Healthy Sleep Patterns, Mortality, and Life Expectancy in Adults With and Without Atherosclerotic Cardiovascular Disease

Abstract

Background

Healthy sleep is a multidimensional behavior critical for chronic disease prevention, yet its long-term impact on mortality and life expectancy—particularly among individuals with atherosclerotic cardiovascular disease (ASCVD)—remains unclear.

Objectives

The purpose of this study was to evaluate the associations between sleep patterns and mortality and life expectancy among adults with and without ASCVD.

Methods

We analyzed data from 148,622 U.S. adults (mean age 48.4 years, 50.6% female) in the National Health Interview Survey (2013-2018), with mortality follow-up through December 31, 2019. A composite sleep score based on 5 self-reported behaviors was constructed to categorize participants as having poor, intermediate, or healthy sleep patterns. Multivariable Cox models estimated hazard ratios (HRs) for all-cause mortality, and life expectancy was calculated using a flexible parametric survival model.

Results

Over a median 4.3 years of follow-up (IQR: 2.8-5.5), 5,643 deaths occurred. Compared with those with poor sleep patterns, participants with healthy sleep patterns had significantly lower all-cause mortality, both among individuals with ASCVD (HR 0.74, 95% CI: 0.62-0.89) and those without ASCVD (HR 0.82, 95% CI: 0.70-0.95; P for additive interaction = 0.03). At age 45, a healthy sleep pattern was associated with an estimated life expectancy gain of 3.0 years (95% CI: 1.1-4.8) among individuals with ASCVD, and 1.5 years (95% CI: 0.3-2.6) among those without.

Conclusions

Multidimensional healthy sleep patterns are associated with lower mortality and increased life expectancy in adults, with greater absolute benefits observed in individuals with ASCVD.

Central Illustration

Atherosclerotic cardiovascular disease (ASCVD) remains the leading cause of death worldwide, contributing substantially to the global burden of noncommunicable diseases.¹ Despite significant advancements in pharmacological therapies that have reduced cardiovascular events and mortality, a considerable residual risk persists. Modifiable lifestyle factors, such as sleep, represent critical targets for the prevention and management of ASCVD.² Sleep health, encompassing both quantity and quality, has emerged as a pivotal determinant of cardiovascular outcomes and overall survival.³ While numerous studies have demonstrated the impacts of sleep duration on mortality, there is growing recognition that sleep patterns, encompassing multidimensional sleep behaviors, provide a more comprehensive framework for optimizing ASCVD outcomes.^4,5

Although the association between sleep and mortality is established for promoting cardiovascular health in the general population, the interplay between sleep health, preexisting ASCVD status, and mortality remains inadequately studied.⁶ ASCVD may amplify the risks associated with poor sleep, while simultaneously serving as a mediator in the relationship between sleep and mortality. Emerging evidence suggests that the benefits of healthy sleep behaviors may be especially pronounced in high-risk populations, such as individuals with ASCVD or diabetes.7, 8, 9 For instance, optimal sleep duration has been linked to significantly lower all-cause mortality in patients with coronary artery disease.⁸ However, the broader associations between multidimensional sleep patterns, mortality, and life expectancy in individuals with ASCVD remain undetermined. It remains unclear to what extent adopting optimal sleep patterns can translate into quantifiable survival benefits for this high-risk group.

To address these gaps, we conducted a prospective cohort study using nationally representative data from the National Health Interview Survey (NHIS) with linkage to National Death Index (NDI) records. We aimed to evaluate the association between multidimensional sleep patterns and all-cause mortality in individuals with and without ASCVD. In addition, we quantified the life expectancy gains associated with optimal sleep patterns, stratified by ASCVD status.

Material and methods

Ethics statement

Institutional ethics approval was not required for this study, as the NHIS data are publicly available as deidentified data.

Study design and population

This prospective cohort study used the pooled publicly deidentified data from 6 consecutive rounds of the U.S. NHIS (2013-2018).¹⁰ A detailed description of the NHIS has been reported elsewhere.¹¹ The NHIS is a nationally representative annual cross-sectional survey that incorporates complex, multistage sampling to provide estimates on the noninstitutionalized U.S. population. The harmonized data was obtained from the Integrated Public Use Microdata Series Health Surveys website (https://nhis.ipums.org/).¹²

In the current analysis, from an eligible sample of 186,661 participants that were linked to the NID,¹³ those with either missing data on all sleep factors (n = 4,718) or ASCVD status (n = 341) were excluded from the analyses. To reduce reverse causation bias, we further excluded those with current chronic conditions such as cancer (n = 18,727), chronic bronchitis (n = 6,308), and emphysema (n = 1,645). In addition, we removed individuals with missing data on covariates including education (n = 547), marital status (n = 272), smoking status (n = 168), alcohol consumption (n = 1,130), physical activity (n = 93), hypertension (n = 99), diabetes (n = 32), and body mass index (BMI, calculated as weight in kilograms divided by height in meters squared; n = 3,959). Therefore, our final analytic sample included 148,622 adults aged 18 years or older (Supplemental Figure 1). The characteristics between the included and excluded participants are shown in Supplemental Table 1.

