Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Oct 3.
Published in final edited form as: Cancer Epidemiol Biomarkers Prev. 2024 Apr 3;33(4):624–627. doi: 10.1158/1055-9965.EPI-23-0983

Sleep and risk of pancreatic cancer in the UK Biobank

Joshua R Freeman 1,*, Pedro F Saint-Maurice 1,2, Ting Zhang 1, Charles E Matthews 1,, Rachael Z Stolzenberg-Solomon 1,
PMCID: PMC10990775  NIHMSID: NIHMS1968073  PMID: 38387085

Abstract

Background:

Light at night, which may cause circadian disruption, is a potential pancreatic cancer risk factor. However, evidence from related exposures such as poor sleep health and shift work remains inconclusive and sparsely investigated.

Methods:

We evaluated associations between self-reported typical sleep duration, chronotype, shift work, insomnia symptoms, snoring, and daytime sleeping and pancreatic ductal adenocarcinomas (PDAC) incidence among 475,286 United Kingdom (UK) Biobank participants. We used Cox proportional hazards models to estimate hazard ratios (HR) and 95% confidence intervals (CI) adjusting for age, sex, body mass index, smoking status, duration, and frequency, alcohol intake, diabetes status, race, and employment/shift work.

Results:

Over 14 years of follow-up, 1,079 adults were diagnosed with PDAC. There were no associations observed between sleep characteristics, including sleep duration (<7 vs. 7-<9 hours; HR: 1.03, 95% CI 0.90–1.19; ≥9 hours; HR: 1.00 [0.81–1.24]), evening chronotype (“definitely” an evening person vs. “definitely” a morning person; HR: 0.99 [0.77–1.29]), shift work, insomnia symptoms, snoring, or daytime sleep and PDAC risk.

Conclusions:

Self-reported typical sleep characteristics and shift work were not associated with PDAC risk.

Impact:

Considering the role of light at night and shift work in circadian disruption and cancer risk, it is plausible that poor sleep health among a general population may be related to cancer risk through similar sleep and circadian disrupting processes. This work may suggest that typical sleep characteristics and shift work are not associated with PDAC, though additional work is needed to confirm these findings.

Keywords: Sleep, Shift Work, Pancreatic Cancer, Pancreatic Ductal Adenocarcinoma, UK Biobank

Introduction

Pancreatic ductal adenocarcinoma (PDAC) etiology is not well understood but may involve modifiable risk factors such as light at night and circadian disruption (1,2). Beyond circadian disruption, short sleep duration has been associated with obesity and type 2 diabetes, both well-established PDAC risk factors (3,4). However, it remains unclear whether poor sleep is also associated with pancreatic cancer. Findings are sparse and to our knowledge no study has evaluated sleep quality and incident PDAC (58).

We evaluated sleep duration, chronotype, shift work, and poor sleep quality in PDAC risk among United Kingdom (UK) Biobank participants. We hypothesized that short and long sleep duration, evening chronotypes, shift worker status, and poorer sleep quality would be positively associated with PDAC risk.

Materials and Methods

The UK Biobank (2006–2010) is a prospective cohort sampled from UK National Health Service (NHS) members who provided electronic consent to participate (9). The study was approved by the National Information Governance Board for Health and Social Care and the North West Multi-Centre Research Ethics Committee (9). We used data from 502,356 participants excluding 27,070 with prevalent cancer, non-imputable data, or invalid follow-up leaving 475,286 adults.

Sleep duration, chronotype, shift work, insomnia/sleeplessness, snoring, and daytime sleeping/dozing were assessed via baseline questionnaire. Questionnaire information is available online (https://biobank.ndph.ox.ac.uk/showcase/search.cgi). Employment, shift work, and night shift work were combined as a single variable.

Participants were followed from enrollment until PDAC, death, or end of follow-up (England and Wales: February 29, 2020; Scotland: January 31, 2021). Incident pancreatic cancer (International Classification of Diseases 10th Revision Codes (ICD-10) C25.0, C25.1, C25.2, C25.3, C25.7, C25.8, C25.9) was ascertained from NHS Digital (England and Wales) and the NHS Central Register (Scotland). We only included cases with International Classification of Diseases for Oncology 3rd Edition (ICD-O-3) histology codes 8000, 8010, 8020, 8140, 8154, 8211, 8453, 8480, 8481, 8490, 8500, 8503, & 8560. Follow-up was calculated from enrollment date to first incident cancer registration, except non-melanoma skin cancer (ICD-10 C44), death, lost to follow-up, or censoring date.

