**Background:** Circadian rhythm disruption is increasingly recognized as a contributor to cognitive decline and neurodegenerative disease. Rest-activity rhythm (RAR) is a key output of the circadian system, but its relationship with specific cognitive domains in older U.S. adults was not well characterized. This study aimed to examine associations between accelerometer-derived RAR parameters and cognitive function in a nationally representative sample.
**Methods:** Data were drawn from the National Health and Nutrition Examination Survey (NHANES) 2011-2014 cycles. The analytic sample included 2,090 adults aged ≥60 years with valid accelerometer data (≥4 days) and cognitive assessments. RAR parameters were calculated using the R package 'nparACT' and included: interdaily stability (IS; range 0-1, higher = more stable day-to-day rhythm), intradaily variability (IV; range 0-2, higher = more fragmented rhythm), relative amplitude (RA; higher = stronger 24-hour oscillation), L5 (mean activity during least active 5 hours), and M10 (mean activity during most active 10 hours). Cognitive function was assessed using three tests: the Consortium to Establish a Registry for Alzheimer's Disease Word List (CERAD W-L; memory, score 0-40), the Animal Fluency Test (AFT; executive function, score = number of animals named in 1 minute), and the Digital Symbol Substitution Test (DSST; processing speed and global cognition, score 0-133). Linear regression models were built with progressive adjustment: Model 1 adjusted for age, sex, and race; Model 2 added BMI, income, education, sleep duration, daily energy intake, exercise, smoking, and drinking; Model 3 further added diabetes, hypertension, hyperlipidemia, and related medication use. Stratified analyses were conducted by age (60-69 vs. ≥70), sex, BMI (<25, 25-29.9, ≥30 kg/m²), and race. Sensitivity analyses excluded participants with less than 9th-grade education, those with peak activity between 23:00-04:00, and those with ≤6 hours sleep.
**Key Results:** The mean age was 69.0 ± 6.7 years; 47.4% were male; 47.1% were non-Hispanic white; mean BMI was 29.11 ± 6.24 kg/m². Mean RAR parameters were: IS 0.52 ± 0.13, IV 0.68 ± 0.22, RA 0.83 ± 0.12, L5 1.09 ± 0.84, M10 11.80 ± 3.70. Mean cognitive scores were: CERAD W-L 25.36 ± 6.48, AFT 16.99 ± 5.41, DSST 47.45 ± 17.25. In fully adjusted models (Model 3): Lower IS was associated with lower CERAD W-L scores (β=2.221, 95%CI: 0.175 to 4.246, P=0.033) and lower DSST scores (β=6.154, 95%CI: 0.999 to 11.308, P=0.019). Lower M10 was associated with lower CERAD W-L scores (β=0.121, 95%CI: 0.044 to 0.199, P=0.002). Weaker RA was associated with lower AFT scores (β=3.406, 95%CI: 1.434 to 5.378, P<0.001) and lower DSST scores (β=20.449, 95%CI: 14.676 to 26.222, P<0.001). Higher L5 was associated with lower AFT scores (β=-0.448, 95%CI: -0.723 to -0.174, P=0.001) and lower DSST scores (β=-2.361, 95%CI: -3.167 to -1.554, P<0.001). IV showed no significant associations with any cognitive test. Sensitivity analyses largely confirmed the main findings, though the association between IS and cognitive scores was less robust in some exclusions.
**Clinical Implications:** This study provides evidence that objectively measured circadian rhythm disruption is associated with poorer cognitive function across memory, executive function, and processing speed domains in older U.S. adults. The findings suggest that RAR parameters—particularly RA, L5, and M10—may serve as early markers of cognitive decline. The authors note that low-risk behavioral interventions (e.g., bright light therapy, consistent sleep-wake schedules, daytime activity) could potentially strengthen circadian rhythms and improve cognitive outcomes, though causal relationships cannot be established due to the cross-sectional design. Limitations include the inability to infer causality, potential residual confounding, and lack of shift work data. Longitudinal and interventional studies are needed to confirm these associations and evaluate the therapeutic potential of circadian rhythm stabilization.