[Reliability of parameters of cognitive evoked P3 potentials]

Abstract

Reliability of event-related P3 parameters. Various parameters of the P3 wave were tested for reliability and test-retest stability. To this end, event-related P3 waves were evoked in 21 subjects during visual and auditory oddball tasks. The protocol was repeated one to two weeks later. Amplitude, latency and area between P3 and baseline were estimated by different procedures. The stability within and between sessions was assessed with the split-half method and a measure of test-retest reliability, respectively. A literature review revealed that latency estimation is most commonly done by computer-aided peak measurement, but determining peak parameters visually after interpellating superimposed artifacts turned out to be significantly more reliable. Interrater reliability for this visual inspection was very high if preestablished rules were employed. Amplitude and area were measured reliably by both computer-aided procedures and visual inspection. The area was quite stable both within and between sessions. Area can serve as a measure of cerebral activation and should therefore be determined. Based on the results recommendations are made about analyzing components of event-related potentials. The need to control for error variance is emphasized. This is the only way to improve methods enough that they will be suitable for use in single case diagnosis.