Professional Documents
Culture Documents
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/239615077
CITATIONS
READS
15
434
5 authors, including:
Richard Graveling
SEE PROFILE
SEE PROFILE
Hilary Cowie
Salim Vohra
3 PUBLICATIONS 45 CITATIONS
SEE PROFILE
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Salim Vohra on 16 December 2013.
The user has requested enhancement of the downloaded file.
Draft Report
13th February 2008
Executive Summary
Background
According to the most recently available Labour Force Survey for 2004-2005,
509,000 individuals in the UK believed they were suffering from stress,
depression or anxiety and these conditions were caused or made worse by
their current or past work. This resulted in an estimated 12.8 million working
days lost with an average of 30.9 days off over a 12 month period. Those
employed in the public sector had some of the highest rates of self-reported
stress, anxiety and depression. In particular those with heavy workloads, tight
deadlines, lack of support at work, in a threatening environment are
recognised as being at an increased risk of stress, depression or anxiety. This
review aims to determine which workplace interventions are effective and
cost-effective in improving mental wellbeing.
Methods
The review considered studies that assessed the effectiveness of specific
workplace interventions aimed at either promoting or improving mental
wellbeing. The concepts involved encompass a wide variety of topics feeding
into the term mental wellbeing. Nineteen databases and twenty-four websites
were searched for relevant research including systematic reviews, reviews
and original research papers.
filtered and those that were irrelevant were discarded. Abstracts were then
screened and papers were obtained for those that fitted the inclusion criteria.
Data extraction and quality assessment were carried out with studies rated as
++, + and -. Results of the data extraction and quality assessment for each
study were presented in structured tables accompanied by a narrative
summary.
Review of Effectiveness
The review of the effectiveness of workplace interventions to improve mental
wellbeing included 66 primary studies. A broad range of interventions were
ii
iii
The
Personal
iv
personal
accomplishment
depersonalisation
(p=0.01).
(p=0.01),
These
exhaustion
small
studies
(p=0.04)
and
indicate
that
concentrate on measures which the individual can control and others include
a degree of organisational intervention.
regarding this are made with caution because, as stated above, these
individual elements have not been formally evaluated for their efficacy.
vi
Evidence Statement 5
Eight studies that were graded positively evaluated different types of
stress-management training1-8 six studies found a positive impact on
mental wellbeing as measured by questionnaire. One Australian
randomised trial found a positive effect that was close to but not
statistically significant (Lindquist et al, 1999 ++) and one study with 54
volunteer German bus drivers (Aust et al 1997) found no significant
effects. The differences amongst studies in interventions, populations
and study quality mitigate against definitive conclusions. However there
is reasonable evidence that multi-faceted training, covering stress
awareness, coping and stress reduction is an effective format.
Six of the eight studies had training programmes involving a trainer or
facilitator of which four found a positive impact on mental wellbeing,
again measured by questionnaire. Two small randomised control trials
(Horan et al. 2002 +) and (Rahe et al. 2002 +) found that small group
sessions have a positive impact on mental wellbeing.
There is evidence from one randomised trial undertaken in the USA
(Cook 2007++) that compared web materials with paper based materials
that paper based training materials are more effective for improving
mental wellbeing.
Counselling and therapy
There is limited research carried out on the impact of counselling and therapy
interventions that are specifically aimed at improving mental wellbeing in the
workplace.
vii
Evidence Statement 6
A UK randomised control trial (Bond et al. 2000 +) with 90 volunteers
from a media company found that three half-day sessions of therapy
and counselling delivered during work time had a positive impact on
mental wellbeing in the short term as measured by questionnaire. A UK
randomised trial with 24 in both the intervention and control groups who
were NHS and Local Authority workers with 10 or more days absence
due to stress, anxiety or depression in the previous 6 months (Grime et
al. 2004 +) found that eight weekly sessions using a computerised
Cognitive Behavioural Therapy programme had a positive impact on
mental wellbeing in the short term as measured by questionnaire.
Exercise and relaxation interventions
The evaluation of interventions using exercise was based on 4 RCTs ranging
in quality from + to ++. Three of the interventions involved aerobic sessions
over periods between 8 and 24 weeks. The results in two of the studies
indicate that aerobic exercise has a positive effect on mental wellbeing. Of
the two studies that were found not to be effective one showed confounding
so the results were unclear (Van Rhenen, 2005) and the other took the form
of a shorter duration office based exercise intervention for those working with
VDUs.
The relaxation training interventions reviewed were a mix of RCT and nonRCT studies ranging in quality between - and ++.
inconclusive in that two did not find an impact on mental wellbeing whereas
one, rated as -, found an improvement in outcome measures.
There is
Two studies evaluated the impact of massage therapy, one a RCT (Field,
1997) rated as ++ and one a Non-RCT rated as ++ (Shulman and Jones,
1996). The studies contradicted each other with the RCT finding no effect
and the Non-RCT finding an impact on the STAI. There were a number of
viii
issues with regard to the Field (1997) study which made it difficult to
determine which, if any, of the multiple interventions had an impact. This is
counter-balanced by the Shulman and Jones (1996) study which evaluated
the intervention immediately post massage thus there is no clarity with regard
to the longer term impact of the massage. There is currently insufficient data
available to support or refute the usefulness of massage therapy in promoting
mental wellbeing.
ix
Evidence Statement 8
A randomised control trial undertaken in Sweden (Hasson et al. 2005 ++)
with 129 receiving the intervention and 174 controls, drawn from
volunteers working for a IT and media company, found that a web based
health promotion and lifestyle training package can improve mental
wellbeing as measured using non-standard questionnaire at baseline
and at 6 months after the web site and related components being
available.
Discussion
This review has covered a very broad area of research relating to mental
wellbeing at work. The general lack of consistency in definition and use of
terms to describe the healthy individuals emotional experience at work is
widely recognised especially when in contrast to the clinically defined mental
disorders or psychological ill-health. This has been reflected in the very broad
range of survey terms needed to ensure comprehensive coverage and, within
individual studies, in the plethora of outcome measures utilised. A similar
comment can also be applied to the interventions themselves with individual
interventions ranging from sending out leaflets on stress, through recreational
music making to massage therapy.
directing the intervention itself (quite correctly as it would have impaired their
ability to carry out an impartial evaluation) and the studies took the form of a
discovered experiment, rather than the individual interventions where the
researchers have had greater control over the intervention.
Conclusions
There are many published papers relating to the general area of interventions
intended to improve mental wellbeing in the workplace. These cover a wealth
of different interventions and outcomes, reflecting a general imprecision in the
descriptive terms used. Despite numerous methodological difficulties and
shortcomings, enough of these papers are of adequate quality to suggest that
there might well be tangible benefits from such interventions, although
generally speaking the papers are not of sufficient quality or number to be
able to make unequivocal evidence statements. It is hoped that it will be
possible to build on the research base identified to provide clearer evidence in
the future.
xi
Table of Contents
Executive Summary ...................................................................................... i
1. Introduction ..................................................................................................1
1.2 The Need for Guidance.......................................................................1
1.3 The Scope of the Review....................................................................2
1.3.1 Areas Covered by the review...........................................................2
1.3.2 Population Groups Covered.............................................................3
1.3.3 Outcomes ........................................................................................4
1.3.4 Research Questions ........................................................................4
2. Methodology ................................................................................................6
2.1 Literature Search ................................................................................6
2.2 Selection of Studies for Inclusion........................................................8
2.3 Search Results .................................................................................10
2.4 Quality Appraisal...............................................................................10
2.5 Study categorisation .........................................................................10
2.6 Assessing applicability ......................................................................12
2.7 Synthesis ..........................................................................................12
3. Summary of Findings .................................................................................13
3.1 Overall summary of studies identified (including numbers, types,
quality, applicability)................................................................................13
3.2 How can work and working conditions be used to promote workrelated mental wellbeing or reduce work-related harm: which
interventions are most effective and cost effective? ...............................15
3.2.1 Organisational Interventions ..........................................................15
3.2.1.1 Changing working/organisational practices to improve mental
wellbeing .............................................................................................15
3.2.1.2 Training supervisors and managers ........................................22
3.2.1.3 Altering shift or work practices ................................................26
3.2.1.4 Support or training to improve skills or job role .......................29
3.2.2 Stress Management Interventions .................................................33
3.2.2.1 Training to cope with stress.....................................................33
3.2.2.2 Counselling and therapy..........................................................56
3.2.2.3 Exercise and Relaxation Interventions ....................................59
3.2.2.4 Health Promotion ....................................................................72
3.2.2.5 Others .....................................................................................75
3.3 What specific characteristics of work and working conditions promote
mental wellbeing effectively and cost effectively?...................................82
3.4
How can organisations support employees who are coping with
stress, anxiety and depression caused by external factors (for example
bereavement, family breakdown or debt)?..............................................82
3.5 How can healthy working conditions be created for different
occupational groups and in different organizational contexts? ...............84
3.6 What help do employers need to review and adapt working practices
and conditions to promote the mental wellbeing of employees?.............88
What are the barriers and facilitators to implementation of any of the
above interventions for both employers and employees?....................88
3.8 Do interventions that promote health equalities also have an impact
on mental wellbeing and productivity? ....................................................91
xii
Index of Tables
xiii
1. Introduction
1.1 Background
The National Institute for Health and Clinical Excellence has been asked by
the Department of Health to develop public health intervention guidance
aimed at promoting employees mental wellbeing. The guidance will provide
recommendations for good practice based on the best available evidence of
effectiveness including cost effectiveness. It is aimed at all employees and
the organisations that represent them. The guidance will support the following
National Service Framework (NSF) Mental Health (DoH, 1999)
The following review assesses the evidence for effectiveness of workplace
interventions that promote mental wellbeing in the workplace.