Assessment of ASCVD

Based on previous research,^14,15 we used a self-reported diagnosis of coronary or cerebrovascular disease to identify those with ASCVD. Specifically, individuals were classified as having ASCVD if they responded positively to ever having been told by a doctor that they had any of the following: coronary heart disease, angina, heart attack (myocardial infarction), and/or stroke. Detailed questionnaires are summarized in Supplemental Appendix I.

Definition of healthy sleep patterns

An index score for the healthy sleep pattern was obtained by summing up the 5 sleep factors, as previously reported.^16,17 In line with our previous research,¹⁷ healthy sleep factors were defined as sleep 7 to 8 h/day; reported never/rarely trouble falling asleep (≤2 times/week); reported never/rarely trouble staying asleep (≤2 times/week); reported waking feeling rested most days or every day (≥5 days/week); and no self-reported sleep medication use. The properties of these sleep questionnaires have been reported^18,19 and are summarized in Supplemental Appendix II. Each sleep factor was scored 1 point if it was optimal and 0 point otherwise. A cumulative score indicating the count of sleep factors was calculated (range 0-5), with a higher score suggesting a healthier sleep pattern. Participants were then categorized into “poor sleep pattern” (healthy sleep score ≤1), “intermediate sleep pattern” (2 ≤ healthy sleep score ≤3), and “healthy sleep pattern” (healthy sleep score ≥4).^16,17

Outcomes

The National Center for Health Statistics has linked data collected from the NHIS surveys with death certificate records from the NDI.^13,20 Mortality status was obtained by probabilistic matching between NHIS data and death certificates up to December 31, 2019.²⁰ Follow-up time was calculated from the date of interview to the date of death or censoring on December 31, 2019.

Covariates

Information on demographics, lifestyles, and medical history were obtained from interview data. The following covariates were adjusted in the multivariable model: age (years), sex (men or women), ancestry (non-Hispanic White, non-Hispanic Black, non-Hispanic Asian, Hispanic, or other), educational attainment (lower than high school, high school diploma/general educational diploma, some college, or more than bachelor degree), marital status (married/living with a partner, widowed/divorced/separated, or never married), smoking status (never, former, or current smoker), alcohol consumption (lifetime abstainer, former, or current drinker), chronic conditions (none, diabetes, hypertension, or both), BMI (<25, 25-29.9, or ≥30 kg/m²), and survey year.

Statistical analysis

Descriptive statistics of the study sample are summarized by ASCVD status and category of healthy sleep score. Baseline characteristics are reported as frequencies (percentages) for categorical variables or mean (SD) for continuous variables. Mortality rates were calculated as events per 1,000 person-years.

First, Cox proportional hazards regression with the duration of follow-up as the timescale was conducted to calculate hazard ratios (HRs) and 95% CIs of all-cause mortality, with the lower category of healthy sleep score as the reference. The proportional hazard assumption was not violated according to the log-minus-log plot (Supplemental Figure 2). Three models were fit to explore the association of healthy sleep score with all-cause mortality risk, stratified by ASCVD status. Model 1 is adjusted for age and sex. Model 2 adjusted for age, sex, ancestry, marital status, and educational attainment. Model 3 additionally included all of the following covariates: smoking status, alcohol consumption, physical activity, chronic conditions, BMI categories, and survey year. To further control for potential nonlinear age effect, age was adjusted using restricted cubic spline functions with 4 knots at the 5th, 35th, 65th, and 95th percentiles (Supplemental Appendix III).²¹ Adjusted cumulative incidence curves and predicted 5-year mortality risks were calculated using a flexible parametric Royston-Parmar model on the log cumulative hazard scale with 3 internal knots (df: 4, default positions: 25%, 50%, 75%) with time-on-study as time scale and presented according to category of healthy sleep score.²² The details in the settings of flexible parametric Royston-Parmar models are provided in the Supplemental Appendix IV.

Moreover, we investigated the joint association of a healthy sleep score and ASCVD status with risk of all-cause mortality, with the healthy sleep pattern and no ASCVD as the reference. We further examined the potential interaction between healthy sleep pattern and ASCVD. Multiplicative interaction was evaluated using the adjusted Wald test by including the multiplicative terms of sleep pattern and ASCVD in the full adjusted model. Biological interaction on an additive scale was tested using relative excess risk due to interaction, attributable proportion, and Synergy Index.²³

Next, to assess the extent to which baseline ASCVD explained the associations of healthy sleep pattern with mortality, mediation proportions of the associations between a healthy sleep pattern and mortality by the mediator (ASCVD) were calculated by SAS MEDIATE Macro (https://ysph.yale.edu/cmips/research/software/analysis-graphics/mediate/).²⁴ Mediation analyses were adjusted for the covariates in Model 3.

