Beskæftigelsesudvalget 2019-20
BEU Alm.del Bilag 93
Offentligt
2128272_0001.png
Journal of Psychosomatic Research 128 (2020) 109867
Contents lists available at
ScienceDirect
Journal of Psychosomatic Research
journal homepage:
www.elsevier.com/locate/jpsychores
Effort-reward imbalance at work and risk of type 2 diabetes in a national
sample of 50,552 workers in Denmark: A prospective study linking survey
and register data
☆
Mads Nordentoft
a,b,
, Naja H. Rod
b
, Jens Peter Bonde
c
, Jakob B. Bjorner
a,d,e
, Ida E.H. Madsen
a
,
Line R.M. Pedersen
a
, Bryan Cleal
f
, Linda L. Magnusson Hanson
g
, Mette A. Nexo
f
, Jaana Pentti
h,i
,
Sari Stenholm
h
, Tom Sterud
j
, Jussi Vahtera
h
, Reiner Rugulies
a,b,k
⁎
a
T
National Research Centre for the Working Environment, Copenhagen, Denmark
Section of Epidemiology, Department of Public Health, University of Copenhagen, Copenhagen, Denmark
c
Department of Occupational and Environmental Medicine, Bispebjerg-Frederiksberg Hospital, Copenhagen, Denmark
d
Optum Patient Insights, Lincoln, RI, USA
e
Section of Social Medicine, Department of Public Health, University of Copenhagen, Copenhagen, Denmark
f
Steno Diabetes Center Copenhagen, Gentofte, Denmark
g
Stress Research Institute, Stockholm University, Stockholm, Sweden
h
Department of Public Health, University of Turku and Turku University Hospital, Turku, Finland
i
Clinicum, Faculty of Medicine, University of Helsinki, Helsinki, Finland
j
The National Institute of Occupational Health, Oslo, Norway
k
Department of Psychology, University of Copenhagen, Copenhagen, Denmark
b
A R TICL E INFO
Keywords:
Diabetes mellitus
Epidemiology
Occupation
Population-based
Psychosocial work factors
Stress
A BSTR A CT
Objective:
To examine the prospective relation between effort-reward imbalance at work and risk of type 2
diabetes.
Methods:
We included 50,552 individuals from a national survey of the working population in Denmark, aged
30–64 years and diabetes-free at baseline. Effort-reward imbalance was defined, in accordance with the litera-
ture, as a mismatch between high efforts at work (e.g. high work pace, time pressure), and low rewards received
in return (e.g. low recognition, job insecurity) and assessed as a continuous and a categorical variable. Incident
type 2 diabetes was identified in national health registers. Using Cox regression we calculated hazard ratios (HR)
and 95% confidence intervals (95% CI) for estimating the association between effort-reward imbalance at
baseline and risk of onset of type 2 diabetes during follow-up, adjusted for sex, age, socioeconomic status,
cohabitation, children at home, migration background, survey year and sample method.
Results:
During 136,239 person-years of follow-up (mean = 2.7 years) we identified 347 type 2 diabetes cases
(25.5 cases per 10,000 person-years). For each one standard deviation increase of the effort-reward imbalance
score at baseline, the fully adjusted risk of type 2 diabetes during follow-up increased by 9% (HR: 1.09, 95% CI:
0.98–1.21). When we used effort-reward imbalance as a dichotomous variable, exposure to effort-reward im-
balance was associated with an increased risk of type 2 diabetes with a HR of 1.27 (95% CI: 1.02–1.58).
Conclusion:
The results of this nationwide study of the Danish workforce suggest that effort-reward imbalance at
work may be a risk factor for type 2 diabetes.
Abbreviations:
ANOVA, analysis of variance; ATC, anatomical therapeutic chemical; BMI, body mass index; CI, confidence interval; ERI, effort-reward imbalance;
HPA, hypothalamus-pituitary-adrenal; HR, hazard ratio; ICD-10, international classification of diseases and related health problems version 10; NRCWE, National
Research Centre for the Working Environment; PCOS, polycystic ovarian syndrome; SES, socioeconomic status; WEHD, Work Environment and Health in Denmark
2012–2020.
