598404

research-article2015
SJP0010.1177/1403494815598404L.L. Andersen et al.Scandinavian Journal of Public Health

Scandinavian Journal of Public Health, 2015; 43: 810–818

Original Article

Effect of physical exercise on workplace social capital: Cluster

randomized controlled trial

Lars L. Andersen1,2, Otto M. Poulsen1, Emil Sundstrup1, Mikkel Brandt1,2,

Kenneth Jay1, Thomas Clausen1, Vilhelm Borg1, Roger Persson3 &
Markus D. Jakobsen1

1National Research Centre for the Working Environment, Denmark, 2Physical Activity and Human Performance group,

SMI, Department of Health Science and Technology, Aalborg University, Denmark, and 3Department of Psychology, Lund
University, Lund, Sweden

Abstract
Aims:While workplace health promotion with group-based physical exercise can improve workers’ physical health, less is
known about potential carry-over effects to psychosocial factors. This study investigates the effect of physical exercise on
social capital at work. Methods: Altogether, 200 female healthcare workers (nurses and nurse’s aides) from 18 departments
at three hospitals were randomly allocated at the department level to 10 weeks of (1) group-based physical exercise at work
during working hours or (2) physical exercise at home during leisure time. At baseline and follow-up, participants replied to
a questionnaire concerning workplace social capital: (1) within teams (bonding); (2) between teams (bridging); (3) between
teams and nearest leaders (linking A); (4) between teams and distant leaders (linking B). Results: At baseline, bonding,
bridging, linking A and linking B social capital were 74 (SD 17), 61 (SD 19), 72 (SD 22) and 70 (SD 18), respectively, on a
scale of 0–100 (where 100 is best). A group by time interaction was found for bonding social capital (P=0.02), where physical
exercise at work compared with physical exercise during leisure time increased 5.3 (95% confidence interval 2.3– 8.2)(effect
size, Cohen’s d = 0.31) from baseline to follow-up. For physical exercise at home during leisure time and exercise at work
combined, a time effect (P=0.001) was found for linking A social capital, with a decrease of 4.8 (95% confidence interval
1.9–7.6). Conclusions: Group-based physical exercise at work contributed to building social capital within teams
at the workplace. However, the general decrease of social capital between teams and nearest leaders during the
intervention period warrants further research.

Key Words: Social capital, human capital, physical activity, well-being, occupational, work-related, randomized trial, nursing

Introduction
Social capital is conceptualized by the Organization
for Economic Co-operation and Development as
informal networks that facilitate cooperation within
or among groups [1]. These networks – characterized
by shared norms, values and understandings –
include, for example, friends, crews and colleagues.
Social capital is a fairly new concept in occupational
safety and health research and in organizational psy-
chology research and can, despite many fluid concep-
tualizations, be seen as an extension of human capital.
A high level of social capital at workplaces is consid-
ered to be reflected in trust and cooperation among
colleagues, individual well-being and efficiency in
production processes [2]. More recent understand-
ings of workplace social capital focus on actual and
potential resources in the social relations at the work-
place that may have an impact on the production pro-
cesses [3]. However, relatively few longitudinal studies
on the association between social capital at work and
well-being, health and productivity exist [4–10].

Correspondence: Lars L. Andersen, National Research Centre for the Working Environment, Lersø Parkalle 105, Copenhagen, 2100, Denmark.
E-mail: lla@arbejdsmiljoforskning.dk

(Accepted 8 July 2015)

© 2015 the Nordic Societies of Public Health
DOI: 10.1177/1403494815598404

Downloaded from sjp.sagepub.com at WASHINGTON UNIV SCHL OF MED on November 16, 2015

