Sleep Duration and Cardiovascular Disease Risk: Epidemiologic and Experimental Evidence

Synopsis

Inadequate sleep has become increasingly pervasive, and the impact on health and quality of life remains to be fully understood. The cardiovascular consequences alone appear to be substantial and significant. This brief review summarizes epidemiologic evidence regarding the association between extremes of sleep duration and the prevalence and incidence of cardiovascular diseases. The adverse effects of experimental sleep loss on physiological functions are discussed, along with those cardiovascular risk factors that may underlie the association with increased morbidity and mortality. Current data support the concept that inadequate sleep duration confers heightened cardiovascular risk. Thus implementation of preventative strategies may be needed to reduce the potential disease burden associated with this widespread high-risk behavior.

Introduction

According to outcome-based recommendations issued recently by the National Sleep Foundation, the appropriate sleep duration for adults lies between 7 to 9 hours per night¹. Notably, only 48% of the US adult population reports a habitual sleep time falling within that range², while 26% average 6 to 7 hours of sleep/night, and 20% sleep less than 6 hours/night.

The time allotted to sleep has gradually declined over the past decades, with similar trends observed in multiple Western countries³. A growing proportion of individuals are curtailing their sleep in response to increasing demands and lifestyle changes, such as prolonged working hours, increased environmental lighting, and introduction of new communication technologies, which enable living “around the clock”.

Nevertheless, changes in sleep habits are not without consequences. Deviations from optimal sleep duration may pose a substantial threat to health, with the detrimental effects of abnormal sleep on physical and psychological well-being only beginning to be unraveled. In this review, available data on the relation between abnormal sleep duration and risk of prevalent and incident cardiovascular disease, the leading cause of morbidity and mortality, will be presented. The putative physiologic mechanisms underlying the observed associations and potential confounders will also be discussed.

Epidemiologic evidence

Hypertension

Hypertension is widespread affecting approximately one-third of the adult population in the US, and is a prominent risk factor for other cardiovascular and cerebrovascular diseases⁴. Epidemiologic evidence indicates that the relationship between customary sleep duration and risk of hypertension is better described as a curvilinear phenomenon, with both extremes of the sleep length distribution independently associated with enhanced likelihood of prevalent hypertension in the general population^5–7.

A U-shaped relationship between self-reported sleep length and hypertension has been documented in a large (N=71,455) national representative sample (National Health Interview Survey, NHIS)⁶, with both ends of the tail exhibiting larger age-standardized prevalences of hypertension (<6 hours/night: 32.4%; ≥10 hours/night: 32.5%) compared to the referent category (8 hours/night, 23.2%). Similarly, the lowest risk of hypertension in the Sleep Heart Health Study ⁷ was observed in those subjects sleeping 7 to 7.9 hours/night, while a progressive increase was seen when moving away from this reference. The greater hazard conferred by both short and long sleep withstood multivariable adjustments for lifestyle, clinical and sleep-related covariates (adjusted OR=1.66, 95% CI=1.35–2.04 for <6 hours/night; OR=1.30, 1.04–1.62 for ≥9 hours/night). Large population studies have replicated this pattern^5,8.

While both ends of sleep duration have been cross-sectionally related to hypertension, the prospective contribution of long sleep is less compelling. As concluded by a recent meta-analysis⁹, current longitudinal data mainly support a role of short sleep as independent marker of incident hypertension.

Indeed, it has been estimated that individuals sleeping less than 6 hours/night are 20–32% more likely to develop hypertension compared to those sleeping 7–8^8,10. Gangwisch¹⁰ studied the incidence of hypertension over a 8 to 10 year time span using the first National Health and Nutrition Examination Survey (NHANES I) dataset. In the pooled cohort comprising 4810 participants, fully-adjusted probability of high blood pressure (BP) was more elevated in those sleeping ≤5 hours/night (adjusted HR=1.32, 1.02–1.71) than in normal sleepers (7–8 hours/night). Nevertheless, age-stratified analysis revealed that those who were 32 to 59 years old and reported habitual sleep ≤5 hours were 60% more likely to develop hypertension than peers sleeping 7–8 hours. Conversely, sleep duration was unrelated to outcomes in the older age strata (60–86 years). Several cohort studies have subsequently replicated these null findings in the geriatric population^8,11,12.