☆
The study was conducted at the National Research Centre for the Working Environment, Copenhagen, Denmark.
⁎
Corresponding author at: National Research Centre for the Working Environment, Lerso Parkallé 105, DK-2100 Copenhagen, Denmark.
E-mail address:
[email protected]
(M. Nordentoft).
https://doi.org/10.1016/j.jpsychores.2019.109867
Received 30 August 2019; Received in revised form 31 October 2019; Accepted 31 October 2019
0022-3999/ © 2019 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/).
 BEU, Alm.del - 2019-20 - Bilag 93: Orientering om resultater fra NFA-artikel om sammenhæng mellem indsatsbelønningsubalance i arbejdet og type 2-diabetes, fra beskæftigelsesministeren
M. Nordentoft, et al.
Journal of Psychosomatic Research 128 (2020) 109867
1. Introduction
Work-related stress may be associated with risk of developing type 2
diabetes [1,2]. Possible physiological pathways include stress-induced
dysregulation of the hypothalamus-pituitary-adrenal (HPA) axis and
the central nervous system [3,4], resulting in increased cortisol levels,
and subsequent disturbance in blood glucose regulation [5], all of
which are associated with risk of insulin resistance and type 2 diabetes
[6]. Work stress may also be related to type 2 diabetes via changes in
health-related behavior, such as sedentary life style and poor diet, and
psycho-physiological changes, such as sleep disturbances and depres-
sive disorders associated with inflammation and cardiometabolic
changes [1,7,8].
The model of effort-reward imbalance (ERI) is an established the-
oretical approach to define and assess stress at the workplace [9]. The
model is built on the notion that a lack of reciprocity in terms of the
efforts that are put into work and the rewards received in return causes
emotional and psycho-physiological stress reactions that subsequently
increase the risk of ill-health. The model emerged in the scientific dis-
cussions in the 1990s [10] and has been tested mostly in relation to
cardiovascular disease and mental health. Recently, two meta-analyses
of prospective cohort studies showed that employees with high ERI
were at higher risk of coronary heart disease [11] and depressive dis-
orders [12].
Research on ERI and risk of type 2 diabetes is scarce. To our
knowledge, only two prospective studies have previously examined the
association between ERI and risk of type 2 diabetes [13,14]. Kumari
et al. (2004) studied 8067 civil servants (70% men) from 20 London-
based departments in the Whitehall II Study and found an association
between high ERI and risk of type 2 diabetes among men, but not
among women [13]. Mutambudzi et al. (2018) studied 1932 workers in
the United States (age > 50 years) from the Health and Retirement
Study and reported an association between high ERI and risk of self-
reported diabetes [14].
Both the study by Kumari et al. (2004) and Mutambudzi et al.
(2018) examined ERI and diabetes in selected occupational groups
(civil servants and older workers) and in relatively small study samples.
These studies also did not investigate if the individual dimensions of
efforts and rewards were associated differently with risk of type 2
diabetes, which may be important to guide future interventions.
Further, it has recently been demonstrated that the impact of other
psychosocial work factors such as long working hours on type 2 dia-
betes risk may depend on socioeconomic status (SES) [15], but whether
the relation between ERI and type 2 diabetes are modified by SES re-
mains unstudied.
In this article, we investigate the prospective relation between ERI
and risk of type 2 diabetes in a national sample of > 50,000 workers in
Denmark [16]. We test both ERI and its subcomponents and examine if
associations differ with regards to sex, age, SES, and migration back-
ground.
