BEU, Alm.del - 2021-22 - Bilag 186: Orientering af BEU om NFA's undersøgelse af sammenhængen mellem følelsesmæssige krav i arbejdet og risiko for hospitals-behandlet depression, fra beskæftigelsesministeren

Beskæftigelsesudvalget 2021-22
BEU Alm.del Bilag 186
Offentligt

Downloaded from www.sjweh.fi on March 09, 2022

Scand J Work Environ Health

Online-first -article

Published online: 09 Mar 2022

doi:10.5271/sjweh.4020

Emotional demands at work and risk of hospital-treated

depressive disorder in up to 1.6 million Danish employees: a

prospective nationwide register-based cohort study

Madsen IEH, Sørensen JK, Bruun JE, Framke E, Burr H, Melchior M,

Sivertsen B, Stansfeld S, Kivimäki M, Rugulies R

Previous studies have linked emotional demands with depression, but

may have been biased by reporting bias or the selection of vulnerable

employees into emotionally demanding occupations. We find an

increased risk of hospital-treated depressive disorder in occupations

with high emotional demands that cannot be explained by reporting

bias or the included risk factors for depression recorded before

workforce entry.

Affiliation:

National Research Centre for the Working Environment,

Lerso Parkalle 105, DK- 2100 Copenhagen, Denmark. [email protected]

Refers to the following texts of the Journal:

2020;46(5):498-507

2012;38(3):187-192 2010;36(6):435-444 2003;29(4):270-279

Key terms:

cohort study; Denmark; depression; depressive disorder;

emotional demand; job exposure matrix; psychosocial; register-based

study; stress

Additional material

Please note that there is additional material available belonging to

this article on the

Scandinavian Journal of Work, Environment & Health

-website.

This work is licensed under a

Creative Commons Attribution 4.0 International License.

Print ISSN: 0355-3140 Electronic ISSN: 1795-990X

BEU, Alm.del - 2021-22 - Bilag 186: Orientering af BEU om NFA's undersøgelse af sammenhængen mellem følelsesmæssige krav i arbejdet og risiko for hospitals-behandlet depression, fra beskæftigelsesministeren

riginal article

Scand J Work Environ Health – online first: 9 March 2022. doi:10.5271/sjweh.4020

This work is licensed under a Creative Commons Attribution

4.0 International License.

Emotional demands at work and risk of hospital-treated depressive disorder in up to 1.6

million Danish employees: a prospective nationwide register-based cohort study

by Ida EH Madsen, PhD,

Jeppe Karl Sørensen, MSc,

Julie Eskildsen Bruun, MSc,

Elisabeth Framke, PhD,

Hermann Burr,

PhD,

Maria Melchior, PhD,

Børge Sivertsen, PhD,

4, 5, 6

Stephen Stansfeld, PhD,

Mika Kivimäki, FMedSci,

8, 9, 10

Reiner Rugulies,

PhD

1, 11, 12

Madsen IEH, Sørensen JK, Bruun JE, Framke E, Burr H, Melchior M, Sivertsen B, Stansfeld S, Kivimäki M, Rugulies R. Emotional

demands at work and risk of hospital-treated depressive disorder in up to 1.6 million Danish employees: a prospective

nationwide register-based cohort study.

Scand J Work Eniviron Health

– online first. doi:10.5271/sjweh.4020

Objective

Previous studies on effects of emotional demands on depression have relied on self-reported exposure

data and lacked control for potential confounding by pre-employment risk factors for depression. This study

used a register-based design to examine the risk of hospital-treated depressive disorder in relation to occupa-

tional levels of emotional demands at work, furthermore addressing the role of risk factors for depression before

workforce entry.

Methods

We analyzed data from two Danish register-based cohorts – Job Exposure Matrix Analyses of Psycho-

social Factors and Healthy Ageing in Denmark (JEMPAD, N= 1 665 798) (17) and Danish Work Life Course

Cohort (DaWCo, N=939 411), which link assessments of emotional demands by job exposure matrices to records

of hospital-treated depressive disorder among employees aged 15–59 years at baseline (average follow up: 9.7

years in JEMPAD, 7.3 years in DaWCo). Potential confounders comprised sociodemographics, job control,

work-related violence and physical demands at work. In DaWCo, we followed individuals from their entry into

the workforce, and also included data on risk factors for depression before workforce entry (eg, parental income,

education, and psychiatric diagnoses).

Results

Employees in occupations with high emotional demands had an increased risk of hospital-treated depres-

sive disorder with confounder-adjusted hazard ratios of 1.32 [95% confidence interval (CI) 1.24‒1.41] and 1.19

(95% CI 1.09‒1.30) in JEMPAD and DaWCO, respectively. This association remained after controlling for risk

factors before workforce entry.

Conclusions

This study suggests that employees in occupations with high emotional demands are at increased

risk of hospital-treated depressive disorder. This increased risk was neither attributable to reporting bias nor

explained by the included risk factors for depression recorded before workforce entry.

Key terms

Denmark; depression; job exposure matrix; psychosocial; register-based study; stress.

1 National Research Centre for the Working Environment, Copenhagen, Denmark.

2 Unit Psychosocial factors and mental health, Federal Institute for Occupational Safety and Health, Berlin, Germany.

3 Sorbonne Université, Inserm, Institut Pierre-Louis d'Épidémiologie et de Santé Publique, IPLESP, Research Group in Social Epidemiology, Paris,

France.

4 Department of Health Promotion, Norwegian Institute of Public Health, Bergen, Norway.

5 Department of Research and Innovation, Helse Fonna HF, Haugesund, Norway.

6 Department of Mental Health, Norwegian University of Science and Technology (NTNU), Trondheim, Norway.

7 Centre for Psychiatry, Barts and the London School of Medicine and Dentistry, Queen Mary University of London, London, UK.

