Circadian rhythms are endogenous cycles in physiological, hormonal, and behavioral processes largely synchronized to the external 24 hr light-dark cycle. The sleep-wake cycle is perhaps the most notable biological process that follows the 24 hr circadian rhythm (Zee and Vitiello, 2009). The timing of light exposure is the primary external driver of circadian rhythms. Therefore, rapid changes in sleep timing can cause circadian misalignment through fluctuating light-dark exposure (Zeitzer et al., 2000).

Circadian misalignment is associated with several age-related diseases, including cancer and cardiovascular disease (CVD) (Torquati et al., 2018; International-Agency-for-Research-on-Cancer, 2010; Janszky and Ljung, 2008). However, the health impacts of irregular sleep-wake timing are still emerging. This remains an important area of study since modern societal and lifestyle trends, including exposure to artificial and blue light at night, longer work hours, shift work, and the 24-7 lifestyle, have blurred the distinction between day and night, increasing the propensity for circadian disruption (Walker et al., 2020). The present study assessed the relationship between sleep regularity and the risk of incident all-cause mortality, cancer mortality, and CVD mortality in the UK Biobank (UKB). We measured sleep regularity via accelerometry to calculate the sleep regularity index (SRI), a new metric sensitive to differences in sleep-wake timing on a circadian timescale.

Table 1 displays sample characteristics. The final sample size was 88,975. There were 3010 all-cause deaths during a median follow-up of 7.1 years (Q1, Q3: 6.6, 7.6). The most common primary cause of death was cancer (n=1701, 57%) followed by CVD (n=616, 20%).

SRI and all-cause mortality

We identified a non-linear association between the SRI and all-cause mortality hazard (p [global test of spline term]<0.001) (Figure 1). Compared to the sample median (SRI = 61), mortality rates were highest among those with the most irregular sleep and decreased almost linearly as SRI approached its median, after which the decrease began to plateau (Figure 1). HRs, relative to the median SRI, were 1.53 (95% CI: 1.41, 1.66) for participants with SRI at the 5th percentile (SRI = 41) and 0.90 (95% CI: 0.81, 1.00) for those with SRI at the 95th percentile (SRI = 75), respectively. Standardized cumulative incidence curves for all-cause mortality are displayed for the SRI at the 5th percentile, median, and 95th percentile in Figure 2. There was little indication that HRs varied according to age (p interaction = 0.48), sex (p=0.36), household income (p=0.62), sleep duration (p=0.47), moderate to vigorous physical activity (p=0.13), history of CVD (p=0.48), or cancer (p=0.29).

Figure 1

Download asset Open asset

Figure 2

Download asset Open asset

There was strong evidence against the proportionality assumption in the discrete-time hazards model (p [time × SRI interaction]<0.001). Time-varying HRs for the 5th and 95th SRI percentiles compared to the median are displayed in the appendix (Appendix 3—figure 2). For the 5th percentile relative to the median, HRs were greatest in the earliest period of follow-up (HRs >2), declining until approximately 2.5 years, after which they remained approximately stable with an HR of around 1.5. There was no clear time variation in the HR for the 95th percentile of SRI vs. the median.

Comparison of SRI with sleep duration SD and sleep onset time SD

The SRI was modestly negatively correlated with the sleep duration SD (–0.32) and sleep onset time SD (–0.42; see correlation matrix in Appendix 3—table 1). Figure 3 displays HRs, relative to the median, for the SRI, sleep duration SD, and sleep onset time SD. For each measure, greater sleep irregularity (i.e., lower SRI or higher SD representing more day-to-day variability) was associated with an increased all-cause mortality rate in independent models (all p [global]<0.001). HRs, for low regularity compared to the median, were largest for the SRI (Figure 3). The addition of the SRI to a model containing both SD metrics (alongside primary model covariates) improved model fit (p [likelihood ratio test]<0.001). Conversely, the addition of sleep duration SD and sleep onset time SD to a model containing the SRI (and primary model covariates) did not meaningfully improve model fit (p [likelihood ratio test]=0.10).

