RADIOLOGICAL PREDICTORS OF MORTALITY IN PATIENTS WITH PRIMARY SPONTANEOUS INTRACEREBRAL HAEMORRHAGE

BACKGROUND Primary spontaneous intracerebral haematoma (ICH) is a deadly form of stroke with very high mortality and morbidity. Outcome prediction is mandatory for optimisation of therapeutic efforts. The objective of our prospective study is to identify reliable radiological predictors of mortality from a plain CT study of brain. MATERIALS AND METHODS We prospectively collected clinical and radiological data from 112 patients with primary ICH. Key measurements were taken from the plain CT scan of brain at admission to collect the radiological parameters. Bivariate analysis of clinical and radiological predictors of poor outcome were undertaken using unadjusted logistic regression analysis. Those radiological variables found to have statistical significance on bivariate analysis were analysed using multivariate logistic regression model. The regression model was validated using ROC (Receiver Operating Characteristic) curve. RESULTS The mortality in our series was 28.6% (N = 32). Radiological factors associated with poor outcome on bivariate analysis were volume of haemorrhage, Midline Shift (MLS), Intraventricular Haemorrhage (IVH) and hydrocephalus. On multivariate analysis IVH (P = .007), MLS (.008) and volume of haemorrhage (P = .029) were significantly associated with poor outcome. Using ROC curve, we had a cut-off volume of 42.5 cm(3) with an optimum sensitivity of 81.2% and specificity of 90%. CONCLUSION Plain CT study of brain obtained within 24 hours of the ictus can be reliably used for predicting the prognosis of subjects with primary supratentorial ICH.