8 Department of Public Health, Clinicum, Faculty of Medicine, University of Helsinki, Helsinki, Finland.

9 Finnish Institute of Occupational Health, Helsinki, Finland.

10 Department of Epidemiology and Public Health, University College of London, London, UK.

11 Department of Psychology, University of Copenhagen, Denmark.

12 Section of Epidemiology, Department of Public Health, University of Copenhagen, Denmark.

Correspondence to: Ida E. H. Madsen, National Research Centre for the Working Environment, Lerso Parkalle 105, DK- 2100 Copenhagen,

Denmark. [E-mail: [email protected]]

Scand J Work Environ Health – online first

Emotional demands at work and risk of hospital-treated depressive disorder

Emotional demands at work, ie, aspects of work that

require sustained emotional effort (1), occur most fre-

quently in occupations where a central part of the work

tasks is the interaction with individuals from outside the

workplace. Particularly, employees in human service

occupations such as healthcare, education, or social

work report high emotional demands at work (2). Such

occupations place high demands on the employees to

empathize with the patients, students or clients they

work with, who may be in difficult, painful or otherwise

distressing situations (3–5). Over time, this emotional

work is thought to potentially lead to compassion fatigue

and increase the risk of depression (6).

Observational studies have linked high emotional

demands at work with an increased risk of depressive

disorder, ascertained by a psychiatric interview (7), hos-

pital-treatment (8), treatment with antidepressants (2),

or a self-administered rating scale (9). A recent review

and meta-analysis on psychosocial stressors at work and

risk of depressive disorders reported a pooled risk of

1.21 (95% CI 1.08–1.36) for emotional demands (10).

However, questions have been raised regarding the

causality of the association. One study (11) found that

emotional demands at work predicted depressive dis-

order when measured with items relating to the overall

subjective perception of emotional demands (eg, “Is

your work emotionally demanding?” but not when

examining task-specific emotional demands measured

with more factual and less subjective questions (eg, “Do

you have to care for the emotional needs of others?”).

This finding suggests that the association between emo-

tional demands and depressive disorder may be inflated

by the self-reported exposure measurement, as the

reported level of emotional demands may be influenced

by the affective state of the respondent (reporting bias),

a long standing methodological concern in psychosocial

work environment research (10, 12).

Another concern is that previous studies have not

accounted for potential confounding by other risk fac-

tors for depressive disorder, which may be unequally

distributed between employees with high compared to

low emotional demands due to the selection of individu-

als at risk of depressive disorder into occupations with

high emotional demands. Such selection may be related

to processes motivating young individuals to enter care

work professions such as higher levels of childhood

parentification (ie, excessively taking care of other fam-

ily members’ needs) (13, 14), eg, due to parental illness

or socioeconomic difficulties. This possible confound-

ing by pre-employment risk factors has been illustrated

by a study showing increased antidepressant treatment

in care work professionals years before entering their

profession (15) and another study showing higher levels

of emotional demands in employees who have suffered

from mental illness in childhood or adolescence (16).

Accordingly, the present study aimed to (i) examine

the longitudinal association between emotional demands

and depressive disorder using an exposure measurement

that is not prone to reporting bias and (ii) control this

association for depression risk factors present before

workforce entry. To this end, we analyzed data from

two Danish register-based cohorts: Job Exposure Matrix

Analyses of Psychosocial Factors and Healthy Ageing in

Denmark (JEMPAD) (17) and Danish Work Life Course

Cohort (DaWCo) (18) which contain measures of emo-

tional demands assessed by job exposure matrices for a

large number of participants, in addition to measures of

clinical diagnoses of depressive disorder through regis-

ters of psychiatric in- and outpatient hospital treatments.

Methods

Study design and populations

We chose to analyze data from the two Danish work-

force cohorts, JEMPAD and DaWCo, in parallel, as

they complement each other; while JEMPAD includes

a wider age range of individuals (aged 30–59 years at

baseline), DaWCo offers the possibility to control for

risk factors of depressive disorder before workforce

entry, by including data on parental socioeconomic

position as well as psychiatric and somatic diagnoses,

but includes only individuals aged 15–30 at baseline.

JEMPAD is a nationwide cohort with information on

employment, psychosocial factors at work, health, labor

market affiliation and socio-demographics. Details of

JEMPAD have been published elsewhere (17). Briefly,

JEMPAD included all employed individuals residing in

Denmark in 2000, 30–59 years old, and with complete

data on gender, age, and migration background, a total

of 1 680 214 individuals. Using the unique Danish civil

registration number, we linked these individuals to other

population-based registers that provided information

on socio-demographics, health services use, diagnoses

for in- and out-patient hospital treatment, and causes of

death. We excluded individuals with diagnosed depres-

sive disorder before or in the year of baseline (N=14

516), yielding an analytic cohort of 1 665 798 individu-

als followed for 16 113 287 person years (mean follow

up: 9.7 years).

DaWCo is an open inception cohort study of all

individuals who first entered the Danish workforce

during the years 1995–2009 and were 15–30 years old

at workforce entry. Details of DaWCo have been pub-

lished elsewhere (18). Briefly, DaWCo was constructed

using population-based Danish registers on employ-

ment, health, demographic and socioeconomic factors to

examine effects of working conditions on health. Work-

Scand J Work Environ Health – online first

Madsen et al

ing conditions were measured repeatedly throughout the

work life by applying annually updated job exposure

matrices. Workforce entry was defined as the first year

with employment as the main source of income (N=979

257). We excluded individuals with missing data on

gender and migration background (N=5 176), and indi-

viduals who died (N=71), emigrated (N=13 087), or

received disability pension (N=361) in their year of

entry, leaving 960 562 individuals in the cohort. To

study incident depressive disorder, we further excluded

individuals with diagnosed depressive disorder before

or in the year of workforce entry (N=4989). To avoid

overlap between the two cohorts, we further excluded 16

162 individuals from the DaWCo population who were

potentially included in the JEMPAD population, as they

entered the workforce during 1995–2000 at the age of

25–30 years. The analytic study population for DaWCo

consisted of 939 411 individuals, followed for 6 825 523

person years (mean follow up: 7.3 years).