2. Methods
2.1. Design and population
The study is a prospective cohort study with register-based follow-
up. Before data linkage of the exposure with the outcome, we published
a study protocol containing detailed descriptions of the design,
methods, data and statistical analyses [16]. Briefly, the study popula-
tion was participants from the biennial survey ‘Work Environment and
Health in Denmark 2012-2020 (WEHD)’ which is a national sample of
the working population in Denmark. Inclusion criteria were a) liable to
pay taxes in Denmark, b) employed with monthly working hours of
≥35 and a monthly income of 3000 Danish kroner (≥$530/€400), c)
aged 18–64 years and d) registered with an address in Denmark. WEHD
consists of 1) a national survey of individuals from the general working
2
population, 2) a supplementary survey (in 2012 and 2016) of in-
dividuals from selected workplaces and 3) a cohort, consisting of those
individuals who responded in the 2012 wave in the general working
population and were invited again in 2014 and 2016. All individuals
received a letter with a link to an online questionnaire containing ap-
proximately 160 questions about work environment, health and life-
style. Non-responders were contacted by mail and telephone and re-
ceived a hard copy of the questionnaire. The overall response rate
across all surveys was 53.9%. Compared to respondents of the survey in
2012, among non-respondents there were more men, they were
younger, had lower education, they were less cohabiting and there were
more with a migration background [17]. To maximize sample size and
statistical power we included all unique responders to the surveys in
2012, 2014 and 2016.
We used a unique personal identification number to link partici-
pants to nationwide registers [18]. According to Danish legislation,
research projects involving surveys with questionnaire and register-
based data only, do not need approval from The National Committee on
Health Research Ethics. The study was approved by The Danish Data
Protection Agency through the joint notification of the National Re-
search Centre for the Working Environment, Copenhagen, Denmark
(no. 2015-57-0074). We obtained register-based information from
Statistics Denmark (no. 706706) and Sundhedsdatastyrelsen (‘The
Danish Health Authority’, no. FSEID-00003251 and no. FSEID-
00003281). All data are stored in a protected server environment
hosted by Statistics Denmark.
A complete list of criteria for inclusion and exclusion is available in
the study protocol [16]. In short, we excluded participants < 30 years
of age because of concerns about misclassification as incident diabetes
is more likely to be type 1 than type 2 diabetes in this age-group [19].
We excluded participants with all types of prevalent diabetes at base-
line and participants with polycystic ovarian syndrome (PCOS) due to
possible overlap with diabetes in drug treatment (n = 2262) assessed
by either a hospital diagnosis, previous purchases of relevant pre-
scription medicine or self-report in WEHD. In addition, we excluded
participants who were pregnant at baseline because of a risk of devel-
oping gestational diabetes (n = 786). We ascertained pregnancy for a
period of 280 days before the date of starting register based maternity
leave [20]. Finally, we excluded participants with missing data about
ERI (n = 1353) and covariates (n = 293), yielding a study sample of
50,552 individuals. See
Fig. 1
for a flowchart detailing the exclusion
process. There were no considerable differences in baseline character-
istics in the population between exclusion stages in the flowchart (see
Table A.1, Appendix A, for a comparison of characteristics at each stage
in the flowchart).
2.2. Effort-reward imbalance
Effort-reward imbalance describes the mismatch between efforts
spent at work and rewards received in terms of financial and career-
related rewards, esteem and job security [21]. As WEHD did not include
items from the original ERI-questionnaire [21], i.e. the scales on efforts
and rewards in WEHD were modified and dissimilar to the original
questions, we used proxy measures to construct scales for efforts and
rewards. Perceived efforts were measured with a scale consisting of six
items, assessing time pressure, high work pace, difficult deadlines,
unexpected tasks, being at disposal outside normal working hours and
overtime work. Perceived rewards were measured with a scale con-
sisting of five items, assessing promotion of professional development,
recognition and appreciation by management, being treated fairly at
work, worries about becoming unemployed and worries about being
transferred to another job (see Table B.1, Appendix B, for the wording
of the effort and reward items).
We calculated sum scale scores for each scale, with high values
indicating high efforts and high rewards, respectively. The internal
consistency of the effort and reward scales was satisfactory with
 BEU, Alm.del - 2019-20 - Bilag 93: Orientering om resultater fra NFA-artikel om sammenhæng mellem indsatsbelønningsubalance i arbejdet og type 2-diabetes, fra beskæftigelsesministeren
2128272_0003.png
M. Nordentoft, et al.
Journal of Psychosomatic Research 128 (2020) 109867
insulins, and sub-levels) or A10AE56 (insulin degludec and liraglutide)
in the Danish National Prescription Registry [18]. Participants who had
purchased these medications but also appeared with ICD-10 code E10
(type 1 diabetes mellitus, and sub-levels) in the Danish National Patient
Register, were not considered as type 2 diabetes cases.