Figure 3

Download asset Open asset

Among 88,975 individuals followed for a median of 7.1 years, there was a non-linear association between sleep regularity and the risk of mortality; mortality rates were highest in persons with the most irregular sleep and decreased approximately linearly as sleep regularity approached its median, after which the decrease began to plateau. Our findings were independent of several illness (including history of cancer and CVD at baseline), sleep duration, sleep fragmentation, and other confounding factors. Overall, these data indicate a relationship between sleep regularity and longevity in a large community-based cohort.

Physiological processes associated with CVD and cancer are under circadian control. Mutations or deletions to circadian clock genes such as CLOCK, PER, and BMAL1 influence BP, endothelial function, and glucose homeostasis (Viswambharan et al., 2007; Curtis et al., 2007; Anea et al., 2009; Rudic et al., 2004). Both major (e.g., chronic shift work) and minor (e.g., daylight savings transitions) stressors to the circadian system have been associated with a higher risk of CVD (Torquati et al., 2018; Janszky and Ljung, 2008). Similarly, circadian misalignment has also been implicated in the pathogenesis of cancer. For example, circadian clocks are critical to the orchestration of cell division (Reddy et al., 2005), and altered clock function can precipitate aberrant cell proliferation (Wang et al., 2016) as well as growth and DNA damage in cancer cells (Gery et al., 2006). Many systems are under circadian influence, including the sleep-wake cycle, and less clear has been the extent to which differences in sleep regularity are related to negative health outcomes. We extend this research by demonstrating that differences in sleep regularity are associated with the risk of mortality from both CVD and cancer.

Whereas sleep regularity has not been examined with respect to incident cancer or mortality, the current findings extend research showing that greater sleep-wake variability, as measured by the SD of sleep onset or duration, was independently associated with a higher risk of incident CVD in the multi-ethnic study of atherosclerosis but not the UKB (Nikbakhtian et al., 2021; Huang et al., 2020) We demonstrate that the SRI contains information about mortality risk beyond that contained in the SD of sleep duration and onset, whereas the converse was not the case. The SRI may be superior to the SD-based metrics because the SRI captures rapid changes in sleep patterns across consecutive days, as compared to the SD-based metrics which only calculate deviation from an individual’s average. Rapid changes in sleep timing have been hypothesized as being principally challenging for the circadian system to accommodate (Phillips et al., 2017) which may, in turn, produce negative health outcomes.

We found evidence that hazard rates across levels of SRI were non-proportional (i.e., varied across the follow-up period) for all-cause and cancer mortality (which accounted for most deaths), though not for CVD mortality. For cancer mortality, HRs for low SRI compared to the median were largest in the earliest follow-up period and decreased thereafter. One plausible interpretation of this finding is that irregular sleep may be a manifestation of the underlying physiological processes of cancer itself or of cancer treatment (i.e., the SRI-cancer mortality association may be due to reverse causation). However, this thesis is challenged by the fact that associations between the SRI and cancer mortality remained similar after adjusting for history of cancer at baseline. In the case of CVD mortality, no such evidence of a decline in HRs over follow-up time was evident; a potential causal role of irregular sleep on CVD death cannot be easily ruled out.

Sleep of insufficient or excessive duration is associated with many adverse health outcomes, including increased mortality risk (Cappuccio et al., 2010). As of 2022, sleep duration was included by the American Heart Association in their Essential Eight guidelines for CVD prevention (Lloyd-Jones et al., 2022). However, sleep is far more complex than its habitual duration and quality, with sleep regularity receiving comparatively little attention. As sleep-tracking wearables become more accessible, objective measurement of sleep regularity has the potential for public and clinical use. Much like sleep duration, replicating the current findings across different samples will be necessary for establishing population norms and clinical targets. Furthermore, identifying the determinants of poor sleep regularity may be of import, not only considering biological factors, but broader social determinants that impact circadian rhythmicity (e.g., racial/ethnic disparities [Chung et al., 2021], neighborhood factors [Richardson et al., 2021]) and consequently overall health.