Finally, we used a flexible parametric Royston-Parmar model on the log cumulative hazard scale with 3 internal knots (df: 4, default positions: 25%, 50%, 75%), using attained age as the underling time-scale, to assess the effects of healthy sleep pattern on life expectancy among people with and without ASCVD (stpm2 command in Stata).^25,26 The area under the survival curve (up to a maximum of 100 years) reflected projected overall survival. Residual life expectancy was first estimated conditional on surviving at ages 30 to 100 years (1-year intervals). Then, years of life gained (difference in average life expectancy) with healthy sleep were estimated as the differences of areas under the survival curves. All analyses were adjusted for sex, ancestry, marital status, educational attainment, smoking status, alcohol consumption, physical activity, chronic condition, BMI categories, and survey year.

We performed several sensitivity analyses. First, to identify subgroups of adults with or without ASVCD in whom the effect of healthy sleep pattern was particularly noteworthy, we stratified analyses by baseline age (<60 vs ≥60 years) and sex (female vs male). Second, a weighted healthy sleep score (ranging from 0 to 5) was constructed to account for the number of available sleep factors per participant (from 1 to 5). Weights were calculated as follows: $\text{weighted}\text{healthy}\text{sleep}\text{score}\frac{5\times healthysleepscore}{totalnumberofitemscompletedoutofthe5}$. Third, we excluded first-year deaths after the baseline survey to reduce potential reverse causation. Fourth, we repeated all analyses by using entry time-stratified Cox model instead of adjustment for survey year. Fifth, because covariate missingness ranged from 0.03% to 2.75% (Supplemental Table 2), we conducted multiple imputation using chained equations to address missing data and evaluated the robustness of results relative to complete case analysis (Supplemental Appendix V). Finally, based on the full adjusted HR and 95% CI, E-value was calculated to assess the potential unmeasured confounding.²⁷

According to the NHIS analytic guidelines, a complex sampling design and sampling weights were considered in our analyses, except for mediation analyses by SAS MEDIATE Macro. All analyses were done with Stata software (version 17.0), SAS software (version 9.4), and R software (version 4.3.1). A two-sided P value of <0.05 was considered statistically significant.

Results

Population characteristics

The final study cohort comprised 148,622 adults (mean [SD] age, 45.0 [17.4] years; 79,146 [50.0%] female; 94,284 [63.2%] White) from 2013 to 2018, of which 11,209 (6.3%) reported preexisting ASCVD. During a median follow-up of 4.3 years (interquartile range: 2.8-5.5; 596,776 person-years, unweighted), a total of 5,643 participants died. Table 1 summarizes the baseline sample characteristics stratified by sleep patterns among adults with and without ASCVD. Of note, 50.5% (95% CI: 49.3-51.7) of the participants with ASCVD had a lower proportion of healthy sleep pattern, as the corresponding proportion was 61.9% (95% CI: 61.5-62.3) among people without ASCVD.

Table 1.

Baseline Characteristics According to Healthy Sleep Pattern in Subjects Without and With ASCVD

