Downloaded from www.sjweh.

fi on June 15, 2023

Scand J Work Environ Health 2022;48(8):621-631

https://doi.org/10.5271/sjweh.4042
Published online: 26 Jun 2022, Issue date: 01 Nov 2022

Workplace psychosocial resources and risk of cardiovascular

disease among employees: a multi-cohort study of 135 669
participants
by Xu T, Rugulies R, Vahtera J, Pentti J, Mathisen J, Lange T, Clark AJ,
Magnusson Hanson LL, Westerlund H, Ervasti J, Virtanen M, Kivimäki M,
Rod NH

Five different clusters of workplace psychosocial resources were

identified, consistent across cohorts with varying distributions of sex,
employment sectors, industries and job types. This multicohort study
from 135 669 employees found high levels of workplace psychosocial
resources across horizontal (culture of collaboration, support from
colleagues) and vertical (leadership quality, procedural justice)
dimensions were associated with a lower risk of incident
cardiovascular disease.

Affiliation: Tianwei Xu, Stress Research Institute, Stockholm

University, 106 91 Stockholm, Sweden. [email protected]

Refers to the following texts of the Journal: 2019;45(4):356-369

2020;46(5):498-507

Key terms: cardiovascular disease; collaboration; leadership quality;

multi-cohort study; procedural justice; psychosocial; psychosocial
resource; social support

This article in PubMed: www.ncbi.nlm.nih.gov/pubmed/35752989

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

Original article This work is licensed under a Creative Commons Attribution
4.0 International License.

Scand J Work Environ Health. 2022;48(8):621–631. doi:10.5271/sjweh.4042

Workplace psychosocial resources and risk of cardiovascular disease among employees:

a multi-cohort study of 135 669 participants
by Tianwei Xu, PhD,1–3 Reiner Rugulies, PhD,2–4 Jussi Vahtera, PhD,5–7 Jaana Pentti, MSc,5, 8, 9 Jimmi Mathisen, MSc,2 Theis
Lange, PhD,2 Alice J Clark, PhD,2, 10 Linda L Magnusson Hanson, PhD,1 Hugo Westerlund, PhD,1 Jenni Ervasti, PhD,9 Marianna
Virtanen, PhD,11, 12 Mika Kivimäki, PhD,8, 9, 13 Naja H Rod, PhD 2

Xu T, Rugulies R, Vahtera J, Pentti J, Mathisen J, Lange T, Clark AJ, Magnusson Hanson LL, Westerlund H, Ervasti J, Virtanen M,
Kivimäki M, Rod NH. Workplace psychosocial resources and risk of cardiovascular disease among employees: a multi-cohort
study of 135 669 participants. Scand J Work Environ Health. 2022;48(8):621–631. doi:10.5271/sjweh.4042

Objective In terms of prevention, it is important to determine effects on cardiovascular disease (CVD) when
some workplace psychosocial resources are high while others are low. The aim of the study was to assess the
prospective relationship between clustering of workplace psychosocial resources and risk of CVD among
employees.
Methods We pooled data from three cohort studies of 135 669 employees (65% women, age 18–65 years and free
of CVD) from Denmark, Finland and Sweden. Baseline horizontal resources (culture of collaboration and support
from colleagues) and vertical resources (leadership quality and procedural justice) were measured using standard
questionnaire items. Incident CVD, including coronary heart and cerebrovascular disease, was ascertained using
linked electronic health records. We used latent class analysis to assess clustering (latent classes) of workplace
psychosocial resources. Cox proportional hazard models were used to examine the association between these
clusters and risk of CVD, adjusting for demographic and employment-related factors and pre-existing physical
and mental disorders.
Results We identified five clusters of workplace psychosocial resources from low on both vertical and horizontal
resources (13%) to generally high resources (28%). High horizontal resources were combined with either inter-
mediate [hazard ratio (HR) 0.84, 95% confidence interval (CI) 0.74–0.95] or high (HR 0.88, 95% CI 0.78–1.00)
vertical resources were associated with lower risks of CVD compared to those with generally low resources.
The association was most prominent for cerebrovascular disease (eg, general high resources: HR 0.80, 95% CI
0.67–0.96).
Conclusions Individuals with high levels of workplace psychosocial resources across horizontal and vertical
dimensions have a lower risk of CVD, particularly cerebrovascular disease.

Key terms collaboration; leadership quality; procedural justice; social support.

1 Stress Research Institute, Stockholm University, Stockholm, Sweden.

2 Department of Public Health, University of Copenhagen, Copenhagen, Denmark.
3 National Research Centre of the Working Environment, Copenhagen, Denmark.
4 Department of Psychology, University of Copenhagen, Copenhagen, Denmark.
5 Department of Public Health, University of Turku, Turku, Finland.
6 The Centre for Population Health Research, University of Turku, Turku, Finland.
7 Turku University Hospital, Turku, Finland.
8 Clinicum, Faculty of Medicine, University of Helsinki, Finland.
9 Finnish Institute of Occupational Health, Helsinki, Finland.
10 Novo Nordisk A/S, Søborg, Denmark.
11 School of Educational Sciences and Psychology, University of Eastern Finland, Joensuu, Finland.
12 Division of Insurance Medicine, Karolinska Institutet, Stockholm, Sweden.
13 Department of Epidemiology and Public Health, University College, London, UK.