Measurement of emotional demands

Emotional demands were measured using a job exposure

matrix (JEM) estimating the average gender- and age-

specific mean levels of emotional demands in occupa-

tions classified according to the Danish version of the

International standard classification of occupations

(DISCO-88) (19) [see supplementary material (www.

sjweh.fi/article/4020)

table S1 for details]. The JEM

was constructed based on survey data from the Dan-

ish Work Environment Cohort Study collected in 2000

(N=8583, response rate 75.0%) and 2005 (N= 12 413,

response rate 62.5%) (20). In the surveys, emotional

demands were measured using a 3-item scale (supple-

mentary table S2). We constructed the JEM from the

survey data using best linear unbiased predictor (BLUP)

estimates of the level of emotional demands as a func-

tion of a person’s DISCO-88 occupation, gender, age

and year of data-collection (2000 or 2005). The models

were based on data from 10 299 first-time respondents

to either DWECS 2000 or 2005. Occupations with <5

survey respondents were collapsed at a less detailed

level of the DISCO-coding with fewer digits resulting

in a JEM covering 246 occupations coded with 4 digits

(210 occupations), 3 digits (22 occupations), 2 digits

(12 occupations) or 1 digit (2 occupations). For further

details on the JEM construction, see (21).

Each individual was assigned an annual level of

emotional demands during 2000–2009 for JEMPAD and

1995–2009 for DaWCo, which was subsequently cate-

gorized by distribution quartiles in the JEMPAD popula-

tion. In years of non-employment (eg, unemployment or

studying), emotional demands were categorized as low.

Supplementary table S3 shows the occupations with the

highest and lowest levels of emotional demands.

Measurement of depressive disorder

Information on depressive disorder was obtained from

the Psychiatric Central Research Register (22) during

1969–1994 and the National Patient Register during

1995–2010 (23). These registers encompass all inpa-

tient psychiatric admissions in Denmark since 1969 and

from 1995 onwards also outpatient admissions (22). We

defined depressive disorder as a main diagnosis of F32

or F33 from ICD-10 (for 1994–2010), and 296.0, 296.2,

298.0, 300.4 from ICD-8 (for 1969–1993). ICD-9 was

never used in Denmark. To exclude individuals with

depressive disorder prior to workforce entry, we addi-

tionally used codes F92.0 (ICD-10) and 308.02 (ICD-8)

for depressive disorder in childhood or adolescence.

Measurement of potential confounders

For both cohorts we included information on calendar

year, gender, age, cohabitation, employment status,

migration background, income, number of health ser-

vices used, job control, risk of work-related violence,

and physical demands at work, and any psychiatric

diagnosis before study baseline (2000 for JEMPAD,

workforce entry for DaWCo). Furthermore, for DaWCo,

additional data were available, including information

on childhood socioeconomic position, maternal and

paternal psychiatric and somatic diagnoses before the

cohort member entered the workforce. The choice of

potential confounders was guided by existing literature

indicating an increased risk of depression in relation to

these factors (24–27) and considerations regarding their

relationship with emotional demands at work.

Supplementary table S1 summarizes the measure-

ment of potential confounders. All sociodemographic

covariates were derived from Danish National Registers

(22, 23, 28–31). For income, we used household income

in the JEMPAD population but personal income in the

younger DaWCo population, where many were likely

still living with their parents in their first years in the

cohort. Data on the number of health services used

were from the Danish National Health services register,

encompassing mainly primary health care services (31).

To avoid adjustment for an intermediate step in a causal

pathway, we included health services use data from the

year preceding the year for measurement of exposure.

Job control, risk of work-related violence, and physi-

cal demands at work were measured using JEM. Job

control and physical demands at work were categorized

by the distribution quartiles of the JEMPAD population.

Risk of work-related violence (yes/no) was dichoto-

mized by predicted risk of work-related violence of

≥2%. This cut-off point was based on the distribution of

the DaWCo population, where it distinguishes the upper

quartile from the three lower quartiles. The correlations

Scand J Work Environ Health – online first

Emotional demands at work and risk of hospital-treated depressive disorder

between the included working conditions are reported

in supplementary table S4.

To measure psychiatric diagnoses before study base-

line, we included diagnoses coded as chapter F (ICD-

10) or 290-315 (ICD-8) recorded as main or subsidiary

diagnoses during the baseline year or before.

Childhood socioeconomic position was measured in

DaWCo by maternal and paternal employment status,

education and income when the cohort member was

15 years. If information was missing, we included data

from preceding years back to the birth of the cohort

member and, if still missing, we included data up until

age 20. Maternal and paternal psychiatric and somatic

diagnoses were included from the Psychiatric Central

Research Register and National Patient Register includ-

ing both main and subsidiary diagnoses recorded before

the cohort member entered the workforce. Linkage to

parental data was available from 1980 onwards and

only for individuals with parents residing in Denmark.

To ensure cohort completeness, individuals with miss-

ing data were assigned to a separate category for those

variables and retained in the analyses. Table S5 shows

the distribution of individuals in relation to the included

pre-employment risk factors for depression.

Age, cohabitation, employment status, income, health

services use, and working conditions were included as

annual time-varying variables, while all the remaining

covariates were considered as time-invariant variables.

Statistical analysis

We analyzed data using Cox proportional hazard regres-

sion models with time-to-first diagnosis of depressive

disorder as the outcome. Fulfillment of the proportional

hazards assumption was assessed by visual inspection

of the log-log hazard plots. We used calendar time as

the time-axis to account for period effects on psychi-

atric treatment (22). We analyzed data longitudinally

with a one year time-lag, relating exposure during year

t to events during year t+1. Individuals were followed

from baseline (1 January 2000 in JEMPAD and date of

workforce entry in DaWCo) until first depressive dis-

order diagnosis, death, emigration, receipt of disability

pension, or end of follow-up (31 December 2010),

whichever came first. We terminated follow-up in the

year 2010 because exposures could not be updated after

2009 due to changes in the occupational classification.