Our study is not without limitations. First, the study was observational. We are, therefore, unable to establish cause and effect. Although we performed extensive analyses to control for confounding, we cannot exclude the possibility that our results are explained by residual confounding. As such, although therapies exist for improving sleep regularity, it’s not clear if these interventions are able to extend the lifespan. Second, sleep and wake were estimated through activity patterns from accelerometry. As compared to polysomnography, there is the potential to misclassify sleep and wake, although accelerometry is more suited to estimate circadian patterns over several days; there are several strengths to using accelerometry (e.g., days of continuous recording, minimal technical apparatus affecting sleep quality), making it the recommended clinical tool for assessing circadian rhythms (Smith et al., 2018). In addition, sleep diaries in the UKB were not available. Consequently, the algorithm we used to determine sleep and wake relied on the identification of a main ‘sleep period time window’ and did not identify napping.

Circadian rhythms have a major influence on health and disease. Although sleep wake timing is under circadian control, research on sleep regularity as a risk factor for mortality was equivocal. These data suggest sleep regularity as an important correlate of longevity, independent of sleep duration, fragmentation, and quality. Future work is needed to determine the underlying mechanisms to inform possible interventions to extend the lifespan.

Data from the UK Biobank are available, pending application approval from: https://www.ukbiobank.ac.uk/. All analysis code is publicly available at: https://osf.io/tf5xu/ under a CC-By Attribution 4.0 International license.

1. 1. Cribb L
   2. Pase M

   (2023) Open Science Framework

   ID tf5xu. Sleep regularity and mortality.

   https://osf.io/tf5xu/

1. 1. Anea CB
   2. Zhang M
   3. Stepp DW
   4. Simkins GB
   5. Reed G
   6. Fulton DJ
   7. Rudic RD

   (2009) Vascular disease in mice with a dysfunctional circadian clock

   Circulation 119:1510–1517.

   https://doi.org/10.1161/CIRCULATIONAHA.108.827477

   • PubMed
   • Google Scholar
2. 1. Cappuccio FP
   2. D’Elia L
   3. Strazzullo P
   4. Miller MA

   (2010) Sleep duration and all-cause mortality: a systematic review and meta-analysis of prospective studies

   Sleep 33:585–592.

   https://doi.org/10.1093/sleep/33.5.585

   • PubMed
   • Google Scholar
3. 1. Chung J
   2. Goodman M
   3. Huang T
   4. Wallace ML
   5. Johnson DA
   6. Bertisch S
   7. Redline S

   (2021) Racial-ethnic differences in actigraphy, questionnaire, and polysomnography indicators of healthy sleep: the multi-ethnic study of atherosclerosis

   American Journal of Epidemiology 1:kwab232.

   https://doi.org/10.1093/aje/kwab232

   • PubMed
   • Google Scholar
4. 1. Curtis AM
   2. Cheng Y
   3. Kapoor S
   4. Reilly D
   5. Price TS
   6. Fitzgerald GA

   (2007) Circadian variation of blood pressure and the vascular response to asynchronous stress

   PNAS 104:3450–3455.

   https://doi.org/10.1073/pnas.0611680104

   • PubMed
   • Google Scholar
5. 1. Fritz J
   2. Phillips AJK
   3. Hunt LC
   4. Imam A
   5. Reid KJ
   6. Perreira KM
   7. Mossavar-Rahmani Y
   8. Daviglus ML
   9. Sotres-Alvarez D
   10. Zee PC
   11. Patel SR
   12. Vetter C

   (2021) Cross-sectional and prospective associations between sleep regularity and metabolic health in the Hispanic Community Health Study/Study of Latinos

   Sleep 44:zsaa218.

   https://doi.org/10.1093/sleep/zsaa218

   • PubMed
   • Google Scholar
6. 1. Fry A
   2. Littlejohns TJ
   3. Sudlow C
   4. Doherty N
   5. Adamska L
   6. Sprosen T
   7. Collins R
   8. Allen NE

   (2017) Comparison of sociodemographic and health-related characteristics of UK Biobank Participants with those of the General Population

   American Journal of Epidemiology 186:1026–1034.

   https://doi.org/10.1093/aje/kwx246

   • PubMed
   • Google Scholar
7. 1. Gery S
   2. Komatsu N
   3. Baldjyan L
   4. Yu A
   5. Koo D
   6. Koeffler HP

   (2006) The circadian gene per1 plays an important role in cell growth and DNA damage control in human cancer cells

   Molecular Cell 22:375–382.

   https://doi.org/10.1016/j.molcel.2006.03.038

   • PubMed
   • Google Scholar
8. Software

   1. Harrell FE
   2. Harrell MFE

   (2019) Package ‘Hmisc, version 5.1-1

   CRAN2018.

   https://cran.r-project.org/web/packages/Hmisc/index.html
9. Book

   1. Hernán MA
   2. Robins JM

   (2010)

   Causal Inference

   CRC Boca Raton.

