# Instability Floors: Separating Bias from Noise in Fairness Audits of Clinical LLM Agents with FairMedAgent

arXiv: https://arxiv.org/abs/2609.03221 · HTML: https://mlsanity.com/paper/2609.03221/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 20 of its 63 references that are in the map.
- First posted on arXiv 2 Sep 2026.

## Builds on (most cited first, 20 of 20)

- [Self-Consistency Improves Chain of Thought Reasoning in Language Models](https://mlsanity.com/paper/2203.11171/) (2022)
- [Equality of Opportunity in Supervised Learning](https://mlsanity.com/paper/1610.02413/) (2016)
- [Datasheets for Datasets](https://mlsanity.com/paper/1803.09010/) (2018)
- [Counterfactual Fairness](https://mlsanity.com/paper/1703.06856/) (2017)
- [State of What Art? A Call for Multi-Prompt LLM Evaluation](https://mlsanity.com/paper/2401.00595/) (2023)
- [Measurement and Fairness](https://mlsanity.com/paper/1912.05511/) (2019)
- [AgentClinic: a multimodal agent benchmark to evaluate AI in simulated clinical environments](https://mlsanity.com/paper/2405.07960/) (2024)
- [Counterfactual Fairness in Text Classification through Robustness](https://mlsanity.com/paper/1809.10610/) (2018)
- [Aequitas: A Bias and Fairness Audit Toolkit](https://mlsanity.com/paper/1811.05577/) (2018)
- [Non-Determinism of "Deterministic" LLM Settings](https://mlsanity.com/paper/2408.04667/) (2024)
- [MedAgentBench: A Realistic Virtual EHR Environment to Benchmark Medical LLM Agents](https://mlsanity.com/paper/2501.14654/) (2025)
- [Understanding and Mitigating Numerical Sources of Nondeterminism in LLM Inference](https://mlsanity.com/paper/2506.09501/) (2025)
- [A Toolbox for Surfacing Health Equity Harms and Biases in Large Language Models](https://mlsanity.com/paper/2403.12025/) (2024)
- [Towards Evaluating and Building Versatile Large Language Models for Medicine](https://mlsanity.com/paper/2408.12547/) (2024)
- [Bias patterns in the application of LLMs for clinical decision support: A comprehensive study](https://mlsanity.com/paper/2404.15149/) (2024)
- [Write It Like You See It: Detectable Differences in Clinical Notes By Race Lead To Differential Model Recommendations](https://mlsanity.com/paper/2205.03931/) (2022)
- [CLIMB: A Benchmark of Clinical Bias in Large Language Models](https://mlsanity.com/paper/2407.05250/) (2024)
- [How Can We Diagnose and Treat Bias in Large Language Models for Clinical Decision-Making?](https://mlsanity.com/paper/2410.16574/) (2024)
- [EQUITRIAGE: A Fairness Audit of Gender Bias in LLM-Based Emergency Department Triage](https://mlsanity.com/paper/2605.03998/) (2026)
- [AgentFairBench: Do LLM Agents Discriminate When They Act?](https://mlsanity.com/paper/2606.16723/) (2026)

## Abstract (arXiv)

Counterfactual fairness audits of clinical language-model agents report a flip rate: how often an action changes when only the patient's demographic descriptor changes. Part of that rate is not demographic. A stochastic agent also changes its own action when nothing changes, and a flip rate cannot be interpreted without knowing how often. We measured it. Re-running one condition ten times over sixteen synthetic vignettes at default sampling changed a clinical agent's action in 8.7 percent of replicate pairs, from 2.2 percent for intensive-care escalation to 17.9 percent for controlled-substance caution, an output given no operational criteria. Across six models from five vendors, pooled floors ranged from 2.5 to 23.7 percent; in this panel neither disclosed size, vendor, nor hosting ordered them. The floor depends on the decoding configuration: majority voting over five draws removed 39 percent of it (95 percent confidence interval (CI) 18 to 64); at temperature 0 three of four locally served models showed no disagreement, but a hosted model still did. We also show that, for a binary action, the flip rate expected under no demographic effect equals the floor and a real effect adds only its square, so a flip rate inside the floor is not evidence of fairness, and direction must be tested with a signed paired test. We give a four-step reporting procedure and release FairMedAgent, the harness, with its protocol, vignettes and analysis scripts, so any team can measure the floor for its own agent.