ABSTRACT NUMBER: ESOC2026A1970 DEVELOPMENT OF A PREHOSPITAL DIAGNOSTIC MODEL TO IDENTIFY INTRACEREBRAL HAEMORRHAGE IN PATIENTS WITH SUSPECTED STROKE

Abstract Background and aims Prehospital identification of patients with intracerebral haemorrhage (ICH) may facilitate rapid referral and timely initiation of acute management. Our aim was to develop a diagnostic model based on clinical characteristics to detect ICH in patients with a suspected stroke in the prehospital setting. Methods We used individual patient data from two prospective cohorts (LPSS (Leiden Prehospital Stroke Study) and PRESTO (Prehospital Triage of Patients With Suspected Stroke)) of patients with suspected stroke transported by ambulance. Nineteen variables were preselected based on clinical expertise and literature. We used stepwise backwards logistic regression for model development. Model performance was assessed, after bootstrap resampling and optimism correction, with concordance (C) statistics. We evaluated the model in a subgroup of patients with a RACE score ≥5, indicating more severe neurological deficits. Results Of 3321 patients (1713 [51.5%] male, median age 73 [IQR 62-82]), 211 patients (6.4%) had a diagnosis of ICH. Four variables were included in the model: systolic blood pressure (OR 1.03,95%CI:1.02–1.03), history of ICH (OR 3.47,95%CI:1.81-6.65), history of diabetes (OR 0.47,95%CI:0.30-0.75) and RACE-score (OR 1.45,95%CI:1.38–1.53). Discrimination showed a C-statistic of 0.82 (95%CI:0.79-0.85). In the RACE ≥5 subgroup, the C-statistic was 0.77 (95%CI:0.75-0.80). Increasing model-predicted probabilities were associated with a higher observed proportion of patients with ICH (Figure 1). Conclusions Our model is based on four easily retrievable clinical characteristics and showed promising results in identifying ICH in the prehospital stroke code setting. External validation is needed to further determine predictive performance. Conflict of interest Naomi de Ruijter: nothing to disclose Figure 1 - belongs to Results