Researchers have developed questionnaires to
measure social capital at workplaces [11,12] but none
of these differentiates between the different types of
social relationships that exist, i.e. bonding, bridging
and linking [13]. Therefore, Borg and co-workers
recently developed and validated a four-dimensional
questionnaire in Danish to measure social relations
within working teams (bonding), between working
teams (bridging), between teams and nearest leaders
(linking A) and between teams and distant leaders
(linking B) [3] . This questionnaire is used in the pre-
sent study to measure the four dimensions of social
capital among healthcare workers at hospitals.
Bonding, bridging and linking refer to distinct types
of social capital and may therefore differentially affect
well-being and productivity in workplaces.
Researchers have investigated the qualitative char-
acteristics of social capital at work [2,14] and a variety
of studies have investigated methods for improving
different aspects of social capital [15–20]. However,
all these approaches attempt to intervene directly and
broadly on work-related social relationships between
employees in teams and between employees and their
superiors. In the present study, we use an indirect
approach by investigating whether improved social
capital emerges in situations that are not related to
the daily work tasks, that is, physical exercise in work
groups. We have previously performed workplace
studies using group-based physical exercise with the
aim of reducing or preventing musculoskeletal disor-
ders. Although neither of these focused on building
social capital, we observed from simple retrospective
questions at follow-up, indications of enhanced social-
izing between colleagues in the exercise groups
[21,22]. These observations led us to investigate
whether social capital at work can increase through
group-based activities such physical exercise.
The aim of this study is to investigate the effect of
physical exercise on workplace social capital in terms
of bonding, bridging and linking. The study uses a
cluster randomized controlled trial design with two
arms – group-based physical exercise at work vs.
home-based physical exercise during leisure time
among healthcare workers at hospitals. We hypothe-
sized that interactions within the groups during phys-
ical exercise at work, in contrast to physical exercise
at home, would improve social capital at work.
Methods
Study design and randomization
This article presents secondary analyses of a trial
focusing on prevention of musculoskeletal disorders.
The study protocol [23] and primary outcome [24]
are reported elsewhere. In brief, we performed a
Downloaded from sjp.sagepub.com at WASHINGTON UNIV SCHL OF MED on November 16, 2015
Scandinavian Journal of Public Health 811
cluster randomized controlled trial with allocation
concealment among 200 female healthcare workers
(nurses and nurse’s aides) from 18 departments at
three hospitals in Copenhagen, Denmark between
August 2013 and January 2014. Each participating
department was defined as a cluster. Using a com-
puter-generated random numbers table in the SAS
statistical software, each department was randomly
allocated to 10 weeks of (1) group-based physical
exercise at work during working hours (WORK,
n=111 in nine clusters) or (2) physical exercise at
home (HOME, n=89 in nine clusters). Immediately
after randomization, participants at each department
and their management were informed by e-mail
about group allocation. The relative participation in
each cluster (i.e. the percentage of workers rand-
omized relatively to the total number of workers at
each department) ranged from 18 to 76% in WORK
and 17 to 71% in HOME. At baseline and follow-up,
participants replied to a questionnaire concerning
social capital in four dimensions: (1) within teams
(bonding); (2) between teams (bridging); (3) between
teams and nearest leaders (linking A); (4) between
teams and distant leaders (linking B). Table I shows
baseline characteristics of the 200 participants.
Participant eligibility and flow
Figure 1 shows the flow of participants through the
trial. The screening questionnaire was administered
by e-mail to 490 healthcare workers (aged 18–67
years) from three Danish hospitals. In total, 314
replied, of which 275 were interested in participating.
Of the 314 respondents, there were 17 men. The ini-
tial inclusion criteria based on the screening ques-
tionnaire were non-pregnant female healthcare
workers with no current cardiovascular or other life-
threatening diseases. Of the 275 interested respond-
ents, 253 met the above criteria and were invited for
a baseline clinical examination, to which 207 showed
up. Exclusion criteria of the clinical exam were: (1)
hypertension (systolic blood pressure (BP) >160,
diastolic BP >100); (2) a medical history of cardio-
vascular diseases (e.g. chest pain during physical
exercise, heart failure, myocardial infarction and
stroke); (3) a medical history of life-threatening dis-
ease; (4) current pregnancy. During the baseline clin-
ical examination and questionnaire survey, seven
workers were excluded (N=200) due to contraindi-
cations: five due to high blood pressure and two due
to blood clot incidence within the last two years.
Ethical approval and trial registration
The study was approved by The Danish National
Ethics Committee on Biomedical Research (Ethical
812 L.L. Andersen et al.
Table I. Demographics, musculoskeletal pain, work and social capital of the 200 female healthcare workers in the HOME and WORK
groups at baseline.