Multivariable-adjusted Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (CI) adjusting for age, sex, body mass index (BMI), alcohol intake, race, diabetes status, employment/shift work, and a 21-level variable for smoking status, duration, and frequency. P-trend was calculated for ordinal sleep and PDAC associations. Multiple imputation was used for missingness. There was no violation of the proportional hazards assumption. Sensitivity analyses were conducted to investigate potential residual confounding by excluding shift workers and participants with diabetes. We censored participants five years after cohort entry and excluded those with <5 years of follow-up time to investigate reverse causation. We evaluated interactions by smoking status, sex, and age with multiplicative interaction terms. Statistical tests were two-sided, and significance was defined as P<0.05. Analyses were conducted using SAS version 9.4 (SAS Institute Inc, Cary, NC; RRID: SCR_008567). This work utilized the computational resources of the NIH HPC Biowulf cluster (http://hpc.nih.gov).

Data Availability Statement

The data underlying this article has been provided by the UK Biobank Resources under Application Number 43456. We do not have permission to share it directly. UK Biobank data are globally available to approved researchers through the UK Biobank research portal (https://www.ukbiobank.ac.uk/; RRID: SCR_012815).

Results

Most (67.2%) participants reported sleeping 7-<9 hours (Table 1). Compared to 7-<9 hours, participants with ≥9 hours sleep duration were older, were more likely to have a higher BMI, be never smokers, and were retired or unemployed.

Table 1.

Baseline participant characteristics by sleep duration categories

Sleep Duration
Missing <7 hours 7-<9 hours ≥9hours
Number of participants 3,630 116,835 319,223 35,598
Variable Mean (SD) Mean (SD) Mean (SD) Mean (SD)
Age (years) 57.2 (8.2) 56.7 (7.9) 56.7 (8.2) 58.8 (8.1)
Follow-up Time (Years) 10.3 (2.4) 10.5 (2.2) 10.5 (2.2) 10.3 (2.5)
n (%) n (%) n (%) n (%)
Sex
 Men 1,387 (38.2) 55,214 (47.3) 146,881 (46.0) 15,664 (44.0)
 Women 2,243 (61.8) 61,621 (52.7) 172,342 (54.0) 19,934 (56.0)
Race
 Missing participants 678 (18.7) 519 (0.44) 980 (0.3) 144 (0.4)
 Asian participants 144 (4.0) 3,342 (2.9) 6,814 (2.1) 866 (2.4)
 Black participants 220 (6.1) 3,505 (3.0) 3,592 (1.1) 469 (1.3)
 Mixed-race participants 31 (0.9) 949 (0.8) 1,649 (0.5) 208 (0.6)
 Other participants 90 (2.5) 1,524 (1.3) 2,471 (0.8) 318 (0.9)
 White participants 2,467 (68.0) 106,996 (91.6) 303,717 (95.1) 33,593 (94.4)
BMI (kg/m 2 )
 Missing 67 (1.9) 700 (0.6) 1,323 (0.4) 268 (0.8)
 <25.0 944 (26.0) 33,882 (29.0) 108,887 (34.1) 9,873 (27.7)
 25.0 - <30.0 1,349 (37.2) 48,488 (41.5) 137,573 (43.1) 14,805 (41.6)
 30.0 - <35.0 794 (21.9) 23,239 (19.9) 52,549 (16.5) 7,259 (20.4)
 35.0 - <40.0 305 (8.4) 7,325 (6.3) 13,957 (4.4) 2,381 (6.7)
 ≥40 171 (4.7) 3,201 (2.7) 4,934 (1.6) 1,012 (2.8)
Diabetes Status 
 Missing 709 (19.5) 489 (0.4) 807 (0.3) 166 (0.5)
 Yes 327 (9.0) 6,908 (5.9) 14,399 (4.5) 3,136 (8.8)
 No 2,594 (71.5) 109,438 (93.7) 304,017 (95.2) 32,296 (90.7)
Alcohol Intake
 Missing 746 (20.6) 221 (0.2) 357 (0.1) 54 (0.2)
 Never 346 (9.5) 6,256 (5.4) 12,489 (3.9) 2,047 (5.8)
 Former 248 (6.8) 5,175 (4.4) 9,636 (3.0) 1,858 (5.2)
 Current drinker, infrequent (<3x/month) 994 (27.4) 29,194 (25.0) 68,270 (21.4) 8,900 (25.0)
 Current drinker, <1 drink/day 645 (17.8) 26,856 (23.0) 82,463 (25.8) 7,848 (22.1)
 Current drinker, 1–3 drink/day 543 (15.0) 38,924 (33.3) 119,835 (37.5) 11,275 (31.7)
 Current drinker, >3 drink/day 108 (3.0) 10,209 (8.7) 26,173 (8.2) 3,616 (10.2)
Smoking Status
 Missing 731 (20.1) 521 (0.5) 1,030 (0.3) 144 (0.4)
 Never 1,621 (44.7) 61,443 (52.6) 178,674 (56.0) 18,090 (50.8)
 Former 810 (22.3) 39,871 (34.1) 108,783 (34.1) 13,020 (36.6)
 Current 468 (12.9) 15,000 (12.8) 30,736 (9.6) 4,344 (12.2)
Employment/Shift Work
 Missing 674 (18.6) 536 (0.5) 1,119 (0.4) 148 (0.4)
 Employed-Mostly Daytime Work 576 (15.9) 55,245 (47.3) 162,513 (50.9) 9,645 (27.1)
 Employed-Job Involves Shift Work 143 (3.9) 7,027 (6.0) 15,310 (4.8) 1,259 (3.5)
 Employed-Job Involves Night Shift Work 223 (6.1) 8,841 (7.6) 14,316 (4.5) 1,264 (3.6)
 Retired 1,105 (30.4) 32,844 (28.1) 101,946 (31.9) 17,603 (49.5)
 Unemployed 909 (25.0) 12,342 (10.6) 24,019 (7.5) 5,679 (16.0)