2.4. Confounders
We identified potential confounders based on the literature using
the method of Directed Acyclic Graphs [22] as detailed in the study
protocol [16]. Using a unique personal identification number we de-
rived information on the confounders from nationwide registers and
linked these data to the participants [18].
The confounders in this study were sex, age, SES, cohabitation,
young children in the household and migration background. Low versus
high SES is associated with a higher risk of type 2 diabetes [23]. The
experience of ERI may follow a social pattern, although there are
conflicting results on the direction, with one study reporting a higher
prevalence of ERI in employees of lower SES [24] whereas another
study reported a higher prevalence of ERI among employees of higher
SES [25]. In this study we defined SES as highest achieved educational
level, divided into three groups
(Low, ≤9 years; Intermediate,
10–12 years; High, ≥13 years).
Cohabitation may be a protective factor
for the experience of and coping with ERI [26] and for the development
of type 2 diabetes [27]. We defined cohabitation as either living alone
or cohabiting. Having young children in the household is a possible
stressor that may affect one's perception of demands at work and may
also affect the ability to recover [28]. We defined young children in the
household as number of children
≤7
years of age in the following ca-
tegories:
0; ≥1.
Migration background is associated with different
predispositions to the development of type 2 diabetes in Denmark [29]
and may also be associated with different levels of perceived work
stress [30]. We categorized migration background as
‘No migration
background’
and
‘Migration background’,
following the categorization by
Statistics Denmark [31].
The relationship between ERI and the risk of developing type 2
diabetes may involve a complex interplay between behavioral factors,
psycho-physiological disturbances and adverse health. ERI is associated
with physical inactivity [32], unhealthy alcohol consumption, smoking,
overweight [33], depression [12], sleep disturbances [28], hyperten-
sion and dyslipidemia [7] which in addition may be potential risk
factors for the development of type 2 diabetes [34–36]. Therefore, these
factors may be potential intermediate variables on the causal pathway
from ERI to type 2 diabetes, and hence should not be controlled for.
However, they may also affect one's perception of efforts and rewards
and therefore be potential confounders. Since we only have baseline
measures of these factors, and therefore cannot distinguish between
mediation and confounding for these variables, we include these vari-
ables not in the main analyses but in supplementary analyses. Detailed
descriptions of the covariates have been published in the study protocol
[16]. In brief, from WEHD we included self-reported physical activity
(categorical in
Table 1:
‘inactive’; ‘low’; ‘medium’; ‘high’,
continuous in
analyses), alcohol consumption (units pr. week) with categories:
‘low
(women: 0-7; men: 0-14)’; ‘medium (women: 8–14; men: 14–21)’; ‘high
(women: > 14; men: > 21)’,
smoking status (‘never’;
‘current/former’),
and BMI categorized into:
‘underweight (< 18.5 kg/m
2
)’; ‘normal weight
(18.5-24.9 kg/m
2
)’; ‘overweight (25-29.9 kg/m
2
)’; ‘obese (≥30 kg/m
2
)’.
Indicators of psycho-physiological disturbances were defined as having
an ICD-10 diagnosis for depressive disorders, sleep disturbances, hy-
pertension or dyslipidemia or by at least two purchases of prescription
medicine for these conditions within a two year period prior to base-
line. We identified depressive disorders with ICD-10 codes F32, F33,
F34.1 and F06.3, with ATC-code N06A and by self-reported depressive
symptoms in WEHD. We identified sleep disturbances with ATC-codes
N05C and N05B and by self-report in WEHD. We defined dyslipidemia
by ICD-10 code E78 and ATC-code C10. Hypertension was defined by
3
Fig. 1.
Flowchart detailing the exclusion of participants in the study.
Cronbach's alpha coefficients of 0.77 and 0.68 for efforts and rewards,
respectively. We computed an ERI-ratio by dividing efforts by rewards.
