The use of amnestic and nonamnestic composite measures at different thresholds in the neuropsychological diagnosis of MCI.

Both amnestic and nonamnestic deficits have been observed in patients with mild cognitive impairment (MCI). Most studies have focused on impairment on single cognitive tests rather than amalgamation of the results of several measures to arrive at a composite impairment index. In this investigation, we examined 20 MCI patients diagnosed as prodromal Alzheimer's disease, AD (mean Mini-Mental State Examination, MMSE = 26.1; SD = 1.7) and determined the extent to which they could be differentiated from 70 normal elderly controls based on composite measures at 1.5-SD and 2.0-SD cutoffs for impairment. At the 1.5-SD cutoff, the median number of memory indices impaired in the MCI-AD group was 5 of 7, whereas at the 2.0-SD cutoff, the median number was 4 of 7. A median of 3 of 7 and 2 of 7 nonmemory indices were impaired at 1.5- and 2.0-SD cutoffs for impairment. Receiver operator characteristics (ROC) analyses indicated that the total number of memory tests impaired at 2.0 SD (sensitivity = 95.0%/specificity = 84.3%) and the composite measure of both impaired memory and nonmemory measures (sensitivity of 85.0%/specificity of 100%) had high levels of discrimination and may have utility as indices of early impairment as well as severity of MCI.