In the main analysis, we adjusted for gender, age,

cohabitation, employment status, migration background,

income, health services use in the preceding year (t-1)

in model 1. Model 2 was additionally adjusted for job

control, work-related violence, and physical demands

at work. For DaWCo, we included a model 3 addition-

ally adjusting for the presence of a psychiatric disorder

before workforce entry (yes/no), childhood socioeco-

nomic position (maternal and paternal employment sta-

tus, maternal and paternal education, and maternal and

paternal income), and maternal and paternal psychiatric

and somatic diagnoses. The applied analytic framework

was based on considerations concerning the potential

causal effects and temporal order of the included vari-

ables and is illustrated in our Directed Acyclic Graph

in figure S1. We included other work-related factors in

model 2 as a separate step because the direction between

emotional demands and theses exposures is uncertain,

whereas there is more certainty regarding the role of the

variables included in model 1 as potential confounders.

As supplementary analyses, we examined the associa-

tion between emotional demands and depressive disorder

separately in men and women. We also examined the

association between emotional demands and depressive

disorder after excluding individuals with any diagnosed

psychiatric disorder before baseline (JEMPAD, N=43

686) or workforce entry (DaWCo, N=48 303). In a post

hoc analysis, we further investigated whether the unex-

pected direction of the association in model 1 for DaWCo

was related to the categorization of emotional demands,

by classifying the exposure according to the quartiles

of the DaWCo population, rather than quartiles of the

JEMPAD population. Finally, we conducted a quantita-

tive bias analysis to estimate the extent of bias caused

by non-differential misclassification of exposure by the

JEM for emotional demands. This analysis accounts for

the imperfect measurement of emotional demands due to

the application of a JEM, and assumes that the measure-

ment error is similar in high and low exposure groups and

for individuals who become cases and for those who do

not become cases. The analysis was conducted using the

methods for quantitative bias analysis proposed by Lash,

Fox, & Fink (32) using the spreadsheet developed for cor-

recting analyses for exposure misclassification available

online (https://sites.google.com/site/biasanalysis). We

applied a sensitivity of 0.53 and a specificity of 0.87 based

on data from the JEM construction. All statistical analyses

were conducted in SAS version 9.4 (SAS Institute, Cary,

NC, USA).

Results

Table 1 shows the characteristics of the study popula-

tions at baseline (2000) for JEMPAD and in the year

of workforce entry for DaWCo. Both populations were

gender balanced, with about 50% women. The mean age

was 44 years in JEMPAD and 20 years in DaWCo. The

members of the DaWCo cohort tended to be more likely

to work in occupations with low emotional demands,

low job control and high physical demands, compared

to the JEMPAD population.

Scand J Work Environ Health – online first

Madsen et al

Table 1.

Baseline characteristics of the study populations.

JEMPAD

DaWCo

% Mean

Total sample

1 665 798 100

939 411 100

Gender

Men

860 073 51.6

476 400 50.7

Women

805 725 48.4

463 011 49.3

Age (years)

43.8

20.1

15–17

103 410 11.0

18–19

379 211 40.4

20–24

378 299 40.3

25–30

78 491 8.4

30–35

360 971 21.7

36–45

587 774 35.3

46–55

550 352 33.0

56–60

166 701 10.0

Cohabitation

Yes, living with

1 238 902 74.4

606 184 64.5

partner or spouse

No, single

423 283 25.4

292 753 31.2

Unknown

3 613 0.2

40 474 4.3

Migration background

1 590 543 95.5

804 526 85.6

Yes

75 255 4.5

134 885 14.4

Education

Primary or lower

387 832 23.3

684 674 72.9

secondary

Upper secondary

764 149 45.9

167 502 17.8

Short cycle tertiary

82 041 4.9

3617 0.4

Bachelor or equivalent 290 729 17.5

11 285 1.2

Master or equivalent

110 171 6.6

3 819 0.4

Doctoral or equivalent

6 687 0.4

12 <0.1

Unknown

23 698 1.4

68 502 7.3

Annual disposable

320 591

101 087

income DKK

Health services (N)

15.3

10.6

Any psychiatric diagno-

43 686 2.6

48 303 5.1

sis before baseline

Emotional demands

Low

406 982 24.4

754 938 80.4

Medium - low

428 309 25.7

47 980 5.1

Medium - high

415 479 24.9

104 223 11.1

High

415 028 24.9

32 270 3.4

Job control

Low

421 253 25.6

722 697 76.9

Medium - low

445 990 26.8

121 778 13.0

Medium - high

383 293 23.0

75 575 8.0

High

415 262 24.9

19 361 2.1

Risk of work-related

violence

Low

1 276 370 76.6

751 761 80.0

High

389 428 23.4

187 650 20.0

Physical demands

Low

394 550 23.7

14 273 1.5

Medium - low

400 085 24.0

51 656 5.5

Medium - high

450 667 27.1

316 698 33.7

High

420 496 25.2

556 784 59.3

Baseline year for JEMPAD cohort members is 2000 and for DaWCo cohort

members their year of workforce entry.

We measured income (Danish kroner) as household income in JEMPAD and

personal income in DaWCo because many individuals in DaWCo shared

households with their parents in the baseline year.

There were 22 378 individuals who became cases

with depressive disorder in JEMPAD and 15 753 became

cases in DaWCo. Amongst the cases, the majority

(69.4% in JEMPAD, 74.4% in DaWCo) were diagnosed

with F32 “Depressive Episode” (data not shown). Table

2 shows the associations between emotional demands

and depressive disorder in the two populations under

study. In model 1, employees in occupations with the

highest level of emotional demands had a hazard ratio

(HR) of depressive disorder of 1.24 [95% confidence

interval (CI) 1.18‒1.30] in JEMPAD and 0.86 (95% CI

0.81‒0.92) in DaWCo. After adjusting for other work-

ing conditions (model 2), the association became more

pronounced in JEMPAD with a HR of 1.32 (1.24‒‒.41).

In DaWCo, adjusting for other working conditions

reversed the association from model 1 and high emo-

tional demands were now also associated with an

increased risk of depression with a HR of 1.19 (95% CI

1.09‒1.30). There was no evidence of a dose–response

association. In JEMPAD, the risk of depressive disorder

amongst employees in occupations with medium levels

of emotional demands was similar to the risk amongst

employees with low emotional demands and, in DaWCo,

the risk of depressive disorder was similar in employees

with high and medium–high emotional demands. When

further adjusting for risk factors for depressive disorder

that pre-existed workforce entry in DaWCo (model 3),

the results remained unchanged.