Correspondence: Tianwei Xu, Stress Research Institute, Stockholm University, 106 91 Stockholm, Sweden. [E-mail: tianwei.xu@psychol-
ogy.su.se]

Scand J Work Environ Health 2022, vol 48, no 8 621

Workplace resources and cardiovascular disease

Early studies on stressful psychosocial working condi- of such clustering across various hierarchical domains in
tions and cardiovascular disease (CVD) risk were pub- order to develop effective multilevel interventions aimed
lished already in the 1960s (1), and subsequent cohort at creating healthier workplaces.
studies have confirmed an association between work To address these limitations, we examined the clus-
stressors and CVD (2). The leading concepts of psy- tering of four key workplace psychosocial resources (ie,
chosocial working conditions include the job–demand– culture of collaboration, social support from colleagues,
control (job strain) model (3–5), the effort–reward leadership quality and procedural justice) and assessed
imbalance model (6), the organizational justice model whether these clusters were associated with the risk of
(7), and the job–demands–resources model (8). More developing CVD (including CHD and CBD) in three
recently, it has been suggested that a focus on potential cohorts with a total of 135 669 men and women from
health-protective resources at work may also be use- Denmark, Finland and Sweden. These cohorts included
ful (9). From a CVD prevention perspective, targeting employees from public and private sectors.
workplace psychosocial resources may complement
traditional workplace interventions, such as wellness
and exercise programs (10).
High levels of workplace psychosocial resources have Methods
been suggested to be linked to a lower risk of mental
health problems including depression (11), the metabolic Study population
syndrome (12), and type 2 diabetes (13, 14), lower lev-
els of inflammatory markers such as C-reactive protein, We used data from the following three prospective
interleukin-6 and tumor necrosis factor alpha (13, 15), cohort studies: The Work Environment and Health in
and lower ambulatory blood pressure (16, 17), which Denmark (WEHD) study, the Finnish Public Sector
are all CVD risk factors. To date, however, the evidence (FPS) study and the Swedish Longitudinal Occupa-
on potential health benefits of workplace psychosocial tional Survey of Health (SLOSH) (figure 1). WEHD is
resources is inconsistent and scarce. Published studies a biennial population-based survey, initiated in 2012 in
have mainly investigated coronary heart disease (CHD) Denmark, with around 58% respondents working in the
but not cerebrovascular disease (CBD), although the lat- private sector (24). FPS is a dynamic cohort of Finnish
ter contributes to 35% of age-standardized CV-related employees with repeated data collections every two to
deaths (18). Three studies found that a higher level of four years initiated in 1998/2000 onwards (25). FPS
workplace psychosocial resources, operationalized as consists of employees in the municipal services of ten
preferable levels of organizational justice (7), leadership Finnish town and 21 public hospitals, who had a job
quality (19), or workplace social support (20, 21), were contract for a minimum of six months. SLOSH is a pop-
associated with a lower risk of incident CHD, but these ulation-based cohort initiated in 2006 in Sweden with
associations were not observed in an earlier study of 19 biennial follow-ups, including 59% participants working
565 full-time employed Swedish women (22), and not for in the private sector (26). A more detailed description of
CBD (21). None of these previous studies considered the these cohorts has been published elsewhere (27).
clustering of workplace psychosocial resources, although According to the data availability and to allow
resources are likely to cluster and through this clustering cross-wave comparability, we included WEHD waves
be differently associated with health outcomes than what 2012–2014, FPS waves 2000–2014, and SLOSH waves
would be expected based on their individual effects. The 2012–2016. Figure 1 depicts the process of baseline
coexistence of workplace psychosocial resources at orga- establishment, including the exclusion criteria, and end
nizational, leadership, and group levels (ie, different hier- of follow-ups. To ascertain incident CVD during the
archical domains) may be dynamic (23). These different follow-up, all CVD cases occurred prior to the baseline
sources of resources may potentially affect each other and were excluded (figure 1).
exert synergistic influences on employees’ health (23). Ethical approval for FPS was obtained from the
Our previous study identified four distinct resource clus- Ethics Committee of the Hospital District of Helsinki
ters among Finnish public sector employees, and some and Uusimaa (25). WEHD was approved by and reg-
clusters were more protective of type-2 diabetes than oth- istered with the Danish Data Protection Agency. Ethi-
ers (14), but it remained unclear whether this workplace cal approval was obtained from the Regional Ethical
resource pattern could be generalized to the wider work- Review Board in Stockholm for SLOSH (26).
ing population, including private sector employees, or to
other health outcomes. The present paper adds new results
Workplace psychosocial resources
towards this end. In terms of prevention, it is important
to explore and understand the clustering of workplace We measured four types of workplace psychosocial
psychosocial resources and the potential health effects resources: (i) culture of collaboration, (ii) support from

622 Scand J Work Environ Health 2022, vol 48, no 8

 Xu et al

Figure 1. Flow chart of

the study population.

colleagues, (iii) leadership quality, and (iv) procedural listening, appreciative, and informative (three items
justice to represent hierarchical dimensions of work- from The Stress Profile and one item from the relational
place psychosocial resources, ie, group (horizontal), justice scale) (32, 33). WEHD included slightly different
leader (vertical) and organizational (vertical) levels, dimensions (8 items including eg, authorization of own
respectively, using standardized items/scales (supple- work and career development) (24). Leadership quality
mentary material, www.sjweh.fi/article/4042, table S1) was categorized into quartiles.
(23). Detailed explanations of choosing the following The variable for procedural justice (fairness in the
categorization can be found in supplementary text S1. principles and processes leading to decision-making and
Culture of collaboration was dichotomized and good the distribution of rewards and benefits) was measured
collaboration was defined as the collaborative efforts to using a modified version of Moorman’s scale (34). Pro-
achieve the best available results or to develop or apply cedural justice was categorized into quartiles in FPS and
new ideas in the workplace. The items from the justice SLOSH. In WEHD, procedural justice was also grouped
and team climate inventories were used as a single item into four levels (one item) and the highest level of pro-
(in WEHD) or as dichotomized by the mean score (in cedural justices was ‘all the time’, followed by ‘often’
SLOSH and FPS) (28, 29). or ‘sometimes’, ‘rarely’ and ‘never’.
Co-worker support on perceived colleagues’ support To understand the heterogeneity of instruments
was dichotomized by whether receiving an affirmative across cohorts, we performed tests for checking the
response to one item (from the Danish Psychosocial correlations and agreements between the single-item
Work Environment Questionnaire in WEHD (30); from instruments and the full scales (supplementary table
the Demand–Control questionnaire in SLOSH (26); S2). There were high correlations (Spearman correla-
and from Statistics Finland working climate questions tion coefficients >0.8) and moderate to strong levels of
in FPS) (31). agreements (0.60<κ<0.90) between the single-item and
Due to the harmonization (14), the leadership vari- scale measures when assessing culture of collaboration.
able in FPS and SLOSH included dimensions on caring, For procedural justice, despite of a high correlation