1.2 The Need for Guidance
Mental wellbeing is a state of wellbeing in which the individual realises his
or her own abilities, can cope with the normal stresses of life, can work
productively and fruitfully, and is able to make a contribution to his or her
community. (WHO 2004). Employees will typically experience short periods
of stress and anxiety in the workplace, without it affecting their mental
wellbeing. However, exposure to chronic stress and anxiety at work can be
detrimental to mental wellbeing. Moreover, a number of diseases and
disorders (e.g. coronary heart disease, musculoskeletal disorders and mental
illness) are related to psychosocial conditions in the workplace (Marmot et al.
1991).
According to the most recently available Labour Force Survey (relating to
2004-05) an estimated 509,000 people in Britain believed they were suffering
from stress, depression or anxiety that made them ill. Furthermore, these
conditions were caused or made worse by their current or past work (Jones et
al. 2006). As a result, an estimated 12.8 million working days (full-day
equivalents) were lost. Each person suffering from stress, depression or
anxiety took, on average, an estimated 30.9 days off over the 12 month period
2004-05 (equivalent to an annual loss of 0.55 days per worker). Stress,
anxiety and depression is thus the second most commonly reported cause of
sickness absence in Britain, behind musculoskeletal disorders.
In 2005 national surveillance schemes put the incidence of work-related
mental health problems in Britain at about 6,400 new cases per year. The
most recent survey of work-related illnesses undertaken for the Health and
Safety Executive (HSE), however, estimates that about 195,000 people
reported that they had first experienced work-related stress, depression or
anxiety in the previous 12 months (Jones et al. 2006). The incidence of workrelated mental health problems in Britain is therefore almost certainly higher
than the figure suggested by national surveillance schemes.
Employees in the public sector in particular, administration, defence,
education, and health and social work - had some of the highest rates of selfreported stress, anxiety and depression (Jones et al. 2006). The main risk
factors include employees with heavy workloads and employees facing tight
deadlines. Employees who receive a lack of support at work, or are being
physically attacked or threatened at work, are also recognised as being at an
increased risk of stress, depression or anxiety (Jones et al. 2006).
There is evidence to suggest that investment in healthy working practices and
the health and wellbeing of employees improves productivity and is cost
effective for business and wider society (Coats and Max, 2005; Dunham,
2001). In addition, given that employees in lower paid jobs are more likely to
experience poor working conditions, improvements in the quality of work and
working conditions may help reduce health inequalities (Siegrist and Marmot,
2004).
1.3 The Scope of the Review
The scope of the review is defined below.
1.3.1 Areas Covered by the review
The review considered interventions that promote mental wellbeing in the
workplace. Interventions were classified as:
Flexible working.
Organisational change
Stress Management
Change Management
Anti-bullying
How can organisations support employees who are coping with stress,
anxiety and depression caused by external factors (for example
bereavement, family breakdown or debt)?
What are the costs and economic benefits to employers: what is the
business case for promoting employees mental wellbeing
2. Methodology
2.1 Literature Search
The review considered intervention studies that promote mental wellbeing in
the workplace. The concepts of mental wellbeing encompass a wide variety
of topics. For example, influences on mental wellbeing include psychosocial
factors such as social support and work life balance issues.
A broad
approach was therefore used in the first instance to ensure that all topics were
considered in the initial searches. However, it was found to be important to
differentiate between those studies which principally examined intervention
strategies and those where the focus is primarily on risk assessment.
Establishing a suitably specific search strategy that avoids these without
excluding potentially valuable material presented particular challenges.
Within this field, randomised controlled trials are unlikely to be a major part of
the research thus all research designs needed to be considered in the initial
stages. In 1993, a benchmark report was published in the UK by the HSE
(Cox, 1993) which served as the basis for much of the risk assessment and
intervention activity in subsequent years.
The concept of mental wellbeing was likely to span a broad range of topic
areas thus the research team recommended searching the databases listed
below.
In addition, the following databases were searched for the economics review.
To ensure a broad range of literature was obtained for the review, the
websites listed below were also searched.
Exclusion Criteria
The exclusion criteria stated within the proposal are reiterated below.
Non-English language
Dissertations
Studies within which mental wellbeing is not the primary purpose of the
intervention
Selection Process
Before requesting the full papers for review a two-stage screening was carried
out. The first stage was a filter process examining the titles of the papers
obtained. Those outside the topic for review were eliminated. The second
stage of the screening process was to screen abstracts against the inclusion
and exclusion criteria. Where it was unclear whether the paper should be
included or excluded, a conservative approach was taken and the full paper
ordered.
At the abstract screening stage, 10% of the abstracts were screened
independently by two reviewers. This allowed a quality assessment to be
carried out and it was found that a 94% consistency rate was achieved by the
reviewers.
On completion of abstract screening, full papers were ordered for review.
Many of the papers were excluded on the basis of not being an intervention
study or the description of an intervention where its effect was not evaluated.
Where any economic papers were identified, these were forwarded to the
economics team for review.
2.3 Search Results
A flow diagram showing the fate of the studies initially identified from the
searches is shown in Figure 1.
Works. Appendix E presents the full papers which were excluded from the
study and an explanation for their exclusion.
2.4 Quality Appraisal
Studies that met the inclusion criteria were evaluated in order to determine the
strength of the evidence for effectiveness. The data from each paper was
extracted using the checklists which are presented in Appendix B. Where a
reviewer was unable to reach a decision on a paper, the paper was reviewed
by a second reviewer and a consensus reached.
2.5 Study categorisation
Each study was categorised on its experimental design and the relevant
checklists used.
10
Search Results
n=7,731
Title filter
Abstract screening
Excluded n=2,661
Excluded n=4,728
Excluded
Included in full paper
screening
n=341
Included in final
review n=66
Total excluded
n= 275
Not focused on mental
wellbeing n=86
Dissertation n=55
No control group within study
n=32
Not an intervention study
n=24
Non-English language n=14
Could not be obtained in time
frame n=12
Article, not original research
n=10
Non-occupational sample
n=9
Reviews assessed for
references n=8
Description of intervention with
no follow-up n=4
Military population n=2
Non-validated outcome
measures n=3
Cross-sectional design n=3
Irrelevant review n=2
Military Population n=2
Evaluation of model of
intervention n=2
Medical Intervention n=1
Confounded study n=1
Repeat paper n=1
Risk management approach
n=1
Measured only physical
outcomes n=1
No pre-intervention measures
made n=1
Recruitment unclear n=1
11
2.7 Synthesis
The data extraction information was synthesised into evidence statements
addressing each of the research questions.
12
3. Summary of Findings
3.1 Overall summary of studies identified (including numbers, types,
quality, applicability)
The review identified 66 studies which met the inclusion criteria and these
were grouped into 2 key areas, organisational interventions and stress
management interventions. In all cases, the review has focussed specifically
on outcome measures directly addressing mental wellbeing.
Additional
13
Number
of
studies
Quality
of
studies
Number
of RCTs
Number with
effective
interventions
Changing
working/organisational
practices to improve
mental wellbeing
Training Supervisors and
Managers
Altering Shift or Work
Practices
Support or training to
improve skills or job role
11
5+
6
4 ++
1+
2+
12++
3+
1-
1++
2+
2++
1/+
Number and
Quality of
studies with
effective
interventions
3++
6+
52+
1-
3
6
Number and
Quality of
studies with
effective
interventions
1+
4-
Number
of
studies
Quality
of
studies
Number
of RCTs
Number with
effective
interventions
22
12
14
11
Health promotion
interventions
Others
5++
11+
61 ++
2+
14++
6+
11++
1+
13++
3+
1++
4+
11++
2++
1+
14
3.2 How can work and working conditions be used to promote workrelated
mental
wellbeing
or
reduce
work-related
harm:
which
(2/+), Landsbergis and Vivona-Vaughan (1995) (2/+), Maes et al. (1998) (2/+)
and Mikkelsen & Saksvik (1999) (2+). Of these four studies only one, DahlJorgensen et al. (2005), had an intervention that was shown to work.
Dahl-Jorgensen et al. (2005) (2/+), carried out a quasi-experimental study
using a participatory approach to organisational change.
The approach
Coopers Job Stress Scale, a scoring system for subjective health complaints,
the Maslach Burnout Inventory and self-reported absenteeism.
The results
15
highlighted by the author who identified that there were doubts about the
quality of the intervention and participation by the workers.
