Fluctuations in Sleep Duration and Timing and Cardiometabolic Risk

Abstract

Sleep behavior has emerged as an important determinant of cardiometabolic health. However, to date, much attention has focused on sleep duration with accumulating evidence resulting in leading medical organizations to include adequate sleep duration in their recommendations for disease risk prevention and health promotion. However, sleep is a multidimensional construct that extends beyond sleep duration and includes factors related to the variability of duration but also to the regularity in its timing across days. These concepts, termed sleep duration variability (day-to-day differences in sleep amounts) and sleep timing regularity (day-to-day differences in sleep timing), can influence the circadian system and have independent health effects beyond sleep duration per se. Herein, we assess the literature evaluating the association of fluctuations in sleep behaviors over time and cardiometabolic risk factors and their potential implications for chronic disease development. We conclude that large-scale population-based studies support an adverse relation between fluctuations in sleep behaviors and cardiovascular disease risk markers but caution that causality should be evaluated in clinical intervention studies.

Graphical Abstract

Introduction

Good sleep is essential for cardiometabolic health. However, suboptimal sleep has only recently been recognized as an important, modifiable cardiometabolic risk factor. In 2010, the American Heart Association (AHA) developed the Life’s Simple 7 concept: 7 factors that should be considered in evaluating cardiometabolic health¹. Those factors included achieving a healthy body weight, maintaining an active lifestyle and eating a healthy diet, not smoking, and having blood pressure, blood glucose, and total cholesterol in the healthy range. Notably, this list did not include any aspect of sleep behavior. In 2022, recognizing the importance of sleep for cardiometabolic health, the AHA added sleep duration to that list, effectively upgrading Life’s Simple 7 to Life’s Essential 8². It is important to note, however, that only sleep duration was considered in developing Life’s Essential 8. A more recent statement from the AHA recognized that other dimensions of sleep were also relevant for cardiometabolic health, noting that, in addition to duration, considerations should be given to regularity, satisfaction, alertness, efficiency, disturbances, and architecture³. Of those dimensions, day-to-day fluctuations in sleep patterns have received considerable attention when large-scale population-based studies noted that individuals who work non-conventional day hours (eg. night or shift workers) had increased risk of cardiometabolic disorders^4,5. Importantly, sleep timing regularity is a stronger predictor of mortality risk than sleep duration⁶, and it does not correlate with sleep duration⁷, suggesting that this sleep metric captures another dimension of sleep as a cardiometabolic risk factor and raises the question whether variable sleep could be a modifiable target for improving cardiometabolic health⁸.

Day-to-day fluctuations in sleep behavior can be defined in terms of variation in the duration of sleep (sleep duration variability) or in terms of variation in the timing of the sleep episode over a period of time (sleep timing regularity) (Figure 1). For duration, the standard deviation (SD) of total sleep time, measured using sleep tracking devices (accelerometers) over multiple days usually defines its variability. However, inconsistency in timing of sleep can be defined by the SD of its onset (bedtime), offset (wake time), or midpoint (timing halfway through the sleep period). It can also be defined as social jetlag: the difference in the timing of sleep midpoint on weekends vs weekdays. Another construct has been proposed to define fluctuations in sleep behavior: the sleep regularity index (SRI)⁷. This index factors sleep and wake periods across the full 24-h day, allowing accounting of naps, and quantifying the overall day-to-day reproducibility of the timing of the sleep or wake state. This index is scored on a scale of 0 to 100 where 0 represents random sleep and wake times and 100 represents sleep and wake times that are identical across days. Finally, other measures of fluctuations in sleep patterns include intradaily variability (IV), a circadian rest-activity metric that refers to frequency of rest and activity periods across a 24-h day, with higher values reflecting less consolidated rest and activity periods; and interdaily stability (IS), which indicates synchronization of behaviors to light and other environmental and behavioral cues across days with higher values reflecting more stable rest and activity over time. Relative amplitude refers to the robustness of the circadian variation of activity levels, with higher values reflecting higher daytime activity, more restful sleep, or both.

Figure 1.

Variability in sleep can be defined in terms of sleep variability (alterations in duration of sleep across days), sleep regularility (alterations in timing of sleep across days), and circadian rest activity rhythms (relative amplitude, interdaily stability, and intradaily variability of rest and activity patterns within and between days). More irregular and variable sleep, low relative amplitude and interdaily stability, and high intradaily variability of activity are associated with worse glycemic control and weight management, which underscore higher risk of cardiovascular disease observed in population-based studies.

Using data from the National Health and Nutrition Examination Surveys (NHANES) 2011–2012 and 2013–2014, Price and colleagues reported that 43% of school-aged children (≥6 y of age) and adults had sleep duration SD >60 min, 20% had sleep midpoint SD ≥60 min, and 43% had social jetlag ≥60 min⁹. In the adult population from the same surveys of NHANES, average SRI was 61.3¹⁰, considered as “irregular sleep”⁷. These population-based analyses show that milder forms of sleep irregularity, distinct from what is induced by night work or transmeridian travel, are prevalent in modern society. The goal of this review is to assess the implications of these common sleep behaviors and report on recent findings (past 5 years) assessing the relation between sleep inconsistencies and cardiovascular risk in the general population.

