Emergency Department Triage Accuracy and Delays in Care for High-Risk Conditions

Abstract

Importance: Emergency department (ED) triage may impact timeliness of care for high-risk conditions.

Objective: To determine the association of ED undertriage with delays in care for patients with subarachnoid hemorrhage (SAH), aortic dissection (AD), and ST-elevation myocardial infarction (STEMI).

Design, setting, and participants: This retrospective cohort study included adult ED patients diagnosed with SAH, AD, or STEMI from January 1, 2016, to December 31, 2020, from a multicenter, community-based health care delivery system. Data analysis were completed in March 2023 to October 2024.

Exposure: Undertriage vs correct triage, defined by operational measures of mistriage.

Main outcomes and measures: Using a lognormal distribution, the outcomes of interest for patients with SAH and AD were adjusted median time to noncontrast computed tomography (CT) (head CT for patients with SAH, chest CT for patients with AD), antihypertensive medication orders (SAH), and β-blocker orders (AD), and ED length of stay (LOS). For patients with STEMI, outcomes of interest were adjusted median time to electrocardiogram (ECG) and troponin orders.

Results: A total of 5929 patients (median [IQR] age, 63.0 [54.0 to 73.0] years; 3876 [65.4%] male) were identified, including 915 with SAH, 480 with AD, and 4534 with STEMI. There were 1129 Asian patients (19.0%), 553 Black patients (9.3%), 889 Hispanic patients (15.0%), and 2906 non-Hispanic White patients (49.0%). Overall, 2175 patients (36.7%) were undertriaged. For patients with SAH, the lognormal estimate for delay in time to head CT was 0.2 (95% CI, 0.0-0.3), or a delay of 2.4 minutes, and for antihypertensive orders, the lognormal estimate was 4.8 (95% CI, 3.6-5.9), or a delay of 33.3 minutes; the lognormal estimate for ED LOS was 0.1 (95% CI, 0.0-0.1), or 7.7 minutes longer. For patients with AD, the lognormal estimate for delays were 0.2 (95% CI, 0.0-0.4), or 8.9 minutes, for chest CT and 0.5 (95% CI, 0.2-0.7), or 17.6 minutes, for β-blocker orders, and ED LOS was 0.2 (95% CI, 0.1-0.3), or 64 minutes longer. For patients with STEMI, differences in time to ECG and troponin orders were not statistically significant, at less than 1 minute, comparing correctly and undertriaged patients.

Conclusions and relevance: In this cohort study of patients diagnosed with SAH, AD, or STEMI, ED undertriage was associated with small but significant delays in key diagnostic and therapeutic orders for patients with SAH and AD but not for patients with STEMI.

Figure 1.. Adjusted Time to Key Diagnostic and Treatment Interventions and Emergency Department (ED) Length of Stay (LOS) Comparing Undertriaged and Correctly Triaged Patients With Subarachnoid Hemorrhage

Difference in time to computed tomography (CT) order: P = .02; difference in time to antihypertensive order: P < .001; difference in ED LOS: P = .16. Model adjusted for age; gender; race and ethnicity; primary language; neighborhood deprivation index; active Kaiser Permanente Northern California health plan membership; comorbidity score; ED arrival mode (ambulance vs walk-in); year of visit; time of visit (office hours vs non–office hours); recent ED, inpatient, and intensive care unit utilization; use of anticoagulant; triage vital signs; and chief complaint (headache or no headache). The time to CT model was a mixed-effects model with clustering by facility only. The time to antihypertensive orders model was a mixed-effects model without clustering (models did not converge with any clustering). The ED LOS model was a mixed-effects model with clustering by patient and facility.

Figure 2.. Adjusted Time to Key Diagnostic and Treatment Interventions and Emergency Department (ED) Length of Stay (LOS) Comparing Undertriaged and Correctly Triaged Patients With Aortic Dissection

Model adjusted for age; gender; race and ethnicity; primary language; neighborhood deprivation index; active Kaiser Permanente Northern California health plan membership; comorbidity score; emergency department (ED) arrival mode (ambulance vs walk in); year of visit; time of visit (office hours vs non–office hours); recent ED, inpatient, and intensive care unit utilization; use of anticoagulant; triage vital signs; and chief complaint (chest pain or no chest pain). Difference in time to computed tomography (CT) order: P = .01; difference in time to β-blocker order: P < .001; difference in ED LOS: P < .001.

Figure 3.. Adjusted Time to Key Diagnostic Interventions Comparing Undertriaged and Correctly Triaged Patients With ST Elevation Myocardial Infarction

Difference in time to electrocardiogram (ECG): P < .001; difference in time to troponin: P = .03. Model adjusted for age; gender; race and ethnicity; primary language; neighborhood deprivation index; active Kaiser Permanente Northern California health plan membership; comorbidity score; ED arrival mode (ambulance vs walk-in); year of visit; time of visit (office hours vs non–office hours); recent ED, inpatient, and intensive care unit utilization; use of anticoagulant; triage vital signs; and chief complaint (chest pain or no chest pain). Both models used mixed effects with clustering by patient and facility.