Lower amplitude of the 24-hour rest-activity rhythm has been linked to cognitive decline and incident dementia in older adults. This cross-sectional analysis examined whether wrist accelerometer–derived relative amplitude is associated with domain-specific cognitive performance during midlife. The specific objective was to evaluate relationships between relative amplitude and both fluid and crystallized cognition among adults aged 35–64 enrolled in the MidCog Study.
The analysis included 851 participants aged 35–64 recruited from 13 primary care clinics, including academic practices and federally qualified health centers. Each participant completed a 14-day wrist accelerometer recording and the NIH Toolbox Cognition Battery. Institutional review board approval was obtained from Northwestern University Feinberg School of Medicine and all participants provided written informed consent.
Relative amplitude was calculated from accelerometer data as the contrast between the most active consecutive 10-hour period and the least active consecutive 5-hour period during the 14-day recording. This metric quantifies the day–night contrast in activity and is used here as a summary measure of rest-activity rhythm amplitude derived from wrist-worn accelerometry.
Cognition was measured with the NIH Toolbox Cognition Battery. Outcomes were expressed as age- and education-adjusted T-scores (normative mean = 50, SD = 10). Primary cognitive outcomes included the Fluid Cognition Composite and its component domains: attention, executive function, episodic memory, working memory, and processing speed. Crystallized cognition outcomes included the Crystallized Cognition Composite, receptive language, and expressive language.
Multivariable linear regression models assessed associations between relative amplitude and cognitive T-scores. Models were built sequentially and adjusted for sociodemographic characteristics, other rest-activity rhythm metrics, established dementia risk factors, average activity level, time in bed, and sleep efficiency. Reported effect estimates correspond to each 1–standard deviation (SD) lower relative amplitude.
In the fully adjusted model, each 1-SD lower relative amplitude was associated with a 2.27-point lower Fluid Cognition Composite T-score (95% CI, 1.15–3.40). This association persisted after controlling for a range of sociodemographic variables, other rest-activity rhythm metrics, dementia risk factors, and sleep characteristics.
Lower relative amplitude was associated with several fluid cognition domains in adjusted analyses. Specifically, a 1-SD lower relative amplitude was associated with:
These estimates indicate modest but statistically significant associations between reduced rest-activity amplitude and worse performance in several fluid cognitive domains.
Relative amplitude was not associated with working memory or with crystallized cognitive outcomes (the Crystallized Cognition Composite and language measures) in the adjusted analyses reported.
Among midlife adults recruited from primary care practices, lower accelerometer-derived relative amplitude was linked to modestly poorer fluid cognitive performance after adjustment for established dementia risk factors and sleep characteristics. The authors propose that relative amplitude may serve as a scalable behavioral correlate of cognitive variability within the normative range in midlife. Because associations were domain-specific—affecting attention, executive function, episodic memory, and processing speed—relative amplitude may reflect aspects of physiology or behavior relevant to fluid cognitive abilities.
As a cross-sectional study, these analyses cannot establish temporality or causality. The authors explicitly state that longitudinal studies beginning in midlife are needed to determine whether relative amplitude predicts future cognitive decline, to assess its predictive value, and to evaluate whether it is modifiable and amenable to interventions.
The Northwestern University Feinberg School of Medicine Institutional Review Board approved the study (STU00214736, STU00221343). All participants provided written informed consent. De-identified individual participant data will be made available to qualified researchers upon reasonable request. Funding sources included awards from the National Institute on Aging and related grants listed in the manuscript. Competing interests and funding declarations are reported by the authors in the source.