Maes et al. (1998) (2/+) carried out a case control study to examine the
effects of combined lifestyle and organisation interventions on health
behaviour, health risks, stress, quality of work and absenteeism. The study
used a battery approach of interventions including advice on exercise, healthy
eating, stress training, social skills, leadership and organisational change.
The intervention was carried out over a 3-year period. The study group were
Dutch household good manufacturers with 175 in the treatment group and
171 in the control group based at another site. Over the period of the study
the numbers reduced to 167, 157 and 134 in the intervention group over 4 test
periods and 169, 157 and 130 in the control group.
measured using the SACL-90 which assesses general stress. There was not
16
There were 37
17
The
participant numbers were 64, 47 and 45 at the three stages in the intervention
group and 71, 35, 14 in the control group. The study identified a reduction in
work stress measures between the treatment and control groups (p<0.05).
Although stress decreased in the intervention group this effect was primarily
mediated through an increase in the control group between pre-test and posttest 1.
The study is
18
employees in each pool but the final analysis identified 55 (intervention) and
24 (control) participants. The results identified that the intervention group post
intervention showed significantly less stress (p<0.05); less negative effect
(p<0.05); less tiredness (p<0.05); more positive energy (p<0.05); and more
relaxation (p<0.05). One weakness in the study is that it is not possible to
identify whether the organisational or individual interventions had more of an
effect. It is also unclear if this study can be generalised.
Reynolds (1997) (2/-) examined the comparative effects of individual and
organisation interventions on psychological wellbeing.
The interventions
The
19
20
Evidence Statement 1
Ten studies, none of which were randomised control trials, evaluated
the effectiveness of interventions involving a participatory approach to
organisational change on mental wellbeing. The studies varied in quality
and there was heterogeneity of interventions, populations and outcomes
evaluated. Four of the ten studies were given a positive quality grading
and one of these Dahl-Jorgensen et al (2005 +) demonstrated that the
intervention improved mental wellbeing. There is currently insufficient
evidence of quality to judge the effectiveness of the use of
organisational participatory interventions in the workplace to improve
mental wellbeing and further research is required.
1. Bourbonnais et al. 2006 Quasi-experimental 2/2. Dahl-Jorgensen et al. 2005 Quasi-experimental (non RCT) 2/+
3. Kawakami, 1997 Quasi-experimental 2/4. Landsbergis & Vivona-Vaughab, 1995 Quasi-experimental 2/+
5. Maes et al. 1998 Quasi-experimental 2/+
6. Mikkelsen & Saksvik, 1999 Non-RCT 2/+
7. Mikkelsen et al , 2000 Non-RCt 2/+
8. Mattila et al. 2006 Quasi-experimental 2/9. Munz et al. 2001 Quasi-experimental 2/10 Reynolds, 1997 Quasi-experimental 2/-
21
reduction found in serum GGT for all participants assessed together (p=0.04)
but not for separate analysis of employees or managers. A number of issues
may have influenced the effectiveness of this intervention including the fact
that 3 other management programmes were simultaneously being carried out
and a fairly high drop-out rate.
22
Takao et al. (2006) (1/++) evaluated the impact of job stress training for
supervisors on psychological distress and job performance on immediate
subordinates. The study involved giving supervisors training which consisted
of a 60 minute lecture and 120 minutes of active listening training.
The
The participants
numbered 154 in the intervention group and 101 in the control group at the
start of the study; this reduced to 134 and 92 on completion of the study.
Outcomes were measured using the Brief Job Stress Questionnaire.
No
significant differences were identified within the study for the whole group;
however sub-group analysis identified a significantly positive effect, for
younger male white collar workers (p=0.012) on psychological distress.
Kawakami et al. (2005) (1/++) carried out a RCT to examine the impact of
web-based training for supervisors and how it affects psychological distress in
subordinate workers. The intervention was a web-based training programme
on work-site mental health and a 4 week training programme.
The
The outcome
measures on the study were the Brief Job Stress Questionnaire using the
sub-scales for vigour, anger/irritability, anxiety and depression. At 4-months
post-intervention there were no significant differences in any of the measures
used for psychological distress.
Kawakami et al, (2006) (1/++) carried out a similar study within a sales and
service company. The intervention again assessed the impact of web-based
training for supervisors on psychological distress in subordinate workers. 23
supervisors received the intervention with 23 controls.
The trained
supervisors had 92 subordinate workers who took part at the beginning of the
study, reducing to 81 at the 4-month assessment. The equivalent figures for
the control supervisors were 114 and 108 respectively. The Brief Job Stress
23
however,
conclusions
of
the
effectiveness
of
specific
Overall, no clear
24
training given in this study was approximately 40 hours over a 6 month period.
This suggests that further research is required evaluating the type of
supervisory training required to impact on mental wellbeing in subordinate
workers.
In addition to this study, one further study examined the impact of training
project managers on their physiological strain. This study aimed to increase
management control using a 10 hour training session, however there was no
impact on wellbeing outcomes post training.
Evidence Statement 2
Four studies evaluated the impact of training for managers and
supervisors on the mental wellbeing of subordinate staff. Two
randomised control trials (Kawakami et al. 2006 ++) and (Kawakami et al.
2005 ++) undertaken in Japan found that web based training to improve
management skills was not effective in improving the wellbeing of
subordinate workers. A randomised control trial in a Japanese brewery
(Takao et al. 2006 ++) and a non randomised control trial in a Swedish
insurance company (Theorell, 2001 ++) used traditional face-to-face
training. The Japanese trial found no significant difference in the whole
group analysis. The Swedish study found that physiological markers for
stress were significantly reduced in the subordinate workers in the
intervention group. The training given in this study was approximately
40 hours over a 6 month period. There is therefore insufficient evidence
to allow any positive statement to be made and further research is
required evaluating the impact of different types of supervisory training
on the mental wellbeing in subordinate workers.
1. Kawakami et al. 2006 RCT 1/++
2. Kawakami et al. 2005 RCT 1/++
3. Takao et al. 2006 RCT 1/++
4. Theorell, 2001 Non-RCT 2/++
25
26
group.
The
27
Evidence Statement 3
There is evidence from one non randomised trial of white collar workers
working for an industrial employer in Israel (Etzion, 2003 +) that taking a
vacation impacts positively on burnout in the short term (immediately
on return from holiday and at three weeks) but stress can significantly
fall on return to work (p<0.01) but at three weeks returns to pre-vacation
levels as measured by a questionnaire.
There is evidence from a UK quasi-experimental study of police officers
(Totterdell, 1992 +) that changing the shift system from 7 day
consecutive shifts to the 35 day Ottawa system can positively impact on
mental wellbeing as measured by a questionnaire.
1. Etzion, 2003 Non-RCT 2/+
2. Totterdell & Smith, 1992 Quasi-experimental 2/+
28
The
intervention was a PSI course involving 3 hour sessions, weekly, for 16 weeks
during working time. The study group were UK mental health workers, mainly
female with an average age of 38 years. 35-40% of the participants were
non-nursing. Participants were nominated to take part in the study and this
resulted in 14 participants in the intervention group and 12 in the control
group. The outcome measures were the Maslach Burnout Inventory and 3
sub-scales measuring personal accomplishment, emotional exhaustion and
depersonalisation. Measures were made pre and immediately post-training.
The study found no significant differences between the groups apart from
personal accomplishment which increased in the training intervention group
(p<0.05).
control group and this was stratified by ward, gender and shift. The outcome
measure of the questionnaire was the Maslach Burnout Inventory including
the subscales of accomplishment, exhaustion and depersonalisation.
The
(p=0.01),
exhaustion
(p=0.04)
and
depersonalisation
29
(p=0.01). This was however a small scale study and the principal investigator
was a member of staff within the unit.
Japanese hospital nurses were the focus of a non-RCT study by Shimizu et
al. (2003) (2/+). The authors examined the impact of communication skills
training on burnout through two 2-day training courses about a month apart.
One male was excluded from the study leaving an exclusively female sample
of 19 treatment and 26 controls reducing to 12 and 14 at completion.
Participants were allocated to the study and then to the groups by their
supervisors who took care to balance the work sections of each group. Initial
allocation to the study referred to those seen (by their supervisors) as having
low-to-moderate communications skills. It is not stated whether this further
influenced allocation to groups. Wellbeing was measured using the Maslach
Burnout Inventory (MBI). This yielded three sub-scales of emotional
exhaustion (EE), depersonalisation (DP), and personal accomplishment (PA).
Assessment of the change in scores five months post intervention showed the
treatment group PA score to have increased while a decrease was shown in
the controls (p<0.05) with neither of the other two groups showing any
significant change. The small group sizes and low retention are both factors of
concern.
Engstrom et al. (2005) (2/+) evaluated the impact of the introduction of new IT
support systems to improve patient monitoring among Swedish nurses. The
intervention was the implementation of new IT support systems to allow
patients with dementia more freedom to move around. The participants in the
study were nurses, with an average age of 40, the majority of whom were
female. There were initially 27 individuals in the intervention group and 32
controls. 12 months post intervention this was 17 intervention and 16 controls.
