Individual variation for cognitive decline: quantitative methods for describing patterns of change.

What are the best quantitative methods for studying cognitive decline? This question was investigated in a sample of 638 individuals aged 50 years and older from the Swedish Adoption/Twin Study of Aging. A battery of cognitive tests tapping multiple domains was administered to each individual from 2 to 7 times over a span of 10 years. Four methods of operationalizing cognitive decline were compared: change scores, a criterion-based method, least squares, and random effects regression (RER). The RER results were most consistent with a significant decline across measures and differences between demented and nondemented individuals. Predicted slopes from the RER model also showed the strongest interrelationships within and across cognitive domains as indicated by factor analysis results and stronger associations with demographic, health, and psychosocial predictors.