Application of multivariate statistics to vestibular testing: Discriminating between Meniere's disease and migraine associated dizziness

Meniere's disease (MD) and migraine associated dizziness (MAD) are two disorders that can have similar symptomatologies, but differ vastly in treatment. Vestibular testing is sometimes used to help differentiate between these disorders, but the inefficiency of a human interpreter analyzing a multitude of variables independently decreases its utility. Our hypothesis was that we could objectively discriminate between patients with MD and those with MAD using select variables from the vestibular test battery. Sinusoldal harmonic acceleration test variables were reduced to three vestibuto-ocular reflex physiologic parameters: gain, time constant, and asymmetry. A combination of these parameters plus a measurement of reduced vestibular response from caloric testing allowed us to achieve a joint classification rate of 91%, by use of a multivariate independent quadratic classification algorithm. Data from posturography were not useful for this type of differentiation. Overall, our classification function can be used as an unbiased assistant to discriminate between MD and MAD and gave us insight into the pathophysiologic differences between the two disorders.