Outcomes were measured using the Satisfaction with Work Questionnaire
and the subscales evaluating sleep disturbance and perceived stress. No
significant differences were found between the groups although perceived
stress had improved in the intervention group.
Schaubroeck et al. (1993) (1/+) carried out a quasi-experimental study to
evaluate the effect of an intervention designed to clarify individual roles at
30
work on subjective strain, physical symptoms or time lost through illness. The
intervention group were given supervisor role clarification training.
The
The
sample was made up of 57% males with a median age of 41 years. The
outcome measures for the intervention were the Mental Health Battery which
was used at 6 months and 10 months post intervention. No significant effects
were found on psychological or physical ill health although role ambiguity and
supervisor dissatisfaction were reduced.
Nielsen et al. (2006) (2/-) carried out a health promotion intervention on
canteen workers including factors such as exercise, empowerment, IT
training, and a workshop on development. At pre-test the numbers in the
canteens were 45 and 26 (intervention groups) and 22 and 25 (control
groups) reducing to 30, 26, 19 and 28. At 20 months post-intervention there
was a significant increase in cognitive stress reaction in the intervention
groups (p<0.05-0.01) and one control group (p<0.01). In addition there was a
significant increase in vitality in the two intervention canteens (p<0.05) and in
a control canteen (p<0.01). Mental wellbeing was not the main focus of this
paper and there were a number of interventions both individual and
organisational carried out within this study. There were also a number of
external changes including a new manager at one site and closures and cuts
at another which were likely to have impacted on the results.
Summary and Evidence Statement
Two high quality studies examined the impact of PSI training on burnout in
mental health care workers.
31
One study, Shimizu et al. (2003), rated 2/+, showed that providing training to
enhance
communication
skills
could
slightly
improve
personal
personal
accomplishment
depersonalisation
(p=0.01).
(p=0.01),
These
exhaustion
small
studies
(p=0.04)
and
indicate
that
32
special affect and a group discussion of a specific affect from the previous list.
An initial sample of 122 employees were randomly selected from amongst
employees in social service, elderly care and education of a large Swedish
municipal authority. From this initial sample, 50 females with high stress levels
(Perceived Stress Questionnaire PSQ) were selected for the study. They
were randomly divided into intervention (n=27) and control (n=23) groups.
Mental well-being was assessed using the PSQ; the Symptoms Check List-90
(SCL-90); and the Global Severity Index (GSI) derived from the SCL-90 with
data collected 6 months prior to intervention, immediately before interventions
and within 5 weeks of the completion of intervention (no explanation is given
for this rather vague description).
33
Cook et al. (2007) (1/++), described the results of a RCT comparing a webbased workplace health promotion programme with a paper-based equivalent
amongst a group of office workers (primarily female, white and degree
educated) in a US-based HR provider. A total of 247 (web) and 233 (paper)
were initially recruited to the study which provided guidance on stress
management, diet and physical activity. Recruitment was voluntary from a
potential population of 5,000, with publicity through a number of channels,
before random allocation to the two treatment groups. However, only 236
(web) and 230 (paper) actually entered the study. With losses during the
study the final analysis was based upon 209 and 210 employees respectively.
The web and paper-based material covered the same topics although the
information provided was not necessarily the same.
Outcome measures
consisted of, Perceived Stress, Symptoms of Distress, and the Brief COPE.
With the exception of the latter these measures were devised for the study
and have not apparently been validated.
Post-treatment results were obtained immediately following a 3-month
intervention period. An ANCOVA revealed no pre-post differences between
the two groups in any of the three outcome measures. However, individual
analyses showed the print version to have more effect than the web-based
34
35
covariate in the analysis. Studying the whole groups (ANCOVA) there were no
significant effects of either of the training modalities on GHQ-12 scores.
However, when those with pre-existing mental health problems were selected
on the basis of a GHQ-12 score of at least 4 at pre-treatment, there was a
significant ANCOVA interaction term (p=0.04) indicating a differential pattern
of effects between treatments. GHQ scores fell for all three at post training;
COG continued to fall, BEH remained down and Control returned to pre-test
levels. Post-hoc tests of change from pre-test to follow up revealed the COG
training was significantly better than the Control (difference 4.70, p=0.005) but
that no other comparisons were significantly different. Applying standard
assessments of effect size the author describe that of the Cognitive
intervention as large and that of the Behavioural as medium. The authors
raise several concerns including that those willing to wait were more likely to
be assigned to the control group and that those who were lost to follow-up
tended to report higher pre-study levels of stress. In addition, the restriction of
apparent benefits to those with pre-existing problems might be regarded as a
drawback (although it could be argued that it provides for more focussing of
attention where it is required). Finally, individuals were assigned to different
treatment groups to some extent as a function of their workplace. Although
more than one course was apparently mounted at some sites this raises the
possibility that non-intervention factors could have had a differential influence
at the various sites thus distorting the results.
Horan (2002) (1/+) reported on the findings of a RCT, designed to examine
the effects of a workplace intervention involving reading, commenting upon
and sharing their own inspirational stories. The intervention lasted for 11
weeks with one, 1 hour meeting per week, away from their work. Control
group members remained at work throughout. A total of 66 employees took
part. These were drawn from a single American company. They were
allocated to one of three treatment groups and three wait-listed control
groups. Although no numbers are given regarding allocation to these groups
and demographic details (mostly married, Caucasian females with children)
are not subdivided they were described as not differing between the treatment
and control groups. Retention figures are not stated. Reference is made to 18
36
37
The results at 3 months show significant positive effects, due to the treatment,
on the GHQ-30; STAI-S and STAI-T; and BDI (all p<0.0005). At 18 months
values remained significantly lower (p<0.005) except for the STAI-S where the
significance was reduced (p<0.002). There was no significant difference in
absence rates between intervention and control groups. The authors
acknowledge that the effects could be attributable to other unintentional
influences such as increased attention, diversion from other cares, or
increased socialisation, rather than the course content itself. The focus on
those with existing problems might limit the generalisability of the findings
although it might also be regarded as assisting to focus help where it is most
needed. The study population of student nurses might restrict the applicability.
Although carried out in a college setting, the nurses were attending following
periods of work in hospitals. However, the timing of the measurement periods
was immediately prior to examinations which might have had some bearing
on reported mental wellbeing although there is no reason to believe that it
38
might have influenced the outcome of the training intervention. The authors
also expressed concern that the subjective measures were not supported by
more objective findings.
McCraty et al. (2003) (1/++) studied the effect on a combination of subjective
and circulatory factors of a stress management intervention, which was
described as a Positive Emotion-focussed stress management programme
(Inner Quality Management). The intervention involved an initial 8 hour course
followed by two 4-hour follow-up courses over a 2 week period. In the
intervention, participants learned positive emotion refocusing and emotional
restructuring techniques. The subjects were volunteers drawn from the
employees of a global information technology company, based in the USA,
who had been diagnosed as suffering from hypertension. From initial baseline
measurements 38 subjects were selected and randomly assigned to
treatment or control status in a crossover design. 21 (treatment) and 17
(controls) were therefore allocated to the two groups. However, two subjects
were unable to participate at their allocated times and therefore exchanged
between groups. As this involved one from each group there was no effect on
group numbers. Attrition meant that 18 (treatment) and 14 (controls) remained
on completion at 3 months after training (before the second cross-over
element). The subjects were predominantly male although demographic
information is only presented for the whole sample. The authors state that
only average age differed between the two groups (the treatment group being
approximately 5 years older on average). Objective assessment was limited to
measurement of blood pressure. Subjective outcome indices utilised the
Personal and Organisational Quality Assessment (POQA) and the Brief
Symptoms Inventory (BSI).
The treatment group showed a significantly larger reduction in systolic blood
pressure than the control group although it is noted that this did not remain
significant when adjusted for demographic factors. This finding was supported
by significant improvements in individual wellbeing including Positive Outlook
(p<0.01); and Stress symptoms (p<0.05) both derived from the POQA
together with depression (p<0.05) and symptoms of phobic anxiety (p<0.05)
39
from the BSI (but not Anxiety). In addition, the Global Severity Index, also
derived from the BSI, showed a significant reduction in the treatment group
(p<0.05).
Mino et al. (2006) (1/+), carried out an RCT into the alleviation of the
symptoms of depression by a training programme with lectures covering the
perception of stress and how to cope with it (2 hours) plus muscle relaxation
(2 hours), coupled with the provision of stress management record sheets and
email advice and counselling. It is not stated whether the course took place in
work time. All 60 employees in a specific department of a Japanese
manufacturing company were invited to participate and 58 did so. They were
randomly allocated to treatment (28) or control (30) groups. Of these 21 and
30 respectively were retained at the 3-month follow-up. Outcome measures
were the GHQ-30, the Centre for Epidemiologic Studies of Depression (CESD) and the Health Status Questionnaire. At follow-up there were no effects on
the GHQ-30 scores but the depression score from the CES-D was
significantly reduced (p=0.003). According to the authors the reported
adherence to the stress management was poor and individuals were reluctant
to utilise the counselling service. Although attrition from the control group was
quite high (25%), analyses of pre-test scores showed no differences between
those retained and those lost.