Inconsistent Sleep Duration and Timing and CVD Incidence

Studies of accelerometer measured sleep have found that sleep variability is strongly related to increased cardiovascular risk. Cardiovascular events such as CVD, myocardial infarction, stroke, and heart failure are more prevalent in individuals with irregular compared to those with regular sleep timing. Using data from the UK Biobank, Huang and colleagues evaluated the prospective association of sleep duration SD with risk of major adverse cardiovascular events (MACE), myocardial infarction, and stroke¹¹. A total of 2,310 MACE occurred over an average 7.5 y of follow-up. The risk of MACE increased starting at >45 min of sleep duration SD, with a linear trend such that individuals in the highest sleep duration SD category (>90 min) had 40% increased risk of CVD (95%CI, 1.18–1.65). For every 1 h increase in sleep duration SD, individuals had 19% increased risk of CVD (95%CI, 1.10–1.27), 23% increased risk of myocardial infarction (95%CI, 1.11–1.35), and 17% increased risk of stroke (95%CI, 1.05–1.29) after adjusting for sociodemographic factors and family history. Further adjustments for lifestyle behaviors, comorbidities, and sleep-related factors slightly attenuated the relation but it remained statistically significant. This is essential as individuals with perception of poor sleep health are generally less optimistic, less likely to engage in healthy activities, and more likely to engage in poor coping mechanisms¹².

Chaput and colleagues also noted that sleep regularity, defined using the SRI, was a strong independent predictor of future MACE¹³. As observed by Huang and colleagues in their evaluation of the association of sleep duration variability with CVD¹¹, including sleep duration in models assessing the relation between sleep regularity and risk of MACE attenuated the association but did not fully account for the increased risk¹³. Interestingly, adequate sleep duration partly offset the increased cardiovascular risk for moderately irregular sleepers (defined as having a SRI between 71.6 and 87.3), but not for irregular sleepers (SRI <71.6)¹³. Together, these studies suggest that sleep duration variability and sleep timing regularity, measured using objective methods, may be more important drivers of cardiovascular risk than sleep duration.

A limitation of the UK Biobank is its lack of racial and ethnic diversity. However, the Multi-Ethnic Study of Atherosclerosis (MESA) provides information on associations between nighttime sleep variability and regularity and cardiovascular disease (CVD) risk in its diverse cohort. Data from the MESA sleep ancillary study, which assessed sleep using wrist actigraphy, revealed that each hour increase in sleep duration SD was associated with a 25% higher risk of CVD (95%CI, 0.98–1.71) and that each hour increase in sleep onset SD was associated with a 10% higher risk of CVD (95%CI, 1.02–1.19)¹⁴. Since a prior study noted interactions between diet quality, sleep disorders, and cardiovascular mortality¹⁵, and the authors previously noted lower dietary quality in individuals with greater sleep duration SD¹¹, the joint association of diet quality and sleep regularity on CVD incidence was examined. When diet quality was also considered in this investigation, the authors noted that individuals with both low diet quality (defined using the Alternate Healthy Eating Index 2010) and variable sleep timing (>60 min SD sleep onset) had 56% higher risk of CVD (95%CI, 1.03–2.37), while low diet quality with variable sleep duration (>90 min sleep duration SD) increased the risk of CVD by 70% (95%CI, 1.09–2.67) when compared to those with a healthy diet and regular sleep¹⁴. This highlights the combined effect of those two lifestyle behaviors on CVD risk.

MESA also provided some insight into the association of sleep variability and sub-clinical markers of CVD risk. In cross-sectional analyses, individuals with variable sleep duration (SD >120 min) were 1.40 times more likely (95%CI, 1.09–1.81) to have a high coronary artery calcium (CAC) burden (score >300) compared to those with less variable sleep duration (SD ≤ 60 min)¹⁶. They were also more likely to have carotid plaque (progressive ration [PR] 1.12; 95%CI, 1.01–1.23) and abnormal ankle brachial index (PR 1.91; 95%CI, 1.12–3.26). Sleep timing regularity was similarly associated with increased sub-clinical markers of CVD: those with more irregular sleep (SD >90 min) had 1.43 times higher likelihood of having high CAC burden (95%CI, 1.10–1.86). Importantly, in sensitivity analyses, these associations remained after adjusting for indicators of sleep quality (sleep duration, severe obstructive sleep apnea, and sleep fragmentation)¹⁶.

Relatedly, circadian rest activity rhythms (CRAR) are critical and understudied factors with multiple health implications. Alterations in sleep/wake patterns are strong determinants of CRAR which may underlie the adverse health consequences of sleep variability and irregularity. Indeed, data from the UK Biobank revealed that lower CRAR was significantly associated with CVD incidence over 6.4 y follow-up¹⁷. In over 92,000 participants with accelerometer data collected over 7 consecutive days, 12,294 CVD diagnoses occurred. The authors reported that relative amplitude of CRAR showed the strongest association with most health outcomes of all CRAR metrics and individuals with low relative amplitude of CRAR had 11% higher risk of developing CVD compared to those with high relative amplitude (95%CI, 1.05–1.16)¹⁷. Similar findings were observed in a UK Biobank investigation restricted to adults with type 2 diabetes¹⁸. In this group, those in the lowest quartile of IS had 40% higher risk of developing CVD compared to those in the lowest quartile (95%CI, 1.04–1.88) while those in the lowest quartile of relative amplitude had a 245% higher risk (95%CI, 1.73–3.49).