As recommended in the literature [21], we used a correction factor
((efforts/rewards)*(5/6)) to take into account the unequal number of
items in the effort and reward scale, so that an ERI-ratio of > 1.0 in-
dicates that efforts exceed rewards. For the regression analyses, we
standardized ERI, efforts, and rewards scores per one standard devia-
tion (SD) increase (mean = 0, SD = 1) and treated them as continuous
measures.
In addition to the continuous ERI-score, we constructed two cate-
gorical measures. First, based on the quartiles of their respective scale
scores we categorized ERI, efforts and rewards into low, medium-low,
medium-high and high. Second, we dichotomized ERI using a cut-off
of > 1.0 to indicate exposure to potential health-hazardous ERI, as
proposed in the literature [9] (see Appendix B, for more details on the
construction of the ERI-measures).
2.3. Type 2 diabetes
We identified incident type 2 diabetes with the International
Classification of Diseases version 10 (ICD-10) code E11 (type 2 diabetes
mellitus with and without complications, and sub-levels) in the Danish
National Patient Register that includes all in- and out-patient hospital
admissions, from death certificates in the Danish Register of Causes of
Death and by purchases of prescription medicine with at least two re-
demptions within a two year period with the Anatomical Therapeutic
Chemical (ATC) codes A10B (blood glucose lowering drugs, excl.
 BEU, Alm.del - 2019-20 - Bilag 93: Orientering om resultater fra NFA-artikel om sammenhæng mellem indsatsbelønningsubalance i arbejdet og type 2-diabetes, fra beskæftigelsesministeren
2128272_0004.png
M. Nordentoft, et al.
Journal of Psychosomatic Research 128 (2020) 109867
Table 1
Baseline characteristics of the study population and relation to effort-reward imbalance (ERI).
Characteristics (missing %)
Total (n = 50,552)
n
Age (0.0%)
30–39
40–49
50–59
60–64
Sex (0.0%)
Women
Men
Migration background (0.0%)
No migration background
Migration background
Education (0.0%)
High (≥13 years of education)
Intermediate (10–12 years of education)
Low (≤9 years of education)
Cohabitation (0.0%)
Yes
No
Young children in the household (0.0%)
No
Yes
Physical activity (1.6%)
Inactive
Low
Medium
High
Alcohol consumption (2.2%)
Low
Medium
High
Smoker (1.0%)
Never
Current/former
BMI (1.5%)
Underweight (< 18.5 kg/m
2
)
Normal weight (18.5–24.9 kg/m
2
)
Overweight (25–29.9 kg/m
2
)
Obese (≥30 kg/m
2
)
Depressive disorders (0.9%)
No
Yes
Sleep disturbances (0.6%)
No
Yes
Hypertension (0.0%)
No
Yes
Dyslipidemia (0.0%)
No
Yes
Parental type 2 diabetes (16.6%)
0
≥1
a
b
ERI-ratio
(%)
a
Mean
b
(SD)
High ERI
c
n
(%)
d
9649
16,754
18,630
5519
26,378
24,174
47,675
2877
22,180
21,952
6420
40,401
10,151
40,513
10,039
5272
14,933
24,851
4665
32,491
10,713
6247
24,477
25,546
502
24,191
18,233
6872
44,400
5686
42,619
7642
43,460
7092
47,211
3341
35,421
6737
(19.1)
(33.1)
(36.9)
(10.9)
(52.2)
(47.8)
(94.3)
(5.7)
(43.9)
(43.4)
(12.7)
(79.9)
(20.1)
(80.1)
(19.9)
(10.6)
(30.0)
(50.0)
(9.4)
(65.7)
(21.7)
(12.6)
(48.9)
(51.1)
(1.0)
(48.6)
(36.6)
(13.8)
(88.6)
(11.4)
(84.8)
(11.4)
(86.0)
(14.0)
(93.4)
(6.6)
(84.0)
(16.0)
0.94
0.93
0.93
0.84
0.92
0.93
0.92
0.95
0.93
0.92
0.91
0.92
0.95
0.92
0.93
0.95
0.93
0.92
0.92
0.92
0.92
0.94
0.91
0.93
0.94
0.91
0.93
0.95
0.90
1.12
0.90
1.07
0.93
0.92
0.93
0.92
0.93
0.94
(0.32)
(0.32)
(0.35)
(0.32)
(0.34)
