Background/purpose: Patients with severe strokes may have different associated medical comorbidities from those with mild strokes. This study evaluated the neurologic and non-neurologic medical predictors of mortality in patients with severe cerebral infarction in the acute stage.
Methods: Patients admitted to a neurologic intensive care unit (ICU) due to cerebral infarction were included. Neurologic and non-neurologic predictors for in-unit mortality were determined by logistic regression analyses. Two models using (A) neurologic factors and (B) combined neurologic and non-neurologic factors as mortality predictors were developed. The performance of the models in predicting overall, neurologic and non-neurologic mortalities was compared by areas under the receiver-operating characteristic curves (AUC) of the derived regressive equations.
Results: Of 231 patients with cerebral infarction admitted to the ICU, 34 (14.7%) died during ICU stay. Conscious state and acute physiologic abnormalities were significant predictors of mortality. The length of ICU stay in patients with non-neurologic mortality was longer than in those with neurologic mortality (p = 0.044). The AUC of Model B was larger than that of Model A in predicting overall (0.768 +/- 0.045 vs. 0.863 +/- 0.033, p = 0.005) and non-neurologic mortalities (0.570 +/- 0.073 vs. 0.707 +/- 0.074, p = 0.009), while there was no difference in predicting death from neurologic causes (0.858 +/- 0.044 vs. 0.880 +/- 0.032, p = 0.217).
Conclusion: Impaired consciousness and acute physiologic abnormalities are independent predictors of mortality for severe ischemic stroke during the acute stage. Neurologic factors predict early mortality from intrinsic cerebral dysfunction, while non-neurologic factors, especially the associated physiologic abnormalities, predict late mortality from medical complications.