Beskæftigelsesudvalget 2022-23 (2. samling)
BEU Alm.del Bilag 251
Offentligt
2748704_0001.png
Articles
Association of workplace violence and bullying with later
suicide risk: a multicohort study and meta-analysis of
published data
Linda L Magnusson Hanson, Jaana Pentti, Mads Nordentoft, Tianwei Xu, Reiner Rugulies, Ida E H Madsen, Paul Maurice Conway,
Hugo Westerlund, Jussi Vahtera, Jenni Ervasti, G David Batty, Mika Kivimäki
Summary
Background
Workplace offensive behaviours, such as violence and bullying, have been linked to psychological
symptoms, but their potential impact on suicide risk remains unclear. We aimed to assess the association of workplace
violence and bullying with the risk of death by suicide and suicide attempt in multiple cohort studies.
Methods
In this multicohort study, we used individual-participant data from three prospective studies: the Finnish
Public Sector study, the Swedish Work Environment Survey, and the Work Environment and Health in Denmark
study. Workplace violence and bullying were self-reported at baseline. Participants were followed up for suicide
attempt and death using linkage to national health records. We additionally searched the literature for published
prospective studies and pooled our effect estimates with those from published studies.
Findings
During 1 803 496 person-years at risk, we recorded 1103 suicide attempts or deaths in participants with data
on workplace violence (n=205 048); the corresponding numbers for participants with data on workplace
bullying (n=191 783) were 1144 suicide attempts or deaths in 1 960 796 person-years, which included data from
one identified published study. Workplace violence was associated with an increased risk of suicide after basic
adjustment for age, sex, educational level, and family situation (hazard ratio 1·34 [95% CI 1·15–1·56]) and full
adjustment (additional adjustment for job demands, job control, and baseline health problems, 1·25 [1·08–1·47]).
Where data on frequency were available, a stronger association was observed among people with frequent exposure to
violence (1·75 [1·27–2·42]) than occasional violence (1·27 [1·04–1·56]). Workplace bullying was also associated with
an increased suicide risk (1·32 [1·09–1·59]), but the association was attenuated after adjustment for baseline mental
health problems (1·16 [0·96–1·41]).
Interpretation
Observational data from three Nordic countries suggest that workplace violence is associated with an
increased suicide risk, highlighting the importance of effective prevention of violent behaviours at workplaces.
Funding
Swedish Research Council for Health, Working Life and Welfare, Academy of Finland, Finnish Work
Environment Fund, and Danish Working Environment Research Fund.
Copyright
© 2023 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY
4.0 license.
Lancet Public Health
2023;
8: e494–503
Stress Research Institute,
Stockholm University,
Stockholm, Sweden
(L L Magnusson Hanson PhD,
T Xu PhD,
Prof H Westerlund PhD);
Clinicum, Faculty of Medicine,
University of Helsinki, Helsinki,
Finland
(J Pentti MSc,
Prof M Kivimäki FMedSci);
Department of Public Health
(J Pentti, Prof J Vahtera PhD)
and
Centre for Population Health
Research, Turku University
Hospital
(J Pentti,
Prof J Vahtera),
University of
Turku, Turku, Finland; Finnish
Institute of Occupational
Health, Helsinki, Finland
(J Pentti, J Ervasti PhD,
Prof M Kivimäki);
National
Research Centre for the
Working Environment,
Copenhagen, Denmark
(M Nordentoft PhD,
Prof R Rugulies PhD,
I E H Madsen PhD);
Department
of Public Health, University of
Copenhagen, Copenhagen,
Denmark
(Prof R Rugulies,
P M Conway PhD);
Department
of Epidemiology and Public
Health
(Prof G D Batty DSc)
and
UCL Faculty of Brain Sciences
(Prof M Kivimäki),
University
College London, London, UK
Correspondence to:
Dr Linda L Magnusson Hanson,
Stress Research Institute,
Stockholm University,
Stockholm 104 05, Sweden
[email protected]
Introduction
Workplace offensive behaviours, such as violence
(ie, behaviours or threats thereof with the objective of
physical, psychological, sexual, or economic harm
1
) and
bullying (a repeated and enduring form of harassment),
are relatively common phenomena. Workplace violence
is prevalent, particularly in industries that involve contact
with patients or clients, such as service and health-care
industries.
2
Globally, the estimated 12-month prevalence
of workplace violence among health-care workers
is 62%.
2
Offensive behaviours are potentially serious stressors
that can have a marked effect on employee health and
wellbeing. Cohort studies suggest an increased risk of
emotional and psychosomatic symptoms, depression,
post-traumatic stress syndrome, diabetes, and cardio-
vascular disease that might partly explain elevated rates
www.thelancet.com/public-health
Vol 8 July 2023
of absenteeism in this group of employees.
3–8
Studies
have also linked employee sexual harassment with
increased suicide risk in women,
9
and workplace bullying
with suicidal ideation
10
and suicidal behaviour.
11
However,
few prospective studies have been published on
workplace violence.
Using data from three prospective cohort studies and
published research identified in a systematic search we
aimed to investigate the association of workplace
violence, workplace bullying, and the risk of subsequent
suicide attempts or death by suicide.
Methods
Study population
In this multicohort study, we used data from two cohort
studies from the individual-participant data meta-
analysis in working populations consortium: the Finnish
e494
 BEU, Alm.del - 2022-23 (2. samling) - Bilag 251: Orientering om nye forskningsresultater på arbejdsmiljøområdet om konsekvenser ved vold og mobning i arbejdet
2748704_0002.png
Articles
Research in context
Evidence before this study
Workplace offensive behaviours, such as violence and bullying,
are a source of emotional stress for employees and a risk factor
for depression and cardiovascular disease, but evidence on the
potential impact on self-harm and suicide is scarce.
We searched the literature (PROSPERO registration CRD
42022301704), Web of Science (including PubMed), Scopus,
PsycInfo, and Embase from database inception to
June 31, 2022, for prospective observational studies published
in English investigating the associations of workplace violence
and bullying with risk of suicide using the search terms
(“violence” OR “bullying”) AND (“work” OR “organization”)
AND (“suicide” OR “suicidal behaviors”). Our search yielded
more than 800 potentially relevant articles, of which only one
fulfilled Population, Exposure, Comparator, and Outcomes
criteria. In a pooled dataset of 98 330 participants from nine
Danish surveys, suicide risk was almost two times higher
among men exposed to workplace bullying than non-exposed
men. In women, suicide risk was imprecisely estimated and no
strong evidence of an association was identified.