Scand J Work Environ Health 2022, vol 48, no 8 623

Workplace resources and cardiovascular disease

between the single-item measurement (Spearman cor- Statistical analysis

relation coefficients >0.8) and the full scale, the agree-
ment was relatively weak (0.40<κ<0.59). Leadership We used latent class analysis, a hypothesis-free data-
items in WEHD were not identical with those in FPS driven approach, to assess clustering of workplace
and SLOSH (the latter two cohorts used exactly same psychosocial resources based on participants’ first eli-
leadership items) and due to lack of shared items, we gible participation (36). In previous study based on
were not able to perform similar validity analyses as for this approach, we have shown that the latent classes
culture of collaboration and procedural justice. observed at baseline (ie, the first eligible participation)
were robust over time and could be extrapolated to
all participants regardless of their participation waves
Assessment of cardiovascular disease
(14). In addition to Bayesian Information Criterion (a
Using the unique personal identification numbers for measurement of model fit), distribution of class member-
each citizen in Denmark, Finland and Sweden, all par- ship probabilities, class sizes and interpretability of the
ticipants were linked to nationwide health, death and classes, we determined the class model according to the
population registers. We used in-patient (all cohorts), comparability across cohorts (supplementary figure S1)
out-patient (SLOSH and WEHD) and death (all cohorts) (36). Although a four-class solution already showed dis-
registers to capture incident CVD (ie, cases occurred tinctive patterns in FPS in a previous study (14), using
prior to baseline were excluded). CVD was identified a five-class model with one additional distinctive latent
if diagnosed with CHD or CBD. We detected CHD class pattern added in WEHD and SLOSH, the three
using the main diagnosis codes of ICD-10 I20.0, I20.1, cohorts were more comparable and we therefore chose
I21–I25 (excluding unspecified angina), and ICD-8/9 to use this five-class solution (supplementary figure S1).
410–414, whereas ICD-10 I60-I69 and ICD-8/9 430– We ran a Cox proportional hazard model with age
438 were used to detect CBD as the main diagnosis. as the underlying time scale. No violation was detected
Subtypes of CVD, including myocardial infarction, for proportional hazard assumption (by using log-log
ischemic stroke and hemorrhagic stroke were also iden- plot or by including interaction terms between time and
tified using ICD codes (supplementary text S2). Incident covariates). The model was adjusted for country of birth
CVD events were identified with the earliest diagnosed (when available), marital status, educational level, type of
date after the baseline, after excluding historical CVD employment contract, and pre-existing mental and physi-
events (figure 1). cal comorbidity. The incidence rate difference was cal-
culated using the Aalen additive hazard model. Subtype
analyses of CHD and CBD were performed. As a supple-
Covariates
ment, we analyzed each individual type of resources with
Confounders were identified prior to data analysis using and without mutual adjustment of the others, using the
directed acyclic graphs based on prior knowledge (36). lowest level of each individual resource as the reference.
Key confounders included age, sex, country of birth We also conducted several sensitivity analyses. In
(Nordic born, other European countries, other conti- order to reduce the risk that employees who had preva-
nents), educational level (≤9, 10–12, ≥13 years), marital lent, albeit undiagnosed CVD might be more likely to
status (unmarried or cohabiting, single, separated or perceive workplace psychosocial resources differently,
divorced and widowed), type of employment contract resulting in reverse causation, a one-year lag-time was
(permanent/non-permanent), pre-existing comorbidities applied. We further restricted the follow-up lengths to
(Charlson Comorbidity Score) and pre-existing diag- the first four years, to rule out the impact of differences
nosed mental disorders (supplementary text S3). These in follow-up lengths across cohorts. Lastly, models were
variables were all extracted from the national registers, additionally adjusted for other covariates, ie, lifestyle
except that marital status in FPS and employment con- and clinical factors assumed to be mediators in the
tract in SLOSH were measured by self-report and there primary analysis.
was no information on country of birth in FPS. To determine potential effect modification, strati-
Other clinical factors including body mass index fied results were carried out for age groups, by sex and
(BMI), self-reported mental health and lifestyle factors educational levels using the Cox model and the additive
including smoking (current smoker/non-smoker), risky hazard model, to estimate interactions on both the mul-
alcohol consumption (yes/no) and physical inactivity tiplicative and additive scales.
(yes/no) were self-reported at the baseline (supplemen- We followed a 2-stage approach in which associa-
tary text S3). We considered them to be potential media- tions were first analyzed in each cohort study separately
tors rather than confounders, as they were measured at and then cohort-specific estimates were combined using
the same time with the exposures. fixed-effect meta-analysis (R package meta version 4.9-
2). We used R package, poLCA version 1.4.1, for latent

