Longitudinal associations between physical activity and cognition post-stroke: the Nor-COAST study

Previous research has been inconsistent regarding the longitudinal relationship between physical activity and cognition after a stroke. In this study, we hypothesized that (i) higher levels of objectively measured daily step counts would be associated with better cognitive function at a group level (between-person level), and (ii) at an individual level, changes from expected step counts would be bi-directionally associated with changes from expected cognitive performance (within-person level) at the same and subsequent time points. We used data from 468 participants of the Norwegian Cognitive Impairment after Stroke study (Nor-COAST), a large multicentre cohort study with follow-up at 3, 18 and 36 months post-stroke. At each follow-up, average step count over 7 days was recorded using accelerometers, and cognitive performance was assessed with the Montreal Cognitive Assessment (MoCA), the Trail Making Test A and B (TMT-A, TMT-B), and the Word List Memory and Recall Test (delayed recall). Adjusted random intercept cross-lagged panel models (RI-CLPMs) were implemented. One RI-CLPM was built for each cognitive outcome, separating between- from within-person associations. To balance baseline characteristics of included and excluded participants in Nor-COAST and to account for selection bias, inverse probability weighting was performed using age, sex, and baseline functional independence. Between-persons, we found a time-invariant association between step count and TMT-A (β = 0.24, 95% CI: 0.09 to 0.40) and TMT-B (β = 0.22, 95% CI: 0.09 to 0.34); after correcting for multiple comparison the association with MoCA was not significant (β = 0.14, 95% CI: 0.01 to 0.28), and there was no association with the Word List Memory and Recall Test (β = 0.09, 95% CI: -0.04 to 0.22). At the within-person level, we observed neither cross-sectional nor cross-lagged associations between step count and any of the cognitive tests. The RI-CLPM fitted the data appropriately and had robust testing models with and without inverse probability weighting. After a stroke, people who, on average, took more steps tended to also perform better on TMT-A and TMT-B scores. However, at an individual level, a person who walked more (or had better cognitive scores) than expected did not have a higher likelihood of also performing better on cognitive tests (or walking more) than their expected average, respectively. Therefore, the observed association between step count and cognitive function after stroke might be explained by consistent individual differences over time, rather than by temporary changes within an individual's typical level of physical activity or cognition.