Modelling the developmental trajectories of cognitive reserve in children

AbstractBackgroundCognitive reserve (CR) is thought to protect cognitive ability against neurodegenerative conditions, such as Alzheimer’s disease, or other neurological insults. Because past research has studied the construct extensively in older adult populations, little is known about its development across the lifespan. The current study generated a dynamic CR index based on a modified residual approach utilised successfully in older adults (Reed et al., 2010; Bettcher et al., 2019), and examined its criterion and construct validity in a paediatric cohort comprising children both with and without ADHD.MethodArchived longitudinal data collected across 2011‐2019 from N = 115 children (MAge = 10.48, SDAge = 0.61, n = 72 [63%] male, n = 43 [37%] with ADHD) was obtained from the Murdoch Children’s Research Institute. Static and dynamic CR were operationalised based on residual variance in (change in) fluid intellect (Matrix Reasoning) not explained by demographics (i.e., sex and age) or MRI‐based brain variables (i.e., grey matter volumes, white matter hypointensities [WMH], and hippocampal volumes). This index was subsequently used to predict relevant outcome variables (i.e., sustained attention, working memory, inhibition, word reading, and mathematical computation) to assess criterion validity at baseline and across time, while construct validity was assessed via interaction effects between CR, brain, and baseline/longitudinal outcomes.ResultsApproximately 8% of the variance in fluid intellect was predicted by selected brain and demographics at baseline, with 14% predicted in the longitudinal model. CR at baseline predicted mathematical computation (p<.001), word reading (p = .012), and working memory ability (p<.001) at baseline. CR change only predicted change in word reading ability (p<.001); it was also found to moderate relationships between WMH volume change and word reading trajectory, and between WMH volume change and change in sustained attention performance.ConclusionOur results suggest that CR can be modelled in children using a residual index, but with less validity than in older adults; weak criterion and construct validity were evident, although this should be tempered by the fact that there was minimal variance explained in the residualised variable. An adjusted model optimised for a paediatric cohort will form the basis of further inquiry into the study of CR in a developmental context.