A Novel Stroke Mimics Prediction Score During In-Hospital Triage For Suspected Stroke Patients: The Stroke Mimics Score
Irene Scala1, Pier Andrea Rizzo3, Marcello Covino2, Giovanni Frisullo1
1Neuroscience, 2Emergency Department, Fondazione Policlinico Universitario A. Gemelli IRCCS, 3Neuroscience, Catholic University of the Sacred Heart
Objective:
First, we aimed to identify predictors of stroke versus Stroke Mimics (SM) among parameters acquired in the Emergency Department (ED). Second, we aimed to develop a diagnostic score predictive of SM based on these parameters.
Background:
Since the time between the onset of stroke symptoms and the start of treatment is critical in determining patient outcome, early differential diagnosis between stroke and SM is a major challenge in the ED.
Design/Methods:
We included adult patients admitted to our ED for suspected stroke between 2015 and 2022 for retrospective enrollment (derivation cohort) and during 2023 for prospective enrollment (validation cohort). The discharge diagnoses were classified as Cerebrovascular Event (CE-stroke+TIA) and SM. Predictors of CE in the derivation cohort were identified through logistic regression analysis. Score points were assigned to each identified predictor based on the β-coefficients obtained in a linear regression model. The score was then validated in the prospective cohort of patients. We then compared the observed events (confirmed CE) with the expected events (suspected CE) at each SMS score level and grouped similar values into three risk groups.
Results:
6998 patients were included in the retrospective cohort, and 1650 in the validation cohort. The SMS was composed of 10 variables easily collected during the triage of patients with suspected stroke (score range: 0-12). The visual calibration of the score allowed us to obtain three different risk categories for diagnosis of CE: SMS≤3 points: low risk (<5%); SMS 4-7 points: medium risk (5-50%); SMS≥8 points: high risk (>50%). In the ROC curves, the score performed excellently in both the derivation (AUC 0.840) and validation cohorts (AUC 0.844).
Conclusions:
The results of our study suggest that a predictive score based mainly on clinical and physiological parameters collected at the time of the triage assessment, namely SMS, has an excellent discriminative ability between CE and SM.
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.