	No ASCVD (N = 137,413)			ASCVD (N = 11,209)
	Poor Sleep (n = 14,525)	Intermediate Sleep (n = 39,052)	Healthy Sleep (n = 83,836)	Poor Sleep (n = 1,993)	Intermediate Sleep (n = 3,545)	Healthy Sleep (n = 5,671)
Age, mean (SD), y	44.9 (15.9)	43.2 (16.3)	43.7 (17.0)	59.4 (16.4)	62.8 (16.3)	66.1 (15.0)
Sex
Men	5,189 (38.1)	16,628 (45.8)	41,496 (52.8)	911 (49.3)	1921 (57.9)	3,331 (63.8)
Women	9,336 (61.9)	22,424 (54.2)	42,340 (47.2)	1,082 (50.7)	1,624 (42.1)	2,340 (36.2)
Ancestry
Non-Hispanic White	9,527 (66.1)	25,596 (65.3)	51,456 (61.1)	1,290 (65.7)	2,497 (71.1)	3,918 (70.7)
Non-Hispanic Black	2,104 (13.9)	5,159 (12.5)	10,567 (12.0)	325 (15.0)	523 (13.2)	861 (12.9)
Non-Hispanic Asian	460 (3.2)	1964 (5.3)	6,163 (7.5)	58 (3.1)	93 (2.9)	216 (4.3)
Hispanic	2,147 (15.1)	5,749 (15.8)	14,459 (18.2)	278 (14.6)	365 (11.2)	600 (11.0)
Other	287 (1.8)	584 (1.2)	1,191 (1.1)	42 (1.6)	67 (1.6)	76 (1.1)
Education
<High school	2046 (13.5)	4,306 (10.3)	10,399 (12.0)	454 (22.4)	780 (21.0)	1,146 (18.4)
High school diploma/GED	3,856 (27.5)	9,434 (24.2)	20,030 (23.9)	610 (30.3)	1,061 (29.7)	1701 (29.2)
Some college	5,241 (34.9)	13,131 (33.5)	24,598 (29.3)	623 (32.3)	1,043 (29.1)	1,515 (27.7)
≥Bachelor degree	3,382 (24.2)	12,181 (32.0)	28,809 (34.8)	306 (15.1)	661 (20.2)	1,309 (24.8)
Marital status
Married or living with a partner	5,477 (46.8)	17,149 (52.3)	38,854 (53.7)	669 (45.5)	1,464 (52.8)	2,498 (57.2)
Widowed, divorced, or separated	4,808 (24.2)	9,884 (17.5)	19,047 (14.7)	1,038 (39.6)	1,647 (35.1)	2,551 (32.5)
Never married	4,240 (29.0)	12,019 (30.2)	25,935 (31.6)	286 (14.9)	434 (12.0)	622 (10.3)
Smoking status
Never smoker	7,610 (54.0)	23,645 (62.6)	56,155 (69.4)	841 (40.5)	1,584 (45.1)	2,693 (47.7)
Former smoker	3,206 (21.2)	8,285 (20.3)	16,544 (18.2)	668 (35.2)	1,309 (35.4)	2,192 (38.8)
Current smoker	3,709 (24.7)	7,122 (17.1)	11,137 (12.4)	484 (24.4)	652 (19.5)	786 (13.5)
Alcohol consumption
Lifetime abstainer	2,270 (15.7)	6,468 (16.4)	18,217 (22.8)	407 (20.0)	758 (20.5)	1,325 (21.8)
Former drinker	2,382 (15.1)	5,160 (12.0)	10,011 (10.6)	624 (30.7)	1,050 (29.1)	1,533 (25.6)
Current drinker	9,873 (69.1)	27,424 (71.6)	55,608 (66.6)	962 (49.3)	1737 (50.4)	2,813 (52.6)
Physical activity
Recommended activity	6,360 (44.6)	19,522 (50.7)	44,866 (54.3)	497 (25.9)	1,063 (31.5)	1980 (36.9)
Insufficient activity	3,351 (23.5)	8,939 (23.3)	16,941 (20.3)	424 (21.8)	754 (21.8)	1,261 (22.2)
Sedentary	4,814 (31.9)	10,591 (26.1)	22,029 (25.3)	1,072 (52.3)	1728 (46.7)	2,430 (40.9)
Chronic condition
None	8,781 (62.6)	26,580 (70.7)	61,060 (75.8)	391 (20.4)	752 (22.3)	1,369 (26.0)
Diabetes	492 (3.4)	1,031 (2.6)	1892 (2.2)	93 (4.2)	155 (4.7)	251 (4.7)
Hypertension	4,068 (26.7)	9,252 (21.7)	17,161 (18.3)	935 (45.5)	1,656 (46.4)	2,760 (47.4)
Hypertension and diabetes	1,184 (7.3)	2,189 (4.9)	3,723 (3.8)	574 (29.9)	982 (26.5)	1,291 (21.9)
BMI, mean (SD), kg/m2	28.9 (7.1)	28.3 (6.4)	27.3 (5.7)	30.2 (8.1)	29.7 (6.9)	28.4 (6.3)
<25	4,586 (31.9)	13,312 (33.6)	32,164 (38.8)	482 (24.2)	840 (22.4)	1,649 (28.5)
25-29.9	4,631 (31.8)	12,955 (33.3)	29,764 (35.3)	636 (31.5)	1,264 (35.4)	2,138 (37.9)
≥30	5,308 (36.3)	12,785 (33.0)	21,908 (25.9)	875 (44.3)	1,441 (42.2)	1884 (33.6)
Survey year
2013	2,664 (16.0)	6,967 (15.9)	15,777 (16.5)	352 (15.6)	605 (16.1)	989 (16.9)
2014	2,665 (15.6)	7,396 (15.7)	16,765 (17.0)	358 (15.4)	648 (15.5)	1,077 (16.3)
2015	2,663 (16.8)	6,991 (17.0)	14,863 (16.4)	334 (14.2)	607 (16.1)	972 (15.4)
2016	2,479 (16.7)	6,799 (16.9)	14,273 (16.6)	367 (18.2)	631 (17.2)	1,068 (18.2)
2017	2043 (16.8)	5,563 (17.3)	11,450 (16.7)	269 (16.4)	496 (16.3)	803 (16.9)
2018	2011 (18.0)	5,336 (17.4)	10,708 (16.7)	313 (20.3)	558 (18.7)	762 (16.3)

ASCVD = atherosclerotic cardiovascular disease; BMI = body mass index; GED = general educational diploma; SD = standard deviation.

