Is using intracerebral hemorrhage scoring systems valid for mortality prediction in surgically treated patients?

While intracerebral hemorrhage (ICH) scoring systems provide mortality and morbidity prediction, the actual mortality rates seem to be lower than those predicted by scoring systems in our clinical impression. To assess the validity of the ICH score and the Surgical Swedish ICH (SwICH) score, we retrospectively reviewed surgically treated ICH patients between 2012 and 2019. Uni- and multivariate analyses were performed to identify variables in predicting 30-day mortality. We identified 203 patients (mean ICH score 2.7; mean SwICH score 2.0). The actual 30-day mortality was 7%, which was significantly lower than those predicted by the ICH and the SwICH scores (55% and 16%, respectively; p < 0.001). Both scores were strongly correlated with the modified Rankin scale (mRS) at discharge (correlation coefficient 0.97 and 0.98; critical value 0.81). The only significant prognostic factors for the 30-day mortality by multivariate analysis were anisocoria (p = 0.03) and preoperative Glasgow Coma Scale (p = 0.03). These two factors also predicted mRS at discharge (p < 0.001). After discharge, 15% of patients improved regarding mRS and 29% of wheelchair-bound patients gained the ability to ambulate. No significant relationship existed between the degree of recovery after discharge and preoperative ICH score (p = 0.25). The ICH and SwICH scores were more valid in predicting morbidity, rather than mortality after surgical intervention for ICH. Anisocoria and Glasgow Coma Scale < 7 were the only two factors that predicted 30-day mortality and morbidity at discharge.