ChatGPT as a Tool for Perioperative Safety: Evaluating Drug Interaction Detection in Anesthetic Regimens

Background Drug-drug interactions (DDIs) are a major contributor to perioperative complications, yet no prior studies have evaluated the performance of large language models in this high-risk setting. This study assessed the accuracy of ChatGPT, GPT-5, free version (OpenAI, San Francisco, USA) in identifying DDIs in anesthetic regimens compared with the gold-standard reference Lexicomp. Methods We conducted an analytical cross-sectional study using 40 synthetic perioperative vignettes generated by OpenEvidence and reviewed by an anesthesiologist. Each vignette was reviewed for clinically significant DDIs using Lexicomp’s “Interactions” feature. ChatGPT was queried twice with standardized prompts, and responses were classified as Correct or Incorrect. Results Across both trials, ChatGPT accurately identified 76 of 80 clinically significant DDIs (sensitivity 95%) with four responses classified as Incorrect. These responses were not entirely erroneous, as each recognized at least one major interaction, yet they omitted additional clinically relevant interactions or failed to provide comprehensive management recommendations. Notably, ChatGPT frequently supplemented its responses with relevant contextual information, including age-specific considerations and perioperative monitoring strategies. Conclusion ChatGPT demonstrated high accuracy in detecting clinically significant DDIs in perioperative anesthetic regimens. Its contextual insights may enhance clinical decision-making; however, occasional omissions and limitations in sourcing warrant cautious integration into practice. Further studies in real-world perioperative settings are needed.