Missingness for analytic variables (Number/475,286): Sleep Duration (3,630/475,286); Chronotype (54,984/475,286); Employment/Shift Work (2,477/475,286); Insomnia/Sleeplessness (1,104/475,286); Snoring (34,608/475,286); Daytime Sleeping (3,235/475,286); Age (0/475,286); Sex (0/475,286); BMI (2,358/475,286); Alcohol Intake (1,378/475,286); Smoking Status, Duration, and Frequency (4,510/475,286); Race (2,321/475,286); Diabetes Status (2,171/475,286).

UK Biobank field codes for variables: Sleep Duration (1160); Chronotype (1180); Employment/Shift Work (6142; 826; 3426); Insomnia/Sleeplessness (1200); Snoring (1210); Daytime Sleeping (1220); Age (21003; 33); Sex (31); BMI (23104; 21001); Alcohol Intake (1558; 1568; 1578; 1588; 1598; 1608; 5364; 20117); Smoking Status, Duration, and Frequency (1239; 1249; 2644; 2877; 2887; 2897; 3446; 3456; 5959; 20116; 21022); Race (21000); Diabetes Status (2443).

During up to 14 years of follow-up, 1,079 PDAC cases were diagnosed. Self-reported sleep characteristics were not associated with PDAC risk (Table 2). Results were unchanged in sensitivity analyses. There were no significant interactions with smoking status, sex, or age (Supplementary Table S1).

Table 2.

