Artificial Intolerance: Stigmatizing Language in Clinical Documentation Skews Large Language Model Decision-Making
This work highlights a critical fairness and robustness vulnerability in LLMs for clinical decision support, which could automate health disparities if unaddressed.
This paper investigates whether large language models (LLMs) inherit and propagate stigmatizing language (SL) from clinical notes, finding that all nine evaluated models exhibit substantial bias, with clinical decisions skewed towards less aggressive patient management. A single SL sentence can alter outputs, and standard mitigation strategies like Chain-of-Thought reasoning show limited efficacy.
Large Language Models (LLMs) are increasingly deployed in high-stakes domains such as clinical decision support and medical documentation. However, the robustness of these models against subtle linguistic variations, specifically stigmatizing language (SL) commonly found in human-authored clinical notes, remains critically under-explored. In this work, we investigate whether frontier LLMs inherit and propagate this human bias when processing clinical text. We systematically evaluate nine frontier LLMs across four stigmatized medical conditions, utilizing clinical vignettes injected with varying intensities and phenotypes of SL (doubt, blame, and maligning). Our results demonstrate that all evaluated models exhibit substantial bias, with clinical decision-making significantly skewed towards less aggressive patient management. Notably, we observe a high sensitivity to linguistic framing, where a single SL sentence is sufficient to alter model outputs, revealing a clear dose-response relationship. Furthermore, we evaluate standard prompt-based mitigation strategies, including Chain-of-Thought (CoT) reasoning and model self-debiasing. These approaches show limited efficacy; models struggle to explicitly identify SL while remaining implicitly influenced by it. Our findings expose a critical vulnerability in current LLMs regarding fairness and robustness in clinical NLP, underscoring the need for rigorous algorithmic guardrails to prevent the automation of health disparities.