Inconsistent Sleep Duration and Timing and Cardiometabolic Risk Factors

Type 2 Diabetes and Glucose Tolerance

Interest in the relation between sleep variability, sleep regularity, and cardiometabolic risk factors began with the findings from cross-sectional analyses that, for every hour increase in sleep duration SD and sleep timing SD, the odds of metabolic syndrome increased by 27% (95%CI, 1.10–1.47) and 23% (95%CI, 1.06–1.42), respectively¹⁹. The study, performed in the context of the MESA sleep ancillary study, also showed similar risk of incidence of metabolic syndrome in prospective analyses from a median follow-up of 6.3 y. When the authors evaluated specific factors of the metabolic syndrome in these middle-aged adults, they found that, after adjusting for age, sex, race, study site, education, and work schedule, the prevalence of elevated fasting plasma glucose increased by 12% (95%CI, 0.97–1.29) for every hour increase in sleep duration variability and by 26% (95%CI, 1.08–1.47) for every hour increase in sleep timing irregularity after adjusting for lifestyle variables, including sleep disorders. In the prospective analyses, with less than half of the sample size of the cross-sectional analyses, no association was observed between sleep duration or sleep timing SD on fasting plasma glucose.

Risk of type 2 diabetes (T2D) was more specifically studied using actigraphy data from the UK Biobank^20,21. In this cohort of middle-aged adults in the UK, sleep was also measured using wrist-worn accelerometry over a 7-d period. Sleep variability was assessed in one study as the SD of sleep duration²¹ while sleep regularity was calculated using the SRI and categorized as irregular (SRI <71.6), moderately irregular (SRI 71.6–87.3), and regular (SRI >87.3)²⁰. Information about the development of T2D was obtained from self-report and medical records over an average follow-up period of approximately 7 y in both studies. In fully adjusted models, individuals with sleep duration SD >60 min had 34% higher risk of developing T2D²¹ while individuals classified as moderately irregular and irregular sleepers had 35% (95%CI, 1.19–1.53) and 38% (95%CI, 1.20–1.59) higher risk of developing T2D, respectively, compared to regular sleepers²⁰. When sleep irregularity was evaluated over a continuous scale, risk reduction was observed starting at SRI of 87.3, which corresponded to 18% lower risk of developing T2D. Risk was reduced by 25% if SRI of 90 was achieved. The authors also evaluated whether sleep duration moderated the association between SRI and T2D incidence risk and noted slightly attenuated risk in those meeting sleep duration recommendations of 7–9 h/day (adults 18–64 y) or 7–8 h/day (adults ≥65 y). In other words, compared to those not achieving sleep duration recommendations, those with adequate sleep duration had lower risk of T2D in each SRI stratum; those in moderate irregular and irregular sleep categories remained at higher risk of T2D compared to those meeting sleep duration recommendations with regular sleep. This study highlights the importance of sleep regularity, above and beyond sleep duration, for T2D prevention.

While these large-scale population-based studies suggest that sleep variability and irregularity are associated with risk factors for T2D¹⁹ and could be involved in the development of T2D²⁰, a case-control study of patients with T2D and age- and gender-matched healthy adults did not find any differences in measures of sleep timing regularity, assessed using accelerometry over an average of 9 d, between cases and controls²². Self-report measure of social jetlag, but not objectively measured social jetlag, was greater in T2D patients compared to controls. Contrary to expectations, within patients, there was a negative association between sleep midpoint irregularity and sleep duration variability and HbA1c. However, this study differed from others in using standard error of the mean rather than standard deviation to define fluctuations in sleep duration and timing and was performed in patients with low overall sleep variability and irregularity. These limitations, along with the small sample size (<30/group), may in part explain the incongruent findings. Furthermore, data differ from those of Brouwer et al.²³ who reported strong associations between sleep duration variability, measured over 7 d of wrist actigraphy, and HbA1c in patients with T2D. Patients in this study had higher HbA1c (average 7.3%²³) than those in the prior study by Kelly and colleagues (47 mmol/L, or 6.4%)²².

A small observational study of approximately 65 university professors in Brazil noted that irregularity in weekday vs weekend timing of sleep, i.e. social jetlag, was correlated with fasting glucose levels²⁴ despite participants having, on average, 1 h social jetlag and sleep duration >7 h/night on both weekdays and weekends. These participants also had sleep efficiency of 91%, showing good general sleep quality. The study was limited in the extent of sleep variables assessed and social jetlag was determined from questionnaire, rather than objective measures. Furthermore, the study did not adjust for any covariates. Thus, while the data are in line with other studies, these results require replication using more robust methodological approaches.

Finally, two observational studies were performed to examine measures of glucose tolerance in groups of young adults^25,26. Culver and colleagues assessed sleep using wrist actigraphy for at least 6 d in 44 healthy adults, on average 23 y of age and body mass index (BMI) 26 kg/m^{2 25}. There was a significant correlation between sleep duration SD and fasting plasma glucose, but this correlation was reduced to a trend after adjusting for sex and BMI. It is important to note that despite of the lack of significance, the strength of the correlation was not affected by the adjustment for age and BMI. Given the small sample size, it is possible that the study was underpowered to detect covariate-adjusted correlations. McNeil and colleagues, on the other hand, did not find any relation between measures of sleep fluctuations, assessed over at least 3 d, and cardiometabolic risk factors, including HOMA-IR, in 147 emerging adults (age 19.4 y)²⁶. Sleep duration, however, was inversely associated with HOMA-IR. This cohort was different from others in that insulin resistance was very high (average 7.3) and sleep duration very short (average 5.4 h). Moreover, fluctuations in sleep duration and timing were defined as a percentage of sleep duration and sleep timing midpoint, respectively, making comparisons with other studies difficult. From those two studies, no consensus can be made on the relation between sleep irregularity and reduced glucose tolerance in young adults.