624 Scand J Work Environ Health 2022, vol 48, no 8

 Xu et al

class analysis, SAS 9.4 procedure, PROC PHREG, for low resources (figure 3A), classes with high horizontal
Cox models and R package timereg, version 1.9.3 for resources combined with either intermediate or high ver-
additive hazard models. Results from Cox models and tical resources were at lower risk of developing incident
additive hazard models were expressed as hazard ratios CVD, corresponding to 3.4 (95% CI -6.7– -0.1) and 2.2
(HR) and incidence rate difference (IRD), respectively (95% CI -5.4–1.0) fewer incident CVD cases per 10 000
with their 95% confidence intervals (CI). Statistical person-year, respectively.
syntax is provided in the supplementary material). Subtype analysis (figure 3B) of 1175 CHD and 1097
CBD incident cases showed the three clusters with high
horizontal resources were associated with a lower risk of
total CBD, especially with hemorrhagic stroke, but not
Results with total CHD. We observed a lower risk of incident
myocardial infarction when perceiving ‘intermediate
Patterns of workplace psychosocial resources vertical+high horizontal’ resource.
Excluding cases during the first year or restrict-
We identified five latent classes of workplace psy- ing to the first four-year follow-up did not change the
chosocial resources, using 57 496 participants from effect estimates (supplementary figure S2). Additional
WEHD, 63 267 from FPS and 14 906 from SLOSH, with adjustments for lifestyle factors and self-reported men-
similar patterns of workplace psychosocial resources tal health did not substantially change the effect sizes
across cohorts (figure 2A). WEHD and SLOSH shared (supplementary figure S3). We did not observe signifi-
similar distribution of resource clusters, while FPS had cant differences across age groups, sex and educational
a larger proportion of ‘general low’ and ‘intermedi- levels (supplementary figure S4).
ate vertical+low horizontal’ and smaller proportion In a supplementary analysis, before mutual adjust-
of ‘low vertical+high horizontal’ and ‘intermediate ment of the individual resources, some associations were
vertical+high horizontal’ resources than WEHD and observed for procedural justice, leadership quality and
SLOSH (figure 2B). co-worker support (supplementary figure S5A). Most of
Three cohorts together (figure 2B), 13% were cat- these associations attenuated after mutual adjustment:
egorized into the ‘general low’ class, in which all the only intermediately high level of procedural justice (HR
four resources were low. The ‘intermediate vertical+low 0.84, 95% CI 0.74–0.96) and high level of support from
horizontal’ class (11%) consisted of mainly intermedi- colleagues (HR 0.87, 95% CI 0.78–0.97) remained asso-
ate (i.e. intermediately high and intermediately low) ciated with a lower risk of CVD after mutual adjustment
levels of vertical resources (procedural justice and (supplementary figure S5B).
leadership quality), but low levels of social support and
culture of collaboration (horizontal resources). The ‘low
vertical+high horizontal’ class (17%) was characterized
by low level of procedural justice and leadership quality Discussion
and high levels of social support from colleagues and
culture of collaboration. The ‘intermediate vertical+high This analysis of individual-level data on almost 140
horizontal’ class (32%) is characterized by an interme- 000 persons from three Nordic cohort studies identified
diate (i.e. intermediately high and intermediately low) a consistent pattern including five classes of workplace
level of procedural justice and leadership quality and psychosocial resource across follow-up waves, employ-
a high level of social support from colleagues and cul- ment sectors and countries. About 13% of the employ-
ture of collaboration. Lastly, in the ‘general high’ class ees experienced low levels of all studied workplace
(28%) individuals reported a relatively high workplace resources, suggesting a potential for improvement. Our
resources across all dimensions. Baseline characteristics findings show a consistent protective effect of workplace
are presented in table 1. resources on overall CVD, most prominently for myo-
cardial infarction and CBD.
A moderate effect on myocardial infarction was
Workplace psychosocial resources and cardiovascular
observed when intermediate level of vertical resources (ie,
disease
procedural justice and leadership quality) were combined
During a mean follow-up of 6.8 years, 2190 incident with high level of horizontal resources (ie, culture of col-
CVD cases (26.8 per 10 000 person-years) were recorded laboration and co-worker support), in line with previous
among 135 669 initially CVD-free participants (mean research on specific aspects of workplace resources and
age: 44 years, proportion of women: 65%) (table 2). hospitalization/death due to myocardial infarction (7, 19,
The results across cohorts were generally homogeneous 20). Lack of social support in general may also be asso-
(I2<0.1%). Compared to the latent class characterized by ciated with cardiac mortality or all-cause mortality (37).

Scand J Work Environ Health 2022, vol 48, no 8 625

Workplace resources and cardiovascular disease

Figure 2. Workplace psychosocial resource pattern in each latent class using data from Work Environment and Health in Denmark study (WEHD: N=57 496),
Finnish Public Sector study (FPS: N=63 267) and Swedish Longitudinal Occupational Survey of Health (SLOSH: N=14 906).

Our findings add to the evidence by comprehensively between CHD and CBD were highly overlapping. These
identifying clustering patterns of resources across several issues require careful investigations in future studies.
key vertical and horizontal psychosocial resources. The underlying mechanisms for the potentially pro-
We found a lower risk of developing CBD when tective vascular effects remain to be uncovered. Earlier
perceiving high level of horizontal resources (ie, culture research have shown a favorable level of workplace
of collaboration and co-worker support). To the best social support (16) and relational justice (17) in con-
of our knowledge, we are the first longitudinal study nection with a lower ambulatory blood pressure. Low
demonstrating an association between these workplace workplace social support has been found to be associ-
resources and risk of CBD. André-Petersson et al (21) ated with a higher level of interleukin-6 (13). Other
found no association between workplace social support possible mechanistic pathways include indirect effect
and CBD, but based on only 58 cases, much less than via health-related behaviors (10, 39–41) and mental
1097 cases in our study. Our findings need to be repli- health problems (42).
cated in other studies.
We found a stronger effect for CBD than CHD,
Public health implications
similar to the one shown in a previous multicohort
study concerning long working hours (38), while the CI It is interesting that there was no evidence of indepen-

