The association between Oxford Acute Severity of Illness Score and mortality in critical hemorrhagic stroke patients: a retrospective cohort study.

Hemorrhagic stroke is a kind of disastrous cerebrovascular disease, Oxford Acute Severity of Illness Score (OASIS) was only evaluated among mixed critically ill patients, its predictive value in hemorrhagic stroke patients remains undiscovered. Evaluate the association between Oxford Acute Severity of Illness Score (OASIS) and mortality in critical hemorrhagic stroke patients. A retrospective cohort study. Data were extracted from a public database named Medical Information Mart for Intensive Care III (MIMIC-III). 1838 critical hemorrhagic stroke patients were included, multivariable logistic regression and receiver operating characteristic (ROC) curves were mainly used to analyze data. ROC curve analyses were also conducted in hemorrhagic stroke patients stratified by Glasgow Coma Scale (GCS) as subgroup analyses. The primary outcome was 30-day mortality. The sample size of this study is 1838 patients. The OASIS was significantly correlated with 30-day mortality (Odds ratio (OR) 1.125 per one-point increase, 95% confidence interval (CI) [1.107-1.144], p < 0.0001), the area under the ROC curve (AUC) of OASIS was comparable to that of Simplified Acute Physiology Score II (SAPSII) for predicting 30-day mortality (AUC: 0.7702 vs. 0.788, P = 0.096). Sensitivity analyses showed the results were stable. In subgroup analyses OASIS also has the similar discriminatory power to predict 30-day mortality for the severe (GCS 3-8) and mild (13-15) hemorrhagic stroke patients, but it has lower discriminatory power to predict 30-day mortality for the moderate (9-12) patients. The OASIS might serve as an alternative choice to predict outcomes of severe and mild hemorrhagic stroke patients in consideration of the practicality. Selection bias was unavoidable because this study was a retrospective observational study.