Cognitive reserve : an investigation of construct validity and relationships with cognitive function in generally healthy older adults

Abstract

This thesis evaluated the construct validity of a multiple-indicator, latent variable model of cognitive reserve (CR), and investigated relationships between this construct and age-related cognitive outcomes in a cohort of generally healthy older adults. CR provides a framework to explain epidemiological and other findings of an association between intellectually enriching experiences across the lifespan and reduced risk of Alzheimer’s disease, dementia, and age-related cognitive decline; measures of these intellectually enriching experiences are typically used as proxy indicators of the hypothetical construct CR. However, inconsistent measurement of CR using a range of proxy indicators has attracted criticism, limits comparability of study findings, and raises concerns about construct validity. Latent variable models of CR that combine multiple proxy indicators have been proposed as one potential solution to this problem, but to date these type of models have not been widely used or evaluated. Chapter 1 introduces CR and related theoretical concepts, and provides an overview of the literature related to cognitive ageing, the effects of CR on cognitive function during normal, non-pathological ageing, and associated research issues. A systematic review of the measurement of CR in studies of investigating cognitive outcomes associated with normal, non-pathological ageing is presented in Chapter 2. This systematic review revealed considerable heterogeneity in proxy indicators used to measure CR across studies, and highlighted the need for theoretical and methodological work to promote a more standardised approach to measurement of the construct. In Chapter 3, a series of structural equation models designed to provide progressively more demanding tests of convergent and discriminant validity were employed to evaluate the construct validity of a multiple-indicator, latent variable model of CR. These models provided consistent support for convergent validity of the latent CR construct and at least partial support for discriminant validity, although there was some overlap between CR and executive function constructs. In Chapter 4, cross-sectional models demonstrated significant direct effects of age and CR on executive function, but the interaction between age and CR had no significant effect on this outcome; there were also no significant relationships between these predictors and episodic memory. In moderation analyses of the longitudinal dataset, there were no significant effects of the interaction between age and CR on decline in the cognitive outcome measures over two years. Methods of neuropsychological assessment that focus on intra-individual differences in cognitive performance, rather than comparison with population norms, may be more sensitive for detecting subtle memory deficits in high-performing individuals with high CR; these individuals may continue to perform within the “normal range” while experiencing a decline for them relative to their premorbid function. In Chapter 5, participants were classified into subgroups using two alternative methods based on intra-individual differences, both of which attempt to “benchmark” memory performance relative to IQ. There were significant differences between subgroups for baseline memory performance; subgroups were not found to differ for baseline performance in other cognitive domains, for subjective memory measures, or for decline in cognitive performance over two years. A general discussion of the findings presented in this thesis, their implications in relation to the existing literature, limitations, and directions for future research is presented in Chapter 6.