HOME WORK

Mean SD Mean SD

Number of participants 89 111

Number of clusters 9 9
Demographics
Age, year 44 10 40 12
Height, cm 168 7 168 6
Weight, kg 69 12 68 12
Body Mass Index, kg.m–2 24 4 24 4
Musculoskeletal
Low back pain lasting 7 days (0–10) 3.0 2.7 2.9 2.7
Work-related
Weekly working hours 34 4 35 4
Duration of healthcare work, years 18 11 15 11
Social capital at work (0–100)
Bonding 73 15 74 18
Bridging 61 19 61 20
Linking A 74 20 70 23
Linking B 71 17 69 19

committee of Frederiksberg and Copenhagen; H-3-
2010-062) and registered in ClinicalTrials.gov
(NCT01921764) prior to enrolment of participants.
Interventions
Participants in WORK performed strength training
during working hours together with colleagues from
the same department and the training was per-
formed at their respective departments. Thus, they
did not train together with colleagues from other
departments. The sessions were conducted in a cir-
cuit-training manner using kettlebells, Swiss balls
and elastic resistance bands (Thera-Band) to
strengthen the legs, back, neck and shoulders.
Designated rooms located at or close to the depart-
ments were equipped prior to the first training ses-
sion. All sessions were supervised by training
instructors. The training instructor was present at
the department at scheduled times each day from
Monday to Friday during the 10 weeks and partici-
pants were encouraged to participate in the training
sessions 5 × 10 minutes per week. In addition, par-
ticipants were offered a total of five group-based
coaching sessions of 30–45 minutes during the 10
weeks with the main goal of increasing motivation
for adherence to the training.
Participants in HOME performed strength train-
ing during leisure time at home. Participants received
a bag with elastic bands and three posters showing
exercises for the shoulder, abdominal and back mus-
cles and recommendations for training technique
and progression. Participants in HOME were also
encouraged to perform the exercises 5 × 10 minutes
per week. This group did not receive supervision
from a training instructor or coach.
Outcomes
At baseline and 10-week follow-up, participants
replied to a questionnaire concerning bonding,
bridging, linking A and linking B social capital. The
questionnaire has been developed and validated
against qualitative interviews by Thomas Clausen
and Vilhelm Borg at the National Research Centre
for the Working Environment in Copenhagen,
Denmark [3]. Two sample questions out of nine
questions for bonding social capital are: ‘In our
team, we agree on what is the most important in our
work tasks’; ‘There is a feeling of unity and cohesion
in my team’. Two sample questions of a total of six
for bridging social capital are: ‘Is there a good work-
ing relationship between your team and the other
teams/departments?’; ‘We have trust in the ability of
the other teams to do the job well’. Two sample
questions out of six questions for linking A social
capital are: ‘Does your nearest leader contribute to
solving everyday problems’; ‘Our nearest leader has
great knowledge and understanding of the work we
do’. Two sample questions out of four questions for
linking B social capital are: ‘Are the employees
involved in decisions about changes at the work-
place?’; ‘There is a common understanding between
the management and employees on how we should
perform our work tasks’.
Participants replied on a horizontally oriented scale
of 0–10, where 0 is ‘no, not at all’ and 10 is ‘Yes, com-
pletely’. For each of the four social capital dimensions,

Downloaded from sjp.sagepub.com at WASHINGTON UNIV SCHL OF MED on November 16, 2015
 Scandinavian Journal of Public Health 813

Figure 1. Flow of participants through the trial. All participants – including those with missing follow-up data – were included in the final
analyses by using the mixed procedure that inherently accounts for missing values.

