Accepted Manuscript

Investigating cortisol production and pattern as mediators in the relationship between

shift work and cardiometabolic risk

Jennifer Ritonja, MSc, Kristan J. Aronson, PhD, Andrew G. Day, MSc, Jill Korsiak,
MSc, Joan Tranmer, PhD

PII: S0828-282X(18)30151-X
DOI: 10.1016/j.cjca.2018.02.006
Reference: CJCA 2740

To appear in: Canadian Journal of Cardiology

Received Date: 14 November 2017

Revised Date: 21 January 2018
Accepted Date: 6 February 2018

Please cite this article as: Ritonja J, Aronson KJ, Day AG, Korsiak J, Tranmer J, Investigating cortisol
production and pattern as mediators in the relationship between shift work and cardiometabolic risk,
Canadian Journal of Cardiology (2018), doi: 10.1016/j.cjca.2018.02.006.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Title: Investigating cortisol production and pattern as mediators in the relationship

between shift work and cardiometabolic risk

Jennifer Ritonjaa MSc, Kristan J. Aronsona,b PhD, Andrew G. Dayc MSc, Jill Korsiaka

PT
MSc, Joan Tranmera,d PhD

RI
a. Department of Public Health Sciences, Queen’s University, Kingston, Ontario, Canada

b. Division of Cancer Care and Epidemiology, Cancer Research Institute, Queen’s

SC
University, Kingston, Ontario, Canada

c. Kingston Health Sciences Hospital Research Institute, Kingston, Ontario, Canada

U
d. School of Nursing, Queen’s University, Kingston, Ontario, Canada
AN
Short title: Cortisol as a mediator between shift work and CMR
M

Word count: 4822

D

Corresponding author:
TE

Dr. Joan Tranmer

Professor, School of Nursing

EP

92 Barrie Street

Queen’s University
C

Kingston, Ontario
AC

K7L3N6

tranmerj@queensu.ca

Telephone: 613-533-6000 ext. 77670

1
 ACCEPTED MANUSCRIPT

Brief Summary

We examined whether total cortisol production and cortisol pattern mediate the

relationship between current shift work and cardiometabolic risk (CMR) among female

PT
hospital employees. Using a cross-sectional study of 326 female employees, we found

RI
that current shift work was associated with a higher (worse) CMR score, lower total

cortisol production and a flatter pattern. Total cortisol production was a partial mediator

SC
in the relationship between shift work and CMR, while cortisol pattern was not.

U
AN
M
D
TE
C EP
AC

2
 ACCEPTED MANUSCRIPT

Abstract

Background: Shift work is a risk factor for many diseases, including cardiovascular

disease. While the biological pathways are still unclear, it is hypothesized that cortisol

disruption during night work is an intermediate. The objective of this study is to

PT
determine whether total cortisol production and cortisol pattern mediate the relationship

RI
between current shift work and cardiometabolic risk (CMR) among female hospital

employees.

SC
Methods: A cross-sectional study was conducted among 326 female employees (166

rotating shift workers, 160 day workers), recruited from a hospital in Southeastern

U
Ontario, Canada, during 2011 to 2014. Participants completed a baseline interview,
AN
questionnaire, and clinical exam. Urine samples were collected over two 24-hour periods

and used to analyze creatinine-adjusted cortisol, which was then used to calculate total
M

cortisol production (AUCG), and pattern (AUCI). Mediation analysis was performed to
D

test the mediating effect of cortisol in the relationship between shift work and a
TE

continuous CMR score.

Results: Current shift work is associated with a 0.52 higher CMR score (95% CI: 0.15,
EP

0.89), a lower cortisol output (AUCG), and a flatter pattern (AUCI) over a 2-day period.

AUCG is a partial mediator in the relationship between shift work and CMR, while AUCI,
C

is not. AUCG is also associated with CMR while controlling for shift work, suggesting
AC

that lower total cortisol production is also linked to CMR in non-shift workers.

Conclusions: Total cortisol production is a partial mediator in the relationship between

rotating shift work and CMR among female hospital employees, while cortisol pattern is

not a mediator.

3
 ACCEPTED MANUSCRIPT

Introduction

Shift work generally refers to a work schedule that is outside of standard daylight

hours, and includes night work and alternating day and night shifts.1 With an increased

demand for 24/7 services, shift work is becoming more prevalent. According to the

PT
Canadian General Social Survey (2010), approximately 33% of the adult working

RI
population in Canada are shift workers2, and the majority of Canadian women working in

rotating and night shift work are in healthcare.3

SC
There is evidence that shift work is a risk factor for cardiovascular disease (CVD).

