INTRODUCTION: Online patient education materials are essential in improving access to and understanding of clinical information. Artificial intelligence (AI), specifically ChatGPT, offers a novel way to synthesize complex information and is increasingly being used for patient education, including in obstetrics and gynecology. This study compares the understandability and actionability of patient education materials about hysterectomy and uterine fibroids generated by ChatGPT with those published by the American College of Obstetricians & Gynecologists (ACOG). METHODS: In a randomized cross-sectional study, participants were allocated to one of four groups: ACOG–Fibroids, ChatGPT–Fibroids, ACOG–Hysterectomy, or ChatGPT–Hysterectomy. The ACOG groups reviewed the corresponding ACOG FAQ webpage, while the ChatGPT groups reviewed responses generated by ChatGPT to the same FAQs. Participants then completed demographic questions and the Patient Education Materials Assessment Tool (PEMAT). The primary outcome was the equivalence of PEMAT scores between ACOG and ChatGPT. RESULTS: Equivalence was not demonstrated between ChatGPT and ACOG for either topic. For fibroid material, ACOG was rated higher on both understandability (86.3 versus 82.6 points, P =.039) and actionability (65.4 versus 53.2 points, P =.003) on superiority testing. For hysterectomy content, no significant difference was observed. No interaction was found between material type and either education level or personal history of fibroids/hysterectomy. CONCLUSIONS/IMPLICATIONS: PEMAT scores for ACOG were superior to ChatGPT for fibroid content, but neither equivalent nor superior for hysterectomy content. While ChatGPT offers a responsive, iterative source of information, its understandability and actionability may depend on topic-specific factors, highlighting both the promise and limitations of AI for patient education.
Derakshani et al. (Thu,) studied this question.