No prospective evidence was available on the association of
workplace violence and suicide risk.
Added value of this study
Our multicohort study showed that the risk of suicide death or
attempt was 1·3 times higher in employees who reported
exposure to workplace violence than those who reported no
exposure. Pooled analyses on workplace bullying suggested the
risk of suicide death or attempt was 1·3 times higher among
individuals who reported exposure to workplace bullying than
those not exposed. However, the association attenuated after
controlling for prevalent mental health problems at baseline.
Implications of all the available evidence
In the present analyses of multiple studies, employees exposed
to workplace violence had an increased risk of suicide. These
findings are consistent with the few existing studies reporting
adverse effects of workplace offensive behaviours on other
health outcomes such as depression and cardiovascular disease.
Although the evidence base is modest in scale, it highlights the
importance of eliminating such behaviours from the workplace
for employee health and supports the establishment of
zero tolerance policies.
See
Online
for appendix
Public Sector Study (FPS)
12
and the Swedish Work
Environment Surveys (SWES;
13
appendix pp 1–2) and a
study from Denmark, the Work Environment and Health
in Denmark study (WEHD).
14
A detailed description of
the included studies is provided in the appendix (p 1).
This study followed the STROBE reporting guideline for
cohort studies.
FPS comprised 151 901 employees with a minimum
6-month employment contract in ten towns and
five hospital districts in Finland. Workplace violence and
bullying were assessed in selected postal surveys
conducted in 1997, 2000, 2004, 2008, and 2012 and
employees were followed up for suicide attempt or death
until Dec 31, 2018. SWES is a representative cross-
sectional survey of the Swedish workforce (individuals
aged 16–64 years), performed every 2 years since 1989.
Data for the present analyses were derived from
self-completion questionnaires completed between
1995 and 2013 and follow-up register data on suicide
attempt or death available until Dec 31, 2016. The eligible
population included 154 677 individuals. WEHD is based
on a random selection of individuals 18–64 years of age
in the national workforce of Denmark. Starting in 2012, a
series of biennial postal or web-based data collections
have been performed. For this study, the eligible
population included 133 924 individuals from the
2012, 2014, and 2016 surveys. Follow-up of suicide
attempts or deaths in the register data was completed
until Dec 31, 2016.
Questionnaire-based and register-based studies do
not require approval from central ethical committees,
according to Danish law; thus the requirement for ethical
approval was waived. Participants provided written
consent by filling out and returning the questionnaires.
Measurement of workplace violence, bullying, and
covariates
In individual participant data, respondents were
considered exposed to any workplace violence if they
reported that they had been targeted with violence or
threats of violence within the past 12 months or within
the current year (FPS). The wording varied slightly
between the cohorts (appendix p 3). To examine any
dose–response associations, we divided SWES parti-
cipants into three groups: individuals with no exposure,
individuals exposed less than once a week (occasionally
exposed), and individuals with weekly exposure
(frequently exposed).
Regarding workplace bullying, participants were asked
whether they were currently being bullied at work (FPS),
or had experienced bullying within the past 12 months
(SWES). We defined exposure to workplace bullying as
giving an affirmative response to the question (appendix
p 3). For the purpose of dose-response analyses, we
identified people with no exposure, occasional exposure
(less than weekly), or frequent exposure (weekly) to
bullying in SWES. However, since bullying data from
WEHD were already published, we restricted analyses
from WEHD to workplace violence only.
Sex, age, family situation (combination of cohabitation
status and presence of children in the household), and
educational level were considered potential confounding
www.thelancet.com/public-health
Vol 8 July 2023
e495
 BEU, Alm.del - 2022-23 (2. samling) - Bilag 251: Orientering om nye forskningsresultater på arbejdsmiljøområdet om konsekvenser ved vold og mobning i arbejdet
2748704_0003.png
Articles
factors.
5–7
In FPS, data were collected from employers
registers (age and sex), surveys (family situation), and
Statistics Finland (education). In SWES, sex, age, family
situation, and educational level were collected from
linked administrative records from Statistics Sweden.
The corresponding WEHD data were collected from
administrative records from Statistics Denmark.
Educational level was classified into three categories:
9 years or younger, 10–12 years, and 13 years or older.
Baseline family situation was categorised as: single,
divorced, separated, or widowed without children;
single, divorced, separated, or widowed with children;
married or living with partner without children; and
married or living with partner with children. Work
characteristics included high job demands and low
control, measured as in previous studies (appendix
pp 4–5).
15,16
Mental health was denoted by history of
physician-diagnosed depression (FPS), baseline reports
of being tired or listless every day
17
(in SWES), and self-
reported treatment for depression in the previous year
(WEHD), or previous mental disorders as indicated by a
diagnosis of International Classification of Diseases
(ICD; version 8–10) F01–F99/290–319 codes in
administrative registers (all cohorts). Somatic health
problems were considered by assessing the presence
or history of any diagnosis within the Charlson's
comorbidity index in the administrative registers.
18,19
the same health registries as suicides, using the
ICD (version 8–10) C00–D49/140–239 codes, including
both benign and malignant cancer tumours. An equally
strong association of workplace violence or bullying with
suicide death or attempt and tumours or neoplasms
was considered informative of potential underlying
confounding or bias, whereas a robust association with
suicide death or attempt combined with no or weak
association with tumours or neoplasms was assumed to
strengthen an observation of an excess risk of suicide
among individuals exposed to workplace violence or
bullying.
