Enhancing Patient Comprehension of Post-Operative Instructions and Education Using AI: A Quality Improvement Project

Background and/or Objectives: Compliance with post-operative instructions is integral for recovery, preventing complications, and avoiding re-injury. However, lack of medical literacy may impede patient comprehension. National surveys indicate the average U.S. resident and the average Medicare beneficiary read at an 8th-grade and 5th-grade reading level, respectively, demonstrating patient literacy as an important consideration. AI programs are potential tools for clarifying medical terminology and simplifying language.

Design: 9 month prospective study.

Setting : Single-Center PM&R & Pain Outpatient Clinic

Participants : 67 patients receiving an interventional procedure under fluoroscopy and local anesthesia.

Interventions: Each patient received two post-operative instructions: original version and AI-revised version. Revision was generated in ChatGPT to a 5th grade reading level. Presentation of post-operative instructions was randomized to control for ordering effects. Patients scored instructions on a 5-point Likert scale for clarity and chose a preferred version based on confidence in adhering to the instructions. Patient insurance information was also collected.

Main Outcome Measures: A mixed-design ANOVA was used to examine main effects of Form Version (within-subjects) and Insurance Type (between-subjects) and also interaction effect on clarity scores.

Results: Form Version (Original vs. AI-Revised) was statistically significant (p=0.00177) with AI-Revised having a higher average score of 0.4. Insurance Type (Medicare vs. PPO) and interaction effect (Form Version-Insurance Type) were not statistically significant (p=0.1426 and p=0.4174). 50 out of 67 patients (74.6%) preferred the AI-revised version. When stratified by insurance types, patient preference for the AI-revised version remained high [16 out of 23 Medicare beneficiaries (69.6%), 34 out of 44 of PPO beneficiaries (77.3%)].

Conclusions: AI-revisions of post-operative instructions led to enhanced patient comprehension, and these findings demonstrate the utility of AI in combating health literacy challenges. Broader integration of AI tools in future adaptations of patient education materials should be considered to reduce health barriers related to poor literacy, improve patient understanding, and ultimately yield better health outcomes through increased compliance.