626 Scand J Work Environ Health 2022, vol 48, no 8

 Xu et al

Table 1. Baseline characteristics of the analytical population by workplace psychosocial resources (N=135 669).
Workplace psychosocial Total General low Intermediate Low vertical+high Intermediate General high
resources (N=135 669) (N=18 040) 13% vertical+low horizontal horizontal vertical+high horizontal (N=37 894)
(N=14 394) 11% (N=22 441) 17% (N=42 900) 32% 28%
% Mean % Mean % Mean % Mean % Mean % Mean
Demographic characteristics
Age (years) 44 45 45 45 44 44
Women 65 67 71 61 64 69
Non-Nordic born a 3 3 4 2 3 3
Low educational level 24 22 19 30 25 23
Married 68 70 68 66 68 70
Clinical characteristics
Comorbidity score 0.15 0.14 0.13 0.16 0.15 0.15
Body mass index b 25 26 25 26 25 25
Mental disorders 2 2 2 3 2 2
Lifestyle characteristics
Current smoker b 19 21 20 21 18 19
Physical inactivity b 26 32 33 24 25 26
Excessive alcohol consumption b 11 12 11 12 12 11
Work-related characteristics
Temporary job contract 12 11 15 9 12 16
a
Based only on the Work Environment and Health in Denmark (WEHD) study and the Swedish Longitudinal Occupational Survey of Health (SLOSH).
b
A total of 36 314 missing existed for these variables, mainly due to the skipped measurement in Finnish Public Sector (FPD) study waves 2000, 2006, 2010.

Table 2. Summary of studies that provided individual participant data used in the analyses for cardiovascular disease, using data from the Work
Environment and Health in Denmark (WEHD) study (N=57 496), Finnish Public Sector (FPS) study (N=63 267) and the Swedish Longitudinal
Occupational Survey of Health (SLOSH) (N=14 906). [CVD=cardiovascular disease; CHD=coronary heart disease; CBD=cerebrovascular disease.].
Cohorts Country Baseline years Follow-up length Baseline age Women,% CVDa CHDa CBDa
(mean, years) (mean, years)
WEHD Denmark 2012–2016 2.0 46 54 32.3 17.4 15.3
FPS Finland 2000–2014 11.8 43 77 22.0 11.5 11.2
SLOSH Sweden 2012–2016 4.0 49 59 30.2 18.6 12.7
All 2000–2016 6.8 44 65 26.8 14.7 12.8
a Incidence rate per 10 000 person-years.

dent associations between each individual resource and may differ in their perception of job, communication for-
CVD in mutually-adjusted models. When clustering was malization and objective-oriented results (43). Male and
considered, the associations with CVD became clear female employees may experience different employment
and similar to those previously found for type-2 diabe- and working conditions (44) and have diverged percep-
tes (14). This emphasizes the importance of exploring tion on some resources (16, 45). A deeper understanding
clustering of resources instead of singling out individual of the distribution of resource clusters in specific set-
effects and is consistent with findings from a systematic tings is needed to develop targeted work-related CVD
review, which highlighted the importance of multi-level preventions in different types of workplaces, eg, among
workplace interventions (23), ie, to intervene on vertical private and public sector employees.
and horizontal dimensions of resources at the same time. Interestingly, the lowest risk of CVD was often
Considering the average annual incidence of 26.8 found in the “intermediate vertical+high horizontal”
per 10 000 persons, the relative differences, such as a resource group rather the highest resource group. This
16% lower risk of developing incident CVD and the suggests that not all resources at work are equally
absolute differences, such as 3.4 lower incidence per important in reducing the risk of CVD. Some studies
10 000 person-years when comparing the ‘intermediate suggest that workplace social support may in some
vertical+high horizontal’ with the ‘general low’, should cases be unhelpful and even trigger stress (46). How-
be interpreted cautiously. The public health importance, ever, the CI between “general high” and “intermedi-
if causal, will depend on the distribution of resource ate vertical+high horizontal” resource groups were
classes across settings. For example, compared with overlapping and thus not statistically different. More
public sector employees (ie, FPS), the general working research is needed to understand whether pursuing the
population (ie, WEHD and SLOSH) contained a smaller highest level of resources at work is always beneficial
proportion of workers in the ‘general low’ resource class for employee health.
and fewer women. Public and private sector employees

Scand J Work Environ Health 2022, vol 48, no 8 627

Workplace resources and cardiovascular disease

Figure 3. Association between clustering of workplace psychosocial resources and incident cardiovascular disease (CVD), after adjustment for age, sex, country
of birth, educational level, marital status, pre-existing comorbidity, pre-existing mental disorders and types of employment contract. Fixed effect meta-analysis.