Association between sleep characteristics and risk of PDACa

Age-adjusted Multivariate adjustedb Excluding Shift Workersc Excluding those with Diabetesd Censored at 5 years follow-upb Excluding those with <5 years follow-upb
Sleep Characteristics Total Person-Years Follow-up PDAC Cases (n=1,079) HR HR HR HR HR HR
(95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI)
Sleep Duration
Sleep <7 hours 116,835 1,225,177 272 1.08 1.03 1.02 1.02 0.96 1.07
(0.94–1.24) (0.90–1.19) (0.88–1.18) (0.87–1.18) (0.75–1.23) (0.90–1.27)
Sleep 7-<9 hours 319,223 3,362,315 698 Ref Ref Ref Ref Ref Ref
Sleep ≥9 hours 35,598 366,741 102 1.11 1.00 1.03 1.10 1.01 0.99
(0.90–1.37) (0.81–1.24) (0.83–1.28) (0.88–1.37) (0.70–1.45) (0.77–1.29)
Chronotype
Definitely “morning” 113,904 1,191,404 256 Ref Ref Ref Ref Ref Ref
More “morning” than “evening” 148,890 1,563,828 336 1.05 1.08 1.09 1.11 1.23 1.01
(0.90–1.23) (0.92–1.26) (0.93–1.29) (0.94–1.31) (0.93–1.63) (0.83–1.23)
More “evening” than “morning” 119,512 1,255,338 276 1.14 1.08 1.08 1.12 1.22 1.02
(0.96–1.35) (0.91–1.28) (0.90–1.29) (0.93–1.34) (0.90–1.64) (0.83–1.25)
Definitely “evening” 37,996 397,529 81 1.17 0.99 0.99 1.03 1.05 0.97
(0.91–1.51) (0.77–1.29) (0.75–1.30) (0.78–1.36) (0.67–1.65) (0.71–1.32)
P-trend 0.08 0.70 0.75 0.45 0.47 0.97
Employment/Shift Work
Employed-Mostly Daytime Work 227,979 2,439,984 353 Ref Ref -- Ref Ref Ref
Employed-Job Involves Shift Work 23,739 253,805 45 1.28 1.21 -- 1.14 0.98 1.32
(0.94–1.75) (0.88–1.64) (0.82–1.59) (0.54–1.78) (0.91–1.90)
Employed-Job Involves Night Shift Work 24,644 264,271 27 0.85 0.74 -- 0.75 0.62 0.79
(0.57–1.26) (0.50–1.09) (0.50–1.14) (0.28–1.34) (0.50–1.25)
Retired 153,498 1,559,976 563 0.96 0.97 -- 0.94 0.78 1.07
(0.82–1.12) (0.82–1.14) (0.80–1.12) (0.60–1.03) (0.88–1.31)
Unemployed 42,949 448,050 85 1.30 1.09 -- 1.04 1.10 1.09
(1.03–1.65) (0.86–1.39) (0.80–1.36) (0.73–1.66) (0.81–1.47)
Insomnia/Sleeplessness
Never/Rarely 115,604 1,218,819 254 Ref Ref Ref Ref Ref Ref
Sometimes 226,289 2,380,323 509 0.92 0.95 0.95 0.91 1.01 0.93
(0.79–1.07) (0.82–1.11) (0.81–1.11) (0.78–1.07) (0.78–1.32) (0.77–1.11)
Usually 132,289 1,381,204 316 0.93 0.92 0.92 0.88 0.88 0.94
(0.79–1.10) (0.78–1.09) (0.77–1.10) (0.74–1.05) (0.65–1.19) (0.77–1.16)
P-trend 0.44 0.35 0.37 0.17 0.37 0.62
Snoring
No 275,726 2,902,522 576 Ref Ref Ref Ref Ref Ref
Yes 164,952 1,729,016 404 1.15 1.01 0.99 1.00 0.96 1.03
(1.02–1.31) (0.89–1.15) (0.86–1.13) (0.88–1.15) (0.76–1.22) (0.88–1.21)
Daytime Sleeping
Never/Rarely 359,301 3,789,459 790 Ref Ref Ref Ref Ref Ref
Sometimes 99,528 1,033,011 250 0.96 0.93 0.92 0.91 0.94 0.92
(0.84–1.11) (0.80–1.07) (0.79–1.06) (0.78–1.06) (0.73–1.21) (0.78–1.10)
Often/All the time 13,222 135,993 31 0.95 0.82 0.84 0.82 1.06 0.71
(0.66–1.36) (0.57–1.18) (0.58–1.22) (0.54–1.24) (0.60–1.86) (0.44–1.13)
P-trend 0.58 0.16 0.17 0.14 0.80 0.13
a

Analyses used multiple imputation; sleep characteristics presented as complete case. Missing level details (PDAC cases/Participants missing sleep exposure data): Sleep Duration (7/3,630); Chronotype (130/54,984); Employment/Shift Work (6/2,477); Insomnia/Sleeplessness (0/1,104); Snoring (99/34,608); Daytime Sleeping (8/3,235).

b

Cox proportional hazards models adjusted for: Age (continuous; years), Sex (binary); BMI (categorical; 5-level), Alcohol Intake (categorical; 6-level), Smoking Status, Duration, and Frequency (categorical; 21-level), Race (binary), Diabetes Status (binary), and Employment/Shift Work (categorical; 5-level) except in models of Employment/Shift Work.

c

Excluding shift workers; cox proportional hazards models adjusted for covariates in b.

d

Excluding participants with diabetes; cox proportional hazards models adjusted for covariates in b, except Diabetes Status (binary).