When analyzing men and women separately, we

found that the association between high emotional

demands and depressive disorder was largely similar

between the genders in both cohorts (table 3). After

excluding individuals with any psychiatric diagnosis

before workforce entry from the DaWCo cohort, the

associations became more pronounced with a HR of

1.29 (95% CI 1.14‒1.45) for employees in occupations

with high emotional demands (supplementary table S6).

In JEMPAD, results were similar to those from the main

analysis after the exclusion of individuals with any psy-

chiatric diagnosis before baseline (table S6).

In a post-hoc analysis, we explored the unexpected

finding that, in model 1 for DaWCo, high emotional

demands were associated with a decreased risk of depres-

sive disorder. When we changed the categorization of

emotional demands in DaWCo to reflect quartiles of

DaWCo, we found associations similar to those of our

main analyses with a HR for model 1 of 0.91 (95% CI

0.85–0.97) for employees in occupations with high emo-

tional demands, and HR of 0.87 (95% CI 0.81‒0.93) and

1.01 (95% CI 0.95–1.07) for employees in occupations

with medium–high and medium–low emotional demands,

respectively (data not shown). The quantitative bias

analysis showed that when we accounted for the misclas-

sification of exposure introduced by the JEM, the HR of

1.31 from JEMPAD increased to 1.90 (data not shown).

Discussion

Using data from two independent Danish register-based

cohorts, this study found that employees in occupa-

tions with high levels of emotional demands are at

Scand J Work Environ Health – online first

Emotional demands at work and risk of hospital-treated depressive disorder

Table 2.

Association between emotional demands and subsequent onset of depression in the JEMPAD and DaWCo cohorts. [CI=confidence interval;

HR=hazard ratio; PY=person years.]

JEMPAD

Emotional demands

Low (reference)

5 405 126

Medium - low

3 564 817

Medium - high

3 598 007

High

3 545 337

Cases

Cases per

10 000 PY

Model 1

95% CI

P-value

<0.001

0.96‒1.05

0.86‒0.95

1.18‒1.30

Model 2

95% CI

P-value

<0.001

1.01–1.12

0.96‒1.08

1.24‒1.41

Model 3

95% CI

P-value

8 878

4 292

3 819

5 389

1.00

0.90

1.24

1.00

1.07

1.02

1.32

DaWCo

Emotional demands

<0.001

Low (reference)

4 770 523 11 066

1.00

Medium - low

670 191 1326

0.81

0.76‒0.86

0.94

0.88‒1.01

0.95

0.89‒1.02

Medium - high

774 719 1898

1.00

0.95‒1.05

1.21

1.13‒1.30

1.20

1.12‒1.29

High

610 090 1463

0.86

0.81‒0.92

1.19

1.09‒1.30

1.17

1.07‒1.27

Adjusted for gender, age, migration background, cohabitation, employment status, health services use, and income.

Adjusted for model 1 + job control, work related violence and physical demands at work.

Adjusted for model 2 + psychiatric diagnosis before workforce entry, maternal and paternal employment status, education, income, and maternal and paternal

psychiatric and somatic diagnoses.

Table 3.

Association between emotional demands and subsequent onset of depression in the JEMPAD and DaWCo cohorts, for men and women

separately. [CI=confidence interval; HR=hazard ratio; PY=person years.]

JEMPAD

Men

Emotional demands

Low (reference)

Medium - low

Medium - high

High

Women

Emotional demands

Low (reference)

Medium - low

Medium - high

High

DaWCo

Men

Emotional demands

Low (reference)

Medium - low

Medium - high

High

Women

Emotional demands

Low (reference)

Medium - low

Medium - high

High

Cases

Cases per

10 000 PY

95% CI

P-value

3 813 939

2 074 823

1 478 224

914 289

1 591 187

1 489 995

2 119 784

2 631 048

5 455

2 314

1 246

1 200

3 423

1 978

2 573

4 189

1.00

1.06

0.97

1.35

1.00

1.04

1.01

1.28

<0.001

1.00–1.14

0.89–1.05

1.23–1.48

<0.001

0.95–1.14

0.92–1.12

1.15–1.42

<0.001

0.73–1.04

1.11–1.50

0.94–1.50

2 920 160

197 405

253 129

108 308

4 586

160

353

127

1.00

0.87

1.29

1.18

<0.001

1 850 363

6 480

1.00

472 785

1 166

0.97

0.89–1.04

521 590

1 545

1.16

1.07–1.24

501 782

1 336

1.20

1.09–1.34

We report fully adjusted estimates, ie, for JEMPAD associations are adjusted for: age, cohabitation, employment status, migration background, income, health ser-

vices use, job control, work related violence and physical demands at work.

We report fully adjusted estimates, ie, for DaWCo associations are adjusted for age, cohabitation, employment status, migration background, income, health servic-

es use, job control, work related violence and physical demands at work, psychiatric diagnosis before workforce entry, maternal and paternal employment status,

education, income, and maternal and paternal psychiatric and somatic diagnoses.

increased risk of depressive disorder. An association

was observed in the JEMPAD population in both mini-

mally and multivariably adjusted models, whereas the

association was masked before adjusting for other work-

ing conditions (low job control, work-related violence

and high physical demands at work) in the younger

DaWCo population. This difference in results between

the cohorts is likely explained by differing distribu-

tions of other working conditions. Members of DaWCo

were more likely than members of JEMPAD to work in

occupations with low control (76.9% versus 25.6% at

baseline). Previous analyses have shown that low con-

trol predicts depression in DaWCo (25), and we found

a positive correlation between emotional demands and

job control (table S4), ie, individuals in occupations

with higher levels of emotional demands on average

Scand J Work Environ Health – online first

Madsen et al

worked in occupations with higher levels of job control.