Theoretical relevance in-patient visits and thus more likely to capture severe
CVD cases, potentially contributing to an underestima-
We selected workplace psychosocial resources at group tion of the effect (47). While the same questionnaires
(eg, team climate), leader (eg, leadership quality) and were used in SLOSH and FPS, WEHD used slightly
organizational levels (eg, perceived organizational sup- different instruments to measure culture of collabora-
port), following a recently proposed theoretical frame- tion, procedural justice and leadership quality. This is an
work for workplace resources by Nielsen et al (23). This unlikely source of major bias because the three cohorts
review also showed that these workplace resources may showed very similar patterns of resource clusters and
be associated with better employees’ job performance associations with CVD. Although the point estimates
and well-being (23). Consistent with existing evidence were higher in SLOSH and WEHD than in FPS, no
of psychosocial resources at work (23), we identified statistically significant heterogeneity was detected in
clustering of the four resources according to vertical cohort-specific effect estimates. We estimated four com-
and horizontal dimensions. Future research may be mon group-, leader- and organization-level workplace
required to disentangle the potential interactions among psychosocial resources using a data-driven approach to
these resources to facilitate the design of cost-effective detect clustering of workplace resources. This approach
multilevel interventions. may be sensitive to the categorization and selection of
the resource items. Future research is therefore needed
to test the robustness of the clusters when using dif-
Limitations and strengths
ferent categorizations. Some more detailed aspects of
Some limitations merit careful consideration. Workplace resources, such as perceiving or receiving co-worker’s
psychosocial resources were measured by self-assess- support was not considered, and may be considered as a
ment at baseline and job changes were not accounted for. limitation (37). While resources tend to highly intercor-
Although omitting the effect of time-varying resources relate, a more comprehensive mapping of resources may
and time-varying confounders may have resulted in an be needed in the future.
underestimation of the association, when we restricted The strengths of our study include the large sample
the follow-up lengths to the first four years, the risk size with long follow-up, which allowed us to perform
estimates remained similar, suggesting this to be of analyses on specific subtypes of CVD as well as conduct
limited concern. FPS contained information only on a range of sensitivity analyses with sufficient statisti-

628 Scand J Work Environ Health 2022, vol 48, no 8

 Xu et al

cal power. We were able to perform a nearly complete Acknowledgement

follow-up by linking survey data to nationwide registries
to identify new cases of definite CHD and CBD. The
inclusion of participants from both the general work- We thank Dr. Constanze Leineweber and Mr. Martin
ing population and public sector employees of three Claeson (Department of Psychology, Stockholm Univer-
Scandinavian countries further provided a sufficient sity) for managing the SLOSH data for analysis.
number of participants from both sexes and ensured the
diversity of industry, job type and employment sectors,
which assures the generalizability of our findings to
similar contexts. References

1. Berkman LF, Kawachi I, Theorell T. Working conditions and

Concluding remarks health. Social epidemiology. 2014;2:153–81.
Our study identified five distinct workplace psychoso- 2. Kivimäki M, Steptoe A. Effects of stress on the development
cial resource clusters with different levels of resources, and progression of cardiovascular disease. Nat Rev
consistent across countries and employment sectors. Cardiol 2018 Apr;15(4):215–29. https://doi.org/10.1038/
nrcardio.2017.189.
Employees with favorable workplace psychosocial
resources, especially intermediate and high vertical and 3. Kivimäki M, Nyberg ST, Batty GD, Fransson EI, Heikkilä
high horizontal resources, were at a lower risk of CVD. K, Alfredsson L et al.; IPD-Work Consortium. Job strain
Further research is needed to determine whether inter- as a risk factor for coronary heart disease: a collaborative
meta-analysis of individual participant data. Lancet 2012
ventions to improve workplace psychosocial resources
Oct;380(9852):1491–7. https://doi.org/10.1016/S0140-
could be beneficial to vulnerable groups with established 6736(12)60994-5.
CVD risk factors.
4. Fransson EI, Nyberg ST, Heikkilä K, Alfredsson L,
Bjorner JB, Borritz M et al. Job strain and the risk of
Funding stroke: an individual-participant data meta-analysis.
Stroke 2015 Feb;46(2):557–9. https://doi.org/10.1161/
The Danish Working Environment Foundation supported
STROKEAHA.114.008019.
this study (grant 13-2015-09). TX was supported by a
grant from the Swedish Research Council for Health, 5. Rugulies R, Framke E, Sørensen JK, Svane-Petersen AC,
Working Life and Welfare (grant 2020-00040). MK Alexanderson K, Bonde JP et al. Persistent and changing job
was supported by research grant from NordForsk (grant strain and risk of coronary heart disease. A population-based
cohort study of 1.6 million employees in Denmark. Scand J
70521, the Nordic Research Programme on Health and
Work Environ Health 2020 Sep;46(5):498–507. https://doi.
Welfare), the UK Medical Research Council (grant org/10.5271/sjweh.3891.
MRC S011676), the US National Institute on Aging
(NIA) (grant R01AG056477), the Academy of Finland 6. Dragano N, Siegrist J, Nyberg ST, Lunau T, Fransson EI,
(grant 329202), and Finnish Work Environment Fund Alfredsson L et al.; IPD-Work consortium. Effort-Reward
(grant 190424). LMH was supported by a grant from Imbalance at Work and Incident Coronary Heart Disease:
A Multicohort Study of 90,164 Individuals. Epidemiology
the Swedish Research Council for Health, Working Life
2017 Jul;28(4):619–26. https://doi.org/10.1097/
and Welfare (grant 2019-01318). MV was supported EDE.0000000000000666.
by Academy of Finland (grant 329201). JV was sup-
ported by the Academy of Finland (grant 321409 and 7. Kivimäki M, Ferrie JE, Brunner E, Head J, Shipley MJ,
329240). JE was supported by the Academy of Finland Vahtera J et al. Justice at work and reduced risk of coronary
(grant 329200), the Finnish Work Environment Fund heart disease among employees: the Whitehall II Study.
Arch Intern Med 2005 Oct;165(19):2245–51. https://doi.
(grant 200097), Government’s Analysis, Assessment
org/10.1001/archinte.165.19.2245.
and Research Activities (grant VN/14606/2019), and
the Academy of Finland Strategic Research Council 8. Demerouti E, Bakker AB, Nachreiner F, Schaufeli WB. The
(grant 336004). job demands-resources model of burnout. J Appl Psychol
2001 Jun;86(3):499–512. https://doi.org/10.1037/0021-
9010.86.3.499.
Disclosures
9. Bakker AB, Demerouti E. Job demands-resources theory:
AJC is an employee at Novo Nordisk A/S and moved taking stock and looking forward. J Occup Health
to Novo Nordisk at the end of the project. Her current Psychol 2017 Jul;22(3):273–85. https://doi.org/10.1037/
ocp0000056.
employer has no role in the study design, analyses and
results interpretation. No other potential conflicts of 10. Mattke S, Liu H, Caloyeras J, Huang CY, Van Busum KR,
interest relevant to this article were declared. Khodyakov D et al. Workplace Wellness Programs Study:

