Sleep – An Affair of the Heart

AN ASSOCIATION BETWEEN EXTREMES OF SELF-REPORTED SLEEP DURATION AND ELEVATED MORTALITY RISK WAS FIRST REPORTED OVER 40 YEARS ago.¹ This U-shaped association with mortality has since been confirmed in over a dozen studies including the Japan Collaborative Cohort Study (JACC). Similar associations have been reported for a number of other outcomes including coronary events, hypertension, diabetes, and obesity.² This consistent pattern, no matter the outcome examined, has led some to suggest that rather than reflecting true effects of too little or too much sleep, extreme responses to a question about habitual sleep may be a proxy for poorly measured predictors of overall health status such as socioeconomic status, mental health or behavioral factors.³

Identifying a specific pathway between altered sleep and mortality would help convince doubters that altered sleep has specific pathophysiologic effects rather than just being a marker of poor health. In this issue of SLEEP, Ikehara et al. use data from the JACC cohort to help address the issue by examining the association between sleep duration and specific causes of death.⁴ If sleep duration is associated with certain causes of death but not others, it would suggest that specific pathways do exist. In fact, the authors found evidence for an association between sleep duration and deaths from cardiovascular disease (CVD) but no association with cancer deaths. This fits well with prior research from the Nurses Health Study (NHS) where sleep duration predicts incident coronary events but not breast cancer.^5,6

There are several other notable findings from the Ikehara study. First, as in many prior studies, long sleep duration appears to be even more strongly associated with adverse outcomes than short sleep. While it has been suggested that long sleep durations may simply reflect terminal illness, the robustness of the associations after excluding the first 5 years of follow-up argue against this explanation. Data from the NHS 2 cohort suggest estimates of the long sleep-mortality relationship may be particularly sensitive to confounding by socioeconomic status, depressed mood and use of medications such as anti-depressants and anxiolytics.⁷ Unfortunately, as in most prior studies in this field, the assessment of these key covariates in JACC is relatively superficial. An important step in understanding whether prolonged sleep does increase mortality risk would be to demonstrate whether individuals who report sleeping 10 or more hours a day actually do sleep more than normal sleepers.

How might extremes of sleep duration impact CVD? Short term experiments have found sleep restriction leads to increased sympathetic tone, hunger dysregulation, and insulin resistance.^8–10 However, these mechanisms don't explain the association with long sleep. One possible mechanism that might explain the increased mortality risk in both reduced and prolonged habitual sleepers is an increase in inflammation. A recent analysis from the Cleveland Family Study found tumor necrosis factor α levels were inversely associated with total sleep time on polysomnography the night prior to sampling while C-reactive peptide and interleukin-6 levels rose with increasing amounts of habitual sleep time.¹¹ Given the key role of inflammation in atherogenesis, further studies of the relationship between sleep duration and inflammatory mediators are needed.

An intriguing finding in the JACC study is the substantial gender differences observed. In particular, women appeared much more susceptible to the adverse cardiac effects of short sleep. Women sleeping less than 5 hours were 2.3-fold more likely to have a fatal coronary event than 7-hour sleepers, while no evidence for an elevated risk in short sleeping men was found. This gender difference replicates findings from Germany where compared to 8-hour sleepers, women sleeping 5 or fewer hours had a 3-fold increase in incident myocardial infarction while no association was observed in men.¹² Similarly, in the Whitehall II cohort, an association between reduced sleep and incident hypertension was observed in women but not men.¹³ Further studies of whether this simply reflects gender differences in the accuracy of self-reported sleep habits or a true biologic difference in cardiac susceptibility to sleep deprivation are needed. Unfortunately, much of the research on the cardio-metabolic effects of short term sleep deprivation has thus far focused exclusively on men.^9,10

Another notable finding from the JACC study is that sleep duration appears even more strongly associated with non-cardiovascular non-cancer deaths than total CVD deaths. Further analyses, though likely underpowered, would be useful to provide insight into which specific causes of death are responsible for this association. Given data linking sleep deprivation with minor infections,¹⁴ more severe infectious complications such as pneumonia and sepsis may be another effect of poor sleep worthy of further study.

Several limitations to the current study should be noted. First and foremost, it relies on self-report rather than objective measures of sleep duration. The accuracy of self-reported sleep duration has been shown to vary by many factors including gender, race, actual sleep duration, sleep apnea and depression, raising concerns that identified associations may have nothing to do with sleep duration per se. However, recent data from the Coronary Artery Risk Development in Young Adults (CARDIA) study confirms a strong association between reduced sleep durations measured actigraphically and incident coronary calcification as a sensitive marker of coronary atherosclerosis.¹⁵ A limitation common to both the JACC and CARDIA studies as well as all prior studies associating extreme sleep durations with mortality has been the lack of accurate assessment of sleep apnea severity, making residual confounding by this prevalent disease a substantial concern.

That extremes of self-reported sleep predict cardiovascular morbidity and mortality has now been confirmed. What is now needed is a comprehensive sleep assessment in a cohort being followed for development of CVD. Such an assessment should include not only objective assessments of sleep duration and sleep apnea severity, but also assessments of the timing and regularity of sleep patterns given the growing link between circadian rhythm alterations and CVD. Until such a full assessment of sleep is performed, dissecting out which aspects of sleep are most relevant for cardiovascular health will remain unclear.

DISCLOSURE STATEMENT

Dr. Patel has indicated no financial conflicts of interest.