Pelletier et al. (1998) (1/+), reports on an intervention conducted by mail and
telephone. Subjects were recruited from amongst the lower middle
management and secretarial job categories amongst the employees of a US
County for a RCT examining the effects of two interventions, differing only in
the use of four follow-up telephone calls by a health educator to supplement a
series of mailings (one every six weeks for a year). A third (control) group
received neither mail nor calls. The mailshots covered stress, relaxation and
individual workplace adjustments and coping strategies. A total of 81
employees, were recruited and allocated equally to the three groups. They are
described as providing a mix of ethnicity, race and gender although no details
are given of the actual demographic make-up. On completion, group numbers
were 21 (Full intervention), 20 (No telephone) and 25 (Control). Wellbeing
40
41
took place during work (average attendance 4 sessions). Both were every
other week for 3 months. The partial intervention participants had mail
feedback of results and no additional contact. Participation rates are unclear.
A total of 501 were initially assigned to groups; 171 (full), 166 (partial) and 164
(control). However, only an overall completers total of 343 is reported (32%
attrition) with no group breakdown. Outcome measures were the Stress
Coping Inventory (SCI); the State-Trait Anxiety Inventory (STAI) and a selfcompleted Quarterly Health Questionnaire (QHQ). In addition, approximately
30% (150) were members of a single Health Maintenance Organisation
(HMO) and data on doctors visits were also collated.
The results apparently show reductions in stress and anxiety measures
across time for all six groups (across the two industries analysed separately).
All were p<0.001 with the exception of Recent Life Changes which was
p<0.05 for the computer industry. However, time by group interactions were
identified which indicate treatment specific effects for the computing industry.
Specifically, negative responses to stress showed changes in the expected
direction with fewer such responses for the Full intervention group, more for
the Partial intervention and most for the Control group (p<0.012). Although the
local authority employees showed the same overall reduction in stress across
all groups there were no group specific effects. What were categorised as
coping measures showed no significant changes across either time or group.
Patterns of self-reported sickness absence across the follow-up year showed
changes in the expected direction with the Full intervention group taking least
time off in both industries. The scores for the city employees approached
significance (p=0.068). This was supported by the data from the HMO which
showed that the city employees who received the Full intervention made
significantly fewer visits to their doctor during the year of follow-up (p=0.04).
Interpretation of the findings is hindered by the inadequacies of the retention
data and no reporting of any characteristics of the HMO sub-sample although
it could be argued that there would be no a priori reason why the subgroup
registered with one particular provider should be any different to others. The
disparity between the subjective data (which essentially showed effects
across all groups regardless of treatment) and objective data (apparently
42
showing benefits from the treatment) are interesting. It appears that those
receiving the full treatment were healthier although they didnt feel any
better.
Sheppard et al. (1997) (1/+) carried out a randomised comparison of two
forms of individual intervention, Transcendental Meditation (TM) and what
was described as a conventional stress management (CSM) programme. The
TM programme involved 5 hours of initial tuition followed by six bi-weekly
sessions at work lasting about an hour each (plus a requirement to practice at
home). The conventional group also had about 11 hours of training/tuition
including regular worksite meetings (but no homework). At the onset, 44
employees volunteered to participate after a preliminary lecture on stress. It is
not known whether the subjects knew that they might be assigned to TM
training beforehand. They were all employed at what is described as a highsecurity US government agency worksite reported to be highly stressful.
Most (85%) were female with a mean age of around 50 years. Participants
were initially, equally assigned randomly to the two groups 17 (TM) and 15
(CSM) remained at completion after a three year follow-up. Outcome
measures of mental wellbeing were STAI and the IPAT Depression scale.
These measures were supplemented by readings of blood pressure.
The results, analysed with an ANCOVA, showed that, after three months,
STAI Trait Anxiety (p=0.05) and IPAT Depression (p=0.025) were significantly
lower in the TM group. The text also refers to STAI State Anxiety being
significantly reduced (p=0.03) at this time although this is not marked in the
results table. From the data this would seem to have been an omission from
the table rather than any error in the text. Neither systolic nor diastolic blood
pressure varied significantly between groups. At the 3-year follow up all three
parameters, State (p<0.025) and Trait-Anxiety (p<0.05) and Depression
(p<0.01) were significantly lower in the TM group alone. Again, no blood
pressure parameters were significant. This appears to be a well-conducted
study which, with the exception of the absence of homework, made real
efforts to equalise the intervention times between the two groups. The specific
tasks to carry out resulted in a very high proportion of TM participants (73%)
43
who were continuing to practice after three years suggesting very strong
motivation (and presumably belief in the benefit derived). Clearly it would
seem not to have been the intervention time itself which had the beneficial
effect. The authors indicate that they were unable to gain acceptance for a
non-intervention control before the study onset.
Shimazu et al, (2005) (2/++) utilised web-based training material (compared to
a wait-listed control group) to examine the effect of cognitive training aimed at
improving coping strategies. It is not stated whether access to the training was
permitted during work time or if they were expected to complete the course
out of work. Recruited from amongst the white-collar employees of a
construction machine company (support staff) in Japan 225 staff were invited
to participate and it appears that all agreed to do so, although 6 (3 from each
group) then dropped out prior to study onset. Leaving 109 (treatment) and 110
(controls) to receive the initial questionnaire. On completion 100 (treatment)
questionnaires were returned. Of these, 6 had not actually received the
intervention but were retained for analysis on an intention to treat basis. There
were 104 questionnaires returned from the control group. Subjects in both
groups were predominantly male with average ages in the early 40s. They
were allocated to groups on the basis of employee number (odds and evens).
There were no significant differences between the two groups at the outset, in
terms of both demographic factors or pre-treatment questionnaire responses.
Outcome
measures
of
wellbeing
utilised
the
Brief
Job
Symptoms
44
not conventionally randomised although this did not have any distorting effect
on baseline measures. The rationale for allocation of such numbers to
employees is not known but might be expected to be chronological on
recruitment. The high volunteer rate and retention levels are impressive.
Despite the absence of any significant main effect the study is of value
because of the clear indication that attitude to the issue of stress can have an
influence on the efficacy of any measures introduced to counter its effects.
The authors indicate that the study had apparently become a subject for
discussion in some workplaces, suggesting that this might have had a spillover effect contaminating the control sample. Additionally, the short time
period between training and evaluation gave little time for any behavioural
change to be implemented and have any influence on outcome measures.
Final follow-up data (which would have allowed for a longer time span) were
not reported because of the absence of any control group at this time.
A brief worksite stress management course, consisting of a single 2-hour
course covering a cognitive-behavioural therapy-based approach to stress
management, was evaluated by Shimazu et al. (2006) (2/++) in a non-RCT
design. All 300 employees of a Japanese construction machinery company
were invited to participate. It appears that all consented as those consenting
were then allocated to the treatment (149) and control groups (151).
Allocation was carried out by managerial staff who stratified them by job
position and work section before assigning them to the two groups (technique
for assignation not given). Four of the treatment group did not then attend the
training and a further one did not answer the second questionnaire leaving
144 who completed all three elements. Attrition from the controls was similarly
low with 143 answering both questionnaires.
45
with this change being significant (p=0.022). These changes were from
baseline values for the treatment group which were higher than those for the
controls (approaching significance, p=0.080). Physical complaints, which had
been significantly higher in the treatment group at baseline (p=0.005) showed
a trend towards a significant change at follow-up (p=0.060) with the control
group showing a greater mean fall (-0.5 cf -0.1). The authors suggest the
adverse effect could be due to an increased awareness of stress in the
treatment group or an increase in strain associated with attempting to
implement the training imparted. They also suggest that the predominantly
male workforce might be a factor although it is not clear why this should have
a differential effect on the two groups.
Aust et al. (1997) (2/+) studied the effects of a theory-based stress
management approach on the wellbeing of a group of German bus drivers.
The intervention consisted of a series of 12 weekly sessions, each lasting 1.5
hours, which took place after work. The courses covered issues such as
muscle relaxation and coping strategies. In addition, ways of improving their
work to reduce stress were discussed but no changes were implemented until
after the study was completed. Participants had an average age of around 50
years and a minimum of 5 years experience. The control group was waitlisted. A total of 54 volunteers were divided into treatment (n=36) and control
(n=28) groups. The basis for the division is not stated. By the end of the study
these numbers had fallen to 22 and 24 respectively. Outcome measures of
wellbeing were a Positive and Negative Mood questionnaire, derived from a
German inventory of the Quality of Life, together with a symptoms
questionnaire derived from a symptoms Inventory. Although there was further
investigation of the two groups, the wait-listed controls received the course
after an intermediate assessment. The case-control element of the study was,
therefore, restricted to this assessment at the end of the first training series.
