Abstracts

Rationale: Status Epilepticus (SE) is associated with significant mortality. Furthermore, there are no outcome scales developed from large prospective populations to predict the probability of mortality in SE. Because of the complex clinical presentations of SE, it has been difficult to develop simple out come scales that accurately predict outcome. Thus, it is important to develop bedside outcome scales to predict mortality for SE that can be readily employed by the clinician in the acute clinical setting. Methods: Patient data from the prospective Richmond SE study were employed to develop this model. Predictors of outcome were identified in 712 cases utilizing multivariate logistic regression analysis. The coefficients for each risk factor was determined and used to calculate the probability of mortality in this model. We used the model to predict outcome without knowing the actual outcome. The predicting model was developed so that it could be programmed into a personal digital assistant (PDA). The model was developed with the intent that the clinician could use this PDA program at the bedside to easily evaluate outcome. The complex calculations needed to evaluate outcome are performed by the PDA with simple input information by the clinician. The clinician through the PDA will be asked a sequence of questions regarding the values for predictors in the model used in computing the probability of outcome. Results: Using the Richmond prospective SE data base we identified a population of 712 SE cases that had the predictors we wanted to analyze. We found the combination of clinical and laboratory predictors produced the best model. To evaluate the model we predicted the probability of mortality for each individual in the data set. We set the mortality to 20%, since that was the observed mortality for SE in the Richmond population. Using the overall 20% mortality rate for SE in the population, we predicted an individual to die (mortality=1) if his/her probability was larger than or equal to 20%. Since we knew the actual mortality for the individuals in the sample, we constructed 2 x 2 tables comparing predicted mortality with actual mortality and calculated the sensitivity and specificity of the prediction. We obtained a high degree of sensitivity and specificity using this model. Conclusions: The results from this study demonstrate that it is possible to develop a bedside SE outcome scale to predict mortality with a high specificity and selectivity. The use of the PDA to perform the complex calculations needed in the model make it user friendly for the clinician to use this scale in the acute clinical setting. The ability to recognize high risk SE cases will offer new hope for the development of novel treatment strategies for high risk cases to prevent death in SE and provide guidelines to counsel families regarding prognosis.