Shift workers have an increased risk of myocardial infarction, ischaemic stroke, and

U
coronary events in comparison to day workers.2,4 In this study, we focused on
AN
cardiometabolic risk (CMR) as our outcome since numerous studies support its strong

association with future CVD.5 There is strong evidence that shift work increases the risk
M

of metabolic syndrome (MetS), a common measure of CMR, in addition to being a risk

D

factor for obesity, body weight gain, and impaired glucose tolerance.6,7
TE

While the evidence supports an association between shift work and CMR, the

exact biological pathways are still unclear. Several mechanisms are hypothesized,
EP

including interrelated psychosocial, behavioral, and physiological mechanisms.8,9 One

hypothesized biological mechanism is that shift work disrupts cortisol production.

C

Cortisol is a circadian-regulated stress hormone with a diurnal pattern of production.10

AC

Cortisol levels peak approximately 30 minutes post-awakening, then decline steeply

throughout the day, reaching their nadir during sleep.10 Shift work may disrupt circadian

mechanisms leading to physiological changes that promote CVD development.9,11

4
 ACCEPTED MANUSCRIPT

There is evidence that shift work is associated with disrupted cortisol, with most

studies suggesting that shift workers have flatter diurnal cortisol patterns and lower 24-

hour cortisol production during night shifts.12–15 There is also evidence that disruptions in

cortisol are associated with CVD and MetS.16–18 However, it is still unclear if cortisol

PT
disruption acts as one of the underlying biological mechanisms in the pathway between

RI
shift work and CMR.

The objective of this study is to assess whether total cortisol production and

SC
pattern mediate the relationship between current shift work and CMR among female

hospital employees.

U
AN
Methods

Study sample
M

A cross-sectional study was conducted among full-time and part-time female

D

hospital employees at an acute-care hospital in Southeastern Ontario, Canada during 2011

TE

to 2014. Participants were recruited through posters, meetings with program managers,

emails, hospital intranet notices, communication books and staff meetings. Pregnant
EP

women, those who gave birth in the previous year, and women with less than one year of

employment were asked to self-exclude. Of the 331 participants originally recruited, four
C

were excluded due to missing cortisol or unknown timing of urine collection, and one
AC

was excluded due to an extreme outlier blood glucose level. For this analysis, 326

participants were included (166 rotating shift workers, 160 day workers), recruited from

inpatient units (e.g. registered nurses), laboratory, diagnostic and support services. This

5
 ACCEPTED MANUSCRIPT

study was approved by the Health Sciences Research Ethics Board at Queen’s University,

and all participants provided written informed consent.

Data Collection

PT
Information on current shift work was obtained through an intake interview.

RI
Participants were classified as either day workers, or rotational shift workers with day

work, and night work between the hours of 24:00 and 05:00. The typical shift work

SC
schedule was two 12-hour day shifts, followed by two 12-hour night shifts, and four to

five days off. Day workers typically worked five consecutive 8-hour days starting at

08:00 or 09:00, followed by two days off.

U
AN
Participants were asked to collect midstream urine at every void over two 24-hour

cycles and to record timing of voids. Day workers collected samples during two day
M

shifts, and rotating shift workers collected samples during one day shift and one night
D

shift, always the second of two consecutive nights. Collection began after awakening, and
TE

an average of seven samples were collected per day. Awakening time was recorded in a

sleep diary and verified with accelerometer data. Samples were assayed using a DiaMetra
EP

urinary “free” cortisol ELISA kit (sensitivity of 2.0 ng/mL). Differences in urinary output

were accounted for by adjusting for creatinine which was assayed using a ParameterTM
C

Jaffe reaction (sensitivity of 0.02 mg/dL). Measures of cortisol and creatinine

AC

concentration were repeated twice, and the duplicate values were compared to ensure

reliability. An external quality assurance test was also completed by comparing

creatinine-adjusted cortisol to an established range for adult females.19

6
 ACCEPTED MANUSCRIPT

To obtain measurements used for CMR scoring, blood samples were drawn after a

nightly fast, and triglycerides, HDL cholesterol and blood glucose were analyzed. Blood

pressure was measured using the BpTRU blood pressure monitor (VSM MedTech Ltd,

Coquitlam, Canada), and a mean was derived from three consecutive readings. Waist

PT
circumference was measured at the midway point between the iliac crest and the lower

RI
rib.

Instead of using the binary MetS definition that is prone to misclassification and

SC
assumes all components contribute equally to risk,20,21 a CMR score for women was used.