Only one intervention study was conducted to investigate the impact on glucose homeostasis of recurrent periods of weekday short sleep with weekend recovery sleep²⁷. Healthy adults underwent one week of stable 8 h sleep and were then randomized to maintained stable 8 h schedule for 16 d or one of 2 periods alternating short sleep for 5 d (average 6 h time in bed) and 8 h sleep for 2 d. In one of those 2 conditions, stable short sleep of 6 h time in bed was maintained for the 5 d periods while in the other, sleep varied in duration (8, 4, 8, 4, and 6 h time in bed). These 16-d periods were done in an inpatient setting with controlled diets designed to maintain body weight. Wake time remained constant in each condition. Glucose homeostasis was measured using a glucose tolerance test. The authors found no group x day interactions for any of the glucose and insulin measures. However, within-group changes revealed modest increases in glucose area under the curve in the variable short sleep condition but not in the stable short sleep condition. The authors concluded that variable short sleep condition produced potential deficits in pancreatic beta cell function. However, the lack of difference between groups is important to note. Furthermore, the study was performed in controlled dietary conditions, in healthy young adults (age 21–35 y, BMI 18.5–24.9 kg/m²) and does not fully reflect real-life conditions of individuals with short and variable sleep.

These findings notwithstanding, currently, there are robust, well-powered, population-based studies showing associations between sleep irregularity and risk of T2D and that sleep variability and regularity are relevant for glycemic control in patients with T2D. However, some findings are not uniformly supported by smaller-scale studies of glycemic control in younger age groups who do not have T2D.

Body Mass Index

Numerous studies have found significant associations between inadequate sleep duration and elevated BMI²⁸, however, variable sleep has not been featured as prominently in the examination of associations between sleep and BMI. Increased risk of incident obesity and larger increase in BMI over 3-y follow-up has been noted in those with higher sleep duration SD participating in the UK Biobank after adjusting for several lifestyle factors, co-morbidities, and sleep duration²⁹. Data from MESA sleep ancillary also revealed a 22% higher incidence of central obesity with every 1 h increase of sleep duration variability but this did not reach statistical significance (95%CI, 0.87–1.72)¹⁹. The UK Biobank prospective cohort study also evaluated the associations of CRAR associations with future health risks and noted the low CRAR group (defined as having relative amplitude <0.802–0.844, depending on age) had greater prevalence of obesity at baseline compared to those in the high CRAR group¹⁷. The low CRAR group also had higher intradaily variability (IV) and was more likely to have sleep duration <7 h. In addition, a study by Makarem et al.³⁰ used a multi-dimensional sleep health (MDSH) score comprising sleep regularity and variability, sleep duration, daytime sleepiness, symptoms of sleep disorders, and difficulty falling asleep to examine its associations with cardiometabolic health and CVD. The authors found that participants with ideal and moderate MDSH had significantly lower BMI and smaller waist circumference compared to those with poor MDSH but neither sleep duration nor sleep regularity/variability was related to BMI. When the sleep regularity/variability score was disaggregated between regularity and variability, only the weekday-weekend difference in sleep duration was related to BMI³⁰.

Sleep variability also emerged as a relevant marker of weight status in the Child Health Checkpoint study in Australia¹². This study recruited children and one of their parents (87% female) who completed 8 days of sleep monitoring using wrist actigraphy. Sleep duration variability was evaluated as the coefficient of variation of the sleep period across days. Having more variable sleep duration was associated with higher BMI even after adjusting for covariates and other objective sleep measures. A pooled analysis of 2 studies noted a mean difference of 2.17 cm (95%CI, 0.61–3.37) in waist circumference between participants with ≥2 h of social jetlag compared to those with <1 h³¹.

Only one study to date is available to determine whether changes in sleep regularity are related to changes in weight status and adiposity, providing some indication of causality³². This study enrolled women and randomized them to sleep restriction vs maintained adequate sleep interventions for periods of 6 wk each. As part of the interventions, women were given guidelines for bedtimes and waketimes to ensure adherence to the sleep duration target. Using data from the adequate sleep arm of the study, the investigators noted differences among women in terms of change in sleep timing regularity relative to a 2-wk baseline screening period. Compared to women who increased or did not change their bedtime regularity, those whose bedtimes became more regular had a reduction in body weight, and total and subcutaneous adiposity, measured using magnetic resonance imaging. No difference in change in skeletal muscle was noted between groups. This study provided some of the first indications that sleep timing regularity could impact weight status in women.

More recently, studies have examined the role of sleep duration variability and sleep timing regularity in weight management. Carvalho et al. noted that greater social jetlag in timing of sleep was associated with lesser weight loss at 6 mo and 1 y post-bariatric surgery in a cohort of mostly young female patients³³. Similarly, in an observational study evaluating the association between social jetlag and change in weight status over time, Minabe et al. noted less weight and body fat loss from the weight loss intervention in those with social jetlag ≥2 h compared to <1 h³⁴. At a higher scale, data from the Prevencion con Dieta Mediterranea (PREDIMED)-Plus trial showed that individuals with higher sleep duration variability, measured using wrist-actigraphy for 8 consecutive days, had smaller decreases in body weight and BMI at 12 mo following the weight loss intervention³⁵. Overall, these findings suggest that fluctuations in sleep behaviors are relevant factors to consider in weight management interventions.

Blood pressure

Sleep regularity has also been evaluated in the context of hypertension and blood pressure. Data from NHANES 2017–2018 showed that the odds of hypertension were lower in those with ideal (0.38; 95%CI, 0.28–0.52) and moderate (0.59; 95%CI, 0.42–0.82) MDSH compared to those with poor MDSH³⁰. There was no association with the sleep regularity component of MDSH (OR 1.19; 95%CI, 0.93–1.51). When examining blood pressure, individuals with ideal, compared to those with poor MDSH, had 4.22 mmHg lower systolic blood pressure (95%CI, −6.95- −1.49) and 3.42 mmHg lower diastolic blood pressure (95%CI, −5.38- −1.47), while those with moderate MDSH had 2.60 mmHg lower systolic (95%CI, −5.06- −0.14)) and 2.61 mmHg lower diastolic pressure (95%CI, −4.33- −0.90) compared to those with poor MDSH. Results indicate a stronger association between poor overall sleep health and systolic and diastolic pressure than for individual components of the MDSH. In fact, sleep regularity scores were only tending to be associated with lower diastolic blood pressure. Despite this, each 1 h increase in weekday-weekend difference in sleep duration (sleep variability) was associated with higher diastolic blood pressure. These findings are in line with those from the UK Biobank cohort showing that individuals with low relative amplitude of CRAR had higher prevalence of hypertension compared to those with high relative amplitude¹⁷.

