Validation of the mSOAR and SOAR scores to predict early mortality in Chinese acute stroke patients

BACKGROUND It is unclear in Chinese patients with acute stroke how the SOAR (stroke subtype, Oxfordshire Community Stroke Project classification, age, and prestrike modified Rankin) and mSOAR (modified-SOAR) scores performed in predicting discharge mortality and 3-month mortality. We aimed to validate the predictability of these scores in this cohort. METHODS Data from the China National Stroke Registry (CNSR) study was used to perform the mSOAR and SOAR scores for predicting the discharge and 3-month mortality in acute stroke patients. RESULTS A total of 11073 acute stroke patients were included in present study. The increased mSOAR and SOAR scores were closely related to higher death risk in acute stroke patients. For discharge mortality, the area under the receiver-operator curve (AUC) of the mSOAR and SOAR scores were 0.784 (95% CI 0.761-0.807) and 0.722 (95% CI: 0.698-0.746). For 3-month mortality, they were 0.787 (95% CI: 0.771-0.803) and 0.704 (95% CI: 0.687-0.721), respectively. The mSOAR and SOAR scores showed significant correlation between the predicted and observed probabilities of discharge mortality (mSOAR: r = 0.945, P = 0.001; SOAR: r = 0.994, P<0.001) and 3-month mortality (mSOAR: r = 0.984, P<0.001; SOAR: r = 0.999; P<0.001). CONCLUSIONS The mSOAR score predicted reliably the risk of death in Chinese acute stroke patients.