Although there were some effects on other aspects of the study there were no
significant effects found on any of the mood or symptom indices. The authors
considered that the volunteer status of the subjects might have influenced the
outcome although the use of a group favourably inclined towards the issue of
46
47
the treatment groups. After initial screening, 189 were allocated to Stress
Management Training (SMT); 165 to Physical Exercise (PE); 162 to an
Integrated Health Programme (IHP); and 344 to Control status. SMT involved
a cognitive-behavioural approach (2 hours weekly for 12 weeks); PE was
described as an aerobic dancing programme (1 hour twice per week for 12
weeks); IHP involved a combination of exercise, stress coping and workplace
examinations (2 hours weekly for 12 weeks). All of these occurred during work
time. The Control group received no stated intervention. Outcome measures
for mental well-being (obtained pre and post-intervention plus at 1 year followup) were the Cooper Job Stress Questionnaire (JSQ) and the Subjective
Health Complaint Inventory (SHC). Data on sick leave were also collected.
The study found no significant effects of any of the interventions on any of the
outcome measures (including sick leave). A further analysis, retaining only
those who attended at least 50% of the training sessions, gave no
improvement. The relatively poor retention to the study, despite the voluntary
nature of involvement and the fact that the sessions took place during work
time is an adverse factor as is the apparently poor adherence to the training
programme (although an attempt was made to analyse for this). These
findings could be suggestive of a poor general attitude which might have
influenced the outcome. In addition, the authors point out that, during the life
of the project, employees were informed of a planned reduction in the number
of post offices by around 40% as one factor at a time of great turmoil
associated with a transfer from state to private ownership.
Lindquist and Cooper (1999) (1/++), examined the effectiveness of a battery
of training and counselling measures on levels of stress amongst office
workers working for the Australian government. The training included stress
and lifestyle training, together with training in coping skills. At the end of the
training period, which comprised four weekly sessions of unspecified duration
(during work time), members of the treatment group were given a 45 minute
individual counselling session during which they were given a personalised
action plan based on their responses to the first (pre-treatment) questionnaire.
During the ensuing eight weeks they were telephoned to provide
48
encouragement and support. Subjects were drawn from the 204 respondents
to an initial survey amongst all 730 employees. They were selected on the
basis of responses to questions on age, gender, lifestyle, coping and stress.
However, the basis for that selection is not given. They were reasonably
balanced for gender (55% female). Demographic details are not given
although the authors state that the two groups did not differ significantly on
any pre-treatment variables. Outcome was determined from responses to the
Organisational Stress Indicator using subscales on work-stress (sources of
stress), coping and physical health. In addition, measurements were obtained
of blood pressure as an objective measure of physical wellbeing.
Using an ANCOVA to adjust for pre-treatment values, no outcomes showed
any
significant
significance.
changes
although
work
stress
(p=0.06)
approached
although blood pressures were higher for the treatment group than the
controls with the difference increasing post-treatment. The authors suggest
that the time between the treatment and evaluation might have been too short
for the benefits of the course to become apparent. In support of this they cite
a follow-up survey which apparently did show significant changes. However,
as by this time the control group had also received training and retention in
the study had fallen to 66% care should be taken in interpreting this
Bunce and West (1996) (2/-) compared a Stress Management Programme
(SMP) and an Innovation Promotion programme (IPP).
Participation was
primarily voluntary, supplemented by others encouraged by their linemanagers to take part. No separate data are provided for these two sample
sources. Participants were allocated to their treatment group on the basis of
work location and the basis for allocation of treatments to locations is not
given. There were a total of 62 in 7 SMP groups; 45 in 6 IPP groups; and 84
controls recruited to the study. At the end of the study, 27 SMP, 20 IPP and
70 controls remained. The number of locations for the control sample is not
given. No demographic details are given for the groups although they are
stated not to differ between groups. An ANCOVA showed no significant main
or interaction effects on the GHQ-12 scores. However one-way ANOVA
49
showed a significant reduction in the SMP group 9 months after the study
onset (p<0.01). The text refers to a trend towards a similar decrease within
the IPP group but no probability is quoted. The issues addressed in the IPP
groups were an eclectic mix, ranging from concerns over dealing with
bereaved relatives to the manual handling of patients. Although the rationale
of adapting course content to the groups perceived needs is understandable
it raises the prospect of some group interventions being less successful than
others, therefore adversely affecting the study outcome.
50
significantly better than the treatment group at the outset. The results showed
that overall stress, tiredness and stress symptoms all fell, although they
remained higher than those for the sham group. These reductions are
described as being statistically significant although no statistics are presented
to support this. In contrast there were no changes in the control group. The
treatment group showed a significant reduction in systolic blood pressure and
an improvement in heart rate variability indices (p<0.001), compared to no
change in the sham group. The self-selection of subjects to the two groups
gives particular cause for concern. The existence of clear pre-treatment
differences between the two groups (not analysed) reinforces that concern.
The absence of any information about attrition is also of concern. It is possible
that there was none but this cannot be determined from the published paper.
Munz et al. (2001) (2/-), utilised a mixture of individual stress management
training and workplace intervention. Participants were drawn from amongst
approximately 150 employees spread between 2 work sites for each group. A
total of 55 (treatment) and 24 (controls) completed all elements of training and
pre-post questionnaires. However, initial numbers are not given and so no
estimate of attrition can be derived. Using an ANCOVA to analyse results at
the end of the 12 week intervention adjusted for pre-test scores showed that
the treatment group had significantly less perceived stress post-intervention
(p<0.05); less depression (p<0.05); less negative effect (negative arousal)
(p<0.05); less tiredness (p<0.05); more positive energy (p<0.05); and less low
negative affect (more relaxation) (p<0.05). The two intervention work units
also showed better increase in productivity and lower absenteeism. These
latter results are more powerful as they reflect overall unit data regardless of
participation in the self-management training. Clearly, as indicated by the
authors, it is not possible to determine the proportional contribution of the two
intervention elements. Inadequate reporting of participation and attrition also
undermines the value of this element of the study although these latter,
objective measures go some way towards rectifying this.
Conventional stress management training was utilised by Shimazu et al.
(2003) (2/-). 24 subjects (12 in each group) joined the study but the number
51
had dwindled to 8 by the end. 64% had been instructed to participate by their
manager and a further 12% joined for other reasons. The subjects were not
allocated to groups randomly but were assigned on the basis of the school
they were attending, ostensibly due to schedules at the various schools. One
week post the conclusion of treatment the follow-up revealed no significant
change in any of these measures. With such a small sample size the study
would have had to have evoked very large changes to have the power to
detect them and this, coupled with the high attrition rate of 33%, means that
these findings must be considered with some scepticism even without the
unusual recruitment pattern.
Walach et al. (2007) (2/+), studied the impact of mindfulness-based training
on coping and wellbeing. 29 volunteers from a staff of 185 took part. The first
12 were allocated to course 1 (treatment) and the remainder (17) to course 2
(control). On completion 11 and 16 remained. The outcome complaints
measures did not reveal any significant changes. It was indicated that the
training had (initially at least) sensitised them to the stresses of their work,
which explained the apparent lack of positive benefits.
Summary and evidence statement
The nature of the studies examining the impact of training to cope with issues
of mental wellbeing have been varied in both the types of interventions used
and the outcome measures assessing impact.
52
concentrate on measures which the individual can control and others include
a degree of organisational intervention.
regarding this are made with caution because, as stated above, these
individual elements have not been formally evaluated for their efficacy.
53
Evidence Statement 5
Eight studies that were graded positively evaluated different types of
stress-management training1-8 six studies found a positive impact on
mental wellbeing as measured by questionnaire. One Australian
randomised trial found a positive effect that was close to but not
statistically significant (Lindquist et al, 1999 ++) and one study with 54
volunteer German bus drivers (Aust et al 1997) found no significant
effects. The differences amongst studies in interventions, populations
and study quality mitigate against definitive conclusions. However there
is reasonable evidence that multi-faceted training, covering stress
awareness, coping and stress reduction is an effective format.
Six of the eight studies had training programmes involving a trainer or
facilitator of which four found a positive impact on mental wellbeing,
again measured by questionnaire. Two small randomised control trials
(Horan et al. 2002 +) and (Rahe et al. 2002 +) found that small group
sessions have a positive impact on mental wellbeing.
There is evidence from one randomised trial undertaken in the USA
(Cook 2007++) that compared web materials with paper based materials
that paper based training materials are more effective for improving
mental wellbeing.
1 (Bergdahl et al., 2005) RCT +
2. (Gardner, 2005) Non-RCT +
3. (Jones et al., 2000) RCT ++
4. (Horan et al., 2002) RCT +
5. (McCraty et al RCT) RCT++
6. (Rahe et al., 2002) RCT +
54
55
In a RCT, Bond and Bunce (2000) (1/+), compared the relative effectiveness
of two forms of intervention, Acceptance and Commitment Therapy (ACT training to accept but not seek to change) and Innovation Promotion
Programme (IPP encouraging changing sources of stress) against a control
group. The training involved three half-day sessions during work time at
weeks 1, 2 and 14 of the training period. With follow-up 13 weeks later. From
a workforce of 309 in a UK-based media company, 90 volunteered to
participate and were randomly allocated to the three groups in equal numbers
stratified for gender. They were mainly graduates with an average age of
around 36 years. Outcome measures of wellbeing were GHQ-12 and Becks
Depression Inventory (BDI). They were obtained prior to each of the treatment
sessions (T1 T3) and at follow-up (T4).
