Specificity and sensitivity of visual evoked potentials in the diagnosis of schizophrenia: Rethinking VEPs

Abstract

Alterations of the visual evoked potential (VEP) component P1 at the occipital region represent the most extended functional references of early visual dysfunctions in schizophrenia (SZ). However, P1 deficits are not reliable enough to be accepted as standard susceptibility markers for use in clinical psychiatry. We have previously reported a novel approach combining a standard checkerboard pattern-reversal stimulus, spectral resolution VEP, source detection techniques and statistical procedures which allowed the correct classification of all patients as SZ compared to controls. Here, we applied the same statistical approach but to a single surface VEP — in contrast to the complex EEG source analyses in our previous report. P1 and N1 amplitude differences among spectral resolution VEPs from a POz-F3 bipolar montage were computed for each component. The resulting F-values were then Z-transformed. Individual comparisons of each component of P1 and N1 showed that in 72% of patients, their individual Z-score deviated from the normal distribution of controls for at least one of the two components. Crossvalidation against the distribution in the SZ-group improved the detection rate to 93%. In all, six patients were misclassified. Clinical validation yielded striking positive (78.13%) and negative (92.69%) predictive values. The here presented procedure offers a potential clinical screening method for increased susceptibility to SZ which should then be followed by high density electrode array and source detection analyses. The most important aspect of this work is represented by the fact that this diagnostic technique is low-cost and involves equipment that is feasible to use in typical community clinics.