Predictors of outcomes after nontraumatic subdural hematoma

We sought to identify predictors of death or discharge to home for patients with nontraumatic subdural hematoma (NT-SDH).

Because neurologists are often involved in evaluation of patients with NT-SDH, we sought to identify factors that would be important in guiding patients and families about outcomes after diagnosis (and therapy) of NT-SDH.

We conducted a retrospective chart review of all cases admitted to Rush-Presbyterian St Lukes Medical Center, Chicago, Ill, from 1997 through 2001. Clinical data, Glasgow Coma Scale (GCS) score, Rankin Scale score, and operation were recorded. We reviewed available computed tomography/magnetic resonance reports and images to determine size and location of lesion. The analysis was restricted to first admission with regression performed of two binary outcomes: death and discharge to home.

There were 98 possible admissions for review with 51 patients meeting study criteria for review. Computed tomography/magnetic resonance images were available for 32 cases. Mean hematoma volume (n = 29) was 97 mL on the left and 82 mL on the right side, and 31 cases had midline shift; 7 also had herniation. In all, 32 (63%) underwent surgical evacuation (2 died); 27 (53%) were discharged home and 10 (20%) died before discharge. Four variables were univariately associated with death: female sex (P = .032); GCS score (P = .028); number of involved lobes (right side only) (P = .035); and no operation (P = .0032). On multivariate analysis, female sex and GCS score predicted likelihood of death. Four variables were univariately associated with discharge home: male sex (P = .036); GCS score (P < .0001); Rankin Scale score (P = .0006); and operation (P < .0001)). On multivariate analysis, GCS score and operation were predictive of discharge home. However, of the 24 patients with GCS score of 15, 20 underwent operation and 18 survived to discharge. Of the 8 nonsurgical deaths, 4 patients were in poor clinical condition and not offered operation.

In this study, GCS score was the major predictor of outcome. Patients with NT-SDH who were male or with good GCS score were less likely to die and more likely to be discharged home. Unlike patients with intracerebral hemorrhage (ICH),lesion size did not predict outcome for patients with NT-SDH. Extent of involvement (on the right side) was a predictor of death.