Scand J Work Environ Health 2022, vol 48, no 8 629

Workplace resources and cardiovascular disease

final Report. Rand Health Q 2013 Jun;3(2):7. 22. Kuper H, Adami HO, Theorell T, Weiderpass E.
11. Finne LB, Christensen JO, Knardahl S. Psychological Psychosocial determinants of coronary heart disease in
and social work factors as predictors of mental distress: a middle-aged women: a prospective study in Sweden.
prospective study. PLoS One 2014 Jul;9(7):e102514. https:// Am J Epidemiol 2006 Aug;164(4):349–57. https://doi.
doi.org/10.1371/journal.pone.0102514. org/10.1093/aje/kwj212.

12. Gimeno D, Tabák AG, Ferrie JE, Shipley MJ, De Vogli 23. Nielsen K, Nielsen MB, Ogbonnaya C, Känsälä M, Saari E,
R, Elovainio M et al. Justice at work and metabolic Isaksson K. Workplace resources to improve both employee
syndrome: the Whitehall II study. Occup Environ well-being and performance: A systematic review and meta-
Med 2010 Apr;67(4):256–62. https://doi.org/10.1136/ analysis. Work Stress 2017;31(2):101–20. https://doi.org/10
oem.2009.047324. .1080/02678373.2017.1304463.

13. Magnusson Hanson LL, Virtanen M, Rod NH, Steptoe A, 24. Sørensen JK, Framke E, Clausen T, Garde AH, Johnsen NF,
Head J, Batty GD et al. Does inflammation provide a link Kristiansen J et al. Leadership Quality and Risk of Long-term
between psychosocial work characteristics and diabetes? Sickness Absence Among 53,157 Employees of the Danish
Analysis of the role of interleukin-6 and C-reactive protein Workforce. J Occup Environ Med 2020 Aug;62(8):557–65.
in the Whitehall II cohort study. Brain Behav Immun 2019 https://doi.org/10.1097/JOM.0000000000001879.
May;78:153–60. https://doi.org/10.1016/j.bbi.2019.01.017.
25. Joensuu M, Kivimäki M, Pentti J, Virtanen M, Väänänen
14. Xu T, Clark AJ, Pentti J, Rugulies R, Lange T, Vahtera J et A, Vahtera J. Components of job control and mortality:
al. Characteristics of workplace psychosocial resources and the Finnish Public Sector Study. Occup Environ Med
risk of diabetes: a prospective cohort study. Diabetes Care 2014 Aug;71(8):536–42. https://doi.org/10.1136/
2022 Jan;45(1):59–66. https://doi.org/10.2337/dc20-2943 oemed-2014-102111.

15. Nakata A, Irie M, Takahashi M. Source-specific social 26. Magnusson Hanson LL, Leineweber C, Persson V, Hyde
support and circulating inflammatory markers among white- M, Theorell T, Westerlund H. Cohort profile: the Swedish
collar employees. Ann Behav Med 2014 Jun;47(3):335–46. longitudinal occupational survey of health (SLOSH).
https://doi.org/10.1007/s12160-013-9555-x. Int J Epidemiol 2018 Jun;47(3):691–692i. https://doi.
org/10.1093/ije/dyx260.
16. Karlin WA, Brondolo E, Schwartz J. Workplace social
support and ambulatory cardiovascular activity in 27. Xu T, Magnusson Hanson LL, Clark AJ, Ersbøll AK,
New York City traffic agents. Psychosom Med 2003 Westerlund H, Madsen IE et al. Onset of Workplace Bullying
Mar-Apr;65(2):167–76. https://doi.org/10.1097/01. and Risk of Weight Gain: A Multicohort Longitudinal Study.
PSY.0000033122.09203.A3. Obesity (Silver Spring) 2020 Nov;28(11):2216–23. https://
doi.org/10.1002/oby.22956.
17. Wager N, Fieldman G, Hussey T. The effect on
ambulatory blood pressure of working under favourably 28. Kouvonen A, Kivimäki M, Vahtera J, Oksanen T, Elovainio
and unfavourably perceived supervisors. Occup Environ M, Cox T et al. Psychometric evaluation of a short measure
Med 2003 Jul;60(7):468–74. https://doi.org/10.1136/ of social capital at work. BMC Public Health 2006
oem.60.7.468. Oct;6(1):251. https://doi.org/10.1186/1471-2458-6-251.

18. Roth GA, Johnson C, Abajobir A, Abd-Allah F, Abera SF, 29. Anderson N, West MA, editors. Team climate inventory:
Abyu G et al. Global, Regional, and National Burden of Manual and user’s guide1994: ASE.
Cardiovascular Diseases for 10 Causes, 1990 to 2015. J Am 30. Clausen T, Madsen IE, Christensen KB, Bjorner JB, Poulsen
Coll Cardiol 2017 Jul;70(1):1–25. https://doi.org/10.1016/j. OM, Maltesen T et al. The Danish Psychosocial Work
jacc.2017.04.052. Environment Questionnaire (DPQ): Development, content,
reliability and validity. Scand J Work Environ Health. 2019
19. Nyberg A, Alfredsson L, Theorell T, Westerlund H, Vahtera
Jul 1;45(4):356–69. https://doi.org/10.5271/sjweh.3793.
J, Kivimäki M. Managerial leadership and ischaemic heart
disease among employees: the Swedish WOLF study. Occup 31. Lehto AM. Quality of working life and equity. Helsinki,
Environ Med 2009 Jan;66(1):51–5. https://doi.org/10.1136/ Finland: Statistics Finland. 1991.
oem.2008.039362.
32. Setterlind S, Larsson G. The stress profile: A psychosocial
20. Clays E, Casini A, Van Herck K, De Bacquer D, Kittel F, approach to measuring stress. Stress Med 1995;11(1):85–92.
De Backer G et al. Do psychosocial job resources buffer https://doi.org/10.1002/smi.2460110116.
the relation between physical work demands and coronary 33. Elovainio M, Kivimäki M, Vahtera J. Organizational
heart disease? A prospective study among men. Int Arch justice: evidence of a new psychosocial predictor of health.
Occup Environ Health 2016 Nov;89(8):1299–307. https:// Am J Public Health 2002 Jan;92(1):105–8. https://doi.
doi.org/10.1007/s00420-016-1165-z. org/10.2105/AJPH.92.1.105.