Some smaller-scale studies have evaluated associations between sleep regularity and variability metrics and blood pressure but have resulted in contradictory findings. Culver and colleagues found that variable sleep duration in young adults (age 18–39 y) was significantly associated with higher brachial systolic and diastolic blood pressure, as well as with aortic systolic blood pressure²⁵. However, contrary to expectations, Kelly et al.²² observed that elevated systolic blood pressure was associated with lower waketime variability, higher IS, and lower IV in healthy participants (ie. more regular sleep). In both T2D patients and healthy individuals, higher systolic blood pressure was associated with having less social jetlag. The causes of contradictions between the two studies remain to be determined.

One earlier clinical intervention study noted blunting of overnight blood pressure dipping in recurrent periods of severe sleep restriction (4 h time in bed) for 3 nights followed by one night of recovery sleep (8 h time in bed)³⁶. Compared to a group maintaining 8 h time in bed for 22 days, participants undergoing 4 blocks of alternating short and adequate sleep had a smaller diastolic blood pressure dipping overnight at each block of sleep restriction. Conversely, a proof-of-concept study evaluated the impact of a 2-wk bedtime regularization intervention on 24-h ambulatory blood pressure in adults with hypertension³⁷. The study showed that reducing bedtime SD from 32 to 7 min, decreased 24-h and nighttime systolic and diastolic blood pressure and reduced daytime diastolic blood pressure with no effect on heart rate. The authors described the extent of the blood pressure reductions to be similar to what is observed with >4 wk of regular exercise or salt reduction. However, the study was short, lacked a control group, and had a small sample size (n=11). Nonetheless, the data are promising, and larger-scale interventions are worth pursuing.

Inflammation and lipid profile

Studies evaluating self-reported and actigraphy measured sleep have found that sleep timing irregularity is associated with inflammation and worse lipid profiles. In evaluating associations between CRAR metrics and cardiometabolic health markers in emerging adults, Hoopes and colleagues found that each 1-SD increase in IS was inversely associated with total (−10.23 mg/dL; 95%CI, −17.71- −2.75 mg/dL) and non-HDL cholesterol (−7.45 mg/dL; 95%CI, −14.44- −0.47 mg/dL) in adjusted analyses³⁸. In addition, each 1-SD increase IV was positively associated with total cholesterol (8.57 mg/dL; 95%CI, 0.71–16.43 mg/dL) and C-reactive protein concentrations (0.38 mg/dL; 95%CI, 0.10–0.66 mg/dL). The study’s findings imply that consistency in day-to-day rest and activity behaviors are associated with better cardiometabolic health in younger adults³⁸. However, these findings are not limited to this population group. Similar findings were observed in patients with T2D, in whom more variable sleep duration, but not high sleep timing irregularity, was associated with elevated C-reactive protein³⁹. In a population-based cohort from the Czech Republic, Sládek and colleagues found that higher social jetlag was significantly associated with both higher total and LDL cholesterol levels, particularly in participants >50 y⁴⁰. Similarly, data from the MESA sleep ancillary study showed that every 1 h increase in sleep onset timing irregularity was associated with a 2.07 higher odds of developing high triglycerides over a median follow-up of 6.3 y (95%CI, 1.38–3.11); the odds of developing high triglycerides for every 1 h increase in sleep duration variability did not reach statistical significance (1.47; 95%CI, 0.98–2.20)¹⁹.

Interestingly, in a cohort of nurses, Slavish et al. reported that greater sleep variability was associated with higher interleukin 6 and interleukin-1β but not C-reactive protein or tumor necrosis factor-α after adjusting for covariates⁴¹. Findings were similar whether sleep was assessed from wrist actigraphy or self-report diaries, and shift work did not moderate any of the associations. Moreover, similar associations were not observed when mean sleep duration was evaluated as a predictor of inflammation, supporting a unique role of sleep variability in modulating inflammatory status⁴¹.

As such, intervening on sleep timing regularity may be a viable option to reduce inflammatory state. St-Onge et al. found that inflammation and adiposity could be reduced, thus improving cardiometabolic health, by implementing more regular bedtime routine while maintaining adequate sleep duration³². In their study, women whose bedtimes became more regular over a 6-wk period had reductions in leukocyte platelet aggregates while women who did not change or had increasingly irregular bedtimes had increased leukocyte platelet aggregates levels, indicative of inflammatory state. These findings suggest that increased inter-night stability could reduce inflammation, lowering cardiometabolic risk.

