Abstract
OBJECTIVES: Sleep disturbances can impair cognition, and cognitive impairment can dysregulate sleep-wake cycles. Despite biological plausibility, disparities in the sleep-cognition relationship are understudied in nationally representative samples. METHODS: Using National Health Interview Survey data from 2011-2018, we investigated cross-sectional associations between sleep and cognition among adults aged ≥40 years overall and by age, sex, race, and ethnicity. Participants self-reported sleep duration (short [<7-hours], long [>9-hours]); insomnia symptoms (trouble falling asleep or staying asleep ≥3 times/week); and non-restorative sleep (waking feeling unrested ≥4 days/week). Cognitive impairment was defined as "not being able to/having a lot of difficulty remembering or concentrating" based on the Washington Group Short Set on Functioning. Adjusted Poisson regression with robust variance estimated prevalence ratios (PRs) and 95% confidence intervals. RESULTS: Among 73,477 adults (mean age ± standard error 58.3 ± 0.07 years), 31.8% reported short sleep, 4.1% long sleep, 35.7% insomnia symptoms, and 35.5% non-restorative sleep. Overall, 2.2% reported cognitive impairment. All sleep dimensions were associated with a higher prevalence of cognitive impairment (range: PR(short-sleep) = 1.23[95% confidence interval:1.07-1.42] to PR(long-sleep) = 4.10[3.48-4.83]). Short sleep, insomnia symptom, and non-restorative sleep were more strongly associated with cognitive impairment among middle-aged adults. Associations with long sleep were stronger among older adults. Differences by sex were not observed. While the prevalence of cognitive impairment was slightly higher among Hispanic/Latine and Black adults, associations with short sleep were strongest among White adults and among Asian adults for long sleep (p-interaction<0.05). CONCLUSIONS: Sleep disturbances were associated with cognitive impairment. Most associations were stronger among middle-aged adults, although future studies with objective measures are warranted.