For more on
Statistics Finland
see https://www.stat.fi/index_
en.html
For more on
Statistics Sweden
see https://www.scb.se/en/
For more on
Statistics Denmark
see https://www.dst.dk/en
Systematic review and assessment
We conducted a systematic search of the literature
following the PRISMA guidelines and the protocol
outlined in PROSPERO (CRD42022301704). We searched
Web of Science (including Medline), Scopus, PsycInfo,
and Embase from database inception to June 31, 2022,
using the search terms “violence”, “bullying”, “work”,
“organization”, “suicide”, and “suicidal behaviors” for
observational studies with a prospective or longitudinal
design. Two authors (LLMH and TX) independently
determined whether the records fulfilled the following
Population, Exposure, Comparator, and Outcomes
(PECO) criteria: (1) included individuals of working age
(15–65 years); (2) reported results regarding associations
between work-related violence or bullying and suicidal
behaviour; (3) used a relevant comparison group
(eg, compared exposed with unexposed or cases with
controls or with the general population); and (4) studied
any type of suicidal behaviour as outcome, defined as
“behaviours that may result in ending of one life, whether
fatal or not”,
23
including suicide attempt and suicide
death. We also screened previous reviews on workplace
bullying and related topics (eg, work stressors) and
suicidality for relevant publications not identified by the
search of original articles. We assessed risk of bias using
ROBINS-E, a tool for non-randomised studies of
exposure, assessing risk of bias due to confounding, in
selection of participants into the study, classification of
exposure, due to missing data, in measurement of
outcomes, and in selection of the reported results. The
overall risk of bias was regarded as high if risk of bias
was considered high in at least one of the domains
(appendix pp 7–8).
Follow-up of suicide death and attempt
Suicide events were identified from hospital discharge
registers (FPS) and causes of death registers (FPS,
SWES, and WEHD), patient registers (including both
inpatient and outpatient data since 2001; SWES), and
the Danish Psychiatric Central Research Register
(WEHD). Events recorded with ICD (versions 8–10)
X60–X84 or E950–959 (self-inflicted harm) or Y10–Y34
or E980–989 (death with undetermined intent) codes in
the Cause of Death register were defined as suicide
death cases, while individuals with a record of self-
inflicted harm or harm with undetermined intent in the
hospital discharge, patient, or psychiatric central
research registries (appendix p 6) were considered cases
of suicide attempt. We identified the first incident
suicide attempt or suicide death occurring after
response to the surveys. The main outcome variable
included both suicide death and suicide attempt.
Separate outcome variables for suicide death and suicide
attempt were also used for two of the studies (SWES
and FPS) with a larger number of events.
Statistical analysis
In analyses of the individual participant data, we first
calculated estimates separately in each cohort. We then
combined study-specific estimates using meta-analysis
(inverse variance method). Considering the small
number of studies, fixed-effect rather than random-effect
meta-analysis was used to pool study-specific esti-
mates.
24,25
The respondents were followed from the date
or year of response to questionnaires to the time of either
first registered suicide attempt or suicide death, death
e496
Assessment of a negative control outcome
We selected incident tumours or neoplasms at follow-up
as a negative control outcome because no or only weak
association (relative risk <1·2) has been observed
between work-related stressors and this endpoint in
meta-analyses and large-scale studies.
20–22
Tumours and
neoplasms were identified from linked records from
www.thelancet.com/public-health
Vol 8 July 2023
 BEU, Alm.del - 2022-23 (2. samling) - Bilag 251: Orientering om nye forskningsresultater på arbejdsmiljøområdet om konsekvenser ved vold og mobning i arbejdet
2748704_0004.png
Articles
from another cause, or end of follow-up, whichever came
first. Since there was no deviation from the proportional
hazard’s assumption (tested using log–log plot and
interaction between time and exposure), we used Cox
proportional hazard regression with age as the underlying
time scale to estimate the risk of suicide death or suicide
attempt. We fitted separate models for exposure to
workplace violence and workplace bullying versus no
exposure in relation to suicide death or attempt
combined, and repeated these analyses for suicide death
and suicide attempt separately. The analyses were
initially adjusted for sex, age, education, and family
situation (basic model), with further control for baseline
job demands and control and prior, baseline mental and
somatic health problems (fully adjusted model).
To examine the robustness of the associations,
we repeated analyses in subgroups according to sex, age,
and educational level,
23,26
categorising age (<40 or ≥40 years)
and educational level (≤12 years or >12 years of education)
A
into two groups to ascertain sufficient numbers in each
group. Due to small case numbers, it was not possible to
do these analyses for WEHD, and thus the analyses were
only done for workplace violence based on FPS and
SWES. Additionally, we did supplementary analyses only
among people without mental health problems at
baseline. We tested for trends across the categories of
violence and bullying (never, occasional, frequent) in
SWES by treating the variable as a continuous predictor
of suicide death or attempt. To evaluate validity using the
negative control outcome, we analysed the association
between workplace violence and tumours or neoplasms
using similar methods as for suicide death or attempt.
We did a fixed-effect meta-analysis to pool study-
specific estimates and published findings identified in
our systematic review. We also did subgroup analyses
according to sex, work sector (mixed occupations
vs
social and health-care workers, based on study
population), and outcome. Due to the small number of
B
Data by cohort study
SWES
154 677 eligible for workplace
violence and bullying
analyses
WEHD
133 924 eligible participants
(workplace violence only)
Systematic review
of published studies
FPS
83 458 eligible participants for
workplace violence analysis
14 896 eligible participants for
workplace bullying analysis
1538 potentially eligible studies
identified through
database search
(June 31, 2022)
19 821 did not respond, had
previously attempted
suicide, or had no exposure
data (workplace violence)
4994 did not respond, had
previously attempted
suicide, or had no exposure
data (workplace bullying)
70 864 did not respond, had
previously attempted
suicide, or had no exposure
data (workplace violence)
70 903 did not respond, had
previously attempted
suicide, or had no exposure
data (workplace bullying)
74 181 did not respond, had
previously attempted
suicide, or had no
exposure data (workplace
violence)
706 duplicates excluded
63 637 participants with exposure
data (workplace violence)
9902 participants with exposure
data (workplace bullying)
83 813 participants with exposure
data (workplace violence)
83 774 participants with exposure
data (workplace bullying)
59 793 participants with
exposure data (workplace
violence)
832 eligible studies identified
through title and abstract
screenng
1060 missing covariate data
(workplace violence)
134 missing covariate data
(workplace bullying)
89 missing covariate data
(workplace violence)
89 missing covariate data
(workplace bullying)
1046 missing covariate data
804 excluded
Final FPS sample
62 577 (workplace violence)
9768 (workplace bullying)
Final SWES sample
83 724 (workplace violence)
83 685 (workplace bullying)
Final WEHD sample
58 747 (workplace violence)
28 full-text studies assessed for
eligibility
27 excluded
205 048 included in analysis of
new data (workplace
violence)
93 453 included in analysis of
new data (workplace
bullying)
1 study included
(workplace bullying;
98 330 participants)
Figure 1:
Flow chart for sample selection by cohort and for previously published data
FPS=Finnish Public Sector Study. SWES=Swedish Work Environment Surveys. WEHD=Work Environment and Health in Denmark study.
e497
www.thelancet.com/public-health
Vol 8 July 2023
 BEU, Alm.del - 2022-23 (2. samling) - Bilag 251: Orientering om nye forskningsresultater på arbejdsmiljøområdet om konsekvenser ved vold og mobning i arbejdet
2748704_0005.png
Articles
studies, heterogeneity was not assessed. We used SAS
(version 9.4) and Stata (version 16.0) to analyse study-
specific associations of workplace violence and bullying
with suicide outcomes, and R (version 4.1.1) was used
to meta-analytically aggregate study-specific effect
estimates.