Between T2 and T3 there was a significant fall in mean GHQ in the ACT
group (p<0.000) which was sustained in T4 (p<0.000). At T3 and T4 the
means were significantly lower than either the IPP or the Control group values
(p<0.001). The IPP treatment group had a significant reduction in BDI from T1
to T2 (p=0.006) whilst, in the ACT group there was a significant fall between
T2 and T3 (p<0.000). In both cases, although remaining lower than previous
values, scores drifted up slightly and were no longer statistically significant at
later time periods. The authors comment that initially high GHQ scores (pretreatment) indicated a propensity for change. They also indicate that analyses
of data from those lost to the study indicated no differences in outcome or
other measures between them.
56
In the CBT group the average score on the Depression subscale was lower at
the end of treatment than it was in the control group (p=0.028). One month
afterwards both Anxiety (p=0.021) and Depression (p=0.040) were lower.
However, these effects did not persist and there was no apparent significant
effect at +3 or +6 months although the adjusted values remained below the
scores at the end of treatment. There was a low attendance rate for the
sessions and adherence to training was seen as a problem (which might have
been due to need to attend a clinic to receive therapy). In addition there were
some apparent problems with employers not allowing time for access. A
number of participants also apparently expressed a preference for their
conventional treatment which was seen as being more tailored and more
face-to-face.
intervention there was a significant interaction between Area (group) and time
(p<0.049). SCL-90R decreased in Area A and increased in both B and C. The
text states that the same effect was seen with the GHQ-12 but the figure
presented also shows a reduction for B. According to the authors, counselling
but not organisational intervention reduced physical and psychological
symptoms. Conflicts in the data reported makes this difficult to interpret. Also
lack of any information on participation rates, attrition and demographic
comparisons diminishes the value of this study.
Evidence Statement 6
A UK randomised control trial (Bond et al. 2000 +) with 90 volunteers
from a media company found that three half-day sessions of therapy
and counselling delivered during work time had a positive impact on
mental wellbeing in the short term as measured by questionnaire. A UK
randomised trial with 24 in both the intervention and control groups who
were NHS and Local Authority workers with 10 or more days absence
due to stress, anxiety or depression in the previous 6 months (Grime et
al.2004 +) found that eight weekly sessions using a computerised
Cognitive Behavioural Therapy programme had a positive impact on
mental wellbeing in the short term as measured by questionnaire.
1 Bond & Bunce, 2000 RCT 1/+
2 Grime, 2004 RCT 1/+
58
59
absenteeism and heart rate, outcome measures were state and trait-anxiety
using the STAI questionnaire. All measures were obtained pre-test 1, then
four weeks later at pre-test 2 followed by post-test 1 at the end of the eight
weeks of exercising and post-test 2, two weeks later. STAI measures were
obtained following each exercise session.
Collapsed across groups, STAI obtained immediately following exercise
sessions showed a significant (p=0.005) reduction in STAI-S. Analysis
showed that the effect was due to aerobic not nonaerobic sessions.
Combined post-test scores, the aerobic treatment resulted in significantly
lower STAI-S scores than the nonaerobic treatment (p=0.018) (effect size
=0.22). Analysis showed the effect to occur mainly at post test 1. STAI-T was
significantly reduced across the study for the aerobic treatment group
(p=0.018) but not the nonaerobic group (effect size = 0.60). Further analysis
showed that only previously non-exercising members of the aerobics group
had a significant reduction in STAI-T (p=0.016). There were no significant
changes in absenteeism in either group. Similarly there were no differences in
resting morning heart rate between the two groups. However, this parameter
was self-recorded and a total of 8 (aerobic) and 9 (nonaerobic) were excluded
for failing to obtain the requisite values. The authors recorded attendance at
exercise classes to be 2.3 out of a required 3 per week. There was no nonintervention group but the differential effect between the two groups suggests
that the results obtained were not due to other, uncontrolled features of the
intervention such as the social benefit of the sessions. It could be
hypothesised that the restriction of the beneficial effect of exercise sessions
on STAI-T to only previous non-exercisers suggests that subjects might have
been anxious about their previous lack of exercise.
Atlantis et al. (2004) (1/+) carried out a RCT amongst shift workers in an
Australian casino. The treatment group participated in a 24 week programme
which combined aerobic exercise (moderate to high intensity) for 20 minutes
on three days per week; weight-training (light to moderate intensity) at least
twice a week and behaviour modification interventions (health education
seminars and health counselling). All this took place out of work time. In
60
Van Rhenen et al. (2005) (1/+) reported on a RCT comparing the effects of
cognitive and physical stress-reducing programmes on psychological
complaints. The interventions consisted of four, 1-hour training sessions over
an 8 week period (with 2wk, 2wk and 4wk breaks) presenting either physical
therapy (exercises, relaxation) or cognitive therapy. The subjects were Dutch
telecommunications workers (apparently white collar), with a mean age of
44.2yrs. 90% of them were men, with an average of 21.1years of work
experience. They were selected on the basis of their 'Distress Score' from the
Four-Dimensional Symptom Questionnaire (4-DSQ) which was in the top
decile of a larger sample of 700 staff. No breakdown of demographic data by
groups is reported. From an initial sample of 396, 59 were allocated to the
cognitive treatment and 71 to the physical treatment with 266 drop outs at this
61
stable. Burnout was reduced in the physical exercise group and exhaustion
raised in the cognitive therapy group but the interaction term was still not
significant. The effect of either intervention on fatigue remained, again with no
significant interaction effect. Thus no differential effect between the two
interventions was demonstrated although both had a positive benefit which
lasted through to follow-up. This was a well-designed study but the lack of an
absolute control group makes extrapolation difficult - was the benefit just from
the breaks from work or a genuine effect of the interactions? The complex
analysis increases the risk of chance significant effects but the persistence
and similarity of the effects between the two groups makes this an unlikely
explanation.
Hinman et al. (1997) (1/++) reported on the effect of a computerised exercise
programme on stress levels amongst female American, office workers who
used video display terminals. The intervention involved computer-directed
exercise breaks twice a day for 15 minutes per break in addition to their
normal breaks. Prompted when to take a break by their computer the exercise
period could be put on hold if it was necessary to interrupt it for work
reasons. 24 received the intervention with 26 controls (3 were excluded from
the intervention group due to complete non-compliance). Self-reported
62
compliance ranged from 3.8% -100% (av=39.5%). Prior to the onset of the
intervention and at the end of the eight week programme wellbeing
assessments were obtained though the four dimensions of stress on the PSQ,
(vocational, psychological, interpersonal, and physical).
Taniguchi et al. (2007) (2/+), carried out a controlled trial of the effects of
relaxation training which consisted of comparing the effects of a one-hour
lecture and 10 minutes relaxation training to those of the lecture alone, the
only difference therefore being the ten minutes relaxation training. Follow-up
measures were obtained immediately post-training meaning that there was no
opportunity for the trained group to practice their relaxation. The study was
carried out in Japan amongst a female group of health care workers. There
were 38 subjects in the treatment group and 41 in the control group with
100% retention. Outcome measures were mood state using the Iceberg
profile of mood states together with the objective measure of salivary
immunoglobulin (s-IgA).
63
Webb et al. (2000) (2/-), examined the effect on blood pressure and personal
strain of a progressive relaxation intervention for African-American women,
selected on the basis of an established risk of hypertension in this population
group. 48 subjects enrolled and 43 completed the study.
The baseline
Three papers evaluated the impact of massage therapy or touch therapy. The
effect of massage therapy on anxiety was studied by Shulman and Jones
(1996) (2/++) in a RCT. The therapy consisted of 15 minutes of at chair
massage once a week for 6 weeks. Controls had a 15 minute break.
Participants were office workers in an American company with more females
than males and an average age of 40 years. A total of 34 subjects
volunteered to participate and were randomly assigned (18 intervention, 16
control). One control group member failed to complete the study. Follow-up
measures were obtained 12 weeks from the start, approximately 3-4 weeks
from the end of the massage period. Wellbeing measures were STAI, using
both the state and trait subscales.
Each data set is formed from the individual means of two sets of
questionnaires completed 1 week apart for the pre-test and delayed post-test
(12 weeks) scores and 3 weeks apart for the immediate post-test (weeks 3
and 6 of treatment period). For state anxiety, all three sets of scores were
significantly different from each other for the massage group (scores fell
significantly and then rose significantly to intermediate value in delayed posttest. There was a significant difference between the two groups at post-test
(during intervention) but not for the delayed post-test. There were no
differences in trait anxiety between the two groups but a general trend for
scores to reduce across the experiment. Subjects had higher than population
norms for STAI (both subscales) suggesting a highly anxious group. This was
explained by the authors in terms of ongoing downsizing within the
65
participating company. Use of a break balanced out the effect of the massage
break to some extent although a break for 'non-work' might not be as
distracting as massage, which might enhance any difference. For example it
is not known if massage allowed conversation. Post-test questionnaires are
believed to have been completed shortly after massage and the data
suggests limited persistence of any effect.