Associations of sleep patterns with mortality stratified by ASCVD status

Table 2 summarizes the associations of sleep patterns with all-cause mortality among adults with and without ASCVD. Compared with reference group (poor sleep), a healthy sleep pattern was associated with 26% (HR = 0.74, 95% CI: 0.62-0.89) and 18% (HR = 0.82, 95% CI: 0.70-0.95) lower risks of all-cause mortality among adults with and without ASCVD in the fully adjusted model. Adjusted cumulative mortality rates according to healthy sleep patterns were plotted in Supplemental Figure 3. Standardized 5-year mortality rates among adults without ASCVD were 4.2% (95% CI: 3.7-4.2) for those in the poor sleep category vs 3.5% (95% CI: 3.3-3.7) for those in the healthy sleep category. Among participants with ASCVD, the standardized 5-year mortality rates were 24.4% (95% CI: 21.7-27.5) among those with a poor sleep and 19.4% (95% CI: 17.9-21.0) among those with a healthy sleep pattern. Absolute adjusted standardized 5-year mortality rate differences between healthy and poor sleep categories are shown in Supplemental Table 3.

Table 2.

Mortality, Hazard Ratios, and Differences in Remaining Life Expectancy According to Healthy Sleep Pattern in Subjects Without and With ASCVD

Measure	No ASCVD (N = 137,413)			ASCVD (N = 11,209)
	Poor Sleep (n = 14,525)	Intermediate Sleep (n = 39,052)	Healthy Sleep (n = 83,836)	Poor Sleep (n = 1993)	Intermediate Sleep (n = 3,545)	Healthy Sleep (n = 5,671)
Mortality
No. of deaths	483	1,132	2,295	273	557	903
Deaths per 1,000 person-year (95% CI)	6.7 (5.9-7.7)	5.3 (4.9-5.7)	5.0 (4.7-5.3)	36.5 (31.4-42.8)	36.8 (33.0-41.2)	38.4 (35.2-41.8)
Model 1, hazard ratio (95% CI)a	Reference	0.80 (0.68-0.94)	0.67 (0.58-0.79)	Reference	0.77 (0.64-0.93)	0.68 (0.57-0.82)
P value		0.006	<0.001		0.008	<0.001
Model 2, hazard ratio (95% CI)a	Reference	0.83 (0.70-0.97)	0.70 (0.60-0.81)	Reference	0.77 (0.64-0.92)	0.67 (0.56-0.81)
P value		0.02	<0.001		0.005	<0.001
Model 3, hazard ratio (95% CI)a	Reference	0.92 (0.78-1.08)	0.82 (0.70-0.95)	Reference	0.81 (0.67-0.98)	0.74 (0.62-0.89)
P value		0.29	0.01		0.03	0.001
Life expectancy
Adjusted difference in remaining life expectancy at 45 years (95% CI), yb	Reference	0.61 (−0.61 to 1.84)	1.47 (0.33-2.62)	Reference	2.02 (0.10-3.93)	2.96 (1.14-4.79)
Adjusted difference in remaining life expectancy at 65 years (95% CI), yb	Reference	0.46 (−0.45 to 1.37)	1.11 (0.25-1.96)	Reference	1.24 (0.06-2.41)	1.83 (0.71-2.95)

Models accounted for the National Health Interview Survey complex design and weights.

ASCVD = atherosclerotic cardiovascular disease.

^aHazard ratio was estimated from Cox model with time-on-study as timescale. Model 1 was adjusted for baseline age; Model 2 was adjusted for baseline age, sex, ancestry, marital status, and educational attainment; Model 3 was additionally adjusted for smoking status, alcohol consumption, physical activity, chronic condition, body mass index categories, and survey year.

^bRemaining life expectancy was calculated as the area under the standardized survival curve, estimated from flexible parametric survival models with attained age as timescale, adjustment for sex, ancestry, marital status, educational attainment, smoking status, alcohol consumption, physical activity, chronic condition, body mass index categories, and survey year.

Joint and interaction analyses of ASCVD status and sleep pattern with mortality

In the joint association analyses, we found that participants with preexisting ASCVD and a poor sleep patten had the highest mortality risk (HR: 2.28, 95% CI: 1.93-2.70; P < 0.001) than participants without ASCVD and adopting a healthy sleep pattern (Figure 1). A statistically significant additive interaction between ASCVD status and sleep pattens on mortality risk was detected, with relative excess risk due to interaction 0.45 (95% CI: 0.04-0.87), attributable proportion 0.19 (95% CI: 0.04-0.35), and Synergy Index 1.52 (95% CI: 1.05-2.19), whereas we did not observe a significant multiplicative interaction (Wald test, F = 0.76, P = 0.47) as shown in Figure 1.

Figure 1.

