Validation of composite proxy measures of cognitive reserve

Abstract

AbstractBackgroundCognitive reserve (CR) enables individuals to maintain cognitive function despite the presence of brain pathology or atrophy [1]. CR is typically measured via proxy variables that reflect exposure to lifetime experiences, such as education [2]. Composite measures of multiple proxies combine contributions from several experiences and display consistent positive relationships with cognitive function [3]. However, the optimal set of proxies for inclusion in composite measures remains to be determined [2]. Here, we aimed to identify a combination of proxies that best reflects the ability to maintain cognitive function despite cortical thinning.MethodAs part of a larger analysis, 277 healthy adult participants provided data for this study (mean age = 69.18, SD age = 6.66, 141 females). Global cognition was defined as average of z‐scores from the Animal Naming task, Colour Trails Test 1 and 2, Immediate and Delayed Recall. Mean cortical thickness was calculated using FreeSurfer using a previously reported method [4]. CR proxies were: educational attainment, occupational complexity, premorbid intelligence, social engagement, cognitively stimulating activity, leisure activity, and physical activity. For each composite, a three‐step hierarchical‐multiple regression was performed to predict global cognition. Independent variables in Step 1 were age, gender, cortical thickness, CR composite was added in Step 2, and an interaction term (cortical thickness*composite) was further added in Step 3. Positive associations between CR composites and global cognition – independent of age, gender and cortical thickness – were ranked on significant R‐squared change values from Step 1 to Step 2. Moderation effects were identified by significant R‐squared change values from Step 2 to Step 3.ResultMost CR composites had positive associations with global cognition, independent of age, gender, and cortical thickness (see Figure 1). A CR composite measuring cognitively stimulating activity, leisure activity, and physical activity had the biggest negative moderation effect (R2 change = 0.013, p = 0.0276) on the cortical thickness‐global cognition relationship.ConclusionA composite measure of cognitively stimulating‐, leisure‐, and physical‐activity, displays construct validity as a measure of CR in addition to a wide range of composites which display positive, independent, associations with cognitive function.