Consequently, in the model where we did not adjust the

association between emotional demands and depression

for the average higher levels of job control associated

with higher emotional demands, the association between

emotional demands and depression was masked. There

are also other possible differences between the cohorts

that could affect the potential effects of their working

conditions on mental health. DaWCo members were all

working during their year of cohort entry, but over time

a proportion became students for a period until returning

into employment (18). It is possible that the jobs held by

DaWCo members in their early years differ to the jobs of

JEMPAD members in terms of working part-time versus

full-time, or the younger DaWCo members could be

more likely to have a more short-term time perspective

on their current position (33).

The association between emotional demands and

depressive disorder was similar in men and women

and did not appear to be explained neither by con-

founding by sociodemographic factors such as gender,

age, income, nor by other working conditions such

as job control, violence or physical demands at work,

or by the included pre-employment risk factors for

depressive disorder included in the DaWCo population.

Furthermore, our quantitative bias analysis suggested

that the true association between emotional demands

and depression may be considerably stronger than that

shown in the present paper, due to the misclassification

of exposure introduced by the JEM. Consequently, the

results of the present paper could be considered conser-

vative estimates of the association between emotional

demands and depression. While we included a range of

confounders in our analyses – the choice of which was

guided by considerations regarding their association

to the examined outcome and potential relation to the

exposure – we did not include analyses of effect modifi-

cation. This type of analyses remains an important topic

for further research.

Comparison with previous studies

The present results add to previous studies linking

high emotional demands at work with an increased

risk of depressive disorder, ascertained by a psychiatric

interview (7), hospital-treatment (8), treatment with

antidepressants (2), or a self-administered rating scale

(9). However, previous findings have been mixed, when

comparing measures of emotional demands that are sub-

jective in nature such as employees’ perceived emotional

demands, to measures that are less subjective, such as

measures focusing on the content of work tasks, and thus

considered less prone to be affected by reporting bias

(11). A recent study comparing different assessments

of emotional demands in relation to risk of long-term

sickness absence found that both content-related and

perceived emotional demands were associated with an

increased risk (34). Furthermore, that study found that

while the association for perceived emotional demands

was attenuated with adjustment for baseline depressive

symptoms, the association for content related emotional

demands remained largely unchanged. Our study adds to

this evidence by demonstrating that emotional demands,

measured at the occupational level, are associated with

an increased risk of depressive disorder. Our findings

suggest that this association cannot be explained by

reporting bias or by the selection of individuals with

higher pre-employment risk of depression according to

the included pre-employment risk factors into occupa-

tions with high emotional demands.

Strengths and limitations

The strengths of this study include its register-based

design enabling this large-scale study with annual expo-

sure assessments of emotional demands and a clinical

measure for depressive disorder based on hospital treat-

ment data as the outcome. Furthermore, a substantial

strength of the study is that we repeated similar analyses

in two independent samples of the Danish working popu-

lation, ensuring the findings’ reproducibility and their

generalizability across all ages of the working population.

Our study also entails some limitations. We did not

measure emotional demands at the individual level, but

used a JEM. While this approach eliminated reporting

bias from the study, it had the drawback of possible

misclassification of exposure as individuals in occupa-

tions with high average levels of emotional demands

may not actually be exposed. This misclassification

may have led to an under-estimation of the association

between emotional demands and depressive disorder

(35), as also indicated by the quantitative bias analysis,

suggesting that the reported results are conservative

estimates of the association between emotional demands

and depression. On the other hand, it should be noted

that there is also the possibility of an overestimation of

the true association between emotional demands and

depressive disorder due to measurement bias caused

by be systematic group level differences in the report-

ing of emotional demands, for instance caused by a

higher prevalence of depression in occupations with

high emotional demands. The magnitude of such bias is

difficult to gauge, but its potential existence should be

kept in mind when interpreting the results. Furthermore,

emotional demands may be related to interactions with

clients and customers, or alternatively be related to

interactions with colleagues, and these sources may have

different mental health effects (36). In the present study,

we could not distinguish between the different sources

of emotional demands.

Scand J Work Environ Health – online first

Emotional demands at work and risk of hospital-treated depressive disorder

Our confounder control was limited to data avail-

able from the registers, and we could not control for

several risk factors for depression such as genetics, life

events, or childhood adversities. However, the adjust-

ment for maternal and paternal psychiatric diagnoses

may to some extent have accounted for genetic differ-

ences, and both life events and childhood adversities are

socioeconomically patterned (37, 38). Consequently, if

associations were substantially biased by lack of data

on these factors, we would expect associations to be

attenuated after adjusting for indicators of childhood

socioeconomic position. This was not the case.

Finally, we included only cases of depressive dis-

order that were hospital diagnosed and the observed

associations may differ from those with symptom-based

measures (39) as many cases of depressive disorder

are untreated or treated exclusively in primary care

(40). Results from a European study (40) showed that

amongst participants with mood disorder during the

past 12 months, there were 36.5% who had consulted

a formal health service for their mental health during

this period. About one third consulted only their GP.

Such cases – in addition to cases who consulted no

health service at all – would not be included in our

outcome measurement. If there are systematic differ-

ences between occupational groups in the likelihood of

receiving hospital treatment when depressed, this could

lead our study to either over- or underestimate the true

association between emotional demands and depressive

disorder. Given this limitation, it is important that the

results of our study are interpreted in light of other pre-

vious studies (eg, 9.) applying outcome measurements

that are not affected by treatment seeking behaviors.

Concluding remarks

This nation-wide register-based study suggests that

employees in occupations with high levels of emotional

demands are at increased risk of hospital-treated depres-

sive disorder. This increased risk was neither attribut-

able to reporting bias nor explained by the included risk

factors for depression recorded before workforce entry.

Conflict of interest

The authors declare no conflicts of interest.

Protection of research participants

This study complies with the Declaration of Helsinki.

No patients were involved. In Denmark, studies that are

based on questionnaire and register-data only do not

require approval from the National Committee on Health

Research Ethics. The Danish Data Protection Agency

approved this study through the joint notification of the

National Research Centre for the Working Environment

(#2015-57-0074). All data are stored in a protected

server environment hosted by Statistics Denmark.