   • Google Scholar
10. 1. Huang T
    2. Redline S

    (2019) Cross-sectional and prospective associations of actigraphy-assessed sleep regularity with metabolic abnormalities: the multi-ethnic study of atherosclerosis

    Diabetes Care 42:1422–1429.

    https://doi.org/10.2337/dc19-0596

    • PubMed
    • Google Scholar
11. 1. Huang T
    2. Mariani S
    3. Redline S

    (2020) Sleep irregularity and risk of cardiovascular events: the multi-ethnic study of Atherosclerosis

    Journal of the American College of Cardiology 75:991–999.

    https://doi.org/10.1016/j.jacc.2019.12.054

    • PubMed
    • Google Scholar
12. 1. Huang WC
    2. Chang SH
    3. Hsueh MC
    4. Liao Y

    (2023) Relationship of sleep regularity with device-based sedentary behavior time and physical activity time in working adults

    Sleep Health 9:86–92.

    https://doi.org/10.1016/j.sleh.2022.10.002

    • PubMed
    • Google Scholar
13. 1. International-Agency-for-Research-on-Cancer

    (2010)

    Painting, firefighting, and Shiftwork

    IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 98:804.

    • Google Scholar
14. 1. Janszky I
    2. Ljung R

    (2008) Shifts to and from daylight saving time and incidence of myocardial infarction

    The New England Journal of Medicine 359:1966–1968.

    https://doi.org/10.1056/NEJMc0807104

    • PubMed
    • Google Scholar
15. 1. Lloyd-Jones DM
    2. Allen NB
    3. Anderson CAM
    4. Black T
    5. Brewer LC
    6. Foraker RE
    7. Grandner MA
    8. Lavretsky H
    9. Perak AM
    10. Sharma G
    11. Rosamond W
    12. American Heart Association

    (2022) Life’s Essential 8: Updating and Enhancing the American Heart Association’s Construct of Cardiovascular Health: A Presidential Advisory From the American Heart Association

    Circulation 146:e18–e43.

    https://doi.org/10.1161/CIR.0000000000001078

    • PubMed
    • Google Scholar
16. 1. Migueles JH
    2. Rowlands AV
    3. Huber F
    4. Sabia S
    5. van Hees VT

    (2019) GGIR: A Research Community–Driven Open Source R Package for Generating Physical Activity and Sleep Outcomes From Multi-Day Raw Accelerometer Data

    Journal for the Measurement of Physical Behaviour 2:188–196.

    https://doi.org/10.1123/jmpb.2018-0063

    • Google Scholar
17. 1. Nikbakhtian S
    2. Reed AB
    3. Obika BD
    4. Morelli D
    5. Cunningham AC
    6. Aral M
    7. Plans D

    (2021) Accelerometer-derived sleep onset timing and cardiovascular disease incidence: a UK Biobank cohort study

    European Heart Journal. Digital Health 2:658–666.

    https://doi.org/10.1093/ehjdh/ztab088

    • PubMed
    • Google Scholar
18. 1. Phillips AJK
    2. Clerx WM
    3. O’Brien CS
    4. Sano A
    5. Barger LK
    6. Picard RW
    7. Lockley SW
    8. Klerman EB
    9. Czeisler CA

    (2017) Irregular sleep/wake patterns are associated with poorer academic performance and delayed circadian and sleep/wake timing

    Scientific Reports 7:3216.

    https://doi.org/10.1038/s41598-017-03171-4

    • PubMed
    • Google Scholar
19. 1. Reddy AB
    2. Wong GKY
    3. O’Neill J
    4. Maywood ES
    5. Hastings MH