the average value of all questions was calculated and during the intervention we chose a 0–10 scale. Prior to
multiplied by 10 (i.e. 0–100). The original scales the intervention we sent out a pilot questionnaire
developed by Borg and Clausen were 5-point scales (n=39) to test the association between the 5-point and
[3], but in relation to detecting possibly small changes 0–10 scale, and found standardized Cronbach’s alpha

Downloaded from sjp.sagepub.com at WASHINGTON UNIV SCHL OF MED on November 16, 2015
814 L.L. Andersen et al.
Table II. Changes in social capital from baseline to 10-week follow-up (differences of least square means and 95% CI).
Within-group change from baseline to follow-up
HOME WORK
Bonding −1.0 (−4.0 to 2.0) 3.8
Bridging 0.1 (−3.9 to 4.1) −1.3
Linking A −4.7 (−8.9 to −0.5) −4.9
Linking B −0.8 (−4.3 to 2.7) −3.2
Note: There was a significant group by time interaction for bonding social capital (i.e. within teams) (P=0.02) and a significant time effect
for linking A social capital (i.e. between teams and nearest leaders) (P=0.001). Numbers marked in bold indicate statistically significant
changes.
of 0.90 (bonding social capital), 0.87 (bridging social
capital), 0.91 (linking A social capital) and 0.88 (link-
ing B social capital).
Sample size calculation
The sample size calculation was based on the primary
outcome (average pain intensity of the back and neck–
shoulder) reported elsewhere and showed that at least
80 participants should be included in each group to
achieve 95% statistical power to detect a between-
group difference of 1 (scale 0–10) in pain intensity at
a P-level of 0.05. We did not perform an a priori sam-
ple size calculation for the outcomes in this article.
Statistical analyses
The change in social capital was evaluated using a lin-
ear mixed model with repeated measures with group
(WORK, HOME), time (baseline, 10-week follow-
up) and group by time (interaction effect) as inde-
pendent variables. Participant nested within
department (cluster) was entered as random effect.
Analyses were adjusted for age and social capital at
baseline. We performed all statistical analyses in
accordance with the intention-to-treat (ITT) principle
using the mixed procedure (Proc Mixed, SAS version
9.3) and including all available data in the analyses,
i.e. including participants with missing follow-up data.
The mixed procedure inherently accounts for missing
values. The estimation method was restricted maxi-
mum likelihood with degrees of freedom based on the
Kenward–Roger approximation. P-levels of 0.05 or
less were accepted as statistically significant. Outcomes
are reported as least mean square differences with
95% confidence intervals (CI).
For outcomes showing significant group by time
interaction, effect sizes (Cohen’s d) were calculated
as the between-group difference at follow-up divided
by the pooled SD at baseline [25]. According to
Cohen, effect sizes of 0.20, 0.50 and 0.80 can be
considered small, moderate and large, respectively.
Downloaded from sjp.sagepub.com at WASHINGTON UNIV SCHL OF MED on November 16, 2015
Between-group difference at follow-up
WORK vs. HOME
(1.1 to 6.6) 5.3 (2.3 to 8.2)
(−5.0 to 2.4) −1.7 (−5.7 to 2.2)
(–8.7 to –1.0) −0.7 (−4.8 to 3.5)
(−6.4 to 0.0) −2.6 (−6.0 to 0.9)
An additional exploratory dose–response analysis
for association between adherence to training (aver-
age number of training sessions per week) and
changes in bonding social capital were performed
using general models (Proc Genmod, SAS version
9.3). Participant nested within department (cluster)
was entered as repeated effect and the analysis was
controlled for baseline bonding social capital.
Results
At baseline, bonding, bridging, linking A and linking
B social capital were 74 (SD 17), 61 (SD 19), 72 (SD
22) and 70 (SD 18) respectively, on a scale of 0–100
(where 100 is best). Table I shows demographics
and social capital for each intervention group at
baseline.
The number of training sessions per week in
WORK and HOME were 2.2 (SD 1.1) and 1.0 (SD
1.2) (averaged for the 10 weeks).
Table II shows results from the randomized con-
trolled trial. A group by time interaction was found
for bonding social capital (P=0.02), where WORK
compared with HOME increased 5.3 (95% CI 2.3 to
8.2) from baseline to 10-week follow-up, which cor-
responds to an effect size of 0.31. For HOME and
WORK combined, a time effect (P=0.001) was found
for linking A social capital, with a decrease of 4.8
(95% CI: 1.9 to 7.6). There was also a tendency for a
time effect (P=0.09) for decreased linking B social
capital.
We also plotted associations at the department
(i.e. cluster) level between some of the outcomes.
Figure 2 shows the change in bonding social capital
vs. the change in linking A social capital from base-
line to follow-up for each cluster in WORK and
HOME. This plot reveals that most clusters in
WORK had a positive development on bonding
social capital (seven out of nine teams). Hence, the
positive effect in WORK seems to be consistent
across departments and not due to large effects in a
few clusters. Figure 2 further shows that: (1) there is