Joint Association of Sleep Pattern and ASCVD Status With Mortality

The reference group was participants without ASCVD and with a healthy sleep pattern. The model has been adjusted for baseline age, sex, ancestry, marital status, educational attainment, smoking status, alcohol consumption, physical activity, chronic condition, body mass index categories, and survey year. Interactions on both the multiplicative and additive scales were evaluated using Wald tests, as well as 3 metrics for additive interaction: RERI, AP, and SI. AP = attributable proportion; ASCVD = atherosclerotic cardiovascular disease; RERI = relative excess risk due to interaction; SI = synergy index.

Mediation proportion of ASCVD between sleep pattern and mortality

In the entire population, compared with healthy sleep pattern, the HRs of a poor sleep pattern for all-cause mortality were 1.24 (95% CI: 1.14-1.35) and 1.28 (95% CI: 1.18-1.39) with (direct effect) and without (total effect) adjusting for ASCVD status, respectively (Table 3). The ASCVD status mediated 12.6% (95% CI: 7.9-19.6) of the association between a poor sleep pattern and all-cause mortality (Table 3).

Table 3.

Associations of Healthy Sleep Pattern With All-Cause Mortality and Mediation Proportion Attributed to Atherosclerotic Cardiovascular Disease

Sleep Pattern	n (%)	Events/Person-Years	HR (95% CI)		Mediation Proportion,b % (95% CI)	P Value
			Without Mediators (Total Effect)	With Mediators (Direct effect)a
Healthy sleep	89,507 (60.2%)	3,198/362,495.4	Reference	Reference
Intermediate sleep	42,597 (28.7%)	1,689/169,006.4	1.20 (1.11-1.30)	1.17 (1.08-1.27)	11.2 (5.6-21.1)	0.0001
Poor sleep	16,518 (11.1%)	756/65,274.1	1.28 (1.18-1.39)	1.24 (1.14-1.35)	12.6 (7.9-19.6)	<0.001

HRs and corresponding 95% CIs are provided for the total and direct effects. Models are adjusted for age, sex, ancestry, marital status, educational attainment, smoking status, alcohol consumption, physical activity, chronic condition, body mass index categories, and survey year. Mediation analysis was conducted via the SAS MEDIATE Macro (https://ysph.yale.edu/cmips/research/software/analysis-graphics/mediate/).

^aDirect effect obtained from a model including atherosclerotic cardiovascular disease.

^bProportion of the association mediated by atherosclerotic cardiovascular disease.

Sleep pattern and life expectancy with or without ASCVD

Supplemental Figure 4 shows the long-term survival according to healthy sleep patterns among populations with and without ASCVD at a starting age from 45 to 65 years. After covariate adjustments, at age 45 years, the estimated life expectancy for participants without ASCVD was 39.8 years (95% CI: 38.6-41.1) with a poor sleep pattern and 41.3 years (95% CI: 40.5-42.1) with a healthy sleep pattern (difference: 1.47 years [95% CI: 0.33-2.62]; Table 2); the corresponding estimates for participants with ASCVD were 29.2 years (95% CI: 27.2-31.2) and 32.2 years (30.7-33.6) (difference: 2.96 years [95% CI: 1.14-4.79]; Table 2). Similarly, in individuals with ASCVD at age 65 years, intermediate and healthy sleep patterns were associated with an average of 1.24 (95% CI: 0.06-2.41) and 1.83 (95% CI: 0.71-2.95) additional life years gained, respectively, compared to poor sleep group; corresponding estimates in those without ASCVD were 0.46 (95% CI: −0.45 to 1.37) and 1.11 (95% CI: 0.25-1.96), respectively (Table 2). The differences of life expectancy for intermediate and healthy sleep vs poor sleep patterns among populations with and without ASCVD at every starting age from 30 to 100 years are displayed in Figure 2 and Central Illustration.

Figure 2.

Life Expectancy Differences Associated With Sleep Pattern by ASCVD Status

(A) Life expectancy differences associated with sleep pattern among people without ASCVD. (B) Life expectancy differences associated with sleep pattern among people with ASCVD. The reference group was participants with a poor sleep pattern. Life expectancy was calculated as the area under the standardized survival curve, estimated from flexible parametric survival models with attained age as timescale, adjustment for sex, ancestry, marital status, educational attainment, smoking status, alcohol consumption, physical activity, chronic condition, body mass index categories, and survey year. ASCVD = atherosclerotic cardiovascular disease.

Central Illustration.

Healthy Sleep Patterns, Mortality, and Life Expectancy in Adults With and Without Atherosclerotic Cardiovascular Disease

This illustration summarizes the associations between multidimensional healthy sleep patterns and subsequent mortality and life expectancy among 148,622 U.S. adults. Healthy sleep was associated with significantly lower all-cause mortality in both those with and without ASCVD. At age 45, healthy sleepers were estimated to live 3.0 additional years with ASCVD and 1.5 additional years without ASCVD compared with poor sleepers. ASCVD = atherosclerotic cardiovascular disease.

