The U-Shaped Association between Sleep Duration, All-Cause Mortality and Cardiovascular Risk in a Hispanic/Latino Clinically Based Cohort

The study investigates whether self-reported sleep duration (total sleep time, TST) is associated with all-cause mortality and 10-year cardiovascular disease (CVD) risk in adults evaluated at a sleep medicine center in Santiago, Chile. The context is grounded in evidence that both short (<7 h) and long (≥9–10 h) sleep durations relate to adverse health outcomes, including cardiometabolic disease and higher mortality. Professional societies (AHA, AASM) recognize sleep as integral to cardiovascular health. Prior research shows U-shaped relationships between sleep duration and mortality/CVD risk, but evidence is scarce in Hispanic/Latino populations residing in their countries of origin. Primary objective: assess association between TST and incident all-cause mortality. Secondary objective: assess relationship between TST and 10-year CVD risk.

Consensus recommendations indicate 7–8 h of sleep for optimal adult health (AASM), with short and long sleep linked to cardiometabolic disease, depression, and increased mortality. OSA is associated with hypertension, cardiovascular compromise, and increased mortality. The AHA has emphasized sleep health’s role in cardiovascular risk. Studies in Hispanic/Latino populations in North America link sleep timing/regularity and sleep patterns to metabolic health, obesity, diabetes, and blood pressure. Large cohort studies (e.g., Ford 2014; Fan 2020; Zhong 2023) report increased CVD risk with short or prolonged sleep versus 7–8 h. Evidence suggests mechanisms including inflammation, autonomic dysregulation, endothelial dysfunction, and socioeconomic/psychiatric factors contributing to adverse outcomes with abnormal sleep duration.

Design: Secondary analysis of a prospective clinically based cohort (SantOSA; ISRCTN62293645) established in 2009 at a tertiary sleep center in urban Santiago, Chile. Adults ≥18 years referred for clinical evaluation of OSA were included; all missing data were excluded. Ethics: Approved by Clinica Las Condes IRB (protocol 10; 9 March 2018); informed consent obtained. Baseline assessments: Standardized questionnaire captured sleep schedule, daytime sleepiness, snoring, witnessed apnea, insomnia, nocturnal suffocation, morning headache; Epworth Sleepiness Scale (ESS, Spanish version). Anthropometrics measured (height, weight) to compute BMI (kg/m²). Sleep testing: Home sleep apnea test (HSAT), type 3 (Embletta MPR, Natus), measuring nasal pressure (airflow), thoracic/abdominal inductance plethysmography, body position, microphone audio, pulse oximetry. Manual scoring by a blinded respiratory specialist per AASM guidelines: apnea = absence of airflow ≥10 s; hypopnea = ≥30% airflow reduction with ≥3% SpO2 drop. Extracted respiratory disturbance index (RDI), time below 90% oxygen saturation (T90%). OSA severity categorized by RDI (non-OSA <5; increasing categories up to severe ≥30 events/h). Exposure: Self-reported TST via the question, “During the past month, how many hours of actual sleep time did you get at night?” TST categorized as short (≤6 h), normal (6–9 h), and long (≥9 h). Outcomes: Primary—incident all-cause mortality determined via Chilean civil registry death certificates (last accessed 26 April 2022). Secondary—10-year CVD risk calculated using the Framingham 2008 general risk profile via R package CVRisk (non-laboratory estimation). Statistical analysis: Continuous variables summarized as medians (IQR); categorical as frequencies (%). Group differences via chi-square (categorical) and ANOVA (continuous). Cox proportional hazards models assessed associations between TST categories and all-cause mortality: Model 1 unadjusted; Model 2 adjusted for baseline CVD risk factors (age, gender, BMI, diabetes mellitus, smoking status, systolic blood pressure, blood pressure medication); Model 3 additionally adjusted for rate of respiratory events. Proportional hazards assumption tested with Schoenfeld residuals scaled against transformed time. For the CVD risk relationship, multivariable linear regression and a third-degree polynomial multiple linear regression were used to evaluate and visualize a U-shaped association between TST and Framingham 10-year CVD risk. Significance threshold p < 0.05. Software: R.

