What information is available online for adults with distal radius fracture? Online literature search and synthesis using generative AI

Introduction Distal radius fractures (DRFs) are the most common adult fracture in the UK, and patients seek online information to support recovery. The quality of such information varies, including in artificial intelligence (AI) generated summaries. This study explored the use of a generative AI model (Gemini) for information searching and synthesis of patient information on DRF recovery. Methods Seven health information questions were co-developed with patients, hand therapists, and surgeons. Each question was searched in Google using three terms: ‘distal radius fracture’, ‘wrist fracture’, and ‘broken wrist’. For each AI Overview, the summary and cited sources were exported. Citations were screened for relevance and checked for accurate representation in the AI summary. Trustworthiness was rated using a National Institutes of Health (NIH) framework. AI summaries were compared with four UK guidelines on DRF management and rehabilitation. Results Across all searches, 277 citations were identified; 37% were irrelevant. Of relevant citations, 57% (n = 99) were highly trustworthy, 40% (n = 70) moderately, and 3% (n = 6) low. Core messages aligned with UK guidelines, but AI summaries often gave fixed recovery timeframes. Discussion Google’s AI Overview produced generally accurate, trustworthy DRF recovery information. However, variable source relevance, differing complexity by search term, and fixed timeframes highlight the need for clinicians to understand the online content patients encounter. Terminology influenced tone, ‘distal radius fracture’ yielded technical summaries, ‘wrist fracture’ mixed technical and practical advice, ‘broken wrist’ offered more lay-friendly guidance. Awareness of these differences can support personalised advice and bridge gaps between AI-generated content and evidence-based practice.