Sensitivity analyses

In subgroup analyses, the associations between healthy sleep and mortality were stronger among people younger than 60 years without ASCVD (Supplemental Table 4). The association between healthy sleep and all-cause mortality was consistent for the weighted healthy score (Supplemental Table 5). Similar findings were observed when excluding individuals who died within 1-year since follow-up (Supplemental Table 6) or repeating analyses by using entry time-stratified Cox models (Supplemental Table 7). Similar findings were observed after multiple imputation (Supplemental Table 8). The estimated E-values for the observed associations between the healthy sleep pattern and mortality among adults with and without ASCVD were 1.39 and 1.23 for the lower 95% CIs and 1.77 and 1.56 for the point estimates, representing the strength of confounding required to reduce the associations to null (Supplemental Figure 5).

Discussion

Through a large, contemporary U.S. cohort, we found that a healthy sleep pattern was associated with significantly lower risk of mortality, which was pronounced among the ASCVD high-risk population. In addition, individuals with ASCVD who adopted a healthy sleep pattern were estimated to live an average 3.0 years longer, which was double than that in those without ASCVD. These findings highlight the potential impacts of sleep behaviors on mortality and life expectancy among adults with or without ASCVD.

The association between a healthy sleep pattern and reduced mortality risk aligns with prior studies emphasizing the important roles of sleep patterns or individual sleep behaviors in promoting cardiovascular health and longevity.^28,29 Previous research has reported a U-shaped relationship between sleep duration and mortality risk, with optimal durations of 7 to 8 hours linked to the lowest mortality among the general population30, 31, 32 or those under different health conditions.^7,33, 34, 35 Moreover, other sleep quality factors, such as difficulty initiating or maintaining sleep, continuity, timing, and irregularity, are reported to be associated with higher risk of mortality, despite heterogeneity exists.^5,28,29,36 Our previous research introduced a multidimensional framework for assessing sleep health patterns by incorporating sleep duration, quality-related factors, and use of sleep medications, which have been assessed in the NHIS samples.¹⁷ Based on this framework, this current study examined the associations between the sleep patterns and all-cause mortality, and we compared the associations between healthy sleep pattens and mortality among individuals with or without ASCVD. These findings suggest that adopting a healthy sleep pattern was associated with a 18% reduced risk of all-cause mortality among adults without ASCVD, while those with ASCVD may sequentially lower the mortality risk by 26%, if the causality confirmed. These findings were consistent with previous studies, although they only examined the associations between sleep duration and mortality among patients with coronary artery disease or stroke and with relatively small sample sizes.^8,35 Our study utilized the nationally representative samples from U.S. NHIS, with 11,209 individuals reporting ASCVD at baseline. These findings underscore the shift from sleep duration to sleep patterns for promoting health outcomes for both the primary and secondary preventions of ASCVD.

Our findings highlight the synergistic associations of poor sleep patterns and ASCVD with mortality. Participants with both ASCVD and poor sleep patterns faced the highest mortality risk as expected, and a significant additive interaction was detected. Previous studies have largely revealed that poor sleep is an independent risk factor of ASCVD, and the Life's Essential 8 have incorporated sleep health as a crucial component to maintain cardiovascular health.^37,38 Our study demonstrates that the ASCVD status may amplify the detrimental impacts of unfavorable sleep behaviors and recommend this high-risk population modify their sleep health status given that only half the population with ASCVD had healthy sleep patterns in our analysis.

To our knowledge, our study is among the first to quantify the life expectancy attributable to adopting a healthy sleep pattern in real-world settings. Individuals with ASCVD who adopted a healthy sleep pattern were estimated to gain double the life years compared to those without ASCVD, which supports the significant interaction between sleep patterns and ASCVD discussed earlier. Consistently, prior research linked optimal sleep duration to longer life expectancy.³⁹ Insufficient sleep and sleep disruption may lead to a loss of life expectancy, particularly in younger populations.⁴⁰ Notably, previous research has highlighted a strong association between sleep behaviors and cardiovascular mortality. Specifically, adopting a healthy sleep pattern was linked to a 6% reduction in all-cause mortality and an 11% reduction in cardiovascular mortality.^41,42 This may help explain why individuals with ASCVD experienced twice the gain in life expectancy from healthy sleep patterns, as a higher proportion of deaths in this group are attributable to cardiovascular causes.

