The Benefit of the Doubt Phenomenon in Emergency Triage Assignment Disparities
Romero Mila, B.; Coggan, H.; Fine, A. M.; Barak-Corren, Y.; Reis, B. Y.; Aysola, J.; Chaudhari, P.; La Cava, W. G.
Show abstract
Emergency department (ED) triage decisions critically impact patient care and are standardized, yet ethnoracial disparities in triage assignment are well documented. We analyzed ethnoracial differences in triage assignments across four U.S. EDs (two adult, two pediatric), comprising 1.4 million encounters from 2011-2025. To better characterize these disparities, we developed an automated triage algorithm that replicates the Emergency Severity Index (ESI) criteria, the standard triage protocol used at each site. The algorithm identifies high-acuity symptoms and danger-zone vital signs that inform triage decisions at the level-2 (emergent) versus level-3 (urgent) boundary. We compared nurse triage assignments across ethnoracial groups, stratified by algorithmic ESI scores, using causal inference methods to adjust for clinical presentation and hospital context. Significant ethnoracial disparities in triage assignment were observed across all sites. Disparities were concentrated among patients algorithmically classified as lower risk but assigned higher acuity by nurses. This pattern is consistent with a "benefit-of-the-doubt" disparity, in which relatively stable, non-Hispanic White patients are more often assigned higher priority than Hispanic and non-Hispanic Black patients with comparable presentations. By contrast, disparities were attenuated or absent among patients deemed high risk by both nurses and the algorithm. Finally, analysis of the projected length-of-stay impact of substituting nurse-assigned with algorithmic triage scores suggests that algorithmic ESI decision support could reduce triage disparities with minimal effects on patient flow.
Matching journals
The top 4 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- The Impact of the “Muslim Ban” Executive Order on Healthcare Utilization in Minneapolis-St. Paul, Minnesota 92%
- Score for Emergency Risk Prediction (SERP): An Interpretable Machine Learning AutoScore–Derived Triage Tool for Predicting Mortality after Emergency Admissions 92%
- Clinical Outcomes, Costs, and Cost-effectiveness of Strategies for People Experiencing Sheltered Homelessness During the COVID-19 Pandemic 90%
Similar papers in this journal
- Development and assessment of a machine learning tool for predicting emergency admission in Scotland 93%
- Novel clinical subphenotypes in COVID-19: derivation, validation, prediction, temporal patterns, and interaction with social determinants of health 92%
- Finding Long-COVID: Temporal Topic Modeling of Electronic Health Records from the N3C and RECOVER Programs 92%
Similar papers in this journal
- Determinants of hospital outcomes for COVID-19 infections in a large Pennsylvania Health System 93%
- Development of a prediction model for 30-day COVID-19 hospitalization and death in a national cohort of Veterans Health Administration patients – March 2022 - April 2023 93%
- Clinical Characteristics and Outcomes for 7,995 Patients with SARS-CoV-2 Infection 93%
Similar papers in this journal
- Ten months of temporal variation in the clinical journey of hospitalised patients with COVID-19: an observational cohort 92%
- Evaluating the effectiveness of rapid SARS-CoV-2 genome sequencing in supporting infection control teams: the COG-UK hospital-onset COVID-19 infection study 89%
- Gendered hiring and attrition on the path to parity for academic faculty 89%
Similar papers in this journal
- Coding Long COVID: Characterizing a new disease through an ICD-10 lens 94%
- Quantifying the impact of hospital catchment area definitions on hospital admissions forecasts: COVID-19 in England, September 2020 - April 2021 91%
- Comparative assessment of methods for short-term forecasts of COVID-19 admissions in England at the local level 90%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.