Figure 2. Change in bonding social capital (x-axis) vs. change in linking A social capital (y-axis) for clusters in intervention groups WORK
and HOME (scale −100 to 100). Size of the dots reflects cluster size.
no clear association between the change over time in
bonding and linking A social capital, indicating that
these social capital dimensions responded differently
to the intervention; (2) the change over time in link-
ing A social capital for most clusters of both WORK
and HOME is negative, indicating that the decrease
in linking A social capital was not driven by a few
large departments.
Figure 3 shows the dose–response association
between adherence to physical exercise at the cluster
level and changes in bonding social capital from
baseline to follow-up (non-significant in both
groups). An interesting observation from this plot is
the small department at the upper left corner. This
cluster had, in spite of no participation in the physi-
cal exercise sessions at all, an increase of more than
30 points in bonding social capital from baseline to
follow-up (three participants). After seeing these
results, we contacted the hospital and were told that
10 staff members of this department were dismissed
during the study period. Performing an additional
exploratory analysis, excluding this particular cluster,
we found a highly significant positive association
between adherence to the physical exercise and the
change in bonding social capital. For each additional
exercise session per week (average for the 10 weeks),
the change in bonding social capital was 3.5 (95% CI
1.2–6.0) in WORK, while no such association was
found for HOME (1.2 [95% CI −1.0 to 3.4]). Also,
the main findings of improved bonding social capital
in WORK compared with HOME remained intact
Downloaded from sjp.sagepub.com at WASHINGTON UNIV SCHL OF MED on November 16, 2015
Scandinavian Journal of Public Health 815
when excluding this cluster in an additional explora-
tory analysis (results not shown).
Finally, at 10-week follow-up, participants replied
to a question asking in retrospect whether they had
received support from their leader to participate in
the physical exercise. In WORK, 79%, 12% and 9%
replied ‘Yes, to a high extent’, ‘Yes, to some extent’
and ‘No’, respectively. Corresponding values for
HOME were 25%, 27%, and 48%, respectively. This
distribution was significantly different between
WORK and HOME (Chi-square, P<0.001).
There were no reported injuries from the training
sessions.
Discussion
Our study showed that group-based physical exercise
at work contributes to building social capital within
teams at the workplace. However, this was seen along
with a general decrease in social capital between
teams and nearest leaders during the 10-week inter-
vention period.
Bonding social capital – i.e. within teams at the
workplace – increased 5 points on a scale of 0– 100 in
WORK compared with HOME. Other studies inter-
vening directly on organizational factors in the con-
ceptual proximity of social capital at work – i.e.
interventions aimed directly at enhancing the work
processes in work groups – found a positive impact on
the collaborative patterns and performance of the
work groups [16,18,20]. In the present study, the
816 L.L. Andersen et al.