Results
205 048 participants had available data on exposure,
outcome, and covariates for workplace violence and were
included in the analysis (62 577 from FPS, 83 724 from
SWES, and 58 747 from WEHD) and 93 453 participants
had available data on exposure, outcome, and covariates
for workplace bullying and were included in the analysis
(9768 from FPS and 83 685 from SWES; figure 1). Our
systematic review identified one study on workplace
bullying,
11
which included 98 330 participants; thus
191 780 individuals were included in the bullying
SWES
Workplace violence Workplace bullying
(n=83 724)
(n=83 685)
40 174 (48·0%)
43 550 (52·0%)
43·1 (11·9)
39 066 (46%)
15 157 (18·1%)
6408 (7·7%)
23 093 (27·6%)
40 138 (48·0%)
43 547 (52·0%)
43·1 (11·9)
39 045 (46·7%)
15 131 (18·1%)
6411 (7·7%)
23 098 (27·6%)
27 617 (47·0%)
31 130 (53·0%)
46·2 (11·3)
2528 (4·3%)
10 438 (17·8%)
25 068 (42·7%)
20 713 (35·3%)
WEHD workplace
violence (n=58 747)
Role of the funding source
The funders of the study had no role in study design, data
collection, data analysis, data interpretation, or writing of
the report.
FPS
Workplace violence Workplace bullying
(n=9768)
(n=62 577)
Sex
Men
Women
Age, years
Family situation
Married or living with partner with
children
Married or living with partner without
children
Single, divorced, separated, or widowed
with children
Single, divorced, separated, or widowed
without children
Education, years
≤9
10–12
≥13
Mental health problems
No
Yes
Somatic health problems*
No
Yes
Job demands
Low
High
Job control
Low
High
Workplace violence
No
Yes
Workplace bullying
No
Yes
··
··
9104 (93·2%)
664 (6·8%)
44 773 (71·5%)
17 804 (28·5%)
··
··
28 353 (45·3%)
34 224 (54·7%)
5168 (52·9%)
4600 (47·1%)
32 380 (51·7%)
30 197 (48·3%)
4794 (49·1%)
4974 (50·9%)
56 224 (89·8%)
6353 (10·2%)
8947 (91·6%)
821 (8·4%)
53 646 (85·7%)
8931 (14·3%)
8943 (91·6%)
825 (8·4%)
5053 (8·1%)
21 921 (35·0%)
35 603 (56·9%)
688 (6·8%)
2606 (26·7%)
6474 (66·3%)
24 408 (39·0%)
22 929 (36·6%)
4194 (6·7%)
11 046 (17·7%)
4694 (48·1%)
2811 (28·8%)
664 (6·8%)
1599 (16·4%)
13 141 (21·0%)
49 436 (79·0%)
45·5 (9·8)
1202 (12·3%)
8566 (87·7%)
42·8 (9·6)
12 253 (14·6%)
40 864 (48·8%)
30 607 (36·6%)
76 408 (91·3%)
7316 (8·7%)
76 584 (91·5%)
7140 (8·5%)
58 168 (69·5%)
25 556 (30·5%)
38 361 (45·8%)
45363 (54·2%)
72369 (86·4%)
11 355 (13·6%)
··
··
12 241 (14·6%)
40 868 (48·8%)
30 576 (36·5%)
76 368 (91·3%)
7317 (8·7%)
76 554 (91·5%)
7131 (8·5%)
58 130 (69·5%)
25 555 (30·5%)
38 384 (45·9%)
45 301 (54·1%)
··
··
76 575 (91·5%)
7110 (8·5%)
7827 (13·3%)
25 780 (43·9%)
25 140 (42·8%)
51 566 (87·8%)
7181 (12·2%)
24 525 (41·7%)
34 222 (58·3%)
33 144 (56·4%)
25 603 (43·6%)
14 087 (24·0%)
44 660 (76·0%)
52 318 (89·1%)
6429 (10·9%)
··
··
Data are n (%) or mean (SD). FPS=Finnish Public Sector Study. SWES=Swedish Work Environment Surveys. WEHD=Work Environment and Health in Denmark study.
*Presence or history of any diagnosis within the Charlson’s comorbidity index including myocardial infarction, congestive heart failure, peripheral vascular disease,
cerebrovascular disease, pulmonary diseases, rheumatic disease, dementia, hemiplegia, diabetes, chronic kidney disease, liver disease, peptic ulcer disease, cancer,
and HIV/AIDS.
Table:
Characteristics of participants from the FPS, SWES, and WEHD cohort studies
www.thelancet.com/public-health
Vol 8 July 2023
e498
 BEU, Alm.del - 2022-23 (2. samling) - Bilag 251: Orientering om nye forskningsresultater på arbejdsmiljøområdet om konsekvenser ved vold og mobning i arbejdet
2748704_0006.png
Articles
analyses. The study populations included 124 116 (61%)
women and 80 932 (39%) men for samples of workplace
violence, and 52 113 (56%) women and 41 340 (44%) men
for analyses of workplace bullying with mean age
ranging from 43 to 46 years (table). A higher proportion
of employees included in FPS (17 804 [28%] of
62 577 participants) had been exposed to workplace
violence than in SWES (11 355 [14%] of 83 724 participants)
and WEHD (6429 [11%] of 58 747 participants; table). The
prevalence of workplace bullying was 7–8% in FPS and
SWES. The risk of bias in these studies was rated as high
because of concerns about residual confounding
(appendix p 8).
