counseling duration, ChatGPT interaction time, and number of questions asked.
Results: The average physician encounter lasted 10.9 minutes in the intervention group and 10.7 minutes in the control group. ChatGPT interactions averaged 6.2 minutes with 4.5 questions per patient. Pre-operative confidence, preparedness, and communication ratings (Q1, Q2, Q6, Q7) were strongly correlated (p < 0.05), confirming internal consistency of the counseling measures. Pre- and post-operative composite satisfaction scores did not differ between groups (p > 0.10), but the control group showed greater understanding of surgical risks (p = 0.012) and satisfaction in communication (p = 0.034). Patients whose questions were fully answered pre-operatively reported significantly higher post-operative satisfaction in both groups (p < 0.05).
Conclusions: Both groups demonstrated that thorough pre-operative communication and resolution of patient questions were key predictors of post-operative satisfaction. ChatGPT integration did not improve overall post-op satisfaction.
Learning Objective: To understand how a conversational AI model (ChatGPT) influences pre-operative communication, patient confidence and satisfaction in otolaryngology surgical settings.
Desired Result: To guide structured deployment of AI-assisted counseling as a complement to physician communication, emphasizing its use for question clarification and expectation management.
Level of Evidence: Level II
Indicate IRB or IACUC: Walter Reed National Military Medical Center IRB was obtained WRNMMC-2023-0423 (approval date 07/31/2024).