Citation reliability of frontier large language models in medical writing and its automated verification
- 1Seoul National University Bundang Hospital
Abstract
Large language models (LLMs) are increasingly used to draft medical manuscripts, yet their citations are unreliable and clinicians lack a validated way to verify them. We evaluated three frontier LLMs, Claude Opus 4.8, GPT-5.5, and Gemini 3.5 Flash, generating 270 cardiology narrative reviews with web search enabled, and verified all 8,050 references against PubMed. Problematic references accounted for 11.5% of GPT-5.5 output, 29.2% of Claude output, and 29.6% of Gemini output (P < 0.001), with no significant gradient across topics of differing publication volume (P = 0.052). Misattribution, a valid PubMed identifier that resolves to a different article, made up 77% of errors, whereas fabrication was rare (0.6%). Against an expert-adjudicated set of 270 references, an LLM-based Chain-of-Verification (CoVe) detected 60 of 62 problematic references (sensitivity 96.8%, specificity 98.6%), including every misattribution and fabrication. LLM-generated citations require identifier-level verification, and CoVe provides it at expert-level accuracy.
Related
- Citation Fabrication and Accuracy in AI-Generated Psychiatric Research: A Comparative Evaluation of Four Contemporary Large Language Models
- Benchmarking ten frontier large language models on 1,477 board style multiple choice questions in hematology
- Accuracy Overstates Evidence Grounding and Abstention Reliability in Mammography Vision-Language Models
- Reliability, Quality, and Readability of Freely Available Large Language Models for Clinical Decision Support in Perforated Peptic Ulcer: A Blinded, Multidimensional Study
- Are Frontier Large Language Models Safer Than Government-Backed Symptom Checkers for Clinical Self-Triage? A Standardised Vignette Evaluation
BibTeX
@misc{shin2026citation,
title = {{Citation reliability of frontier large language models in medical writing and its automated verification}},
author = {R. Shin and J.-M. Lee and J. Park and J.-S. Kwun and H.-W. Cho and S.-H. Kang and K.-H. Jeon},
howpublished = {medRxiv},
year = {2026},
month = aug,
doi = {10.64898/2026.08.31.26361806},
url = {https://www.medrxiv.org/content/10.64898/2026.08.31.26361806v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Citation reliability of frontier large language models in medical writing and its automated verification
AU - R. Shin
AU - J.-M. Lee
AU - J. Park
AU - J.-S. Kwun
AU - H.-W. Cho
AU - S.-H. Kang
AU - K.-H. Jeon
PY - 2026
DA - 2026/08/31
PB - medRxiv
DO - 10.64898/2026.08.31.26361806
UR - https://www.medrxiv.org/content/10.64898/2026.08.31.26361806v1
AB - Large language models (LLMs) are increasingly used to draft medical manuscripts, yet their citations are unreliable and clinicians lack a validated way to verify them. We evaluated three frontier LLMs, Claude Opus 4.8, GPT-5.5, and Gemini 3.5 Flash, generating 270 cardiology narrative reviews with web search enabled, and verified all 8,050 references against PubMed. Problematic references accounted for 11.5% of GPT-5.5 output, 29.2% of Claude output, and 29.6% of Gemini output (P < 0.001), with no significant gradient across topics of differing publication volume (P = 0.052). Misattribution, a valid PubMed identifier that resolves to a different article, made up 77% of errors, whereas fabrication was rare (0.6%). Against an expert-adjudicated set of 270 references, an LLM-based Chain-of-Verification (CoVe) detected 60 of 62 problematic references (sensitivity 96.8%, specificity 98.6%), including every misattribution and fabrication. LLM-generated citations require identifier-level verification, and CoVe provides it at expert-level accuracy.
ER -