    (2005) Circadian clocks: neural and peripheral pacemakers that impact upon the cell division cycle

    Mutation Research 574:76–91.

    https://doi.org/10.1016/j.mrfmmm.2005.01.024

    • PubMed
    • Google Scholar
20. 1. Richardson AS
    2. Troxel WM
    3. Ghosh-Dastidar M
    4. Hunter GP
    5. Beckman R
    6. Collins R
    7. Brooks Holliday S
    8. Nugroho A
    9. Hale L
    10. Buysse DJ
    11. Buman MP
    12. Dubowitz T

    (2021) Violent crime, police presence and poor sleep in two low-income urban predominantly Black American neighbourhoods

    Journal of Epidemiology and Community Health 75:62–68.

    https://doi.org/10.1136/jech-2020-214500

    • PubMed
    • Google Scholar
21. 1. Rudic RD
    2. McNamara P
    3. Curtis AM
    4. Boston RC
    5. Panda S
    6. Hogenesch JB
    7. Fitzgerald GA

    (2004) BMAL1 and CLOCK, two essential components of the circadian CLOCK, are involved in glucose homeostasis

    PLOS Biology 2:e377.

    https://doi.org/10.1371/journal.pbio.0020377

    • PubMed
    • Google Scholar
22. Book

    1. Singer JD
    2. Willett JB

    (2003) Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence

    Oxford University Press.

    https://doi.org/10.1093/acprof:oso/9780195152968.001.0001

    • Google Scholar
23. 1. Smith MT
    2. McCrae CS
    3. Cheung J
    4. Martin JL
    5. Harrod CG
    6. Heald JL
    7. Carden KA

    (2018) Use of Actigraphy for the Evaluation of Sleep Disorders and Circadian Rhythm Sleep-Wake Disorders: An American Academy of Sleep Medicine Systematic Review, Meta-Analysis, and GRADE Assessment

    Journal of Clinical Sleep Medicine 14:1209–1230.

    https://doi.org/10.5664/jcsm.7228

    • PubMed
    • Google Scholar
24. 1. Stensrud MJ
    2. Hernán MA

    (2020) Why test for proportional hazards?

    JAMA 323:1401–1402.

    https://doi.org/10.1001/jama.2020.1267

    • PubMed
    • Google Scholar
25. 1. Torquati L
    2. Mielke GI
    3. Brown WJ
    4. Kolbe-Alexander T

    (2018) Shift work and the risk of cardiovascular disease. A systematic review and meta-analysis including dose-response relationship

    Scandinavian Journal of Work, Environment & Health 44:229–238.

    https://doi.org/10.5271/sjweh.3700

    • PubMed
    • Google Scholar
26. 1. van Hees VT
    2. Sabia S
    3. Anderson KN
    4. Denton SJ
    5. Oliver J
    6. Catt M
    7. Abell JG
    8. Kivimäki M
    9. Trenell MI
    10. Singh-Manoux A

    (2015) A novel, open access method to assess sleep duration using a wrist-worn accelerometer

    PLOS ONE 10:e0142533.

    https://doi.org/10.1371/journal.pone.0142533

    • PubMed
    • Google Scholar
27. 1. van Hees VT
    2. Sabia S
    3. Jones SE
    4. Wood AR
    5. Anderson KN
    6. Kivimäki M
    7. Frayling TM
    8. Pack AI
    9. Bucan M
    10. Trenell MI
    11. Mazzotti DR
    12. Gehrman PR
    13. Singh-Manoux BA
    14. Weedon MN

    (2018) Estimating sleep parameters using an accelerometer without sleep diary

    Scientific Reports 8:12975.

    https://doi.org/10.1038/s41598-018-31266-z

    • PubMed
    • Google Scholar
28. 1. Viswambharan H
    2. Carvas JM
    3. Antic V
    4. Marecic A
    5. Jud C
    6. Zaugg CE
    7. Ming X-F
    8. Montani J-P
    9. Albrecht U
    10. Yang Z

    (2007) Mutation of the circadian clock gene Per2 alters vascular endothelial function