Women have been found to be more susceptible to the pressor effects of abnormal sleep. Estimates for both prevalent and incident hypertension derived from the Whitehall II study¹³ were higher in middle-aged women who were sleeping ≤5 hours/night compared to those sleeping 7 hours, but not in men. Sex-specific associations have been confirmed and further detailed in a transversal examination of the Western New York Health Study¹⁴, where a sub-analysis of the female sample classified by menopausal status unveiled significantly higher odds only in premenopausal women.

Differential vulnerability to abnormal sleep length has also been reported across ethnicities. Event rates for hypertension were higher in Black subjects from the NHIS who were sleeping <6 hours/night or >8 hours/night, compared to their White counterparts¹⁵. These data are in line with findings from the Coronary Artery Risk Development in Young Adults (CARDIA) study¹⁶, where objective sleep duration, as quantified from actigraphy, related to greater surges in BP in African-Americans.

The moderating effect of demographic variables on the link between sleep duration and hypertension interestingly parallels the increased prevalence of this condition in women after 65 years old and in the African-American population⁴.

When BP is treated as continuous variable, again a U-shaped relation with sleep hours can be described, although more robust for systolic values^12,17. Abnormal sleep duration is also associated with altered diurnal BP rhythmicity. Both excess sleep and curtailed sleep have been linked with attenuated nocturnal dipping in BP^17,18, which is a sensitive prognostic marker for cardiovascular disease¹⁹.

Coronary heart disease

Coronary heart disease (CHD), which comprises a spectrum of acute and chronic manifestations, remains the major cause of death worldwide²⁰, with rising prevalence⁴. Abnormal sleep duration has been identified as a risk factor for CHD on the basis of epidemiologic studies, which show a cross-sectional relation consistent with a U-shaped curve^21–24. Weighted prevalence of total CHD was higher in respondents of the Behavioral Risk Factor Surveillance System (BRFSS) survey reporting either ≤6 hours/night (11.1%, 95% CI: 10.1–12.1) or ≥10 hours/night (14.8%, 12.0–17.6) than in the reference group sleeping 7 to 9 hours (7.9%, 7.3–8.5)²². When the clinical presentations of CHD are considered separately, a heterogeneous pattern of risk emerges. Both sleep lengths of 6 or less hours/night and 9 or more hours/night were associated with heightened odds for history of myocardial infarction in a Finnish population²³ and in the NHIS cohort²⁴, but only short sleepers has a significant adjusted prevalence of myocardial infarction using the 2007–2008 NHANES dataset²¹. Findings are discrepant for angina, although this condition seems more closely related to short sleep duration^23,24.

Similarly to prevalence, incidence of fatal and non-fatal CHD events is greater with habitual sleep duration above or below 7 to 8 hours, as indicated by pooled relative disease risk of 1.48 (1.22–1.80) for short sleepers and 1.38 (1.15–1.66) for long sleepers²⁵. This trend is in line with data on all-cause mortality²⁶, which depict a curvilinear relationship between time asleep and death estimates, with best life expectancy achieved by those reporting sleeping habitually 7–8 hours/night.

In a longitudinal study comprising 71,617 female participants enrolled in the Nurses’ Health Study²⁷, multivariate-adjusted relative risks of coronary heart disease at 10-year follow-up were 1.39-fold higher in women reporting ≤5 hours/night and 1.37-fold higher in those sleeping ≥9 hours compared to those sleeping 8 hours/night. A similar pattern has been described for CHD deaths in a community-based Chinese cohort²⁸, while likelihood of future fatal and non-fatal CHD was better predicted by the short sleep category in the Framingham Offspring Cohort²⁹. Compared to 7–8 hours/night, self-reported sleep of ≤5 hours or ≥10 hours induced, respectively, a 25% and a 43% age- and race-adjusted raised risk for CHD in a large national cohort of postmenopausal women³⁰. However, the estimates were markedly attenuated when controlling for behavioral, socioeconomic and clinical covariates and fell out of statistical significance. Comparable results were noted in a prospective investigation on a Dutch cohort³¹. Disparities in CHD risk have been evident in age- and sex-stratified analyses, although with divergent results^32–35.