Figure 3. Adherence, i.e. average number of training sessions per week (x-axis) and changes in bonding social capital (y-axis) for clusters
in intervention groups WORK and HOME (scale −100 to 100). Size of the dots reflects cluster size. Excluding the WORK cluster in the
upper left corner (n=3) who experienced dismissal of 10 employees and did not participate in training at all, there was a significant asso-
ciation between adherence and the change in bonding social capital in WORK. For each additional exercise session per week (average for
the 10 weeks), the change in bonding social capital was 3.5 (95% CI 1.2–6.0) in WORK, while no such association was found for HOME.

physical exercise programme was designed to
strengthen physical fitness of the healthcare workers;
thus, any effect on social capital can be considered
indirect. Based on the present findings, we cannot
reveal the underlying mechanisms of the improvement
in social capital. However, in a previous study with
group-based workplace physical exercise among office
workers [26], one participant explained in a subse-
quent interview performed by the Danish Knowledge
Centre for the Working Environment (Videncenter for
Arbejdsmiljø) that:
‘it’s funny doing it together (i.e. the exercise), talking
about things which have nothing to do with our work’.
Another explained:
‘We see each other in another role than usual. It’s
liberating in the sense that we use our bodies. It’s nice to
feel that you laugh a little at them and they laugh at you.
It’s at different atmosphere than what may be
professionally. It catches on when you just have laughed
at something together and we later may have something
where we come across each other, then it may be
easier to get over it.’ (http://www.youtube.com/watch?
v=XtZU-b1x400)
Thus, according to this narrative, performing physi-
cal exercise together in groups at the workplace may
feel liberating, enhance the social relationships between
team members and indirectly facilitate cooperation
during work tasks among members of the work team.
Another study mapping help preferences among
employees in the Danish police force found that the
option of exercising in a social/collegial context was the
third most common preference (31%) after the possi-
bility to do physical exercise at work (46%) and offers
of free exercise (44%), respectively [27]. Possible
underlying factors of the present results may be a com-
bination of: (1) positive feeling of being selected; (2)
improved intercommunication and collaboration
about work tasks (behavioural processes); (3) increased
shared mental models (cognitive processes); (4)
improved solidarity and collective self-efficacy (moti-
vational processes); (5) enhanced quality of relations
(trust, respect, recognition). Another influential mech-
anism may be the creation and categorization of
employees belonging to a ‘physical exercise group’
introducing so called ‘in-group bias’ [28–30]. From
this perspective, positive relationships within the group
(i.e. bonding) are strengthened and maintained by the
process of social comparisons with people outside the
group, such as leaders and other teams. An exploratory
dose–response analysis also suggested that higher
adherence to physical exercise in the WORK group was
associated with a larger increase in bonding social capi-
tal, suggesting that the amount of time spent together

Downloaded from sjp.sagepub.com at WASHINGTON UNIV SCHL OF MED on November 16, 2015