Mechanisms of Increased Cardiovascular and Metabolic Risk in Unhealthy Sleep

Nightly fluctuations in sleep schedules lead to desynchronization of sleep-wake timing and circadian disruption^38,42–44, which can interrupt all major cardiovascular functions, including vascular tone and endothelial function, thereby increasing the risk of CVD^45–47. Inflammatory markers, including white blood cell count, interleukin-6 and C-reactive protein, are elevated in a cross-sectional sample of individuals with irregular and variable sleep^48–50. Fluctuations in sleep duration and timing are associated with endothelial dysfunction, an early step in the development and progression of CVD^51,52. Prolonged, mild sleep restriction achieved by a bedtime delay, a prominent feature of irregular sleep⁷, increased endothelial cell (EC) inflammation and oxidative stress while unexpectedly failing to activate antioxidant responses in healthy women^53,54. Increased endothelial oxidative stress in the absence of appropriate upregulation of antioxidant responses after mild sleep restriction in healthy women caused endothelial dysfunction that over time increases CV risk^53,54. These findings are in stark contrast with model organisms whose antioxidant responses are strongly activated after insufficient and disturbed sleep, thereby highlighting the importance of studying the effects of unhealthy sleep on vascular function directly in humans^55–58.

Interestingly, simply adhering to a stable bedtime for 6 wk while maintaining habitual normal sleep duration of 7–9 h/night tends to reduce oxidative stress and inflammation in ECs³². This beneficial effect was observed inadvertently in healthy women randomized to maintain habitual sleep duration during a crossover trial of mild sleep restriction vs. habitual adequate sleep. To ensure sleep duration of 7–9 h/night was maintained, the investigators advised participants to follow a fixed bedtime and wake schedule^32,53,54. Those who adhered to a stable sleep schedule had reduced levels of endothelial oxidative stress and inflammation compared to baseline³². In contrast, participants who maintained their irregular sleep schedule had no change in levels of endothelial oxidative stress and inflammation compared to baseline. Importantly, both groups maintained adequate, habitual sleep duration. These findings suggest that irregular sleep timing, a highly prevalent yet frequently overlooked behavioral pattern, activates endothelium independently from sleep duration and that a stable sleep schedule, a simple, low-cost lifestyle intervention, may alleviate these alterations. In fact, maintaining a stable sleep schedule is a key component of clinical sleep hygiene recommendations⁴³. Emerging data suggest that these recommendations, which were developed to improve sleep quality, may also be useful as a cardiometabolic health promotion strategy.

Mechanisms underlying increased metabolic risk in variable sleep in humans remain unclear as most insights into weight gain, decreased insulin sensitivity and impaired compensatory insulin secretion are derived from animal models with organ-specific silencing of clock genes or mice fed under conditions of central and peripheral pacemaker misalignment^59–67. These models do not mimic well variable sleep in humans. Two large-scale population-based studies of older adults have noted associations of rest-activity rhythms with metabolites from various biochemical pathways implicated in chronic diseases, particularly immune and inflammatory diseases^68,69, which could shed some light into mechanisms. Thus, future research in this area is needed. Moreover, studies should be done that take into consideration dietary and physical activity patterns, which may be associated with sleep variability and regularity, and could be implicated in the mechanistic pathway to adverse cardiometabolic health.

Conclusion

Observational studies using subjective and objective metrics of sleep behaviors suggest a role of fluctuations in sleep behaviors in the development of CVD. Greater sleep variability and irregularity are associated with a higher risk of type 2 diabetes, obesity, and pro-atherogenic lipid and inflammatory profile in most studies. Not surprisingly, inconsistent findings have been reported in specific populations, including young adults with very short sleep duration and patients with very well controlled T2D. The effects of sleep variability and regularity on blood pressure remain unclear. However, despite accumulating observational evidence that variable and irregular sleep are detrimental for cardiometabolic health, the causal evidence is lacking and the molecular mechanisms underlying such associations are just emerging. Fluctuations in sleep duration and timing do not necessarily reflect circadian disruptions and require further investigation. Furthermore, agreement in measures and methods to establish sleep behaviors as variable (duration) or regular (timing) are needed. For example, marked heterogeneity in methods used to calculate the SRI have been shown to result in differing classification of participants and rate estimates for health outcomes, which call into question the interpretation of the findings⁷⁰.

Finally, future studies should evaluate mechanisms and causality, aiming to isolate the independent effects of sleep duration variability from sleep duration and to also evaluate the impact of sleep irregularity on cardiovascular risk factors. A study of patients with heart failure suggest that moderately irregular sleep (SRI ≤87) post-event is associated with worse comorbidity and increased risk of experiencing an adverse clinical event (emergency room visit, hospitalization, death)⁷¹. These studies should also consider the sociodemographic context of the participants, as those factors could be confounding associations observed in population-based studies.

Table 1.

Summary of study characteristics and results by cardiometabolic risk factor.