(0.33)
(0.33)
(0.38)
(0.31)
(0.34)
(0.38)
(0.33)
(0.37)
(0.34)
(0.31)
(0.38)
(0.34)
(0.33)
(0.31)
(0.34)
(0.32)
(0.35)
(0.32)
(0.35)
(0.38)
(0.32)
(0.34)
(0.37)
(0.30)
(0.49)
(0.31)
(0.44)
(0.33)
(0.35)
(0.33)
(0.35)
(0.33)
(0.36)
2481
4251
4779
894
6350
6055
11,616
789
5400
5420
1585
9584
2821
9874
2531
1451
3707
5929
1093
7898
2539
1661
5625
6625
129
5607
4545
1911
9694
2571
9346
2987
10,684
1721
11,583
822
8797
1755
(25.7)
(25.4)
(25.7)
(16.2)
(24.1)
(25.0)
(24.4)
(27.4)
(24.3)
(24.7)
(24.7)
(23.7)
(27.8)
(24.4)
(25.2)
(27.5)
(24.8)
(23.9)
(23.4)
(24.3)
(23.7)
(26.6)
(23.0)
(25.9)
(25.7)
(23.2)
(24.9)
(27.8)
(21.8)
(45.2)
(21.9)
(39.1)
(24.6)
(24.3)
(24.5)
(24.6)
(24.8)
(26.1)
Column percentage.
P-values from one-way ANOVA tests for differences in characteristics of participants and means of effort-reward imbalance were < 0.001 for most character-
istics, except hypertension (p = .026), dyslipidemia (p = .10) and parental type 2 diabetes (p = .012).
c
4
th
quartile of the ERI-ratio.
d
Row percentage.
ICD-10 codes I10-I13 and I15 and purchases of antihypertensive drugs
[16].
2.5. Statistical analyses
Using Cox proportional hazard models with age as the underlying
time axis, we calculated hazard ratios (HR) and 95% confidence in-
tervals (95% CI) for the prospective association between ERI and in-
cident type 2 diabetes and assessed the effect of the individual di-
mensions of efforts and rewards. We also investigated whether the
4
estimate attributable to ERI was explained by an interaction between
efforts and rewards (ERI) or by one of these dimensions by analyzing if
the interaction between efforts and rewards was associated with a
higher risk of developing type 2 diabetes when adjusting for efforts and
rewards. Interactions were tested in a Cox model (deviation from
multiplicative interaction) and in Aalen's additive regression model
(deviation from additive interaction). In the study protocol we planned
to analyze categorical measures of ERI, efforts and rewards [16]. To
increase statistical power we added analyses of continuous measures of
ERI, efforts and rewards.
 BEU, Alm.del - 2019-20 - Bilag 93: Orientering om resultater fra NFA-artikel om sammenhæng mellem indsatsbelønningsubalance i arbejdet og type 2-diabetes, fra beskæftigelsesministeren
M. Nordentoft, et al.
Journal of Psychosomatic Research 128 (2020) 109867
Participants were followed from the date they filled in the ques-
tionnaire until incident type 2 diabetes or censoring due to emigration,
death by other causes than diabetes, incident gestational diabetes or
PCOS, pregnancy, or at end of follow-up (31 December 2016), which-
ever came first. We calculated crude estimates, sex- and age-adjusted
estimates (model 1) and estimates further adjusted for cohabitation,
young children in the household, SES, migration background and for
survey and sample method (i.e. whether participants were from the
samples of the general working population or the supplementary
workplace samples) (model 2).
We assessed the proportional hazards assumption of the Cox model
by testing interactions of the covariates with log(time), by visual in-
spection of log(−log(survival)) curves and by testing and visually in-
specting the interactions between Schoenfeld residuals of the covariates
with log(time). We found no major violations of the proportional ha-
zards assumption.
