Abstract
Abstract Background The Hispanic Community Health Study/Study of Latinos (HCHS/SOL) showed that self‐reported cognitive decline (SCD) is associated with objectively assessed cognition. In HCHS/SOL, sleep phenotypes using empirically derived latent class/cluster analyses are associated with brain vascular outcomes. Since these measures may predict cognitive decline and dementia, we examined the associations between sleep phenotypes, SCD and mild cognitive impairment (MCI) among a diverse sample of Hispanics/Latinos living in the United States. Method HCHS/SOL is prospective cohort study designed using a multisite (Bronx, NY, Chicago, IL, Miami, FL, San Diego, CA) multistage probability sample. SCD was measured as part of SOL‐INCA, an HCHS/SOL ancillary study. SOL‐INCA obtained repeated cognitive tests, seven‐years apart and the Everyday Cognition Short Form (ECog‐12). The ECog‐12 is a self‐rated questionnaire of current multiple cognitively relevant functional abilities compared to 10‐years ago administered to 5,563 SOL‐INCA participants aged ≥ 50 years. We used latent class analysis using 15 common sleep related symptoms, home sleep apnea test and four comorbidities to derive three sleep phenotypes: (1) Minimally Symptomatic group (49.6%), (2) Disturbed Sleep group (17.1%), and (3) Daytime Sleepiness group (33.3%). We used survey regression to evaluate associations between visit‐1 sleep phenotypes and the ECog‐12 (linear) cognitive domains (memory, language, executive function, and global) and MCI (logistic) using the minimally symptomatic sleep group as the reference. Models were adjusted for age, education, sex, background, field center, insurance, body mass index, cigarette use, alcohol use, cardiovascular disease, language preference, and respiratory event index. Result Average age was 64.34±8.25 years and 55% were female. In fully adjusted models, participants with disturbed sleep had worse global cognition β = 0.18 [0.14;0.22], memory β = 0.21 [0.16;0.26], language β = 0.23 [0.17;0.29], and executive function β = 0.16 [0.12;0.21]. Compared to the reference, we observed stronger associations in participants with daytime sleepiness; global cognition β = 0.22 [0.16;0.27], memory β = 0.26 [0.19;0.32], language β = 0.26 [0.19;0.34], and executive function β = 0.20 [0.15;0.26]. Adults with disturbed sleep, OR = 1.44 [1.08;1.93], and daytime sleepiness OR = 1.67 [1.16;2.40] had higher odds of MCI. Conclusion Empirically derived sleep phenotypes capturing disturbed sleep and sleepiness had associations with self‐reported cognitive function and MCI in this sample of U.S. Hispanic/Latino adults.