For the analysis of workplace violence, the mean follow-
up was 9·1 years (SD 3·3) for FPS, 12·8 years (SD 5·5)
for SWES, and 2·7 years (1·7) for WEHD. 1103 suicide
deaths or attempts were recorded during 1 803 496 person-
years at risk (figure 2; appendix pp 9–11). After adjustment
for age, sex, education, and family situation, the risk of
suicide attempt or death was 1·3 times higher in
participants exposed to workplace violence than those
who were unexposed (hazard ratio [HR] 1·34 [95% CI
1·15–1·56]) and the effect estimates were directionally
N (total)
All participants
FPS cohort
SWES cohort
WEHD cohort
Adjustments
Unadjusted (age as time scale)
Sex, education, and family situation
Work characteristics
Somatic health problems
Mental health problems
Subgroup analyses
Sex*
Men
Women
Age, years*
<40
≥40
Level of education*
Low
High
Exposure to violence†
Never
Occasional
Frequent
Outcome*
Suicide attempt
Suicide death
146 301
146 301
29 159
29 159
953
163
72 369
9197
2158
72 369
9197
2158
744
111
39
80 091
66 210
14 036
15 123
785
300
50 681
95 620
10 950
18 209
456
629
55 315
92 986
7125
22 034
524
561
205 048
205 048
205 048
205 048
205 048
35 588
35 588
35 588
35 588
35 588
1103
1103
1103
1103
1103
205 048
62 577
83 724
58 747
n (exposed)
35 588
17 804
11 355
6429
n (cases)
1103
191
894
18
consistent across the three cohorts and across outcomes
in the two cohorts with data on both suicide attempts (1·26
[1·06–1·49]) and suicide deaths (1·76 [1·21–2·54]). The
association was also robust to additional adjustment for
work characteristics (1·33 [1·15–1·56]) and for mental
and somatic health problems (1·25 [1·08–1·47]). In
SWES, frequent exposure to workplace violence (weekly)
was associated with a higher risk of suicide death or
attempt (1·75 [1·27–2·42]) than occasional exposure (less
than weekly; 1·27 [1·04–1·56]; trend across frequency
categories, p<0·0001 for suicide attempts and deaths).
This dose-response relation was observed for suicide
deaths and suicide attempts (appendix p 12). Subgroup
analyses showed that the association between workplace
violence and suicide was stronger in men (1·65
[1·30–2·10]) than women (1·14 [0·93–1·40]; p=0·02),
whereas no difference was observed between
younger (1·41 [1·17–1·70]) and older (1·24 [1·01–1·52])
participants (p=0·16), or by educational level (high 1·10
[0·82–1·47]); low 1·41 [1·17–1·70]; p=0·16).
Analyses of the negative control outcome showed no
association between workplace violence and risk of
tumours or neoplasms (HR adjusted for age, sex,
HR (95% CI)
1·34 (1·15–1·56)
1·15 (0·85–1·56)
1·37 (1·15–1·64)
3·28 (1·14–9·45)
1·28 (1·10–1·48)
1·34 (1·15–1·56)
1·33 (1·15–1·56)
1·26 (1·08–1·47)
1·25 (1·08–1·47)
p value
0·0002
0·3732
0·0005
0·0275
0·0012
0·0000
0·0003
0·0033
0·0046
1·65 (1·30–2·10)
1·14 (0·93–1·40)
1·41 (1·17–1·70)
1·24 (1·01–1·52)
1·41 (1·17–1·70)
1·10 (0·82–1·47)
1·00 (ref)
1·27 (1·04–1·56)
1·75 (1·27–2·42)
1·26 (1·06–1·49)
1·76 (1·21–2·54)
0·50
1·00
5·00
0·0000
0·2109
0·0037
0·0388
0·0003
0·5329
··
0·018
0·001
0·0078
0·0029
Reduced risk Increased risk
Figure 2:
Association of workplace violence with risk of suicide attempt or death in three cohort studies
HRs and 95% CIs pooled from FPS, SWES, and WEHD and adjusted for age, sex, education, and family situation, unless otherwise stated. HR=hazard ratio.
FPS=Finnish Public Sector Study. SWES=Swedish Work Environment Surveys. WEHD=Work Environment and Health in Denmark study. *Data pooled from FPS and
SWES, therefore the number of individuals in these categories does not sum to the total for all participants. †Data from SWES.
e499
www.thelancet.com/public-health
Vol 8 July 2023
 BEU, Alm.del - 2022-23 (2. samling) - Bilag 251: Orientering om nye forskningsresultater på arbejdsmiljøområdet om konsekvenser ved vold og mobning i arbejdet
2748704_0007.png
Articles
education, and family situation, 1·02 [95% CI 0·98–1·05];
appendix p 12).
Analyses on workplace bullying were based on
960 attempted suicides or deaths by suicide in
1 246 998 person-years at risk among 93 453 participants
(mean follow-up 17·6 years [SD 1·8] for FPS, 12·8 years
[5·6] for SWES). The basic adjusted HR was 1·22
(95% CI 0·99–1·51), with little change after further
adjustment for work characteristics and somatic
diseases (figure 3). However, additional adjustments for
mental health problems at baseline attenuated this
estimate to 1·07 (0·86–1·33). A supplementary analysis
excluding participants with mental health problems
at baseline confirmed the null finding. In this sub-
population, the basic adjusted HR was 0·87 (0·21–3·60)
in FPS and 1·06 (0·81–1·38) in SWES. No association
was identified between workplace bullying and tumours
or neoplasms: the pooled HRs varied between 1·02 and
1·03, depending on which factors were adjusted for
(appendix p 13).
In the systematic review, we identified 832 articles,
but only one fulfilled the PECO criteria (figure 1A).
The included study from Denmark consisted of
98 330 participants (62 582 [63·6%] women and 35 748
[36·4%] men; mean age 44·5 years [SD 11·2]) of whom
10 259 (10·4%) reported workplace bullying at baseline.
The study population consisted of participants from
several Danish surveys,
11
and WEHD, in which bullying
N (total)
All participants
FPS cohort
SWES cohort
Adjustments
Unadjusted (age as time scale)
Sex, education, family situation
Work characteristics
Somatic health problems
Mental health problems
Subgroup analyses
Sex
Men
Women
Age, years
<40
≥40
Level of education
Low
High
Outcome
Suicide attempt
Suicide death
93 453
93 453
7774
7774
850
138
0·50
56 403
37 050
4570
3204
697
263
36 286
57 167
2741
5033
437
523
41 340
52 113
3466
4308
473
487
93 453
93 453
93 453
93 453
93 453
7774
7774
7774
7774
7774
960
960
960
960
960
93 453
9768
83 685
n (exposed)
7774
664
7110
n (cases)
960
64
896
was measured between 2004 and 2014 depending on
cohort and collection wave, and suicide follow-up was
obtained from linked electronic health records. The
risk of bias in this study was judged as high because of
concerns of unmeasured confounding (appendix p 8).