doing such activities is important. As expected, this
dose–response association was not present for the
HOME group as they trained alone at home.
Bridging social capital – i.e. between teams at the
workplace – remained unchanged. This would also
be expected as training was performed in groups with
colleagues from the same department and no inter-
action due to the project was facilitated between dif-
ferent departments. Thus, training was performed in
designated training rooms located at or close to the
departments. The finding of unchanged bridging
social capital during the study period also suggests
that bonding within groups is not occurring by dero-
gating other teams.
Linking A social capital – i.e. between teams and
the nearest leader – decreased five points in both
groups combined from baseline to follow-up, with no
significant between-group difference. Although not
measured, it can be speculated that participants of
both groups may have had high expectations of sup-
port from their nearest leader to participate actively
in the project. The context of the project was that
employees could participate if they could find the
time, i.e. the nearest leader did not allocate addi-
tional personnel to help participating departments
fulfil their work tasks. An imbalance between expec-
tations and realities may have caused a general
decrease in social capital between teams and nearest
leaders, which again may have fuelled discussions
within the teams, making the social comparisons
towards the nearest leader more relevant. On the
other hand, this change may not at all be related to
the workplace activities in the project, as one group
trained at home. Furthermore, based on the follow-
up questionnaire replies, 79% of the participants in
WORK experienced a high degree of support from
their leader to participate. As expected, this was not
the case for the HOME group, most likely because
they exercised at home and their leader was not
expected to be involved. However, because we can
only speculate about the mechanisms of the present
findings, future workplace intervention studies
should investigate this aspect in more detail. If work-
place interventions are capable of causing negative
effect on social capital between teams and leaders,
this should be taken seriously and considered an
adverse organizational event.
Strengths and limitations
The cluster randomized controlled trial design pro-
tected against contamination between intervention
groups and allowed for a more ‘real-world’ setting,
where colleagues from the same department had the
opportunity to participate in the same intervention.
Although participants from different departments
Downloaded from sjp.sagepub.com at WASHINGTON UNIV SCHL OF MED on November 16, 2015
Scandinavian Journal of Public Health 817
and intervention groups may have met and talked
about the project, e.g. during lunch breaks, the risk of
between-group contamination is far less in cluster
randomized trials than individually randomized tri-
als. There was a low loss of participants to follow-up
and all randomized participants were included in the
ITT analyses, which strengthens the validity of the
estimated effects. A limitation is that blinding of par-
ticipants was not possible due to the behavioural
intervention design. To minimize the potential bias
from lack of blinding, we offered participants of both
groups active interventions – i.e. work vs. home-
based training. Further, as the intervention focused
on physical exercise and prevention of musculoskel-
etal disorders, neither of the intervention groups may
have had any particular outcome expectations regard-
ing changes in social capital. A limitation is that we
cannot conclude whether the decrease in social capi-
tal between teams and nearest leaders was caused by
the project or some other underlying factors at the
hospitals. Based on this single study, the generaliza-
bility of the results is for now limited to female
healthcare workers.
Conclusion
In conclusion, group-based physical exercise at work
contributed to building social capital within teams at
the workplace. However, the general decrease of
social relationships between teams and nearest lead-
ers during the intervention period warrants further
research.
Funding
This study was supported by a grant from the Danish
Parliament (SATS 2004) and the Danish Working
Environment Research Fund (Grant no. 48-2010-
03). These sponsors had no role in the study other
than providing funding.
Conflict of interest
None declared.
References
[1] OECD. The Wellbeing of Nations: The Role of Human and
Social Capital, Education and Skills. Paris, France: OECD
Centre for Educational Research and Innovation, 2001.
[2] Hasle P and Møller N. From conflict to shared develop-
ment: social capital in a Tayloristic environment . Econ Ind
Democracy 2007;28:401–29.