In sensitivity analyses we investigated the possibility of mis-
classifying prevalent cases as incident cases by applying a washout
period where we excluded incident cases during the first year of follow-
up. In addition, we investigated if an age cut-off point of ≥40 years
instead of ≥30 years for inclusion into the study would have an influ-
ence on the results. Further, we assessed if the association between ERI
and type 2 diabetes depended on type of ascertainment of prevalent and
incident diabetes.
We investigated possible effect modification of the association be-
tween ERI, efforts and rewards and risk of developing type 2 diabetes
by sex, age, SES and migration background in stratified analyses.
We further adjusted model 2 for potential behavioral and psycho-
physiological mediators, and parental type 2 diabetes as suggested in
previous studies [13,14]. The first model included physical activity,
alcohol consumption, smoking, BMI, sleep disturbances, depressive
disorders, hypertension and dyslipidemia. In the second model we
linked participants to register data of their parents and included par-
ental type 2 diabetes as a covariate (no parent with type 2 diabetes
versus one or two parents with type 2 diabetes).
All analyses were carried out using the statistical software R version
3.5.1. All statistical tests were two-sided with a significance level of 5%.
3. Results
efforts and rewards, and risk of type 2 diabetes. The HR of incident type
2 diabetes was 1.03 (95% CI: 0.93–1.15) per one SD increase in the
effort-score and 0.91 (95% CI: 0.82–1.01) per one SD increase in the
reward-score. When using reward-quartiles in the analyses, medium-
high rewards, compared to low rewards were associated with a lower
risk of type 2 diabetes in the fully-adjusted analysis (HR: 0.70, 95% CI:
0.52–0.93) (Table C.1, Appendix C).
We found no interaction between efforts (continuous, 1 SD increase)
and rewards (continuous, 1 SD increase) when tested on a multi-
plicative scale (p
interaction
= 0.84) or on an additive scale
(p
interaction
= 0.77) (results not shown).
3.4. Sensitivity analyses
In the sensitivity analyses we took into account a one year washout
period, an alternative age cut-off point, different exclusion criteria at
baseline and types of outcome ascertainment. The estimates for the
association between ERI, its two components and risk of incident type 2
diabetes in the sensitivity analyses were similar to the estimates in the
main analyses (Fig.
2).
3.5. Subgroup analyses
The subgroup analyses are presented in
Fig. 3.
Overall, there were
no large differences between men and women or in relation to migra-
tion background. Regarding age and SES groups, the association esti-
mates for ERI, efforts and rewards were more heterogeneous (Fig.
3).
High rewards were associated with a lower risk of type 2 diabetes in the
group with the highest SES (HR: 0.75, 95% CI: 0.60–0.93), but not in
the intermediate or low SES groups.
3.6. Adjustment for behavioral and psycho-physiological mediators and
parental type 2 diabetes
When we further adjusted the estimates in the main analyses for
potential mediators and parental type 2 diabetes, the estimates for ERI,
efforts and rewards remained virtually the same as in the main analyses
(Table E.1, Appendix E).
4. Discussion
3.1. Baseline characteristics
4.1. Summary of findings
Table 1
shows the characteristics of the study sample at baseline.
There were no differences in mean ERI between men (mean = 0.93,
SD = 0.33) and women (mean = 0.92, SD = 0.34) or between SES
groups. High ERI was more prevalent among those below 60 years of
age, among participants with migration background and among those
living alone.
3.2. ERI and incident type 2 diabetes
During 136,239 person-years of follow-up (mean = 2.7 years) we
identified 347 cases of incident type 2 diabetes (25.5 cases per 10,000
person-years). A one SD higher ERI-score was associated with a 9%
increased risk of incident type 2 diabetes in the sex- and age-adjusted
analysis (HR: 1.09, 95% CI: 0.98–1.21,
Fig. 2,
model 1). Adjustment for
all confounders did not change the estimate (HR: 1.09, 95% CI:
0.98–1.21,
Fig. 2,
model 2). Using ERI categorized in quartiles, instead
of the continuous ERI, yielded similar results (Table C.1, Appendix C).
Using ERI dichotomized into exposure versus no exposure, based on an
ERI-ratio > 1.0, yielded a HR of 1.27 (95% CI: 1.02–1.58) (Table D.1,
Appendix D).
