Prospective Validation of the Canadian Syncope Risk Score in a United States Community Setting: An Interim Analysis

The Canadian Syncope Risk Score (CSRS) can be used to identify which emergency department (ED) patients with unexplained syncope are likely to develop a serious post-ED event but has not been widely validated in U.S. community EDs. We sought to address this evidence gap.

We undertook a prospective pragmatic observational study to validate the CSRS in 5 U.S. community EDs from 03/2022 through 10/2024 (30 of 32 months have been analyzed). We included health plan members ≥16 years of age with syncope whose treating physician activated a clinical decision support tool. We excluded patients with a serious cause identified in the ED (e.g., intestinal bleeding). The tool helped calculate the CSRS, assigned a risk category with an estimated 30-day incidence of serious outcomes (our primary outcome, adopted from the original CSRS studies), and provided risk-based recommendations for monitoring and consultation. Serious outcomes were collected using extraction from clinical databases combined with manual health records review and were each adjudicated. We calculated calibration and discrimination characteristics for CSRS validation.

We included 3,642 patients with median age 68 years (IQR 51-79); 2,031 (55%) were female. Overall, 98 (2.7%) experienced a 30-day composite outcome, including 6 patients (0.16%) who died. Only 1 outcome occurred among 603 patients under 40 years of age. With the total score as the only predictor in a logistic regression model, the calibration slope was 0.82 with good fit at low values (where accurate prediction of absolute risk is clinically meaningful). The area under the receiver operating characteristic curve was 0.73 (95% CI: 0.68-0.77). The proportion of patients with a composite outcome increased from 7 of 1,257 (0.6%) in the very-low-risk group to 11 of 125 (8.8%) in the very-high-risk group (Cochran-Armitage trend test P<0.001). None of the very low and low-risk patients died. At a threshold score of −1 (2,385 of 3,642 patients [65.5%] had a risk score ≥-1), the CSRS sensitivity and specificity were 92.8% (95% CI, 85.8%-97.0%) and 35.3% (95% CI, 33.7%-36.9%), respectively.

In this interim analysis of a multicenter validation study of the CSRS in U.S. community EDs, the tool showed good fit at low values and good discrimination. Outcomes were rare in younger adults. We await study completion and subsequent tool expansion to 16 affiliated EDs.