LO09: Role of hospitalization for detection of serious adverse events among emergency department patients with syncope: a propensity-score matched analysis of a multicenter prospective cohort

Abstract

Introduction: Selecting appropriate patients for hospitalization following emergency department (ED) evaluation of syncope is critical for serious adverse event (SAE) identification. The primary objective of this study is to determine the association of hospitalization and SAE detection using propensity score (PS) matching. The secondary objective was to determine if SAE identification with hospitalization varied by the Canadian Syncope Risk Score (CSRS) risk-category. Methods: This was a secondary analysis of two large prospective cohort studies that enrolled adults (age ≥ 16 years) with syncope at 11 Canadian EDs. Patients with a serious condition identified during index ED evaluation were excluded. Outcome was a 30-day SAE identified either in-hospital for hospitalized patients or after ED disposition for discharged patients and included death, ventricular arrhythmia, non-lethal arrhythmia and non-arrhythmic SAE (myocardial infarction, structural heart disease, pulmonary embolism, hemorrhage). Patients were propensity matched using age, sex, blood pressure, prodrome, presumed ED diagnosis, ECG abnormalities, troponin, heart disease, hypertension, diabetes, arrival by ambulance and hospital site. Multivariable logistic regression assessed the interaction between CSRS and SAE detection and we report odds ratios (OR). Results: Of the 8183 patients enrolled, 743 (9.0%) patients were hospitalized and 658 (88.6%) were PS matched. The OR for SAE detection for hospitalized patients in comparison to those discharged from the ED was 5.0 (95%CI 3.3, 7.4), non-lethal arrhythmia 5.4 (95%CI 3.1, 9.6) and non-arrhythmic SAE 6.3 (95%CI 2.9, 13.5). Overall, the odds of any SAE identification, and specifically non-lethal arrhythmia and non-arrhythmia was significantly higher in-hospital among hospitalized patients than those discharged from the ED (p < 0.001). There were no significant differences in 30-day mortality (p = 1.00) or ventricular arrhythmia detection (p = 0.21). The interaction between ED disposition and CSRS was significant (p = 0.04) and the probability of 30-day SAEs while in-hospital was greater for medium and high risk CSRS patients. Conclusion: In this multicenter prospective cohort, 30-day SAE detection was greater for hospitalized compared with discharged patients. CSRS low-risk patients are least likely to have SAEs identified in-hospital; out-patient monitoring for moderate risk patients requires further study.