Multivariable risk scores for predicting short term outcomes for emergency department patients with unexplained syncope: a systematic review
OBJECTIVES Emergency department patients with unexplained syncope are at risk of experiencing an adverse event within 30 days. Our objective was to systematically review the accuracy of multivariate risk stratification scores for identifying adult syncope patients at high and low risk of an adverse event over the next 30 days.
METHODS We conducted a systematic review of electronic databases (MEDLINE, Cochrane, Embase and CINAHL) from database creation until May 2020. We sought studies evaluating prediction scores of adults presenting to an emergency department with syncope. We included studies that followed patients for up to 30 days to identify adverse events such as death, myocardial infarction, stroke, or cardiac surgery. We only included studies with a blinded comparison between baseline clinical features and adverse events. We calculated likelihood ratios and confidence intervals.
RESULTS We screened 13,788 abstracts. We included 17 studies evaluating nine risk stratification scores on 24,234 patient visits, where 7.5% (95% CI 5.3-10%) experienced an adverse event. A Canadian Syncope Risk Score of 4 or more was associated with a high likelihood of an adverse event (LR score=4 or more 11 [95% CI 8.9-14). A Canadian Syncope Risk Score of 0 or less (LR Score = 0 or less 0.10 [95% CI 0.07-0.20]) was associated with a low likelihood of an adverse event. Other risk scores were not validated on an independent sample, had low positive likelihood ratios for identifying patients at high risk, or had high negative likelihood ratios for identifying patients at low risk.
CONCLUSION Many risk stratification scores are not validated or not sufficiently accurate for clinical use. The Canadian Syncope Risk Score is an accurate validated prediction score for emergency department patients with unexplained syncope. Its impact on clinical decision making, admission rates, cost or outcomes of care is not known.