In addition to established coronary disease, sleep length has been examined in relationship with early indices of subclinical atherosclerotic disease, and findings are compatible with normal sleep duration being protective against later cardiovascular events.

Ultrasonographic measurement of carotid intima-media thickness (IMT) is often adopted as a non-invasive indicator of atherogenic vascular damage³⁶. Data from the Study of Health in Pomerania³⁷ depict a J-shaped association between habitual hours of sleep and carotid wall thickening, with the lowest carotid IMT values detected in subjects sleeping 7–8 hours. A monotonic relation was instead observed in the CARDIA study³⁸, where actigraphy-derived sleep duration predicted carotid IMT in middle-aged men, but not in women, with each additional hour of sleep associated with 0.026 mm decrease in IMT. A longitudinal examination sampling 495 individuals from the same cohort³⁹ confirmed the contribution of curtailed sleep to incident coronary risk. Using CT-assessed coronary artery calcification as an early predictor of CHD⁴⁰, and accounting for conventional risk factors, risk of calcification at 5 year follow up declined progressively as sleep time lengthened.

Cerebrovascular disease

Cerebrovascular disease is among the leading causes of long-term disability⁴¹ and responsible of 16.4% of cardiovascular deaths⁴. As outlined recently in a meta-analysis by Ge and Guo⁴², enhanced cross-sectional and prospective vulnerability to cerebrovascular events is associated with both ends of the sleep duration distribution. Pooled ORs (95% CI) for prevalent stroke were 1.71 (1.39–2.02) for short sleep duration and 2.12 (1.51–2.73) for long sleep duration. The corresponding pooled HRs for incident stroke were 1.13 (1.02–1.25) and 1.40 (1.16–1.64). The concept of a U-shaped relation is corroborated by results from large population-based studies including the BRFSS survey²² and the NHIS²⁴. There are also data reflecting a more robust association of stroke with excessive sleep^35,43,44.

An early prospective investigation on how sleep pattern affects the likelihood of stroke was conducted using the NHANES I study population⁴³, which comprised 7,844 participants followed for 10 years. This analysis revealed that the independent relative risk of ischemic or hemorrhagic stroke was 1.5 times higher in individuals reporting >8 hours/night than in those who were sleeping 6–8 hours.

Chen et al⁴⁴ prospectively examined a cohort of 93,175 postmenopausal women and found both self-reported short (≤6 hours/night) and long sleep (≥9 hours/night) to be significant predictors of ischemic stroke (including fatal and non-fatal events) at 7.5 year follow-up in the age- and race-adjusted analysis. While curtailed sleep was no longer an independent determinant in multivariate models including socioeconomic, lifestyle, psychiatric and medical covariates (HR: 1.14, 0.97–1.33), long sleep retained its predictive strength (HR: 1.70, 1.32–2.21). In addition, an association with short sleep was unmasked when the analysis was restricted to women without cardiovascular disease at baseline. Conversely, in a 4-year examination of a population of Japanese elderly hypertensives who underwent brain scans for detection of silent cerebral infarcts⁴⁵, insufficient sleep predicted future stroke only in those with brain lesions at study entry.

Similarly to cardiovascular disorders, the impact of extreme sleep lengths on cerebrovascular disease risk appears to be modified by demographic characteristics. Among the 154,599 participants of the 2006–2011 NHIS study⁴⁶, age standardized-prevalences of stroke were 2.78%, 1.99% and 5.21% in respondents reporting sleeping ≤6, 7–8 and ≥9 hours/night, respectively. Risk of stroke remained more elevated in short and long sleepers in the pooled sample compared to the reference (7–8 hours/night) after controlling for pertinent covariates. Nonetheless, further stratification by age and gender revealed that the association with short sleep persisted in young women (18–44 years) while the association with long sleep was maintained in both male and female older adults (≥65 years). Notably however, the increased crude hazard for incident stroke symptoms found in US employed individuals aged ≥45 years sleeping <6 hours/night⁴⁷ was abolished in adjusted models.

Putative biological mechanisms and moderators

Observational findings from epidemiologic investigations are supplemented by experimental, laboratory-based evidence, which provide mechanistic insights into the biological substrate underlying the heightened cardiovascular risk associated with inadequate sleep.

