Combinations of evoked potential amplitude measurements in relation to psychiatric diagnosis.

The purposes of this investigation were (1) to determine the extent to which single nonredundant measures, derived from optimal combinations of evoked potential (EP) amplitude measurements, could differentiate between various groups of psychiatric patients and between patients and nonpatients, and (2) to assess the replicability of such discriminations. Somatosensory, visual, and auditory EPs were recorded from 15 locations in 253 unmedicated patients and 99 nonpatients. Multivariate statistical methods were used to reduce the amplitude measurements to sets of factor scores that met specified criteria for entry into discriminant analyses between pairs of the following groups: nonpatients, neuroses, personality disorders, schizophrenias, schizotypal/borderlines, major depressives, and manics. Sensitivity, specificity, and predictive value were assessed for discriminant scores. The discriminant scores yielded many differences between groups, most of which were replicable in split-half analyses. Among these were differences between all six patient groups and nonpatients, and between schizophrenics and nonpsychotics. Split-half discriminant analyses were also performed with 12 measures that were obtained by taking the means of factor scores grouped by sensory modality, time, and spatial location; these variables, used without preselection, provided several replicable diagnostic discriminations.