[3] Borg V, Mateu NC and Clausen T. Udvikling af en ny metode
til undersøgelse af social kapital på arbejdspladsen. Copenhagen:
The National Research Centre for the Working Environ-
ment, 2014.
[4] Kompier MA and Taris TW. Understanding the causal
relations between psychosocial factors at work and health
- a circular process. Scand J Work Environ Health 2011;
37:259–61.
818 L.L. Andersen et al.
[5] Oksanen T, Kouvonen A, Vahtera J, et al. Prospective study
of workplace social capital and depression: are vertical and
horizontal components equally important? J Epidemiol Com-
munity Health 2010;64:684–9.
[6] Vaananen A, Kouvonen A, Kivimaki M, et al. Work-
place social capital and co-occurrence of lifestyle risk fac-
tors: the Finnish Public Sector Study. Occup Environ Med
2009;66:432–7.
[7] Kouvonen A, Oksanen T, Vahtera J, et al. Low workplace
social capital as a predictor of depression: the Finnish Public
Sector Study. Am J Epidemiol 2008;167:1143–51.
[8] Oksanen T, Kouvonen A, Kivimaki M, et al. Social capital at
work as a predictor of employee health: multilevel evidence
from work units in Finland. Soc Sci Med 2008;66:637–49.
[9] Lindstrom M. Psychosocial work conditions, social partici-
pation and social capital: a causal pathway investigated in a
longitudinal study. Soc Sci Med 2006;62:280–91.
[10] Liukkonen V, Virtanen P, Kivimaki M, et al. Social capital
in working life and the health of employees. Soc Sci Med
2004;59:2447–58.
[11] Kouvonen A, Kivimaki M, Vahtera J, et al. Psychometric
evaluation of a short measure of social capital at work. BMC
Public Health 2006;6:251.
[12] Olesen KG, Thoft E, Hasle P, et al. Virksomhedens Sociale
Kapital. Copenhagen: The Danish Working Environment
Council & The National Research Centre for the Working
Environment, 2008.
[13] Szreter S and Woolcock M. Health by association? Social
capital, social theory, and the political economy of public
health. Int J Epidemiol 2004;33:650–67.
[14] Nilsson K, Hertting A, Petterson IL, et al. Pride and confi-
dence at work: potential predictors of occupational health in
a hospital setting. BMC Public Health 2005;5:92.
[15] Long JC, Cunningham FC and Braithwaite J. Bridges, bro-
kers and boundary spanners in collaborative networks: a sys-
tematic review. BMC Health Serv Res 2013;13:158.
[16] Gittell JH, Godfrey M and Thistlethwaite J. Interprofes-
sional collaborative practice and relational coordination:
improving healthcare through relationships. J Interprof Care
2013;27:210–3.
[17] Kukkurainen ML, Suominen T, Rankinen S, et al. Organi-
zational vision: experience at the unit level. J Nurs Manag
2012;20:868–76.
Downloaded from sjp.sagepub.com at WASHINGTON UNIV SCHL OF MED on November 16, 2015
[18] Buljac-Samardzic M, Dekker-van Doorn CM, van Wijn-
gaarden JD, et al. Interventions to improve team effective-
ness: a systematic review. Health Policy 2010;94:183–95.
[19] Stajkovic AD, Lee D and Nyberg AJ. Collective efficacy,
group potency, and group performance: meta-analyses of
their relationships, and test of a mediation model. J Appl
Psychol 2009;94:814–28.
[20] Salas E, DiazGranados D, Klein C, et al. Does team training
improve team performance? A meta-analysis. Hum Factors
2008;50:903–33.
[21] Mortensen P, Larsen AI, Zebis MK, et al. Lasting effects
of workplace strength training for neck/shoulder/arm pain
among laboratory technicians: natural experiment with
3-year follow-up. Biomed Res Int 2014;2014:845851.
[22] Jay K, Frisch D, Hansen K, et al. Kettlebell training for
musculoskeletal and cardiovascular health: a randomized
controlled trial. Scand J Work Environ Health 2011;37:
196–203.
[23] Jakobsen MD, Sundstrup E, Brandt M, et al. Effect of
workplace- versus home-based physical exercise on pain in
healthcare workers: study protocol for a single blinded clus-
ter randomized controlled trial. BMC Musculoskelet Disord
2014;15:119.
[24] Jakobsen MD, Sundstrup E, Brandt M, et al. Effect of work-
place- versus home-based physical exercise on musculoskel-
etal pain among healthcare workers: a cluster randomized
controlled trial. Scand J Work Environ Health 2015;41(2):
153–163.
[25] Cohen J. Statistical Power Analysis for the Behavioral Sciences,
2nd ed. New Jersey: Lawrence Erlbaum Associates, 1988.
[26] Andersen LL, Saervoll CA, Mortensen OS, et al. Effective-
ness of small daily amounts of progressive resistance train-
ing for frequent neck/shoulder pain: randomised controlled
trial. Pain 2011;152:440–6.
[27] Persson R, Cleal B, Jakobsen MO, et al. Help preferences
among employees who wish to change health behaviors.
Health Educ Behav 2013;41:376–86.
[28] Brewer MB. The importance of being we: human nature and
intergroup relations. Am Psychol 2007;62:726–38.
[29] Hewstone M, Rubin M and Willis H. Intergroup bias. Annu
Rev Psychol 2002;53:575–604.
[30] Tajfel H. Experiments in Intergroup Discrimination. Sci Am
1970;223:96–102.