Prognostic predictors of thalamic hemorrhage

Abstract

Background and objective: There is no study evaluating the role of clinical, evoked potential and radiological parameters in the prognosis of thalamic hemorrhage employing multivariate logistic regression analysis, thus we aimed to evaluate the role of these parameters in predicting the 3 month outcome following thalamic hemorrhage. Setting: Tertiary care referral teaching hospital. Methods: Fifty-three patients with CT proven thalamic hematoma were evaluated. Conscious level was assessed using the Glasgow Coma Score (GCS), severity of stroke by the Canadian Neurological Scale (CNS), while muscle tone, tendon reflexes and power were also recorded. Hematoma size and type, and evidence of ventricular extension were obtained from the CT scan. Hematomas were classified as (A) thalamic with postero-lateral extension or (B) thalamic without postero-lateral extension. Central motor conduction to upper limb and median somatosensory evoked potentials (SEP) were recorded. Outcome was defined at 3 months on the basis of the Barthel Index (BI) with good being a BI of 12 or greater and poor a BI of less than 12. Best predictors of outcome were evaluated by single variable logistic regression analysis followed by multivariate logistic regression. Results: Age ranged between 35 and 85 years; 18 were women. Mean GCS was 10.4 and CNS was 3.9. Thirty-one patients had type A hematomas and 22 type B. The hematoma was small in 5, medium in 35 and large in 13 patients. Ventricular extension was present in 34 patients. Motor evoked potentials were unrecordable in 36 and central motor conduction time was prolonged in 8 patients. Median SEP was unrecordable in 37 and N9-N20 conduction time was prolonged in 2 patients. At 3 months, 8 patients had died, 24 had good and 21 had poor outcome. On univariate logistic regression analysis diabetes mellitus, GCS, pupillary asymmetry, CNS score, type and size of hematoma and motor and somatosensory evoked potentials were significant in relation to outcome. On multivariate logistic regression analysis, the best predictors of outcome at 3 months were the type of hematoma and CNS score. Conclusion: CNS score and CT appearance of hematoma are the best predictors of 3 month outcome following thalamic hemorrhage. The proposed model for outcome assessment is simple and easy to apply and could have wide clinical application.