21. André-Petersson L, Engström G, Hedblad B, Janzon 34. Lallukka T, Halonen JI, Sivertsen B, Pentti J, Stenholm
L, Rosvall M. Social support at work and the risk of S, Virtanen M et al. Change in organizational justice as
myocardial infarction and stroke in women and men. Soc a predictor of insomnia symptoms: longitudinal study
Sci Med 2007 Feb;64(4):830–41. https://doi.org/10.1016/j. analysing observational data as a non-randomized pseudo-
socscimed.2006.10.020. trial. Int J Epidemiol 2017 Aug;46(4):1277–84. https://doi.
org/10.1093/ije/dyw293.

630 Scand J Work Environ Health 2022, vol 48, no 8

 Xu et al

35. Greenland S, Pearl J, Robins JM. Causal diagrams for 42. Magnusson Hanson LL, Rod NH, Vahtera J, Peristera P,
epidemiologic research. Epidemiology 1999 Jan;10(1):37– Pentti J, Rugulies R et al. Multicohort study of change in
48. https://doi.org/10.1097/00001648-199901000-00008. job strain, poor mental health and incident cardiometabolic
disease. Occup Environ Med 2019 Nov;76(11):785–92.
36. Hulman A, Witte DR, Vistisen D, Balkau B, Dekker
https://doi.org/10.1136/oemed-2018-105595.
JM, Herder C et al. Pathophysiological characteristics
underlying different glucose response curves: a latent 43. Kim J. Strategic Human Resource Practices: Introducing
class trajectory analysis from the prospective EGIR-RISC Alternatives for Organizational Performance Improvement
study. Diabetes Care 2018 Aug;41(8):1740–8. https://doi. in the Public Sector. Public Adm Rev 2010;70(1):38–49.
org/10.2337/dc18-0279. https://doi.org/10.1111/j.1540-6210.2009.02109.x.

37. Barth J, Schneider S, von Känel R. Lack of social support 44. Campos-Serna J, Ronda-Pérez E, Artazcoz L, Moen
in the etiology and the prognosis of coronary heart disease: BE, Benavides FG. Gender inequalities in occupational
a systematic review and meta-analysis. Psychosom health related to the unequal distribution of working and
Med 2010 Apr;72(3):229–38. https://doi.org/10.1097/ employment conditions: a systematic review. Int J Equity
PSY.0b013e3181d01611. Health 2013 Aug;12(1):57. https://doi.org/10.1186/1475-
9276-12-57.
38. Kivimäki M, Jokela M, Nyberg ST, Singh-Manoux
A, Fransson EI, Alfredsson L et al. Long working 45. Jepsen DM, Rodwell JJ. Justice in the workplace: the centrality
hours and risk of coronary heart disease and stroke: a of social versus judgmental predictors of performance varies
systematic review and meta-analysis of published and by gender. Int J Hum Resour Manage 2009;20(10):2066–83.
unpublished data for 603,838 individuals. Lancet. 2015 Oct https://doi.org/10.1080/09585190903175639.
31;386(10005):1739–46. https://doi.org/10.1016/S0140-
6736(15)60295-1. 46. Hughes IM, Freier LM, Barratt CL. “Your help isn’t helping
me!” Unhelpful workplace social support, strain, and the
39. Kouvonen A, Oksanen T, Vahtera J, Väänänen A, De Vogli role of individual differences. Occup Health Sci 2022
R, Elovainio M et al. Work-place social capital and smoking Mar;1–37. https://doi.org/10.1007/s41542-022-00115-x.
cessation: the Finnish Public Sector Study. Addiction 2008 47. Xu T, Magnusson Hanson LL, Lange T, Starkopf L,
Nov;103(11):1857–65. https://doi.org/10.1111/j.1360- Westerlund H, Madsen IEH, et al. Workplace bullying and
0443.2008.02315.x. violence as risk factors for type 2 diabetes: a multicohort
study and meta-analysis. Diabetologia. 2018 Jan;61(1):75–
40. Colell E, Sánchez-Niubò A, Benavides FG, Delclos GL,
83. https://doi.org/10.1007/s00125-017-4480-3.
Domingo-Salvany A. Work-related stress factors associated
with problem drinking: A study of the Spanish working
population. Am J Ind Med 2014 Jul;57(7):837–46. https://
doi.org/10.1002/ajim.22333.

41. Griep RH, Nobre AA, Alves MG, da Fonseca MJ, Cardoso
LO, Giatti L et al. Job strain and unhealthy lifestyle: results
from the baseline cohort study, Brazilian Longitudinal Study
of Adult Health (ELSA-Brasil). BMC Public Health 2015
Received for publication: 14 March 2022
Mar;15(1):309. https://doi.org/10.1186/s12889-015-1626-4.

Scand J Work Environ Health 2022, vol 48, no 8 631