The trusted source for
healthcare information and
Can We Predict the Outcome of Bacterial Meningitis?
Abstract & Commentary
By Joseph E. Safdieh, MD Assistant Professor of Neurology, Weill Medical College, Cornell University Dr. Safdieh reports that he has received grant/research support from the American Academy of Neurology.
Synopsis: The risk of adverse outcomes in bacterial meningitis can be estimated upon initial patient presentation using six clinical features.
Source: Weisfelt M, et al. A risk score for unfavorable outcome in adults with bacterial meningitis. Ann Neurol 2008;63:90-97.
Bacterial meningitis is a serious and life-threatening disease. The ability to predict poor outcome in this disease would assist the clinician with risk stratification and potentially with management. The authors of the study used a logistic regression analysis of data from a previously published cohort study examining 696 cases of community-acquired bacterial meningitis to develop a risk score predicting unfavorable outcome (anything less than full recovery with independent function). The authors selected 22 clinical variables to test as potential predictors of poor prognosis in bacterial meningitis. In total, 34% of patients in the prospective cohort suffered a poor outcome. Six of the 22 proposed variables demonstrated significant correlation with poor prognosis (Glasgow Outcome Scale < 1), and were included in the scoring system. These included age, heart rate, Glasgow Coma Scale score, cranial nerve palsy, low CSF white blood cell count (<1000 cells/ mm3), and the presence of gram-positive cocci in CSF gram stain.
The risk score is calculated by adding the values of each subscore for the six determinants. The individual subscores are obtained using a table that assigns values to each of the six clinical determinants. As an example, for cranial nerve (CN) palsy, the subscore is 0 if CN palsy is absent and 9 if it is present. The 6 subscores are added up to obtain the risk score. The risk score is then converted to a percentage risk of unfavorable outcome. For example, a risk score of 30 corresponds to a 40% risk of unfavorable outcome.
After using one cohort (the derivation cohort) to develop the risk score, the authors went further by validating this risk score, using another previously published cohort of patients with bacterial meningitis (the validation cohort). The authors calculated the risk score for each patient and compared the actual outcome to the predicted outcome. The concordance index for the risk score and actual outcome was 0.81, suggesting a robust prediction.
Neurologists rely on risk scores to predict recovery in a number of serious neurological diseases, most notably the Hunt-Hess score for subarachnoid hemorrhage (SAH). Bacterial meningitis is comparable to SAH in that a wide range of outcomes is possible, from full recovery to death. Clinicians should be helped by an easy-to-use scoring system to predict outcome in bacterial meningitis within one hour of presentation to the emergency department. Patients with a higher predicted risk may need to be monitored more closely, with more aggressive treatment for brain edema and vasculitis. Randomized controlled trials are needed to assess whether more aggressive treatment for patients at higher risk might actually lower the expected risk for poor outcome.