Mechanistic pathways underlying the relationship between sleep and mortality further illustrate the importance of sleep health in ASCVD populations. Healthy sleep is essential for cardiovascular health as it regulates inflammation, supports metabolic processes, and maintains autonomic and vascular function. It reduces proinflammatory cytokines like interleukin-6 and tumor necrosis factor-α, alleviating systemic inflammation that drives atherosclerosis progression.⁴³ Sleep also improves insulin sensitivity and glucose metabolism, reducing the risk of diabetes and hyperlipidemia, which are major contributors to ASCVD.⁴⁴ Furthermore, it balances the autonomic nervous system, lowers sympathetic overactivity, and stabilizes blood pressure, all of which are critical for heart health.⁴⁵ Sleep also aligns circadian rhythms and improves endothelial function, helping maintain vascular health and reducing cardiovascular risks.⁴⁶ In individuals with ASCVD, poor sleep exacerbates inflammation, metabolic dysregulation, and autonomic imbalance, which may amplify cardiovascular stress and lead to increased mortality risk.⁴⁷ These findings emphasize the need to prioritize healthy sleep patterns as a preventive measure to reduce cardiovascular events and improve health outcomes in high-risk populations.

Strengths and limitations

Despite several strengths of this study, including large representative samples, long-term follow-up with linkage to the NDI, and extensive sensitivity analyses, several limitations should be acknowledged. The reliance on self-reported sleep data and ASCVD status may introduce the potential for recall bias and misclassification. However, the questionnaire used in the NHIS is a validated instrument delivered by trained interviewers to minimize bias. The use of self-reported ASCVD is reasonable in epidemiological studies, as prior validations have shown good accuracy and minimal bias.⁴⁸ Nevertheless, the validity and use of self-reported ASCVD conditions have been widely evaluated in the NHIS.^14,15 Sleep patterns were assessed only at baseline. Future studies incorporating objective sleep measures, such as actigraphy or polysomnography,⁴⁹ can provide more accurate assessments of sleep health and capture the dynamic sleep behavior change over time, as well as other aspects of sleep health. In addition, as an observational study, this research cannot establish causal relationships. The calculated E-values (1.56 and 1.77) suggest that relatively modest unmeasured confounding could potentially account for the observed associations; therefore, despite extensive adjustment for known risk factors, residual confounding cannot be excluded, and the findings should be interpreted with appropriate caution. People with optimal sleep behaviors may concurrently adopt other health-promoting behaviors, such as lifestyles, medication adherence, and healthy diet, although we have adjusted many lifestyle factors in our analyses. However, dietary information was not collected in the NHIS due to its survey design. Nevertheless, randomized controlled trials evaluating the effects of sleep interventions on ASCVD outcomes and mortality are needed to confirm these findings. Finally, this study did not account for specific sleep disorders, such as sleep apnea, which are unavailable in the NHIS data and may have unique impacts on cardiovascular health and mortality. Future research should extend the sleep patterns by involving more sleep behaviors to deepen our understanding regarding sleep and health.

Conclusions

Through a large, contemporary U.S. cohort, the findings demonstrate that only half the U.S. individuals with preexisting ASCVD maintain a healthy sleep pattern. This study reveals the independent associations between a healthy sleep pattern and all-cause mortality both in individuals with and without ASCVD. In addition, this study provides quantified insights as to the extended life expectancy by adopting a healthy sleep pattern for the primary and secondary ASCVD prevention, enhancing the current lifestyle recommendations and interventions for the ASCVD high-risk population.

Data availability statement

The NHIS data are publicly available on request (https://www.cdc.gov/nchs/nhis/index.html). The final datasets generated for this study are available on reasonable request to the corresponding author.

Ethics approval

The National Center for Health Statistics Disclosure Review Board reviews and approves the NHIS (additional information on ethics approvals and procedures for informed consent are available from the National Center for Health Statistics).

Perspectives.

COMPETENCY IN PATIENT CARE AND MEDICAL KNOWLEDGE: Multidimensional healthy sleep patterns—including optimal sleep duration, sleep continuity, and absence of sleep disorders—are associated with reduced all-cause mortality and increased life expectancy. Notably, individuals with atherosclerotic cardiovascular disease (ASCVD) derive even greater survival benefits from healthy sleep, highlighting the potential of sleep optimization as an effective adjunct in secondary cardiovascular prevention.

TRANSLATIONAL OUTLOOK: Incorporating comprehensive sleep assessment and counseling into routine clinical care for individuals with or at risk of ASCVD could improve long-term outcomes. Future interventions targeting sleep health may offer a novel strategy to extend life expectancy in high-risk cardiovascular populations.

Funding support and author disclosures

This work was supported by funding from the Capital’s Funds for Health Improvement and Research (CFH2024-2G-2036), the National Natural Science Foundation of China (72474142 and 82103942), the R&D Program of Beijing Municipal Education Commission (KM202210025015), the Beijing Nova Program (20250484858), the Clinical Research Incubation Project, Beijing Chao-Yang Hospital, Capital Medical University (CYFH202310), the Talent development plan for the future in Medical-Engineering Integration by BRA-CDCHE and ZTA (MBRC0012025021), the National Clinical Key Specialty Construction Project: Cardiology (Coronary Artery Disease) (“1 + N” Discipline Cluster Model), and the National Clinical Key Specialty Construction Project. The authors have reported that they have no relationships relevant to the contents of this paper to disclose.