A causative link between sleep deficiency and aberrant cardiovascular function is supported by multiple studies which applied models of short-term, total sleep deprivation. Along with these traditional protocols, extended sleep restriction paradigms have been more recently implemented to better simulate the chronic, partial sleep loss commonly experienced in everyday life.

Acute surges in systolic and diastolic BP occur following 24–88 hours of sustained wakefulness^48–53. In the absence of sizable increases in diurnal resting BP^54–56, healthy individuals exposed to prolonged partial sleep curtailment exhibit elevations in nocturnal BP along with dampened nocturnal dipping⁵⁷ and an amplified morning surge⁵⁸.

A sympathoexcitatory effect of sleep loss has been proposed to account for the rise in BP. Research on hemodynamics and neural circulatory control largely supports the enhanced release of plasma and urinary norepinephrine after sustained sleep restriction^54,59 but not in response to acute sleep loss^49,50,60. While results on heart rate are mixed, being either unaffected^48,49,54,55 or accelerated^50–52 after sleep loss, muscle sympathetic nerve activity has been consistently found to be inhibited^48,49,60. An altered arterial baroreflex functioning in terms of resetting towards a higher BP level has thus been postulated, substantiated by findings of a rightward and downward shift in the operating point following sleep restriction^48,60. Results on baroreflex sensitivity are less clear, showing an increase,⁶¹ decrease,⁶² or no change⁴⁸. On the contrary, enhanced cardiac sympathetic drive as estimated from heart rate variability, a non-invasive index of cardiac autonomic modulation predictive of adverse events⁶³, has been more consistently documented after total sleep loss^50,62 and after 5 days of sleep truncation⁵⁴.

Electrocardiographic abnormalities in cardiac conduction and repolarization resembling a proarrhythmogenic profile have also been detected after acute experimental sleep deprivation^64,65.

In addition, sleep debt provokes early deterioration of vascular structure and function that may promote cardiovascular risk. Exposure to one night of sleep deprivation increases arterial stiffness as estimated by brachial-ankle pulse-wave velocity⁵³. Interestingly, these experimental data are in contrast with those gathered from population studies, which linked greater arterial stiffness to long sleep duration in men^66,67. Impaired endothelial function reflects poor vascular health and is thought to be a precursor of atherosclerosis⁶⁸. Individuals undergoing experimental sleep curtailment exhibit reduced endothelium-dependent vasodilation, indicated by diminished acetylcoline -induced cutaneous^50,55 and venous vasodilation⁵⁴ and decreased brachial flow-mediated vasodilation⁶⁹. Sleep loss also damages the coronary microcirculation⁷⁰. The concomitant enhanced release of cellular adhesion molecules and selectins from activated endothelial cells^50,71 further supports the concept of systemic endothelial dysfunction promoted by sleep deprivation. Findings of abbreviated prothrombin and thrombin times,⁷² consistent with development of a pro-thrombotic state, have also been reported.

In addition to direct effects on cardiac and vascular regulation, sleep deficiency has been showed to adversely impact numerous physiological functions that are implicated in increased cardiovascular risk. There is substantial evidence that sleep loss impairs glucose homeostasis and insulin sensitivity^59,73–75. Compared to 10 hours/night, one week of sleep restriction at 5 hours/night elicited a 20% decrease in insulin sensitivity⁷³. Exposure to sleep truncation generated a positive energy balance leading ultimately to weight gain^76–79. Current research indicates that the increase in body weight achieved with sleep restriction in the setting of ad libitum access to food is primarily driven by overeating, as results on energy expenditure are inconsistent. Likewise, the underlying molecular pathways accompanying excess food intake show discrepant responses in energy balance regulatory hormones such as leptin and ghrelin^76,77. On the other hand, compelling epidemiologic data confirm the link between inadequate sleep and metabolic disorders such as obesity and type 2 diabetes^80,81, both major precursors of cardiovascular disease. Additional manifestations of endocrine derangements triggered by experimental sleep curtailment comprise reduced testosterone in men^48,55 and upregulation of the hypothalamic-pituitary-adrenal axis^82,83. Enhanced secretion of inflammatory cytokines such as C-reactive protein and interleukin-6 occurs in response to acute or cumulative sleep loss^50,51,84,85, reflecting stimulation of pro-inflammatory pathways. Immune functions are also compromised under sleep debt conditions^56,72.

