The validation of auditory event-related potentials in differential diagnosis of dementia

Background. This study is designed to explore whether or not the auditory ERPs can be a clinically effective tool for differential diagnosis in demented patients. Methods. The auditory ERPs were elicited by an auditory oddball paradigm, linked ear reference, in 30 normal aging subjects, 114 patients with dementia, 38 patients with schizophrenia, 69 patients with mood disorder, 29 patients with cerebrovascular disease (CVD), 22 patients with other neurological diseases (such as epilepsy, parkinsonism) and 30 patients with organic mental disorder (OMD). Results. There were no differences in the N100 and P200 among each group. The mean latencies of N200 and P300 for demented patients were significantly longer when compared to those of other groups. The mean latency of N200 for patients with nondemented organic mental disorder (including CVA, OMD and other neurological diseases) was significantly longer when compared to normal aging subjects. To predict the probability of a demented patient, the study groups were separated into two groups (demented vs. nondemented), then using the logistic regression analysis to predict the probability. It showed the age, N200 latency and P300 latency as the most contributing factors, however, the older in age and the longer the N200 and P300 latency, the higher the probability to be a demented patient with an overall 84.64% correct classification. Conclusions. It is concluded that according to the above predict equation, the sensitivity and the specificity are acceptable as an effective tool in clinical diagnosis of dementia.