Reference	Participant population	Study type/Length of follow-up (years)	Findings
Cardiovascular disease
Huang et al. 2025 (11)	N=86,219 57.5% females Age 47–76 y	Longitudinal Median 7.5 y	HR for major cardiovascular event by sleep duration variability (SD) compared to ≤30 min: 31–45 min, 1.10 (95% CI, 0.97–1.26) 46–60 min, 1.19 (1.04–1.37) 61–90 min, 1.33 (1.15–1.53) >90 min, 1.19 (1.18–1.65)
Chaput et al. 2025 (13)	N=72,269 57.1% females Mean age 62.1 y	Longitudinal Mean 7.8 y	HR for major cardiovascular event by sleep regularity (SRI) compared to regular sleepers: Moderately irregular, 1.08 (1.01–1.70) Irregular, 1.26 (1.16–1.37)
Potts et al. 2025 (14)	N=1,782 55% females Mean age 68 y	Longitudinal Median 8.8 y	HR for cardiovascular disease compared to regular sleep timing (SD <30 min): 30-<60 min, 0.92 (0.61–1.40) 60-<90 min, 1.08 (0.70–1.69) ≥90 min, 1.24 (0.80–1.91) HR for cardiovascular disease compared to non-variable sleep duration (SD <60 min): 60-<90 min, 1.15 (0.80–1.66) 90-<120 min, 1.23 (0.82–1.85) ≥120 min, 1.52 (0.96–2.41)
Full et al. 2023 (16)	N=1,782 55% females Mean age 68.6 y	Cross-sectional	PR for coronary artery calcium >300 compared to regular sleep timing (SD ≤30 min): 31–60 min, 1.29 (1.01–1.65) 61–90 min, 1.23 (0.93–1.62) >90 min, 1.39 (1.07–1.82) PR for ankle brachial index <0.9 compared to regular sleep timing (SD ≤30 min): 31–60 min, 1.12 (0.64–1.98) 61–90 min, 1.18 (0.64–2.17) >90 min, 1.58 (0.92–2.74) PR for coronary artery calcium >300 compared to non-variable sleep duration (SD ≤60 min): 61–90 min, 1.26 (0.99–1.59) 91–120 min, 1.32 (1.05–1.67) >120 min, 1.33 (1.03–1.71) PR for ankle brachial index <0.9 compared to non-variable sleep duration (SD ≤60 min): 61–90 min, 1.50 (0.86–2.59) 91–120 min, 1.62 (0.98–2.70) >120 min, 1.75 (1.03–2.95)
Feng et al. 2023 (17)	N=92,614 56.4% females Median age 64 y	Longitudinal Median 6.4 y	HR for incident cardiovascular diseases relative to better circadian rest activity rhythms metric: Relative amplitude, 1.11 (1.05–1.16) Interdaily stability, 0.96 (0.92–1.01) Intradaily variability, 0.98 (0.94–1.03)
Yang et al. 2023 (18)	N=3,147 40% females Mean age 65.2 y Type 2 diabetes	Cross-sectional	HR for incident cardiovascular diseases for worse, relative to best, circadian rest activity rhythms metric: Relative amplitude, 2.45 (1.73–3.49) Interdaily stability, 1.11 (1.05–1.16) Intradaily variability, 1.31 (0.97–1.75)
Type 2 diabetes and glucose tolerance
Huang and Redline, 2019 (19)	N=2,003 53.6% females Mean age 69.5 y	Cross-sectional Longitudinal	Higher odds of elevated blood glucose with lower sleep onset timing regularity No association of sleep duration variability and sleep onset timing regularity and incident elevated blood glucose
Chaput et al. 2024 (20)	N=73,630 56.1% females Mean age 62.2 y	Longitudinal Mean 7.7 y	HR for incident type 2 diabetes by sleep regularity (SRI) compared to regular sleepers: Moderately irregular, 1.35 (1.19–1.53) Irregular, 1.38 (1.20–1.59)
Kianersi et al. 2024 (21)	N=84,421 57.4% females Mean age 62.3 y	Longitudinal Median 7.5 y	HR for incident type 2 diabetes compared to non-variable sleep duration (SD ≤30 min): 31–45 min, 1.08 (0.93–1.25) 46–60 min, 1.10 (0.94–1.27) 61–90 min, 1.20 (1.03–1.40) ≥91 min, 1.16 (0.97–1.39)
Kelly et al. 2022 (22)	N=27 healthy N=30 with type 2 diabetes 49.2% females Mean age 54.1 y	Between-group comparisons	Compared to healthy adults, adults with type 2 diabetes had greater social jetlag No difference in circadian rest activity rhythm metrics, sleep variability, or sleep regularity between groups
Brouwer et al. 2020 (23)	N=172 with type 2 diabetes	Cross-sectional	High sleep duration variability, but not low sleep timing regularity, was associated with worse HbA1c
Galeno et al. 2023 (24)	N=103 60.2% females Mean age 44 y	Cross-sectional	Social jetlag was associated with blood glucose concentrations
Culver et al. 2022 (25)	N=44 45.5% females Mean age 23 y	Cross-sectional	Sleep duration variability was not associated with blood glucose concentration
McNeil et al. 2024 (26)	N=147 59% females Mean age 19.4 y	Cross-sectional	Sleep duration variability and sleep timing regularity were not associated with HOMA-IR
Cheung et al. 2025 (27)	N=48 50% females Mean age 22.8 y	Clinical intervention 3 groups: Stable short sleep Variable short sleep Control adequate sleep	No effect of group x day interaction or main effect of group on plasma glucose and insulin (both for fasting and in response to glucose load) and Matsuda index
Body mass index
Kianersi et al. 2025 (29)	N=10,572 52.5% females Mean age 63.4 y	Longitudinal 3 y	Relative BMI change percentage (standardized over 3 y) compared to non-variable sleep duration (SD ≤30 min): 31–45 min, 0.20 (−0.38–0.77) 46–60 min, 0.73 (0.13–1.32) >60 min, 0.68 (0.09–1.27) RR for incident obesity compared to non-variable sleep duration (SD ≤30 min): 31–45 min, 1.10 (0.77–1.57) 46–60 min, 1.40 (0.99–1.98) >60 min, 1.60 (1.14–2.24)
Feng et al. 2023 (17)	N=92,614 56.4% females Median age 64 y	Longitudinal Median 6.4 y	Compared to high relative amplitude, those with low relative amplitude had higher prevalence of obesity.
Makarem et al. 2022 (30)	N=4,555 51.1% females Mean age 49 y	Cross-sectional	Greater weekday-weekend differences in sleep duration, but not timing, was associated with higher BMI
Matricciani et al. 2022 (12)	N=1,045 87% females	Cross-sectional	Greater sleep duration variability was associated with higher BMI
St-Onge et al. 2020 (32)	N=37 100% females Mean age 34.9 y	Clinical intervention 6-week observation	Women who spontaneously reduced their bedtime variability over a 6-wk period (n=29), compared to women who did not, had a reduction in whole body volume (measured by magnetic resonance imaging)
Carvalho et al. 2025 (33)	N=122 77.0% females Median age 33.0 y	1 y following bariatric surgery	There was a negative association between social jetlag and reduction in BMI at 6 and 1 y after bariatric surgery
Minabe et al. 2024 (34)	N=11,829 68.8% females Mean age 40.4 y	Clinical intervention Mean engagement in weight loss program 154.8 d	Social jetlag was inversely associated with degree of reduction in BMI
Papandreou et al. 2020 (35)	N=1,986 47% females Mean age 65 y	Weight loss intervention 12 mo follow-up	Compared to those in the first tertile of sleep variability (less variable), those in second and third tertile had smaller reductions in BMI
McNeil et al. 2024 (26)	N=147 59% females Mean age 19.4 y	Cross-sectional	Sleep duration variability and sleep timing regularity were not associated with BMI or fat mass index
Blood pressure
Feng et al. 2023 (17)	N=92,614 56.4% females Median age 64 y	Longitudinal Median 6.4 y	Compared to high relative amplitude, those with low relative amplitude had higher prevalence of hypertension.
Huang and Redline, 2019 (19)	N=2,003 53.6% females Mean age 69.5 y	Cross-sectional Longitudinal	No association between sleep duration variability or sleep onset timing regularity and elevated blood pressure No association of sleep duration variability and sleep onset timing regularity and incident elevated blood pressure
Makarem et al. 2022 (30)	N=4,555 51.1% females Mean age 49 y	Cross-sectional	Greater weekday-weekend differences in sleep duration was associated with higher diastolic blood pressure
Culver et al. 2022 (25)	N=44 45.5% females Mean age 23 y	Cross-sectional	Sleep duration variability was associated with brachial systolic and diastolic blood pressure and with aortic systolic blood pressure
Yang et al. 2017 (36)	N=45 51.1% females Mean age 31 y	Intervention study 4 cycles of 3 d of 4 h time in bed followed by 1 d of 8 h vs maintained 8 h time in bed (control)	Diastolic blood pressure dipping was reduced in the sleep restriction group but not in the control group Average systolic blood pressure was increased in first cycle and diastolic blood pressure in first, second, and fourth cycles of sleep restriction compared to control
Thosar et al. 2025 (37)	N=11 with hypertension 63.6% females Mean age 53 y	Intervention study Bedtime regularization for 2 wk	Bedtime regularization reduced 24-h and nighttime systolic and diastolic blood pressure; daytime diastolic, but not systolic, blood pressure was reduced
Inflammation and lipid profile
Hoopes et al. 2021 (38)	N=52 59.6% females Mean age 20 y	Cross-sectional	Interdaily stability was inversely associated with total cholesterol and non-HDL cholesterol Intradaily variability was associated with CRP
Reutrakul et al. 2024 (39)	N=35 with type 2 diabetes 54.3% females Mean age 54.3 y	Cross-sectional	Greater sleep duration variability, but not lower sleep regularity, was associated with higher CRP
Sladek et al. 2023 (40)	N=1282 Age 18–97 y	Cross-sectional	Social jetlag was associated with higher total and LDL cholesterol
Huang and Redline, 2019 (19)	N=2,003 53.6% females Mean age 69.5 y	Cross-sectional Longitudinal	No association between sleep duration variability or sleep onset timing regularity and elevated triglycerides Increased odds of incident elevated triglycerides with lower sleep onset timing regularity
Slavish et al. 2020 (41)	N=392 92% females Mean age 39.5 y	Cross-sectional	Greater intraindividual variability in sleep time was associated with higher IL-6 and IL-1β but not CRP or TNF-α
McNeil et al. 2024 (26)	N=147 59% females Mean age 19.4 y	Cross-sectional	Sleep duration variability and sleep timing regularity were not associated with CRP
St-Onge et al. 2020 (32)	N=37 100% females Mean age 34.9 y	Cross-sectional	Women who spontaneously reduced their bedtime variability over a 6-wk period (n=29), compared to women who did not, had a reduction in leukocyte platelet aggregates

