Text

Objectives and questions: In December 2024, ChatGPT attracted more than 1.6 billion views, with approximately 180.5 million users. With its substantial progress in development and increasing application in the medical field, ChatGPT has become a valuable tool for providing health-related information. This study aims to evaluate the accuracy, reliability, and ease of comprehension of ChatGPT’s responses to frequently asked questions about knee osteotomies.

Material and methods: The ten most frequently asked questions were obtained from an informational brochure published by the German Society of Knee Surgeons. The questions were entered in German, and ChatGPT provided responses in the same language. The chatbot was prompted to answer in layman’s terms, use the best available knowledge, avoid fabricating information, and limit each response to 100 words. All questions were entered independently in a single session to ensure consistency in ChatGPT’s responses and minimize variations that could arise between different interactions.

The responses were then analysed by two orthopaedic surgery residents for clarity, accuracy, and completeness using the DISCERN technique, under the supervision of a board-certified knee surgeon. Readability was assessed using the German-adapted Flesch Reading Ease Score.

Results: The average DISCERN score for the ten responses was 45/80, with an average rating of 2.81 ±0.52 out of 5. This indicates a moderate level of reliability but highlights a lack of clear references and discussion of alternative treatments. While the information was generally easy to comprehend, it lacked a detailed risk-benefit analysis.

Subjectively, the answers appeared easy to understand and avoided complex clinical terminology. However, the Flesch Reading Ease Score averaged 33.8 ±6.23 suggesting the text is relatively difficult to read and is more suitable for individuals with an advanced level of education, equivalent to a university degree.

Discussion and conclusions: ChatGPT provides a moderate to high level of information in response to frequently asked questions about knee osteotomies. However, its responses lack evidence-based references, research citations, and a thorough discussion of alternative treatment options. This underscores the importance of tailored patient education provided by orthopedic surgeons, balancing cited guidelines, quantitative patient-reported outcomes, and simplified explanations ensuring that AI-generated responses complement but do not replace expert medical guidance, particularly in shared decision-making in the preoperative setting.