Sample: n = 1385; 78% male; median age 53 years (IQR 42–64); median BMI 29.5 kg/m² (IQR 26.7–33.1); median ESS 8 (IQR 5–12); median TST 6.5 h (IQR 6–7.5); median AHI 18.7 events/h (IQR 8.7–35.6); median T90% 3.6% (IQR 0.4–17.6); median Framingham 10-year CVD risk 13.6% (IQR 6.7–27.2). TST categories: short ≤6 h (n = 632), normal 6–9 h (n = 639), long ≥9 h (n = 114). Follow-up: median 6.6 years (IQR 5.1–9.4); deaths = 146 (10.5%). Cause of death distribution: CVD 26.4%, cancer 19.4%, other 54.0%. Mortality associations (Cox models; reference = normal TST): Model 3 (fully adjusted for baseline CVD risk factors + respiratory event rate)—short TST HR 2.51 (95% CI 1.48–4.25), p = 0.01; long TST HR 3.97 (95% CI 1.53–10.29), p = 0.04. Proportional hazards assumption not violated (p = 0.88). Kaplan–Meier curves demonstrated lower survival for short and long TST versus normal. CVD risk: Framingham 10-year risk medians—short 14.7% (IQR 7.44–27.1), normal 11.8–12.8% (reported IQR ~6.26–27.1), long 23.4% (IQR 8.11–30.0). Third-degree polynomial regression showed a U-shaped association between TST and 10-year CVD risk, with increased risk at both short and long sleep durations.

Self-reported short (≤6 h) and long (≥9 h) sleep durations were independently associated with higher all-cause mortality and increased 10-year CVD risk compared with normal (6–9 h) sleep duration, addressing the primary and secondary research questions in a Hispanic/Latino clinically based cohort. These relationships persisted after adjustment for baseline cardiovascular risk factors and respiratory event rates (OSA severity), indicating that sleep duration exerts an effect beyond OSA-related pathology. Findings align with prior literature demonstrating U-shaped associations between sleep duration and adverse outcomes. Potential mechanisms include autonomic dysregulation, inflammation, endothelial dysfunction, sleep fragmentation, and socioeconomic/psychiatric factors influencing sleep patterns. Clinically, recognizing abnormal sleep duration may inform risk stratification and motivate comprehensive interventions (sleep hygiene, management of sleep disorders, lifestyle modification) to mitigate cardiovascular and mortality risk.

In this Chilean Hispanic/Latino clinically based cohort, both short (≤6 h/night) and long (≥9 h/night) sleep durations were significantly associated with increased all-cause mortality and higher predicted 10-year cardiovascular risk, demonstrating a U-shaped relationship. These associations remained after adjusting for baseline cardiovascular risk factors and respiratory event rates. Future research should utilize objective sleep measures (e.g., actigraphy/polysomnography) to refine sleep duration and architecture assessment, explore causal pathways and mediators (inflammation, autonomic function, socioeconomic and psychiatric factors), evaluate targeted interventions (psychoeducation/psychosocial programs, OSA treatment, lifestyle modification), and expand to broader, community-based Hispanic/Latino populations to improve generalizability.

Generalizability may be limited: all participants were Hispanic/Latino and recruited from a sleep center, where sleep disorders are prevalent. Total sleep time was self-reported rather than measured objectively (actigraphy/PSG), introducing potential misclassification. The CVD risk assessment used the non-laboratory Framingham 2008 calculator, which, while practical, may differ from laboratory-based estimates. Missing data were excluded, potentially introducing selection bias. The CVD risk analysis was cross-sectional (prevalent risk), limiting causal inference. HSAT, while validated, may be less comprehensive than polysomnography for OSA characterization.