Prediction of outcome after subarachnoid hemorrhage: timing of clinical assessment

Abstract

OBJECTIVE Currently, early prediction of outcome after spontaneous subarachnoid hemorrhage (SAH) lacks accuracy despite multiple studies addressing this issue. The clinical condition of the patient on admission as assessed using the World Federation of Neurosurgical Societies (WFNS) grading scale is currently considered the gold standard. However, the timing of the clinical assessment is subject to debate, as is the contribution of additional predictors. The aim of this study was to identify either the conventional WFNS grade on admission or the WFNS grade after neurological resuscitation (rWFNS) as the most accurate predictor of outcome after SAH. METHODS This prospective observational cohort study included 1620 consecutive patients with SAH admitted between January 1998 and December 2014 at our university neurovascular center. The primary outcome measure was a poor modified Rankin Scale score at the 2-month follow-up. Clinical predictors were identified using multivariate logistic regression analyses. Area under the receiver operating characteristic curve (AUC) analysis was used to test discriminative performance of the final model. An AUC of > 0.8 was regarded as indicative of a model with good prognostic value. RESULTS Poor outcome (modified Rankin Scale Score 4–6) was observed in 25% of the patients. The rWFNS grade was a significantly stronger predictor of outcome than the admission WFNS grade. The rWFNS grade was significantly associated with poor outcome (p < 0.001) as well as increasing age (p < 0.001), higher modified Fisher grade (p < 0.001), larger aneurysm size (p < 0.001), and the presence of an intracerebral hematoma (OR 1.8, 95% CI 1.2–2.8; p = 0.002). The final model had an AUC of 0.87 (95% CI 0.85–0.89), which indicates excellent prognostic value regarding the discrimination between poor and good outcome after SAH. CONCLUSIONS In clinical practice and future research, neurological assessment and grading of patients should be performed using the rWFNS to obtain the best representation of their clinical condition.