References

de Jonge J, Dormann C. The DISC model: Demand-Induced

Strain Compensation mechanisms in job stress. In: Dollard

MF, Winefield HR, Winefield AH, editors. Occupational

Stress in the Service Professions. London: Taylor & Francis

Ltd; 2003. p. 43–74.

Madsen IE, Diderichsen F, Burr H, Rugulies R. Person-

related work and incident use of antidepressants: relations

and mediating factors from the Danish work environment

cohort study. Scand J Work Environ Health 2010

Nov;36(6):435–44.

https://doi.org/10.5271/sjweh.3049.

Kim H. Empathy in the Early Childhood Classroom:

Exploring Teachers’ Perceptions, Understanding and Practices.

Indiana University 2017.

Sinclair S, Raffin-Bouchal S, Venturato L, Mijovic-

Kondejewski J, Smith-MacDonald L. Compassion fatigue:

A meta-narrative review of the healthcare literature. Int

J Nurs Stud 2017 Apr;69:9–24.

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

ijnurstu.2017.01.003.

Eriksson K, Englander M. Empathy in Social Work. J Soc

Work Educ 2017;53(4):607–21.

https://doi.org/10.1080/104

37797.2017.1284629.

Showalter SE. Compassion fatigue: what is it? Why does

it matter? Recognizing the symptoms, acknowledging the

impact, developing the tools to prevent compassion fatigue,

and strengthen the professional already suffering from the

effects. Am J Hosp Palliat Care 2010 Jun;27(4):239–42.

https://doi.org/10.1177/1049909109354096.

Niedhammer I, Malard L, Chastang JF. Occupational factors

and subsequent major depressive and generalized anxiety

disorders in the prospective French national SIP study. BMC

Public Health 2015 Feb;15(1):200.

https://doi.org/10.1186/

s12889-015-1559-y.

Wieclaw J, Agerbo E, Mortensen PB, Burr H, Tüchsen

F, Bonde JP. Psychosocial working conditions and the

risk of depression and anxiety disorders in the Danish

workforce. BMC Public Health 2008 Aug;8:280.

https://doi.

Acknowledgements

This paper was funded by the Danish Working Envi-

ronment Research Fund (grant numbers 17-2014-03,

27-2017-03 and 10-2019-03) and NordForsk (grant

number 75021). The funders had no role in the collec-

tion, analysis, and interpretation of data; in the writing

of the report; or in the decision to submit the paper for

publication.

Scand J Work Environ Health – online first

Madsen et al

org/10.1186/1471-2458-8-280

Kim IH, Noh S, Muntaner C. Emotional demands and the

risks of depression among homecare workers in the USA.

Int Arch Occup Environ Health 2013 Aug;86(6):635–44.

https://doi.org/10.1007/s00420-012-0789-x.

ilo.org/public/english/bureau/stat/isco/isco88/publ4.htm

[Accessed: 09-04-2019].

20. Burr H, Bjorner JB, Kristensen TS, Tüchsen F, Bach E.

Trends in the Danish work environment in 1990-2000

and their associations with labor-force changes. Scand J

Work Environ Health 2003 Aug;29(4):270–9.

https://doi.

org/10.5271/sjweh.731.

21. Framke E, Sørensen JK, Alexanderson K, Farrants K,

Kivimäki M, Nyberg ST et al. Emotional demands at

work and risk of long-term sickness absence in 1·5 million

employees in Denmark: a prospective cohort study on effect

modifiers. Lancet Public Health 2021 Oct;6(10):e752–9.

https://doi.org/10.1016/S2468-2667(21)00185-7.

22. Mors O, Perto GP, Mortensen PB. The Danish

Psychiatric Central Research Register. Scand J

Public Health 2011 Jul;39(7 Suppl):54–7.

https://doi.

org/10.1177/1403494810395825.

23. Lynge E, Sandegaard JL, Rebolj M. The Danish National

Patient Register. Scand J Public Health 2011 Jul;39(7

Suppl):30–3.

https://doi.org/10.1177/1403494811401482.

24. Alonso J, Angermeyer MC, Bernert S, Bruffaerts R,

Brugha TS, Bryson H et al.; ESEMeD/MHEDEA 2000

Investigators, European Study of the Epidemiology of

Mental Disorders (ESEMeD) Project. Prevalence of mental

disorders in Europe: results from the European Study of the

Epidemiology of Mental Disorders (ESEMeD) project. Acta

Psychiatr Scand Suppl 2004;109(420):21–7.

25. Svane-Petersen AC, Holm A, Burr H, Framke E, Melchior

M, Rod NH et al. Psychosocial working conditions and

depressive disorder: disentangling effects of job control

from socioeconomic status using a life-course approach. Soc

Psychiatry Psychiatr Epidemiol 2020 Feb;55(2):217–28.

https://doi.org/10.1007/s00127-019-01769-9.

26. Madsen IE, Svane-Petersen AC, Holm A, Burr H, Framke

E, Melchior M et al. Work-related violence and depressive

disorder among 955,573 employees followed for 6.99

million person-years. The Danish Work Life Course Cohort

study: work-related violence and depression. J Affect

Disord 2021 Jun;288:136–44.

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

jad.2021.03.065.

27. Hannerz H, Holtermann A, Madsen IE. Musculoskeletal

pain as a predictor for depression in the general working

population of Denmark. Scand J Public Health 2021

Aug;49(6):589–97.

28. Pedersen CB. The Danish civil registration system. Scand

J Public Health 2011 Jul;39(7 Suppl):22–5.

https://doi.

org/10.1177/1403494810387965.

29. Baadsgaard M, Quitzau J. Danish registers on personal income

and transfer payments. Scand J Public Health 2011 Jul;39(7

Suppl):103–5.

https://doi.org/10.1177/1403494811405098.

30. Jensen VM, Rasmussen AW. Danish education registers.

Scand J Public Health 2011 Jul;39(7 Suppl):91–4.

https://

doi.org/10.1177/1403494810394715.