No published studies were identified on workplace
violence and suicide.
Figure 4 shows results from analyses combining the
findings from two cohorts with those of the published
study. The combined analysis included data for
191 783 participants among whom 1144 suicides or
suicide attempts were recorded during 1 960 796 person-
years at risk. In a model with basic adjustments, the
pooled HR for workplace bullying was 1·32 (1·09–1·59).
No major differences in these estimates were identified
between men and women, cohorts of mixed occupations
versus social and health workers, or in relation to
suicide death or suicide attempts. Adjustment for
mental health attenuated the findings (1·16 [0·96–1·41];
appendix p 14).
Discussion
This analysis of prospective population-based cohort
studies from Finland, Sweden, and Denmark found a
1·3 times higher risk of suicide death or attempt among
employees exposed to workplace violence and a
potentially increased risk in those exposed to workplace
bullying when compared with unexposed individuals.
HR (95% CI)
1·22 (0·99–1·51)
2·05 (0·97–4·32)
1·17 (0·94–1·46)
1·23 (0·99–1·52)
1·22 (0·99–1·51)
1·20 (0·99–1·51)
1·20 (0·97–1·49)
1·07 (0·86–1·33)
p value
0·0644
0·0592
0·1628
0·0580
0·0644
0·0902
0·0956
0·5543
1·21 (0·90–1·64)
1·24 (0·92–1·56)
1·06 (0·75–1·50)
1·33 (1·01–1·76)
1·13 (0·87–1·46)
1·43 (0·99–2·07)
1·21 (0·96–1·52)
1·59 (0·97–2·62)
1·00
5·00
0·2148
0·1102
0·7547
0·0438
0·3607
0·0569
0·1037
0·0669
Reduced risk Increased risk
Figure 3:
Association of workplace bullying with risk of suicide attempt or death in two cohort studies
HRs and 95% CIs pooled from FPS and SWES and adjusted for age, sex, and family situation, unless otherwise stated. HR=hazard ratio. FPS=Finnish Public Sector
Study. SWES=Swedish Work Environment Surveys.
www.thelancet.com/public-health
Vol 8 July 2023
e500
 BEU, Alm.del - 2022-23 (2. samling) - Bilag 251: Orientering om nye forskningsresultater på arbejdsmiljøområdet om konsekvenser ved vold og mobning i arbejdet
2748704_0008.png
Articles
The excess risk associated with workplace violence
followed a dose-response pattern and was not attributable
to differences in age, sex, education, family situation,
other work characteristics, or mental or physical health
between exposed and unexposed employees. The
association was observed in subgroups by age and
education, but sex-specific analyses indicate that risk was
greater among men than women. Although we observed
variation in cohort-specific effect estimates, the findings
were directionally consistent such that the HRs indicated
increased risk in all three Nordic countries. In pooled
analyses, including published research, workplace
bullying was associated with an increased risk of suicide
attempt and death by suicide. However, the excess suicide
risk associated with workplace bullying attenuated after
adjustment for mental health at baseline, suggesting that
this association was not robust between workplace
violence and suicide.
We are not aware of previous large-scale studies on
workplace violence or bullying in relation to suicide and
identified only one previously published paper on
bullying. Thus, the present study fills an important
knowledge gap. A 2018 narrative overview found
associations of low socioeconomic position, social
isolation, low scores on tests of intelligence, and mental
health problems with elevated suicide rates, but found
a paucity of studies on psychosocial stress.
27
A
2017 systematic review suggested an association between
workplace bullying and suicidal ideation,
10
but that review
identified only one study on workplace bullying and
suicide attempts, which was a case-control investigation
based on 69 participants,
28
and no studies on death by
suicide. A 2022 Danish study, which was included in our
meta-analysis, found that exposure to workplace bullying
was associated with an elevated risk of suicidal behaviour
N (total)
All participants
Current study
Conway et al (2022)
Subgroup analyses
Sex
Men
Women
Work sector†
Mixed occupations
Social and health-care workers
Outcome
Suicide attempt
Suicide death
191 783
191 783
18 033
18 033
999
177
182 015
9768
17 369
664
1080
64
77 088
114 695
NA*
NA*
536
608
11
among men, but not women.
11
However, the analysis did
not account for potential confounding by family situation
and work stressors, such as high work demands or low
job control, which might also affect suicide risk.
29,30
Our
analysis, which controlled for a wider set of potential
confounders, suggests that part of the observed excess
suicide risk associated with workplace bullying might be
attributable to prevalent mental health problems that
could confound the association.
Adjustment for baseline mental health might be
relevant since individuals with mental health problems
might be more likely to be bullied or perceive bullying
and be at increased risk of suicide. Similarly, other
stressful working conditions could drive the association
between workplace bullying and suicide.
29
Adjustments
for work characteristics could therefore reduce the risk
that the association between workplace bullying and
suicide is explained by reverse causation or health
selection.
10
Both mental health and work characteristics
could also act as mediators of the relationship between
workplace bullying and suicide and controlling for these
factors could be considered an overadjustment and the
fully adjusted multivariate model could underestimate
the real association. Future longitudinal research with
repeated data collection is needed to clarify the role of
mental health in this association.
Our results suggested a stronger association between
workplace violence and suicide attempts among men
than women. This is consistent with the findings
indicating a greater risk for suicide among men than
women exposed to job stressors.
29
A possible explanation
could be that men have been found to be more susceptible
to external social and economic stressors
31
and less likely
to seek health care for psychological disorders than
women. Additionally, due to use of higher lethality
HR (95% CI)
1·22 (0·99–1·51)
1·77 (1·15–2·70)
1·16 (0·96–1·41)
p value
0·0644
0·0087
0·1303
n (exposed)
18 033
7774
10 259
18 033
n (cases)
1144
960
184
1144
191 783
93 453
98 330
1·32 (1·09–1·59) 0·0040
Additional adjustment for mental health 191 783
1·52 (1·17–1·97)
1·17 (0·90–1·52)
1·28 (1·05–1·55)
2·05 (0·97–4·32)
1·33 (1·09–1·62)
1·78 (1·16–2·73)
0·50
1·00
5·00
0·0017
0·2427
0·0129
0·0592
0·0048
0·0083
Reduced risk Increased risk
Figure 4:
Association of workplace bullying with risk of suicide attempt or death in two cohort studies and previously published data
HRs and 95% CIs pooled from the current study and Conway et al,
11
and adjusted for age, sex, education, socioeconomic status, and family situation. HR=hazard ratio.