The continuous score was based on a principal component analysis of CMR indicators

U
among 2,557 women aged 30-65 in France, and uses blood pressure, fasting blood
AN
glucose, triglycerides and waist circumference to determine a woman’s risk of

cardiometabolic outcomes.21 The CMR score was standardized to have a mean of zero
M

(standard deviation of one) in the original French development score, and each standard
D

deviation increase in the score is associated with 1.7-fold (95% CI: 1.0, 2.7) increase in
TE

risk of developing CVD in women (see Supplementary Material 1 for more details on

calculating the score).21 Currently, there is no validated CMR score for use on an adult
EP

Canadian population. This score was chosen since it allowed adequate variation on our

sample, which maximized the statistical power in comparison to other CMR scores
C

tested.22,23 Since our sample represents a relatively young population, CVD risk scores
AC

could not be used since most participants did not meet the cut-offs for increased CVD

risk.24,25

Potential confounders such as age, education, menopausal status, and lifestyle

behaviours were measured through interviews and questionnaires. Participants also

7
 ACCEPTED MANUSCRIPT

provided information on work characteristics, such as type of employment and past shift

work.

Statistical Analysis

PT
Baseline demographic characteristics of shift and day workers were compared

RI
using unadjusted Chi-Square tests and t-tests. Since shift workers were on average

younger than day workers, and age is a strong risk factor for cardiometabolic outcomes

SC
and cortisol, multiple regression adjusting for age was used to facilitate baseline

comparisons. Fasting blood glucose and triglycerides were log-transformed to ensure

U
normality of residuals and are presented as percent changes. Two quantitative summary
AN
measures of 24-hr cortisol were calculated using the trapezoidal rule for area under the

curve (AUC). AUC with respect to the ground (AUCG) represents the 24-hr quantity of
M

cortisol production (i.e. total cortisol output26,27). AUC with respect to the increase
D

(AUCI) represents the diurnal cortisol pattern,26,27 and more negative values are
TE

interpreted as flatter curves. The arithmetic means of AUCI and AUCG from the two

cycles were used in all analyses as a more reliable estimate of average diurnal cortisol
EP

production (see Supplementary Material 1 for more details).

Mediation analysis was performed using multivariable linear regression. AUCG

C

and AUCI were considered as mediators (Figure 1). To test the significance of the
AC

mediation effect (i.e. the indirect effect of shift work on CMR via cortisol), 95%

confidence intervals around the indirect effect were constructed using bootstrapping.28

Since behavioural and lifestyle factors are hypothesized as steps along the casual

pathway linking shift work and CVD risk, only variables not on the causal pathway

8
 ACCEPTED MANUSCRIPT

between shift work and cortisol or CMR were considered as potential confounders.

Physical activity, caffeine intake, smoking status and alcohol intake may additionally

confound the relationship between cortisol and CVD, and were assessed as confounders

for this pathway (path b). All confounders were assessed empirically using a change-in-

PT
estimate approach. Final models adjusted for age, education, menopausal status and past

RI
shift work in all pathways; caffeine intake and smoking status were also adjusted in path

b. Of the 326 participants, one participant was excluded from AUCG analyses and two

SC
participants were excluded from AUCI analyses due to outliers. To test the robustness of

our results, we performed 2 sensitivity analyses; 1) using a validated CMR score created

U
on 6870 participants in the National Health and Nutrition Examination Survey
AN
(NHANES), and 2), using MetS as a binary outcome of CMR.22 All data analyses were

performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, 2013).
M
D

Results
TE

Characteristics of study population

In our study population, shift workers are younger (38.8 years vs. 44.8 years) and
EP

are more likely to be nurses than day workers (Table 1). After adjusting for age, shift

workers have a lower mean AUCG in comparison to day workers, while mean AUCI is
C

not different (Table 2). In terms of cardiometabolic components (Table 2), shift workers
AC

have higher mean fasting blood glucose levels and higher mean triglyceride levels after

adjusting for age. There is a trend among shift workers to have higher mean waist

circumferences and blood pressure levels in comparison to day workers. In addition, shift

9
 ACCEPTED MANUSCRIPT

workers have a higher mean CMR score in comparison to day workers, as well as a

higher prevalence of MetS (21.0% vs. 11.9%).

Mediation analysis

PT
Table 3 presents the results of the mediation analysis. After adjusting for age,

RI
education, menopausal status, and past shift work experience, shift work is associated

with a 0.52 (95% CI: 0.15, 0.89) higher CMR score in comparison to day workers (path

SC
c). In terms of cortisol, shift work is associated with a lower average AUCG (β: -157.19,

95% CI: -228.50, -85.88) and a lower average AUCI (β: -113.50, 95% CI: -215.26, -

U
11.74) (path a). This indicates that shift workers produce lower total cortisol production
AN
and have a flatter cortisol pattern over a 2-day period in comparison to day workers.

Average AUCG, but not average AUCI is associated with the CMR score when
M

controlling for shift work and confounding variables (path b). For every one standard
D

deviation of AUCG, the CMR score decreases by 0.21 (95% CI: -0.38, -0.04). This
TE

suggests that lower levels of cortisol output, independent of shift work, are associated

with an increase in the CMR score. When examining the mediating effect (path ab), only
EP

average AUCG has an indirect effect (β: 0.11, 95% CI: 0.02, 0.23). Further, the

association between shift work and CMR (path c’) is attenuated when controlling for
C

AUCG, indicating that average AUCG is a mediator. In terms of the mediating effect,
AC

average AUCG mediates 17% of the total effect. There is no evidence that AUCI is a

mediator, since the criteria for mediation are not met and there is no indirect effect.

