Abstract
Neuropsychological data from 134 patients diagnosed with probable Alzheimer's disease (AD) were studied retrospectively to investigate whether subgroups of patients with qualitatively distinct profiles could be identified. Three empirical classification approaches were undertaken in this regard: Q-type factor analysis, hierarchical agglomerative cluster analysis, and iterative partitioning. Three subgroups were consistently identified across the clustering methods. Subgroup 1, the largest of the groups, was marked by moderate to severe anomia and constructional dyspraxia. Individuals in subgroup 2 displayed relatively spared visual-perceptual/constructional functioning but severe anomia. Members of subgroup 3 exhibited intact naming and nonverbal reasoning and moderate difficulty in copying overlapping figures. The three subgroups did not differ with respect to age, age at disease onset, duration of illness, educational level, or Hamilton depression rating. Detailed description of the data analyses are provided as a tutorial outlining subtyping methodology. Results are discussed in terms of the subgroup and the stage model approaches to the conceptualization of AD.