Development of a prehospital assessment to identify stroke mimic conditions

Background Despite routine use of pre-hospital identification instruments, approximately 30% of suspected stroke admissions are stroke mimics (SM). Early identification may allow “false positive” SM patients to be directed to appropriate care and improve healthcare resource utilisation. Methods A retrospective database of ambulance records containing a paramedic impression of stroke was linked to hospital specialist diagnosis data from 01/06/13 to 31/05/16. Logistic regression identified clinical features predictive of SM. An assessment score was constructed prioritising specificity over sensitivity. Results 1650 patients (mean age 75.3, 47% male, 40% SM) were included. 1520 (92%) were Face Arm Speech Test (FAST) positive. Table 1 describes the characteristics in the SM assessment. Each characteristic scores 1 point if present. Table 1 Stroke mimic characteristics 86% (66/77) of suspected stroke patients scoring 1 were SM. 100% (6/6) of patients scoring >1 characteristic were SM. A score ≥1 identified SM with 11% (95% CI, 8–13) sensitivity, 99% (95% CI, 98–99) specificity, positive predictive value of 87% (95% CI, 79–94), negative predictive value of 62% (95% CI, 60–64) and a diagnostic odds ratio of 11 (95% CI, 6–20, p