Outcome-grounded effect of clinically stigmatizing information on large language model emergency triage prioritization

Emergency triage decisions can be affected by stigmatizing information, potentially compromising the equity and timeliness of care. This study examined how clinically stigmatizing information influenced large language model triage prioritization, using outcome-grounded scenarios. The findings indicate that some stigmatizing formulations can lead to harmful reprioritization. The study raises concerns about bias in AI-supported triage, although simulated evaluations cannot establish the frequency of these effects in actual clinical practice.