Absolute and relative reliability of several response parameters used in vestibular assessment

This experiment investigated the reliability of five indices used to express the magnitude of induced vestibular nystagmus. The investigation was undertaken because information concerning the reliability of the various response parameters when caloric testing is repeated on the same subject is limited. Electronystagmographic assessments of the nystagmus response to the Fitzgerald-Hallpike test were obtained from 16 subjects on three occasions with equal intervals of time between occasions. The data were statistically treated with an analysis of variance technique that allows the generation of correlation coefficients. The findings showed that speed of the slow phase, total beats, culmination frequency, and total amplitude reliably express the magnitude of induced nystagmus on repeated tests. Duration of the induced reaction was not as reliable as the other indices on repeated testing. In addition, difference scores were more repeatable than absolute scores and the caution that must be maintained when using correlation coefficient analyses to estimate reliability is highlighted.