Abstract
INTRODUCTION: Evidence suggests that engaging in a high volume of daily or weekly physical activity may reduce the risk of cognitive impairment due to Alzheimer's disease and related dementias (ADRD). However, activity variability and its associations with cognition are underexplored. We examined whether activity variability, measured at the minute level, may capture movement patterns that provide insight into cognitive impairment risk. METHODS: Using a cross-sectional sample of 711 older adults (mean [standard deviation] age = 79.1 [6.0] years]) from the National Health and Aging Trends Study in 2021, we calculated a novel metric of activity variability, defined as the standard deviation of minute-to-minute changes in accelerometer activity counts across the 7 day wear period. We investigated associations between activity variability and other activity metrics with cognitive impairment using logistic regression models. RESULTS: Lower activity variability was associated with increased odds of cognitive impairment (odds ratio [OR] = 2.23; 95% confidence interval [CI: 1.26-3.97]), even after accounting for total activity counts and other covariates. In receiver operating characteristic curve analyses, activity variability was the strongest predictor of prevalent cognitive impairment compared to other activity metrics (area under the curve = 0.787; 95% CI [0.742-0.832]). DISCUSSION: Activity variability, measured at the minute level, may capture complexity of daily activities and may serve as a novel indicator of cognitive health. Low activity variability may signal underlying neurological processes that contribute to ADRD risk. Further investigation into potential biological mechanisms that may explain the observed associations between activity variability and cognition is warranted. HIGHLIGHTS: We created a novel metric using variability of minute-level changes in activity counts.Low activity variability was associated with a greater prevalence of cognitive impairment.Activity variability outperformed other activity metrics in area under the receiver operating characteristic curve analyses.Future work should examine underlying neurological mechanisms.