NA=not applicable. *Data not reported in Conway et al.
11
†Data for mixed occupations are from the Swedish Work Environment Surveys and Conway et al,
11
and data
for social and health-care workers are from the Finnish Public Sector Study.
e501
www.thelancet.com/public-health
Vol 8 July 2023
 BEU, Alm.del - 2022-23 (2. samling) - Bilag 251: Orientering om nye forskningsresultater på arbejdsmiljøområdet om konsekvenser ved vold og mobning i arbejdet
2748704_0009.png
Articles
suicide methods among men,
23
suicide attempts among
men might be more likely to be recorded in hospital
registers. Generally, there is a higher risk for under-
ascertainment of suicide attempts owing to absence of
recorded clinical care. Furthermore, since this is an
observational study, we cannot exclude the possibility
that unaccounted confounding is the true cause of the
effect modification.
Although absolute differences in suicide risk are small
in employee populations such as that included in this
study, the moderate relative differences combined with
an observed dose-response gradient for workplace
violence strengthen the plausibility of workplace
offensive behaviours contributing to suicidal behaviour.
The excess risk remained increased after adjustment for
multiple covariates including baseline mental health and
working conditions, suggesting that the observed
associations were not attributable to several known risk
factors for suicide. A null finding for a negative control
outcome further strengthens the likelihood that our
findings are not attributable to bias.
The large sample size, use of data from three different
settings, the register-based case definition, and inte-
gration of our results on workplace bullying into a meta-
analysis of published data are important strengths of this
study. In contrast to many previous case-control and
other self-report studies in the field, recall bias and
stigma did not affect the results. The data on exposure
and outcome derived from independent sources also
reduce the risk of common method variance, a common
problem that affects validity in observational studies.
Although the inclusion of deaths with undetermined
intent might increase the number of false positives
suicides,
32
it has been found to reduce bias due to spatial
and secular trends in detecting and classifying cases of
suicide when intent was indeterminable.
33
This study has several limitations. It remains unclear to
what extent our findings are generalisable to countries or
settings other than Finland, Sweden, and Denmark,
which have strong occupational health and safety
legislations. We used single-item measures of workplace
violence and bullying, which might be less accurate than
multi-item measures.
34
The length of follow-up varied
between participants due to multiple baselines and might
have been too short or too long for some individuals to
optimally detect effects of workplace violence and
bullying. Since this was an observational study, we cannot
exclude the possibility that there are other unmeasured
factors contributing to confounding. For example,
childhood adversities could be associated both with
workplace bullying and suicidal behaviour.
35
We used
linked health-care records to ascertain attempted suicides,
although such data might in some cases indicate self-
harm rather than the patient’s intention to die.
32
Similar
to a previous review,
10
our literature search used a
relatively narrow set of predefined terms and therefore it
is possible that some relevant studies were missed.
www.thelancet.com/public-health
Vol 8 July 2023
In conclusion, observational data from three Nordic
countries suggests that workplace violence or bullying
might subsequently result in an increased risk of
suicide attempt or death. These findings highlight the
potentially serious consequences of offensive workplace
behaviours.
Contributors
LLMH
developed the hypotheses, acquired funding for the study, and
wrote the first draft of the report and designed the study together with
MK and GDB. LMH, JP, MN, TX, RR, IEHM, PMC, HW, JV, JE, GDB,
and MK contributed to the design of the study, generation of hypotheses,
interpretation of the data, and critical review of the paper. LMH, JP, and
MN analysed cohort data, had full access to the data, and take
responsibility for the integrity of the data and the accuracy of the data
analysis. LMH is the guarantor. The final responsibility for the decision
to submit for publication was shared by all the authors.
Declaration of interests
MN changed employment during the conduct of this study and is now
an employee of Novo Nordisk A/S. All other authors declare no
competing interests.
Data sharing
In the Finnish Public Sector Study, pseudonymised questionnaire data as
used in this study can be shared by request to the investigators (jenni.
[email protected]). Linked health records require separate permission from
the National Institute of Health and Welfare and Statistics Finland.
SWES data are not publicly available due to legislative or ethical
restrictions, but can be requested for research directly from Statistics
Sweden, while linked health records can be requested from the Swedish
National Board of Health and Welfare. The WEHD study is based on
anonymised microdata available from Statistics Denmark. Access to data
can only be permitted through an affiliation with a Danish authorised
environment.
Acknowledgments
This study was supported by the Swedish Research Council for Health,
Working Life and Welfare (2019–01318, PI LMH), the Finnish Work
Environment Fund (200097), and the Danish Working Environment
Research Fund (10–2019–03). JV received funding from the Academy of
Finland (321409, 329240). JE received funding from the Finnish Work
Environment Fund (200097). MK received funding from the UK Medical
Research Council (MRC S011676), the Wellcome Trust (221854/Z/20/Z),
the Academy of Finland (350426, 329202), and Finnish Work
Environment Fund (200097, 190424).
References
1
International Labour Organization. Eliminating violence and
harassment in the world of work. Switzerland: International Labour
Organization, 2019.
2
Liu J, Gan Y, Jiang H, et al. Prevalence of workplace violence against
healthcare workers: a systematic review and meta-analysis.
Occup Environ Med
2019;
76:
927–37.
3
Hogh A, Viitasara E. A systematic review of longitudinal studies of
nonfatal workplace violence.
Eur J Work Organ Psychol
2005;
14:
291–313.
4
Nyberg A, Kecklund G, Hanson LM, et al. Workplace violence and
health in human service industries: a systematic review of prospective
and longitudinal studies.
Occup Environ Med
2021;
78:
69–81.
5
Xu T, Magnusson Hanson LL, Lange T, et al. Workplace bullying
and workplace violence as risk factors for cardiovascular disease:
a multi-cohort study.
Eur Heart J
2019;
40:
1124–34.