An almost inevitable companion of sleep deprivation is circadian disruption, as demonstrated by distortion of diurnal rhythmicity of metabolic, cardiovascular, and immunologic function during experimental sleep manipulation^{57–59,82,84}. Given that a 24-hour pattern of oscillation has been described in most physiological processes⁸⁶, it is not surprising that circadian misalignment may severely compromise health status. In this regard, hazards of shift work, in which circadian misalignment is often combined with chronic sleep debt, have been become increasingly apparent⁸⁷.

It is noteworthy that experimental evidence corroborating harmful cardiovascular effects of excessive sleep duration is lacking, which contributes to making any causal association between long sleep and cardiovascular disease more elusive. Controversy has indeed been raised as to whether prolonged sleep duration is truly an independent risk predictor. It has been argued that excessive sleep duration may be simply an epiphenomenon of poor health, being secondary to undiagnosed illnesses or subclinical conditions. Cross-sectional data indeed support poor perceived health in long sleepers^14,88. In addition, prolonged sleep is often reported in conjunction with a constellation of accepted cardiovascular risk factors such as older age^24,29,34, low socioeconomic status^30,34, depressive mood^7,12,24,30, hypnotic usage³⁴, prevalent sleep disorders⁷, metabolic syndrome components^22,24,34, and high-risk behaviors such as smoking^6,24, excess alcohol consumption⁶, and low physical activity^6,24,32. However, consistent with a U-shaped relationship, these potential confounders are also prevalent in those who report insufficient sleep^{6,12,24,29,30,32,34}, and therefore could similarly bias the link between short sleep and cardiovascular risk.

In agreement with this hypothesis, as mentioned above, a number of studies have unmasked modifier effects exerted by gender, age and ethnicity. Poor self-rated health⁸⁸, insomnia^30,89, sleep apnea⁹⁰, and use of hypnotics/tranquilizers⁹¹ have been found to moderate the association between sleep duration and cardiovascular outcomes.

Nevertheless, it should be noted that in numerous large cohort studies, sleep duration retained its predictive role after maximal adjustment^7,22,24, although controlling for potential confounders and mediators generally attenuated the estimates. These findings still favor a pathogenic contribution of inadequate sleep to cardiovascular vulnerability.

It is important to acknowledge that while sleep duration is a relatively crude measure of a complex entity such as sleep, other more refined indices of sleep quality and quantity such as sleep efficiency, sleep fragmentation, or slow wave sleep, have been found to be significant predictors of cardiovascular risk, independent of sleep length^16,92. Nevertheless, as these measures cannot be derived from questionnaires, but require technical instrumentations such as polysomnography or actigraphy, their suitability for epidemiologic research is limited.

Conclusions

Observational and experimental data converge to indicate that inadequate sleep duration poses a substantial hazard for cardiovascular morbidity and mortality. Sleep deficiency seems to induce adverse, sustained, and systemic alterations which conceivably act in concert to ultimately predispose to cardiovascular disease. The concomitant occurrence of conventional risk factors in individuals exposed to chronic sleep debt is likely to further amplify the risk of poor outcomes. Conversely, while long sleepers exhibit a similar pattern of predominant risk factors, the putative pathophysiological pathways leading to overt disease are still largely unknown. More mechanistic research is warranted to address this gap.

Given the pervasive and escalating prevalence of inadequate sleep and mostly sleep deficiency, the potential future burden on public health cannot be ignored. As sleep curtailment is largely voluntary and therefore modifiable, the cardiovascular complications may be preventable and plausibly reversible. Preventative strategies should therefore be undertaken to raise awareness of the possible deleterious sequelae in the general population, paralleled by interventional studies aimed at clarifying whether adherence to optimal sleep durations may improve cardiovascular risk profiles.

Key points.

• Inadequate sleep has become increasingly pervasive, and its impact on health and quality of life remains to be fully understood.

• Both extremes of sleep duration have been associated with increased prevalence and incidence of cardiovascular diseases including hypertension, coronary heart disease, and stroke.

• Aberrations in physiological functions induced by abnormal sleep may explain this association, along with enhanced prevalence of established cardiovascular risk factors.