3.3. ERI components and incident type 2 diabetes
Fig. 2
also shows the association between the ERI components,
5
In this prospective study of a national sample of 50,552 workers in
Denmark we found a suggestive, albeit not statistical significant asso-
ciation between an increase in ERI-score at baseline and a 9% increased
risk of developing type 2 diabetes. When we analyzed ERI as a di-
chotomized variable, exposure to ERI predicted risk of type 2 diabetes
with a hazard ratio of 1.27, which was statistically significant. Analyses
of the ERI components suggested that the association between high
rewards and a lower risk of type 2 diabetes was more pronounced than
the association between high efforts and higher risk of type 2 diabetes.
4.2. Comparison with previous studies on ERI and type 2 diabetes
To our knowledge, this is the first study to investigate the pro-
spective association between ERI, efforts and rewards and type 2 dia-
betes in a national workforce. Our study is also by far the largest study
on ERI and type 2 diabetes to date. Previously, Mutambudzi et al.
(2018) reported an association between ERI and self-reported diabetes
among 1932 workers, 50 years or older in the United States [14]. Ku-
mari et al. (2004) examined 8067 British civil servants and found an
association between ERI and type 2 diabetes among men, but not
among women [13]. We did not find such sex differences in our study.
Comparisons of the results from our study with these two previous
studies have to be viewed with caution, as we examined a sample from
 BEU, Alm.del - 2019-20 - Bilag 93: Orientering om resultater fra NFA-artikel om sammenhæng mellem indsatsbelønningsubalance i arbejdet og type 2-diabetes, fra beskæftigelsesministeren
2128272_0006.png
M. Nordentoft, et al.
At risk
n
Incident type 2 diabetes
Cases/
pr. 10,000 person−years
ERI
(continuous, 1 SD)
HR (95% CI)
Efforts
(continuous, 1 SD)
HR (95% CI)
Rewards
(continuous, 1 SD)
HR (95% CI)
Main analyses
Crude
Model 1
Model 2
50,552
50,552
50,552
347/25.5
347/25.5
347/25.5
1.07 (0.96−1.19)
1.09 (0.98−1.21)
1.09 (0.98−1.21)
0.95 (0.85−1.06)
0.97 (0.87−1.07)
1.03 (0.93−1.15)
0.87 (0.79−0.97)
0.86 (0.78−0.96)
0.91 (0.82−1.01)
Sensitivity analyses
One year washout
Age cut−off >=40 years
Not excluding self−reported
prevalent diabetes
6
50,450
40,903
50,701
245/18
320/28.7
392/28.7
1.16 (1.03−1.31)
1.09 (0.98−1.21)
1.09 (0.99−1.20)
1.06 (0.93−1.20)
1.02 (0.91−1.14)
1.04 (0.94−1.15)
0.85 (0.75−0.96)
0.90 (0.81−1.00)
0.90 (0.82−0.99)
Outcome ascertainment
Diagnoses + death
Medication + death
Diagnoses + medication
(including insulin) + death
50,552
50,552
50,552
64/4.7
335/24.6
369/27.1
1.06 (0.83−1.34)
1.10 (0.99−1.22)
1.08 (0.97−1.19)
0.80
1.0
1.25
←
Decreased risk Increased risk
→
0.93 (0.72−1.20)
1.04 (0.93−1.16)
1.02 (0.92−1.14)
0.80
1.0
1.25
←
Decreased risk Increased risk
→
0.87 (0.69−1.11)
0.90 (0.81−1.00)
0.92 (0.83−1.02)
Journal of Psychosomatic Research 128 (2020) 109867
0.80
1.0
1.25
←
Decreased risk Increased risk
→
Fig. 2.
Main analyses and sensitivity analyses of the relation between effort-reward imbalance (ERI), efforts and rewards and risk of developing type 2 diabetes.
Model 1: Adjusted for sex and age. Model 2: Adjusted for sex, age, cohabitation, young children in the household, SES, migration background, survey year and sample method. Sensitivity analyses: Adjusted for covariates
of model 2. At risk: Participants without diabetes at baseline. Incident type 2 diabetes: Participants who developed type 2 diabetes during follow-up.