6
Rudkjoebing LA, Bungum AB, Flachs EM, et al. Work-related
exposure to violence or threats and risk of mental disorders and
symptoms: a systematic review and meta-analysis.
Scand J Work Environ Health
2020;
46:
339–49.
7
Birkeland Nielsen MB, Einarsen S. What we know, what we do not
know, and what we should and could have known about workplace
bullying: an overview of the literature and agenda for future
research.
Aggress Violent Behav
2018;
42:
71–83.
8
Harvey SB, Modini M, Joyce S, et al. Can work make you mentally ill?
A systematic meta-review of work-related risk factors for common
mental health problems.
Occup Environ Med
2017;
74:
301–10.
e502
For more on the
Swedish
National Board of Health
and Welfare
see
www.socialstyrelsen.se
For more on
data access
see
https://www.dst.dk/en/TilSalg/
Forskningsservice
 BEU, Alm.del - 2022-23 (2. samling) - Bilag 251: Orientering om nye forskningsresultater på arbejdsmiljøområdet om konsekvenser ved vold og mobning i arbejdet
2748704_0010.png
Articles
9
10
11
12
13
14
15
16
17
18
19
20
21
Magnusson Hanson LL, Nyberg A, Mittendorfer-Rutz E,
Bondestam F, Madsen IEH. Work related sexual harassment and
risk of suicide and suicide attempts: prospective cohort study.
BMJ
2020;
370:
m2984.
Leach LS, Poyser C, Butterworth P. Workplace bullying and the
association with suicidal ideation/thoughts and behaviour:
a systematic review.
Occup Environ Med
2017;
74:
72–79.
Conway PM, Erlangsen A, Grynderup MB, et al. Workplace bullying
and risk of suicide and suicide attempts: a register-based
prospective cohort study of 98 330 participants in Denmark.
Scand J Work Environ Health
2022;
48:
425–34.
Joensuu M, Kivimaki M, Pentti J, et al. Components of job control
and mortality: the Finnish Public Sector Study.
Occup Environ Med
2014;
71:
536–42.
Corin L, Pousette A, Berglund T, Dellve L, Hensing G, Björk L.
Occupational trajectories of working conditions in Sweden:
development trends in the workforce, 1997–2015.
Scand J Work
Environ Health
2021;
47:
335–48.
Johnsen NF, Thomsen BL, Hansen JV, et al. Job type and other
socio-demographic factors associated with participation in a
national, cross-sectional study of Danish employees.
BMJ Open
2019;
9:
e027056.
Magnusson Hanson LL, Theorell T, Bech P, et al. Psychosocial
working conditions and depressive symptoms among Swedish
employees.
Int Arch Occup Environ Health
2009;
82:
951–60.
Magnusson Hanson LL, Rod NH, Vahtera J, et al. Multicohort study
of change in job strain, poor mental health and incident
cardiometabolic disease.
Occup Environ Med
2019;
76:
785–92.
Bech P. Depressed mood as a core symptom of depression.
Medicographia
2008;
94:
9–13.
Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for
defining comorbidities in ICD-9-CM and ICD-10 administrative
data.
Med Care
2005;
43:
1130–39.
Ludvigsson JF, Appelros P, Askling J, et al. Adaptation of the
Charlson Comorbidity Index for register-based research in Sweden.
Clin Epidemiol
2021;
13:
21–41.
Lipsitch M, Tchetgen Tchetgen E, Cohen T. Negative controls: a tool
for detecting confounding and bias in observational studies.
Epidemiology
2010;
21:
383–88.
Heikkilä K, Nyberg ST, Theorell T, et al. Work stress and risk of
cancer: meta-analysis of 5700 incident cancer events in 116,000
European men and women.
BMJ
2013;
346:
f165.
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Yang T, Qiao Y, Xiang S, Li W, Gan Y, Chen Y. Work stress and the
risk of cancer: a meta-analysis of observational studies.
Int J Cancer
2019;
144:
2390–400.
Turecki G, Brent DA, Gunnell D, et al. Suicide and suicide risk.
Nat Rev Dis Primers
2019;
5:
74.
Hedges LV, Vevea JL. Fixed- and random-effects models in meta-
analysis.
Psychol Methods
1998;
3:
486–504.
von Hippel PT. The heterogeneity statistic I(2) can be biased in
small meta-analyses.
BMC Med Res Methodol
2015;
15:
35.
Crump C, Sundquist K, Sundquist J, Winkleby MA.
Sociodemographic, psychiatric and somatic risk factors for
suicide: a Swedish national cohort study.
Psychol Med
2014;
44:
279–89.
Batty GD, Kivimäki M, Bell S, et al. Psychosocial characteristics as
potential predictors of suicide in adults: an overview of the evidence
with new results from prospective cohort studies.
Transl Psychiatry
2018;
8:
22.
Lac G, Dutheil F, Brousse G, Triboulet-Kelly C, Chamoux A. Saliva
DHEAS changes in patients suffering from psychopathological
disorders arising from bullying at work.
Brain Cogn
2012;
80:
277–81.
Milner A, Witt K, LaMontagne AD, Niedhammer I. Psychosocial job
stressors and suicidality: a meta-analysis and systematic review.
Occup Environ Med
2018;
75:
245–53.
Almroth M, Hemmingsson T, Kjellberg K, et al. Job control, job
demands and job strain and suicidal behaviour among three million
workers in Sweden.
Occup Environ Med
2022;
79:
681–89.
Möller-Leimkühler AM. The gender gap in suicide and premature
death or: why are men so vulnerable?
Eur Arch Psychiatry Clin Neurosci
2003;
253:
1–8.
Swain RS, Taylor LG, Braver ER, Liu W, Pinheiro SP,
Mosholder AD. A systematic review of validated suicide outcome
classification in observational studies.
Int J Epidemiol
2019;
48:
1636–49.
Linsley KR, Schapira K, Kelly TP. Open verdict v. suicide—
importance to research.
Br J Psychiatry
2001;
178:
465–68.
Nielsen MB, Matthiesen SB, Einarsen S. The impact of
methodological moderators on prevalence rates of workplace
bullying. A meta-analysis.
J Occup Organ Psychol
2010;
83:
955–79.
Hughes K, Bellis MA, Hardcastle KA, et al. The effect of multiple
adverse childhood experiences on health: a systematic review and
meta-analysis.
Lancet Public Health
2017;
2:
e356–66.
e503
www.thelancet.com/public-health
Vol 8 July 2023