Comprehensive predictions of outcome in closed head-injured patients. The development of prognostic equations

Abstract

A comprehensive diagnostic evaluation was administered to 162 closed head-injured patients within 1 to 21 days (mean, 7.5 days) after injury. Each evaluation consisted of (1) power spectral analyses of electroencephalogram (EEG) recorded from 19 scalp locations referenced to age-matched norms, (2) brainstem auditory evoked potentials, (3) computed tomography (CT)-scan, and (4) Glasgow Coma Score (GCS) at time of admission (GCS-A) and at time of EEG test (GCS-T). Functional outcome at one year following injury was assessed using the Rappaport Disability Rating Scale (DRS), which measures the level of disability in the six diagnostic categories of (1) eye opening, (2) best verbal response, (3) best motor response, (4) self-care ability for feeding, grooming, and toileting, (5) level of cognitive functioning, and (6) employability. The ability of the different diagnostic measures to predict outcome at one year following injury was assessed using stepwise discriminant analyses to identify patients in the extreme outcome categories of complete recovery versus death and multivariate regression analyses to predict patients with intermediate outcome scores. The best combination of predictor variables was EEG and GCS-T, which accounted for 74.6% of the variance in the multivariate regression analysis of intermediate outcome scores and 95.8% discriminant accuracy between good outcome and death. The best single predictors of outcome in both the discriminant analyses and the regression analyses were EEG coherence and phase. A gradient of prognostic strength of diagnostic measures was EEG phase greater than EEG coherence greater than GCS-T greater than CT-scan greater than EEG relative power. The value of EEG coherence and phase in the assessment of diffuse axonal injury was discussed.