31. Andersen JS, Olivarius NF, Krasnik A. The Danish National

Health Service Register. Scand J Public Health 2011 Jul;39(7

Scand J Work Environ Health – online first

10. Mikkelsen S, Coggon D, Andersen JH, Casey P, Flachs

EM, Kolstad HA et al. Are depressive disorders caused by

psychosocial stressors at work? A systematic review with

metaanalysis. Eur J Epidemiol 2021 May;36(5):479–96.

https://doi.org/10.1007/s10654-021-00725-9.

11. Vammen MA, Mikkelsen S, Hansen AM, Bonde JP,

Grynderup MB, Kolstad H et al. Emotional Demands at

Work and the Risk of Clinical Depression: A Longitudinal

Study in the Danish Public Sector. J Occup Environ Med

2016 Oct;58(10):994–1001.

https://doi.org/10.1097/

JOM.0000000000000849.

12. Rugulies R. Studying the effect of the psychosocial

work environment on risk of ill-health: towards a more

comprehensive assessment of working conditions. Scand J

Work Environ Health 2012 May;38(3):187–91.

https://doi.

org/10.5271/sjweh.3296.

13. DiCaccavo A. Investigating individuals’ motivations

to become counselling psychologists: the influence

of early caretaking roles within the family. Psychol

Psychother 2002 Dec;75(Pt 4):463–72.

https://doi.

org/10.1348/147608302321151943.

14. Nikcević AV, Kramolisova-Advani J, Spada MM. Early

childhood experiences and current emotional distress: what

do they tell us about aspiring psychologists? J Psychol 2007

Jan;141(1):25–34.

https://doi.org/10.3200/JRLP.141.1.25-

34.

15. Madsen IE, Aust B, Burr H, Carneiro IG, Diderichsen F,

Rugulies R. Paid care work and depression: a longitudinal

study of antidepressant treatment in female eldercare

workers before and after entering their profession. Depress

Anxiety 2012 Jul;29(7):605–13.

https://doi.org/10.1002/

da.21940.

16. Witt K, Milner A, Chastang JF, LaMontagne AD,

Niedhammer I. Impact of lifetime compared to adolescent-

onset mental illness on psychosocial employment quality

in adulthood: analysis of a nationally representative French

cohort. Int Arch Occup Environ Health 2018 Oct;91(7):887–

900.

https://doi.org/10.1007/s00420-018-1331-6.

17. Rugulies R, Framke E, Sørensen JK, Svane-Petersen AC,

Alexanderson K, Bonde JP et al. Persistent and changing job

strain and risk of coronary heart disease. A population-based

cohort study of 1.6 million employees in Denmark. Scand J

Work Environ Health 2020 Sep;46(5):498–507.

https://doi.

org/10.5271/sjweh.3891.

18. Svane-Petersen AC, Framke E, Sørensen JK, Rugulies R,

Madsen IE. Cohort profile: the Danish Work Life Course

Cohort study (DaWCo). BMJ Open 2019 Nov;9(11):e029658.

https://doi.org/10.1136/bmjopen-2019-029658.

19. ISCO-88. Summary of major groups International Labour

Organization (ILO); 2004. Available from:

http://www.

Emotional demands at work and risk of hospital-treated depressive disorder

Suppl):34–7.

https://doi.org/10.1177/1403494810394718.

32. Lash TL, Fox MP, Fink AK. Applying Quantitative

Bias Analysis to Epidemiologic Data. Chapter 6:

Misclassification, pp 79-108. Springer Publishing Company,

Incorporated; 2009.

33. Nielsen ML, Dyreborg J, Kines P, Nielsen KJ, Rasmussen

K. Exploring and expanding the category of’young workers’

according to situated ways of doing risk and safety-a case

study in the retail industry. Nord J Working Life Stud

2013;3(3):219.

https://doi.org/10.19154/njwls.v3i3.3019.

34. Framke E, Sørensen JK, Nordentoft M, Johnsen NF,

Garde AH, Pedersen J et al. Perceived and content-

related emotional demands at work and risk of long-term

sickness absence in the Danish workforce: a cohort

study of 26 410 Danish employees. Occup Environ Med

2019 Dec;76(12):895–900.

https://doi.org/10.1136/

oemed-2019-106015.

35. Hoffmann S, Guihenneuc C, Ancelet S. A cautionary

comment on the generation of Berkson error in

epidemiological studies. Radiat Environ Biophys 2018

May;57(2):189–93.

https://doi.org/10.1007/s00411-018-

0737-6.

36. Duarte J, Berthelsen H, Owen M. Not All Emotional

Demands Are the Same: Emotional Demands from Clients’

or Co-Workers’ Relations Have Different Associations with

Well-Being in Service Workers. Int J Environ Res Public

Health 2020 Oct;17(21):7738.

https://doi.org/10.3390/

ijerph17217738.

37. Doom JR, Mason SM, Suglia SF, Clark CJ. Pathways

between childhood/adolescent adversity, adolescent

socioeconomic status, and long-term cardiovascular disease

risk in young adulthood. Soc Sci Med 2017 Sep;188:166–

75.

https://doi.org/10.1016/j.socscimed.2017.06.044.

38. Glasscock DJ, Andersen JH, Labriola M, Rasmussen K,

Hansen CD. Can negative life events and coping style

help explain socioeconomic differences in perceived stress

among adolescents? A cross-sectional study based on

the West Jutland cohort study. BMC Public Health 2013

Jun;13:532.

https://doi.org/10.1186/1471-2458-13-532.

39. Thielen K, Nygaard E, Andersen I, Rugulies R, Heinesen

E, Bech P et al. Misclassification and the use of register-

based indicators for depression. Acta Psychiatr Scand

2009 Apr;119(4):312–9.

https://doi.org/10.1111/j.1600-

0447.2008.01282.x.

40. Alonso J, Angermeyer MC, Bernert S, Bruffaerts R,

Brugha TS, Bryson H et al.; ESEMeD/MHEDEA 2000

Investigators, European Study of the Epidemiology of

Mental Disorders (ESEMeD) Project. Use of mental health

services in Europe: results from the European Study of the

Epidemiology of Mental Disorders (ESEMeD) project. Acta

Psychiatr Scand Suppl 2004;109(420 S420):47–54.

Received for publication: 21 November 2021

Scand J Work Environ Health – online first