Predictors of mortality in stroke patients admitted to an intensive care unit

OBJECTIVE: Improved pathophysiologic insight and prognostic information regarding in-hospital risk of mortality among stroke patients admitted to an intensive care unit. DESIGN: Retrospective analysis. SETTING: Neurology/neurosurgery intensive care unit in a tertiary care university medical center. PATIENTS: A total of 63 consecutive ischemic stroke patients. INTERVENTIONS: Patients were classified according to in-hospital mortality. Charts were reviewed to retrospectively generate an admitting Acute Physiology and Chronic Health Evaluation (APACHE) II score. The APACHE II score and its individual components were assessed for predicting subsequent death. MEASUREMENTS AND MAIN RESULTS: Of 63 patients, 13 died and 50 survived to either discharge or surgical intervention. The mean admitting APACHE II score of survivors (6.9) was lower than that of patients who died (17.2; p < .0001). None of the 33 patients with a score <9 died, compared with 43% of those with a score > or =9. A score > or =18 was uniformly associated with fatal outcome (n = 8). Univariate analysis identified APACHE II total score, Glasgow Coma Scale score, temperature, pH, and white blood cell count as significant predictors of death. Among multivariate logistic regression models examining the components of the APACHE II score, the model containing white blood cells, temperature, and creatinine best predicted death. CONCLUSIONS: Several features of the APACHE II score are associated with risk of death in this patient population. The findings suggest particular physiologic derangements that are associated with, and may contribute to, increased mortality in critically ill patients with acute ischemic stroke.