Healthcare Emergency Triage: Quantifying Safety, Throughput, and Equity Risk Under Uncertainty, Crowding, and Decision Latency

Abstract

This article proposes an engineering-oriented reliability framework for emergency triage that models end-to-end uncertainty propagation from initial symptoms and vital signs through decision rules, reassessment intervals, queueing dynamics, and escalation policies into distributional outcomes that matter for safety and service performance, including probability of under-triage, time-to-provider exceedance probability, adverse clinical event risk before definitive evaluation, over-triage burden, and an operational cost and harm index that integrates patient risk with resource strain. A scenario-based quantitative study is developed for a generic high-volume emergency department serving mixed-acuity adult patients, comparing four triage architectures: baseline scale-based triage with static rules and discretionary reassessment, expanded screening with more data but without governance, calibrated risk scoring with static thresholds, and a governance-optimized two-tier system that combines calibrated risk scoring, explicit uncertainty handling, capacity-aware dynamic thresholds, staged reassessment with trigger-based escalation, and safety-bounded operational controls. Results show that adding data or algorithms without governance can increase volatility and over-triage under crowding, that calibrated scoring improves stability but is fragile when queueing regimes shift, and that the two-tier governed architecture reduces under-triage and time-to-provider exceedances while stabilizing resource use and reducing inequitable failure patterns under documentation noise and surge conditions. Three copy-ready tables and complete prompts for data-driven figures are provided for Techne submission.