To test the robustness of our results, we used a validated CMR score created on

6870 participants from the NHANES cohort. The results are similar to our original model

10
 ACCEPTED MANUSCRIPT

(Supplementary Materials 2, Table S1), indicating that our results are robust to the use of

another CMR score. Thus, the original CMR score was appropriate to use on our sample.

We also conducted an exploratory sensitivity analysis using MetS as the outcome, instead

of the CMR score (Table S2). Although limited in statistical power, the results suggest

PT
that total cortisol production, but not cortisol pattern, is a mediator in the relationship,

RI
supporting our original results with the CMR score.

SC
Discussion

In this study, women in rotating shift work positions have lower total cortisol

U
production and flatter cortisol patterns averaged across two 24-hour periods than day
AN
workers, and a higher CMR score. Investigating the mediating effect of cortisol, we

found that average total cortisol production partially mediates the relationship between
M

current shift work and the CMR score. There is no evidence of mediation for cortisol
D

pattern. These results provide evidence that total cortisol production is an intermediate in
TE

the pathway between shift work and CMR, while cortisol pattern is not.

A number of biological mechanisms may explain why decreased cortisol

EP

production increases the likelihood of CMR. Cortisol exerts a multitude of effects on the

body, with roles in glucose and free fatty acid metabolism, autonomic nervous system,
C

inflammatory system, and sex and growth hormones.29,30 Disruption in cortisol regulation
AC

can lead to dysfunction of these systems, and HPA-axis dysfunction is proposed to

contribute to CVD development.29 Cortisol plays a crucial role in the regulation of lipid

metabolism, and changes in cortisol are linked to changes in lipolysis, free fatty acid

production, and fatty accumulation in the liver.31,32 It is hypothesized that changes in lipid

11
 ACCEPTED MANUSCRIPT

metabolism by cortisol contribute to the development of atherosclerosis.33,34 Further,

there is evidence that sleep duration mediates the association between shift work and

MetS,35 and previous studies show that sleep restriction disrupts cortisol.36,37 Acute sleep

restriction due to shift work, which has also been linked to a number of CMR indices,38

PT
could serve as a mechanism that disrupts cortisol.

RI
Among rotating shift workers, we found the CMR score is 0.52 (95% CI: 0.15,

0.89) higher than day workers after adjusting for age, education, menopausal status, and

SC
past shift work experience. According to this score, this corresponds to a 32% increase in

the odds of incident CVD among shift workers over a 9-year period.21 While, this score is

U
not validated in a Canadian population, a similar finding was found when using MetS as
AN
the outcome, although limited in statistical power. In addition, the use of another

validated CMR score did not change our findings, indicating the CMR score originally
M

used was appropriate.

D

In our study, shift work is associated with lower total cortisol production and a
TE

flatter cortisol pattern over a 2-day period, which is consistent with past studies.12–15 We

also found that total cortisol production is associated with CMR when controlling for
EP

shift work. This suggests that a lower cortisol output, independent of shift work, is

associated with a higher CMR score. This highlights that cortisol production could be
C

involved in the pathogenesis of CMR, even in non-shift working women. There is no

AC

evidence that cortisol pattern is associated with CMR, after controlling for shift work.

This is consistent with a recent cross-sectional study that found diurnal cortisol pattern

was not associated with MetS.18

12
 ACCEPTED MANUSCRIPT

In this study, cortisol partially mediates the relationship between shift work and a

CMR score, where 17% of the total effect is mediated by total cortisol production. A

similar finding was also found when using MetS as the outcome, supporting the role of

total cortisol production as a mediator between shift work and CMR. Overall, the small

PT
mediating effect suggests that total cortisol production is only a partial mediator in the

RI
relationship between shift work and CMR. This highlights the complexity and

multifactorial nature of this pathway, and suggests that other intermediates should be

SC
considered.

To our knowledge, this is the first study to investigate the mediating role of

U
cortisol in the relationship between shift work and CMR. In comparison to laboratory
AN
studies, the observational study design allowed us to capture changes in cortisol

production during participants’ regular working schedule. In addition, the use of

M

bootstrapping to calculate 95% confidence intervals for the indirect effect provides a
D

robust estimate of the mediating effect, in comparison to methods that assume normality
TE

in the sampling distribution.28

The cross-sectional study design prevents assessment of temporality. However,

EP

there is evidence that shift work precedes cortisol disruption, and that cortisol disruption

precedes CVD risk, so analysis of mediation on a cross-sectional design is valid. These

C

results may not be generalizable to men, other occupations, or other shift work schedules.
AC

It is also uncertain what cortisol production and pattern represent, as they could represent

both circadian rhythm and stress responses. Lastly, there is some potential for selection

bias. If more shift workers participated because of their concern for CVD, this could

overestimate the effect of shift work on CMR.

