Multivariate Analysis of Patients with Head Injury Using Quantification Theory Type II

—With Special Reference to Prediction of Patient Outcome—

Abstract

Quantification theory type II was used to predict the outcome of 63 patients with head injury. Age, sex, two factors based on neurological examination, and seven factors based on findings of skull radiography and computed tomography of the head were selected as predictors. Patient outcome was evaluated 6 months after injury and assigned to good recovery, severe disability, or death. Discriminant analysis of patient outcomes was performed using the 11 factors. Two category scores were obtained for each category, since the highest correlation ratios were 0.869 and 0.252, and others were less than 10^-15. For each patient, a pair of sample scores was then obtained by simple summation of the 11 category scores. Pairs of sample scores were plotted, and the three groups of patient outcomes were clearly distinguishable without exception. These findings show that the outcome of patients with head injury can be accurately discriminated by quantitative analysis of qualitative data.