Establishing the limits and characteristics of normal age-related cognitive decline

Research into ageing-related pathology relies not only on exploration of disease aetiology, but also a clear understanding of the normal ageing process. The present study aims to examine the characteristics of elderly subjects who lie on the borderline between normal and pathological ageing.

Cognitive functioning is examined using computerized neuropsychometric assessment in a population of 833 normal elderly from which a cohort of 397 subjects with sub-clinical cognitive impairment are followed over three years. Subjects receive a standardized neurological examination and ApoE genotypes are established.

Analysis of covariance revealed no cross-sectional age differences for syntax comprehension (p = 0.19), articulation (p = 0.46), semantic matching (p = 0.12), reading (p = 0.79), and implicit memory (p = 0.21) while explicit memory, language skills and visuospatial skills were found to deteriorate both in the cross-sectional age comparisons and across time. An overall intellectual ability factor, derived from Principal Components Analysis, was found by regression to decline principally in persons with low education, and a high initial IQ level was observed to provide a protective effect over age 75. Persons with higher levels of education show relative stability over time on language and secondary memory tasks but deteriorate as rapidly as persons with low education on visuospatial tasks. Five separate patterns of sub-clinical cognitive deficit were isolated by cluster analysis. Two groups, with differing clinical profiles (of which only one manifested the ApoE4 allele), were found to have an increased risk of developing senile dementia (OR = 4.4 and 3.9). A third group had a high prevalence of depressive illness, and the remaining two showed very little change.

Ageing does not affect all cognitive functions uniformly: a high initial education level slowing rate of decline for certain tasks. Separate patterns of cognitive change are observed in early senile dementia, benign change and changes related to depressive illness. Results suggest the need for more stringent selection of normal control groups.