13
 ACCEPTED MANUSCRIPT

Conclusions

These results provide evidence that total cortisol production is one mechanism by

which shift workers are at an increased likelihood for CMR, while cortisol pattern is not a

PT
mediator. The relatively small mediating effect of cortisol production suggests that other

RI
factors contribute to higher CMR in shift workers. Future research should consider a

longer collection period of urine to measure cortisol within a multifactorial approach to

SC
understanding the association between shift work and CVD, including consideration of

potential mediators such as sleep, lifestyle behaviors, and other circadian disruption

measures such as melatonin.39

U
AN
M
D
TE
C EP
AC

14
 ACCEPTED MANUSCRIPT

Acknowledgments

We thank study participants and the investigative team (L. McGillis-Hall, I. Janssen, C.

Collier) and staff (R. Corbin). This study was completed in partial fulfillment of degree

PT
requirements for JR’s Master of Science in Epidemiology at Queen’s University,

RI
Kingston, Ontario, Canada.

SC
Funding Sources: This study was funded by the Canadian Institutes of Health Research

and the Workplace Safety and Insurance Board (Ontario).

U
AN
Disclosures: None.
M
D
TE
C EP
AC

15
 ACCEPTED MANUSCRIPT

References

1. Stevens RG, Hansen J, Costa G, et al. Considerations of circadian impact for

defining “shift work” in cancer studies: IARC Working Group Report. Occup

Environ Med. 2011;68(2):154-162. doi:10.1136/oem.2009.053512.

PT
2. Vyas M, Garg A, Iansavichus A, et al. Shift work and vascular events: systematic

RI
review and meta-analysis. BMJ. 2012;345(e4800). doi:10.1136/bmj.e4800.

3. Wong IS, McLeod CB, Demers PA. Shift work trends and risk of work injury

SC
among Canadian workers. Scand J Work Environ Heal. 2011;37(1):54-61.

doi:10.5271/sjweh.3124.

4.
U
Vetter C, Devore EE, Wegrzyn LR, et al. Association between rotating night shift
AN
work and risk of coronary heart disease among women. JAMA.

2016;315(16):1726-1734. doi:10.1001/jama.2016.4454.
M

5. Mottillo S, Filion KB, Genest J, et al. The metabolic syndrome and cardiovascular
D

risk a systematic review and meta-analysis. J Am Coll Cardiol. 2010;56(14):1113-

TE

1132. doi:10.1016/j.jacc.2010.05.034.

6. Wang F, Zhang L, Zhang Y, et al. Meta-analysis on night shift work and risk of
EP

metabolic syndrome. Obes Rev. 2014;(8):1-12. doi:10.1111/obr.12194.

7. Proper KI, van de Langenberg D, Rodenburg W, et al. The relationship between

C

shift work and metabolic risk factors: a systematic review of longitudinal studies.
AC

Am J Prev Med. 2016;50(5):e147-e157. doi:10.1016/j.amepre.2015.11.013.

8. Knutsson A, Bøggild H. Shiftwork and cardiovascular disease: review of disease

mechanisms. Rev Environ Health. 2000;15(4):359-372.

doi:10.1515/REVEH.2000.15.4.359.

16
 ACCEPTED MANUSCRIPT

9. Puttonen S, Harma M, Hublin C. Shift work and cardiovascular disease –

pathways from circadian stress to morbidity. Scand J Work Environ Health.

2010;36(2):96-108. doi:10.5271/sjweh.2894.

10. Fries E, Dettenborn L, Kirschbaum C. The cortisol awakening response (CAR):

PT
Facts and future directions. Int J Psychophysiol. 2008;72(1):67-73.

RI
doi:10.1016/j.ijpsycho.2008.03.014.

11. Archer SN, Oster H. How sleep and wakefulness influence circadian rhythmicity:

SC
effects of insufficient and mistimed sleep on the animal and human transcriptome.

J Sleep Res. 2015;24(5):476-493. doi:10.1111/jsr.12307.

12.
U
Hung EWM, Aronson KJ, Leung M, Day A, Tranmer J. Shift work parameters and
AN
disruption of diurnal cortisol production in female hospital employees. Chronobiol

Int. 2016;33(8):1045-1055. doi:10.1080/07420528.2016.1196695.

M

13. Fekedulegn D, Burchfiel CM, Violanti JM, et al. Associations of long-term shift
D

work with waking salivary cortisol concentration and patterns among police
TE

officers. Ind Health. 2012;50(6):476-486. doi:10.2486/indhealth.2012-0043.

