Clinical predictors of Japanese encephalitis.

Over a 5-year period, virological investigations for Japanese encephalitis (JE) were conducted in children presenting with acute encephalopathic illness. Clinical features of JE-positive patients (n = 116) were compared with patients in whom the diagnosis could be excluded (n = 57). Multivariate analysis by logistic regression revealed that two clinical signs--central hyperpneic breathing pattern and extrapyramidal signs--were significant predictors of the diagnosis. Application of the model yielded a sensitivity of 41.3% and a specificity of 80.7% with positive and negative predictive values of 81.3 and 40.3%, respectively. This indicates that the model may be helpful in making the diagnosis but not in excluding it. The model should be further validated in different areas where the disease is prevalent.