Discussion

Among 475,286 UK adults, self-reported sleep characteristics were not associated with incident PDAC. Previous work found suggestive, though non-significant evidence for short sleep duration and greater pancreatic cancer risk or having 6–8 hours sleep and lower risk (5,6,8). The null findings for chronotype and shift work are generally consistent with prospective studies in the International Agency for Research on Cancer report (2,7). This work expands on these prior studies through evaluating sleep quality and PDAC risk. Study strengths include a large sample size, long follow-up, and many incident PDAC cases. Limitations include that this study used baseline self-reported typical sleep but not workday and non-workday-specific sleep, which may underestimate average sleep. Sleep duration was based on total sleep in a 24-hour period including naps. We were limited in evaluating detailed shift work classifications. Sleep characteristics and shift work may have changed during follow-up. Lastly, the UK Biobank included predominantly White adults which may limit generalizability to other populations.

Our findings do not support an association between self-reported sleep characteristics and PDAC risk. Future research should evaluate repeated measures of sleep using both self-report and actigraphy to better understand how change in sleep and circadian patterns over time may influence PDAC etiology.

Supplementary Material

1

Acknowledgements

Funding

J.R. Freeman and T. Zhang were supported by the Intramural Research Program Cancer Research Training Award, National Cancer Institute, National Institutes of Health (Z99 CA999999).

Footnotes

Conflict of Interest Disclosure Statement: The authors declare no potential conflicts of interest.

Supplementary data for this article are available at Cancer Epidemiology, Biomarkers & Prevention Online.

References

  • 1.Xiao Q, Jones RR, James P, Stolzenberg-Solomon RZ. Light at Night and Risk of Pancreatic Cancer in the NIH-AARP Diet and Health Study. Cancer Res 2021;81(6):1616–22 doi 10.1158/0008-5472.CAN-20-2256. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Night Shift Work. In IARC Monographs on the Identification of Carcinogenic Hazards to Humans Volume 124. 2019. <https://publications.iarc.fr/593>. April 30, 2023. [Google Scholar]
  • 3.Grandner MA, Seixas A, Shetty S, Shenoy S. Sleep Duration and Diabetes Risk: Population Trends and Potential Mechanisms. Curr Diab Rep 2016;16(11):106 doi 10.1007/s11892-016-0805-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Eibl G, Cruz-Monserrate Z, Korc M, Petrov MS, Goodarzi MO, Fisher WE, et al. Diabetes Mellitus and Obesity as Risk Factors for Pancreatic Cancer. J Acad Nutr Diet 2018;118(4):555–67 doi 10.1016/j.jand.2017.07.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Luu HN, Paragomi P, Wang R, Jin A, Brand RE, Koh WP, et al. Composite Score of Healthy Lifestyle Factors and the Risk of Pancreatic Cancer in a Prospective Cohort Study. Cancer Prev Res (Phila) 2022;15(1):29–36 doi 10.1158/1940-6207.CAPR-21-0205. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Titova OE, Michaelsson K, Vithayathil M, Mason AM, Kar S, Burgess S, et al. Sleep duration and risk of overall and 22 site-specific cancers: A Mendelian randomization study. Int J Cancer 2021;148(4):914–20 doi 10.1002/ijc.33286. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Yuan S, Mason AM, Titova OE, Vithayathil M, Kar S, Chen J, et al. Morning chronotype and digestive tract cancers: Mendelian randomization study. Int J Cancer 2023;152(4):697–704 doi 10.1002/ijc.34284. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Gu F, Xiao Q, Chu LW, Yu K, Matthews CE, Hsing AW, et al. Sleep Duration and Cancer in the NIH-AARP Diet and Health Study Cohort. PLoS One 2016;11(9):e0161561 doi 10.1371/journal.pone.0161561. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Fry A, Littlejohns TJ, Sudlow C, Doherty N, Adamska L, Sprosen T, et al. Comparison of Sociodemographic and Health-Related Characteristics of UK Biobank Participants With Those of the General Population. Am J Epidemiol 2017;186(9):1026–34 doi 10.1093/aje/kwx246. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1

Data Availability Statement

The data underlying this article has been provided by the UK Biobank Resources under Application Number 43456. We do not have permission to share it directly. UK Biobank data are globally available to approved researchers through the UK Biobank research portal (https://www.ukbiobank.ac.uk/; RRID: SCR_012815).

RESOURCES