14. Niu S-F, Chung M-H, Chu H, et al. Differences in cortisol profiles and circadian
EP

adjustment time between nurses working night shifts and regular day shifts: A

prospective longitudinal study. Int J Nurs Stud. 2015;52(7):1193-1201.

C

doi:10.1016/j.ijnurstu.2015.04.001.
AC

15. Charles LE, Fekedulegn D, Burchfiel CM, et al. Shiftwork and diurnal salivary

cortisol patterns among police officers. J Occup Environ Med. 2016;58(6):542-

549. doi:10.1097/JOM.0000000000000729.

16. Vogelzangs N, Beekman ATF, Milaneschi Y, Bandinelli S, Ferrucci L, Penninx

17
 ACCEPTED MANUSCRIPT

BWJH. Urinary cortisol and six-year risk of all-cause and cardiovascular mortality.

J Clin Endocrinol Metab. 2010;95(11):4959-4964. doi:10.1210/jc.2010-0192.

17. Kumari M, Shipley M, Stafford M, Kivimaki M. Association of diurnal patterns in

salivary cortisol with all-cause and cardiovascular mortality: Findings from the

PT
Whitehall II study. J Clin Endocrinol Metab. 2011;96(5):1478-1485.

RI
doi:10.1210/jc.2010-2137.

18. Desantis AS, Diezroux A V, Hajat A, et al. Associations of salivary cortisol levels

SC
with metabolic syndrome and its components: the Multi-Ethnic Study of

Atherosclerosis. Psychoneuroendocrinology. 2014;49:260-271.

doi:10.1210/jc.2011-0483.
U
AN
19. Mayo Clinic. COCRU - Clinical: Cortisol/Cortisone, Free, Random, Urine.

20. DeBoer MD, Gurka MJ. Clinical utility of metabolic syndrome severity scores:
M

consideration for practitioners. Diabetes, Metab Syndr Obes Targets Ther.

D

2017;10:65-72. doi:10.2147/DMSO.S101624.
TE

21. Hillier TA, Rousseau A, Lange C, et al. Practical way to assess metabolic

syndrome using a continuous score obtained from principal components analysis.

EP

Diabetologia. 2006;49:1528-1535. doi:10.1007/s00125-006-0266-8.

22. Gurka MJ, Lilly CL, Oliver MN, et al. An examination of sex and racial/ethnic
C

differences in the metabolic syndrome among adults: A confirmatory factor

AC

analysis and a resulting continuous severity score. Metabolism. 2014;63(2):218-

225. doi:10.1016/j.metabol.2013.10.006.

23. Agarwal S, Jacobs DR, Vaidya D, et al. Metabolic Syndrome Derived from

Principal Component Analysis and Incident Cardiovascular Events: The Multi

18
 ACCEPTED MANUSCRIPT

Ethnic Study of Atherosclerosis (MESA) and Health, Aging, and Body

Composition (Health ABC). Cardiol Res Pract. 2012;2012:1-9.

doi:10.1155/2012/919425.

24. D’Agostino, Sr RB, Grundy S, Sullivan LM, Wilson P, for the CHD Risk

PT
Prediction Group. Validation of the Framingham coronary heart disease prediction

RI
scores. Jama. 2001;286(2):180. doi:10.1001/jama.286.2.180.

25. Ridker PM, Buring JE, Rifai N, Cook NR. Development and validation of

SC
improved algorithms for the assessment of global cardiovascular risk in women.

JAMA. 2007;297(6):611. doi:10.1001/jama.297.6.611.

26.
U
Pruessner JC, Kirschbaum C, Meinlschmid G, Hellhammer DH. Two formulas for
AN
computation of the area under the curve represent measures of total hormone

concentration versus time-dependent change. Psychoneuroendocrinology.

M

2003;28(7):916-931. doi:10.1016/S0306-4530(02)00108-7.
D

27. Fekedulegn DB, Andrew ME, Burchfiel CM, et al. Area under the curve and other
TE

summary indicators of repeated waking cortisol measurements. Psychosom Med.

2007;69(7):651-659. doi:10.1097/PSY.0b013e31814c405c.
EP

28. Preacher KJ, Hayes AF. Asymptotic and resampling strategies for assessing and

comparing indirect effects in multiple mediator models. Behav Res Methods.

C

2008;40(3):879-891. doi:10.3758/BRM.40.3.879.
AC

29. Lemche E, Chaban OS, Lemche A V. Neuroendorine and epigentic mechanisms

subserving autonomic imbalance and HPA dysfunction in the metabolic syndrome.

Front Neurosci. 2016;10:142. doi:10.3389/fnins.2016.00142.

30. Chung S, Hoon Son G, Kim K. Circadian rhythm of adrenal glucocorticoid: Its

19
 ACCEPTED MANUSCRIPT

regulation and clinical implications. BBA - Mol Basis Dis. 2011;1812:581-591.

doi:10.1016/j.bbadis.2011.02.003.