Abbreviations: BMI, body mass index; CRP, C-reactive protein; HDL, high-density lipoprotein; HR, hazard ratio; IL, interleukin; LDL, low-density lipoprotein; OR, odds ratio; PR, prevalence ratio; TNF-α, tumor necrosis factor α

Highlights.

• Sleep timing irregularity and duration variability are associated with a higher risk of CVD, independent of sleep duration and other indicators of sleep quality like obstructive sleep apnea and sleep fragmentation.

• Population-based studies show associations between sleep irregularity and risk of T2D.

• Sleep variability and regularity are relevant for glycemic control in patients with T2D and weight management in adults with obesity.

• Studies are needed to provide unified definitions and cutpoints for sleep irregularity and variability and evaluate mechanisms.

List of abbreviations

BMI

body mass index

CAC

coronary artery calcium

CRAR

circadian rest activity rhythms

CVD

cardiovascular disease

EC

endothelial cell

HbA1c

hemoglobin A1c

HOMA-IR

homeostatic model assessment of insulin resistance

IS

interdaily stability

IV

intradaily variability

MACE

major adverse cardiovascular events

MDSH

multi-dimensional sleep health

MESA

Multi-Ethnic Study of Atherosclerosis

NHANES

National Health and Nutrition Examination Survey

PREDIMED

Prevencion con Dieta Mediterranea

SRI

sleep regularity index

T2D

type 2 diabetes