    Circulation 115:2188–2195.

    https://doi.org/10.1161/CIRCULATIONAHA.106.653303

    • PubMed
    • Google Scholar
29. 1. Walker WH
    2. Walton JC
    3. DeVries AC
    4. Nelson RJ

    (2020) Circadian rhythm disruption and mental health

    Translational Psychiatry 10:28.

    https://doi.org/10.1038/s41398-020-0694-0

    • PubMed
    • Google Scholar
30. 1. Wang Q
    2. Ao Y
    3. Yang K
    4. Tang H
    5. Chen D

    (2016) Circadian clock gene Per2 plays an important role in cell proliferation, apoptosis and cell cycle progression in human oral squamous cell carcinoma

    Oncology Reports 35:3387–3394.

    https://doi.org/10.3892/or.2016.4724

    • PubMed
    • Google Scholar
31. 1. Zee PC
    2. Vitiello MV

    (2009) Circadian Rhythm Sleep Disorder: Irregular Sleep Wake Rhythm Type

    Sleep Medicine Clinics 4:213–218.

    https://doi.org/10.1016/j.jsmc.2009.01.009

    • PubMed
    • Google Scholar
32. 1. Zeitzer JM
    2. Dijk DJ
    3. Kronauer R
    4. Brown E
    5. Czeisler C

    (2000) Sensitivity of the human circadian pacemaker to nocturnal light: melatonin phase resetting and suppression

    The Journal of Physiology 526 Pt 3:695–702.

    https://doi.org/10.1111/j.1469-7793.2000.00695.x

    • PubMed
    • Google Scholar
33. 1. Zuraikat FM
    2. Makarem N
    3. Redline S
    4. Aggarwal B
    5. Jelic S
    6. St-Onge MP

    (2020) Sleep regularity and Cardiometabolic Heath: is variability in sleep patterns a risk factor for excess Adiposity and Glycemic Dysregulation?

    Current Diabetes Reports 20:1–9.

    https://doi.org/10.1007/s11892-020-01324-w

    • PubMed
    • Google Scholar

Author details

1. Lachlan Cribb

   Turner Institute for Brain and Mental Health, School of Psychological Sciences, Monash University, Melbourne, Australia

   Contribution

   Data curation, Formal analysis, Methodology, Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

   Additional information

   Joint first authors

   "This ORCID iD identifies the author of this article:" 0000-0001-5276-5249

2. Ramon Sha

   Turner Institute for Brain and Mental Health, School of Psychological Sciences, Monash University, Melbourne, Australia

   Contribution

   Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

   Additional information

   Joint first authors

3. Stephanie Yiallourou

   Turner Institute for Brain and Mental Health, School of Psychological Sciences, Monash University, Melbourne, Australia

   Contribution

   Supervision, Methodology, Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

   Additional information

   Joint first authors

   "This ORCID iD identifies the author of this article:" 0000-0002-2433-6372

4. Natalie A Grima

   Turner Institute for Brain and Mental Health, School of Psychological Sciences, Monash University, Melbourne, Australia

   Contribution

   Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

5. Marina Cavuoto

   1. Turner Institute for Brain and Mental Health, School of Psychological Sciences, Monash University, Melbourne, Australia
   2. National Ageing Research Institute, Melbourne, Australia

   Contribution

   Data curation, Writing – review and editing

   Competing interests

   No competing interests declared

6. Andree-Ann Baril

   Douglas Mental Health University Institute, McGill University, Montreal, Canada

   Contribution

   Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

   Additional information

   Joint senior authors

7. Matthew P Pase

   1. Turner Institute for Brain and Mental Health, School of Psychological Sciences, Monash University, Melbourne, Australia
   2. Harvard T.H. Chan School of Public Health, Harvard University, Boston, United States

   Contribution

   Conceptualization, Supervision, Funding acquisition, Project administration, Writing – review and editing

   For correspondence

   matthew.pase@monash.edu

   Competing interests

   No competing interests declared

   Additional information

   Joint senior authors

   "This ORCID iD identifies the author of this article:" 0000-0002-4143-8485

• 5,525

  views

• 189

  downloads

• 2

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.