31. Arnaldi G, Scandali VM, Trementino L, Cardinaletti M, Appolloni G, Boscaro M.

Pathophysiology of dyslipidemia in Cushing’s syndrome. Neuroendocrinology.

PT
2010;92 Suppl 1(Suppl. 1):86-90. doi:10.1159/000314213.

RI
32. Macfarlane DP, Forbes S, Walker BR. Glucocorticoids and fatty acid metabolism

in humans: fuelling fat redistribution in the metabolic syndrome. J Endocrinol.

SC
2008;197(2):189-204. doi:10.1677/JOE-08-0054.

33. Dekker MJ, Koper JW, van Aken MO, et al. Salivary cortisol is related to

U
atherosclerosis of carotid arteries. J Clin Endocrinol Metab. 2008;93(10):3741-
AN
3747. doi:10.1210/jc.2008-0496.

34. Troxler R, Sprague E, Albanese R, Fuchs R, Thompson A. The association of

M

elevated plasma cortisol and early atherosclerosis as demonstrated by coronary

D

angiography. Atherosclerosis. 1977;26(2):151-162. doi:10.1016/0021-

TE

9150(77)90098-3.

35. Korsiak J, Tranmer J, Day A, Aronson K. Sleep duration as a mediator between an

EP

alternating day and night shift work schedule and metabolic syndrome among

female hospital employees. Occup Environ Med. 2017;(Online first).

C

doi:10.1136/oemed-2017-104371.
AC

36. Buxton OM, Cain SW, O’Connor SP, et al. Adverse metabolic consequences in

humans of prolonged sleep restriction combined with circadian disruption. Sci

Transl Med. 2012;4(129):129ra43. doi:10.1126/scitranslmed.3003200.

37. Rao MN, Blackwell T, Redline S, et al. Association between sleep duration and

20
 ACCEPTED MANUSCRIPT

24-hour urine free cortisol in the MrOS Sleep Study. PLoS One. 2013;8(9):e75205.

doi:10.1371/journal.pone.0075205.

38. Van Cauter E, Spiegel K, Tasali E, Leproult R. Metabolic consequences of sleep

and sleep loss. Sleep Med. 2008;1(9 Suppl 1):S23-8. doi:10.1016/S1389-

PT
9457(08)70013-3.

RI
39. Leung M, Tranmer J, Hung E, Korsiak J, Day AG, Aronson KJ. Shift work,

chronotype, and melatonin patterns among female hospital employees on day and

SC
night shifts. Cancer Epidemiol Prev Biomarkers. 2016;25(5):830-838.

doi:10.1158/1055-9965.EPI-15-1178.

40.
U
Karasek R. Job Content Questionnaire User’s Guide. 1985.
AN
41. Alberti K, Eckel RH, Grundy SM, et al. Harmonizing the Metabolic Syndrome A

Joint Interim Statement of the International Diabetes Federation Task Force on

M

Epidemiology and Prevention for the Study of Obesity. Circulation.

D

2009;120(16):1640-1645.
TE
C EP
AC

21
 ACCEPTED MANUSCRIPT

List of Tables

Table 1: Baseline characteristics of the study population.

Table 2: Differences in cardiometabolic risk factors and cortisol by current shift work

PT
status, adjusted for age.

Table 3: Mediation analysis of the relationship between shift work and CMR score by

RI
cortisol.

U SC
List of Figures
AN
Figure 1: Conceptual model of the relationship between shift work, cortisol, and

cardiometabolic risk (CMR).

M
D
TE
C EP
AC

22
 ACCEPTED MANUSCRIPT

Tables

Table 1: Baseline characteristics of the study population.

Shift Day workers p-value

workers (n=160)

PT
(n= 166)

RI
Age 38.8 (11.5) 44.8 (10.0) <0.01*

Post-menopausal 51 (30.7) 59 (36.9) 0.24

SC
Education

High-school/post-secondary (diploma, 87 (52.4) 96 (60.0) 0.15

certificate)
U 79 (47.6) 64 (40.0)
AN
Undergraduate/graduate/other
M

Employment category

Nursing 141 (84.9) 76 (47.5) <0.01*

D

Othera 25 (15.1) 84 (52.5)

TE

Past shift work experience (years) 11.6 (9.5) 7.8 (8.4) <0.01*

High life stress

EP

Present 20 (12.0) 32 (20.0) 0.04*

C

Not present 138 (83.1) 120 (75.0)

AC

Missing 8 (4.9) 8 (5.0)

Job strainb

High 50 (30.1) 37 (23.1) 0.15

Active 39 (23.5) 29 (18.1)

Passive 41 (24.7) 54 (33.8)

 ACCEPTED MANUSCRIPT

Low 36 (21.7) 40 (25.0)

