Commingling analysis of memory performance in elderly men.

Smalley et al. [(1992) Genet Epidemiol 9:333-345] found evidence of a mixture of two distributions in memory performance among offspring of patients with dementia of the Alzheimer type (DAT), suggesting that these groups reflect genotypic subgroups of carriers and non-carriers of a putative DAT gene. One prediction of this hypothesis is that, in the general population, two distributions of memory performance are present, with a smaller proportion of subjects in the low-scoring cluster than that found among the offspring sample, but similar to the prevalence of DAT in the elderly community-at-large. Memory performance was investigated in a large sample of normal elderly males (N = 1,149; mean age = 71.4 +/- 4.7 years) to test this hypothesis. Commingling analyses of performance on the Benton visual retention test demonstrated significant negative skewness in the distribution of memory performance, requiring a transformation to fit a single normal distribution. In the absence of a transformation, two distributions fit better than one, with 6% of subjects falling into a "low"-scoring cluster. These findings are consistent with the hypothesis that memory performance may represent a premorbid or morbid difference in those who go on to develop, or currently have, DAT, possibly allowing identification of at-risk carriers of a putative single major gene for DAT.