Achieved physical activity recommendations

Yes 75 (45.2) 28 (39.4) 0.21

No 70 (42.2) 128 (49.4)

PT
Missing 21 (12.6) 4 (11.2)

RI
Smoking status

Current 10 (6.0) 21 (13.1) 0.01*

SC
Past smoker 34 (20.5) 43 (26.9)

Never 122 (73.5) 96 (60.0)

Alcohol intake
U
AN
Drinks >2 times/week 37 (22.3) 57 (35.6) 0.03*
M

Drinks 1-4 times/month 74 (44.6) 61 (38.1)

Rarely/never drinks 54 (32.5) 41 (25.6)

D

Missing 1 (0.6) 1 (0.6)

TE

Caffeine intake

Usually/often/sometimes 91 (54.8) 78 (48.7) 0.25

EP

Rarely/never 75 (45.2) 82 (51.3)

*p<0.05
C

CVD, cardiovascular disease.

AC

Data are presented as mean (SD) or n (%) for proportions, with p-values estimated by two-

sample t-tests and Chi-Squared tests, respectively

 ACCEPTED MANUSCRIPT

a. Nutrition services, lab technician, radiation therapist, environmental services, ergonomist,

education, research, total compensation association, abilities claim specialist, concurrent

reviewer, child life specialist

b. Classified according to Karasek’s job strain model41

PT
RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

Table 2: Differences in cardiometabolic risk factors and cortisol by current shift work status,

adjusted for age.

Shift worker Day worker Age-adjusted Age adjusted

PT
(n= 166) (n=160) difference (shift- p-value

day) (95% CI)

RI
AUCG 996.6 (286.1) 1139.2 (291.2) -116.1 <0.01*

(-180.5, -51.6)*

SC
AUCI 3.2 (378.5) 33.8 (468.2) -40.0 0.40

U
AN (-137.1, 55.3)

Blood pressure (mmHg)

Systolic 113.0 (13.8) 113.1 (14.8) 3.0 (-0.03, 6.0) 0.05

M

Diastolic 72.6 (8.4) 71.6 (9.2) 1.9 (-0.1, 3.8) 0.06

Waist circumference (cm) 87.9 (15.0) 85.6 (14.5) 3.1 (-0.2, 6.4) 0.07
D

Serum triglycerides 1.1 (0.9) 0.9 (0.7) 14.4 (0.9, 29.9)*a 0.04*
TE

(mmol/L)

HDL cholesterol (mmol/L) 1.6 (0.4) 1.6 (0.4) -0.03 (-0.1, 0.1) 0.60
EP

Fasting blood glucose 5.1 (0.8) 5.0 (0.6) 2.7 (0.1, 5.4)*b 0.04*
C

(mmol/L)
AC

CMR score 0.16 (1.6) -0.10 (1.5) 0.40 (0.2, 0.8)* <0.01*

Presence of MetSc 35 (21.0) 19 (11.9) 2.71 (1.4, 5.2)*d <0.01*

AUCG, area under the curve with respect to the ground; AUCI, area under the curve with respect

to the increase; CMR, cardiometabolic risk; HDL, high density lipoproptein; MetS, metabolic

syndrome.
 ACCEPTED MANUSCRIPT

*p<0.05

Data are presented as mean (SD) or n (%), age-adjusted p-values and differences derived from

multiple linear/logistic regression

a. Percent change in triglycerides

PT
b. Percent change in fasting blood glucose

RI
c. Defined according to Alberti et al., 200942

d. Odds ratio

U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

Table 3: Mediation analysis of the relationship between shift work and CMR score by cortisol.

Mediator n Path a Path ba Path c’ Path c Path ab

(β, 95% CI) (β, 95% CI) (β, 95% CI) (β, 95% CI) (β, 95% CI)

AUCG 325 -157.19 -0.21 0.43 0.52 0.11

PT
(-228.50, - (-0.38, -0.04)*c (0.05, (0.15, (0.02, 0.23)*

RI
85.88)*b 0.80)* 0.89)*

AUCI 324 -113.50 -0.10 0.51 0.52 0.03

SC
(-215.26, - (-0.28, 0.09)c (0.14, (0.15, (-0.02, 0.10)

11.74)*b 0.88)* 0.89)*

U
AUCG, area under the curve with respect to the ground; AUCI, area under the curve with respect
AN
to the increase; CMR, cardiometabolic risk.
M

All models adjusted for age, education, menopausal status, and past shift work experience

a. Models additionally adjusted for caffeine intake and smoking status

D

b. Per µmol/mol of creatinine

TE

c. Per 1-unit increase in standard deviation of mediator

*p<0.05
C EP
AC
 ACCEPTED MANUSCRIPT

Figures

PT
RI
U SC
Figure 1: Conceptual model of the relationship between shift work, cortisol, and
AN
cardiometabolic risk (CMR).
M
D
TE
C EP
AC