McElligott et al. (2003) (1/+), utilised a RCT to examine the effects of touch
therapy on relaxation and anxiety amongst nurses. Touch therapy was
compared with laying on of hands in single 45 minute sessions (plus 15 mins
for test administration). A total of 12 commenced the intervention study, with 8
controls. Although all touch recipients remained in the study only three touch
controls remained on completion. Subjects were mainly (75%) female, aged in
their 30's and degree educated. Before and after each treatment session only
66
(no follow-up) anxiety was recorded on a single Visual Analogue Scale and
measures were obtained for blood pressure, heart rate, pulse oximetry and
respiration rate.
67
The relaxation training interventions reviewed were a mix of RCT and nonRCT studies ranging in quality between - and ++.
inconclusive in that two did not find an impact on mental wellbeing whereas
one, rated as -, found an improvement in outcome measures.
There is
Two studies evaluated the impact of massage therapy, one a RCT (Field,
1997) rated as ++ and one a Non-RCT rated as ++ (Shulman and Jones,
1996). The studies contradicted each other, with the RCT finding no effect
and the Non-RCT finding an impact on the STAI. There were a number of
issues with regard to the Field (1997) study which made it difficult to
68
69
Evidence Statement 7
A randomised trial comparing aerobic and nonaerobic exercise
(Altchiler and Motta, 1994+) found that aerobic exercise had a positive
impact on anxiety and other questionnaire-based stress measures. A
randomised control trial with Australian casino workers evaluated a 24
week out of work time programme which combined aerobic exercise
(moderate to high intensity) for 20 minutes on three days per week;
weight-training (light to moderate intensity) at least twice a week
exercise; and behaviour modification interventions (health education
seminars and health counselling) found mental health and other health
benefits when measured at the end of the programme.
There is currently insufficient research available to support the use of
relaxation training to improve mental wellbeing further research
required.
There is currently insufficient evidence to support the use of massage
therapy in promoting mental wellbeing further research required.
One US randomised trial comparing transcendental meditation with a
more conventional stress management programme (Sheppard et al.
1997 +) found a positive impact on mental wellbeing in the longer term
further research required.
1 Altchiler and Motta, 1994 RCT 1/+
2 Atlantis et al. 2004 RCT 1/+
3. de Lucio et al. 2000 RCT 1/+
4. Taniguchi et al. 2007 Non-RCT 2/++
5. Webb et al. 2000 Non-RCT 2/6. Field, 1997 RCT 1/++
7. Shulman & Jones, 1996 Non-RCT 2/++
8. McElligott et al. 2003 RCT 1/+
70
71
In a RCT, Hasson et al. (2005) (1/++) examined the effects on mental and
physical well-being and stress-related biological markers of a web-based
health promotion and stress management initiative. However, the control
group had access to the website but not to the lifestyle training & chat
elements meaning that these were effectively the controlled elements under
examination. A total of 129 (treatment) and 174 (controls) were drawn from
volunteers working for a Swedish IT and media company. It is not known what
the total number of potential participants was although there is an estimate
given of 80% take-up overall. The treatment group had a slight excess of
males over the control group and both had a mixed age distribution. No
separate analyses of pre-test demographics were reported although these
and other demographic variables were allowed for in an ANCOVA. 121
(treatment) and 156 (controls) remained at the end and data from those who
dropped out was analysed on an intention to participate basis. Outcome
measures included physiological markers (Cardiovascular and lifestyle; Stress
related; Recovery related; Immune markers and neuropeptides) although only
those regarded as stress-related were reported in this paper. In addition, a
self-rated health and stress questionnaire (non-standard) was administered.
Data were collected at baseline and at 6 months after the web site and related
components became available.
72
There was no significant group by time interaction in the analysis of the blood
pressure data. For the STPI scales only one subscale (entitled curiosity)
showed any significant effect (p<0.05) with the treatment group having an
increased score whilst the control group mean decreased. This was a
comprehensive Health Promotion initiative of which mental well-being was
only a small aspect of outcome measurement (although stated as a key part
of the study). The only statistically significant emotional measure was that of
curiosity which is of questionable relevance to wellbeing.
74
Evidence Statement 8
A randomised control trial undertaken in Sweden (Hasson et al. 2005 ++)
with 129 receiving the intervention and 174 controls, drawn from
volunteers working for a IT and media company, found that a web based
health promotion and lifestyle training package can improve mental
wellbeing as measured using a non-standard questionnaire at baseline
and at 6 months after the web site and related components being
available.
1. Hasson et al, 2005 RCT 1/++
3.2.2.5 Others
Six studies were assessed which examined very different modes of stress
reduction from any of the general groups above or each other. Alford et al.
(2005) (2/++) used an approach entailing written expression of emotions;
Kawakami et al. (1999) (1/+) mailed advice on dealing with stress; Martin and
Sanders (2003) (1/+); utilised a Positive Parenting project to reduce work
stress; Nhiwatiwa (2003) (2/++); used a leaflet to reduce the extent of distress
following assault; Wachi et al. (2007) (1/++) assessed the benefits of
Recreational Music Making; and Wilson et al. (2001) (1/+) used Eye
Movement Desensitisation and Reprocessing. All of them received positive
ratings. Those by Alford et al. (2005); Wachi et al. (2007); and Wilson et al.
(2001), had positive effects although the results from Wachi et al. (2007), are
difficult to interpret because they utilised a crossover design and the
significant effects were only obtained with the second group who had
previously served as the control group. The other three (Kawakami et al.
(1999); Martin and Sanders (2003); and Nhiwatiwa, (2003)) failed to show any
effect.
Alford et al (2005) (2/++), reported on a controlled exploration of whether
written expression of emotions by employees in stressful occupations helped
to minimise stress reactions. The intervention required participants to write
about recent stresses, emotions, related thoughts and plans in a journal for
75
15-20 minutes per day, for 3 consecutive days. The control group had no such
requirement. The study was carried out with a group of child protective service
officers, in Queensland, Australia, 85% of whom were women with an average
age of 35.2 years. At the onset 34 were assigned to the treatment group and
31 to the controls. The first week involved completion of questionnaires
regarding demographic variables together with pre-treatment versions of the
outcome measure questionnaires. The intervention then started at the
beginning of week 2. At the end of week 2 treatment and control group
members were asked to complete the post-treatment questionnaires, 31
(treatment) and 30 (controls) did so. Although the intervention continued, the
comparative element of it effectively ceased at this point as subsequent
assessments only involved the treatment group. Wellbeing outcome
measures were obtained using the GHQ-12; PANAS (Positive & Negative
Affect Schedule); and the JIG (Job in General) scale.
Kawakami et al. (1999) (1/+), used a RCT to explore the effects of a mailed
advice leaflet on stress reduction on psychological distress, blood pressure,
serum lipids and sick leave. Following a large scale Health Risk Assessment
(HRA) programme, the results of individual HRAs were mailed out to
participants. Those allocated to the treatment group also received advice on
dealing with stress in the same mailing plus written advice tailored to the
individual on physical activity, nutrition, breakfast, alcohol, relaxation, etc. This
therefore constituted the intervention. The study was carried out amongst
employees of a Japanese manufacturing company who obtained a GHQ-12
score of 3 or more in the initial HRA. A total of 113 were initially selected in
two groups. However, other exclusions at baseline reduced numbers to 91
(intervention) and 88 (control). At one year 81 & 77 remained of whom 48 and
76
No effect of training on work stress (scale referred to not given) was found
although post-training levels fell in the treatment group and rose in control
group. Training had a beneficial effect on child behaviour and other measures
not of concern here, but these did not manifest themselves in terms of stress
although a positive trend was apparent. Analyses of the data obtained from
those who dropped out suggest no difference to those remaining but the high
rate of attrition is of concern.
At follow-up the treatment group had higher levels of distress (IES) postintervention (change = +2, compared to -6 in controls; p<0.03). There was no
difference in GHQ-28 although mean scores went up in the treatment and
down in the control group. The study took no account of previous number of
assaults experienced or of the severity of any assault which might have
influenced the outcomes. There is poor reporting of the sampling strategy
used and of the numbers involved, or potentially involved, at different stages.
Wachi et al. 2007; (1/++;) examined the effect of recreational music making
on natural killer cell activity, cytokines and mood states of employees of a
Japanese manufacturer of electrical products in a RCT crossover design
intervention. Recreational Music Making (RMM) involved one 3-hour
intervention in which participants followed a pre-determined programme,
mainly involving the use of drumming with a variety of instruments. Controls
undertook 3 hours of leisurely reading, choosing from a selection of the latest
issues of magazines and newspapers. Potential subjects were excluded if
they regularly listened to drumming music or had previously participated in
drumming sessions. There were 20 (treatment) and 20 (controls) in the first
part of the crossover study which provided the main case-control comparison.
In the second part there were 20 and 19 respectively. There were six months
between phases. Outcome was assessed through use of the Mood States
questionnaire (POMS) and measures of natural killer (NK) cell activity.
because, by this stage, the control group had already received the treatment
(albeit with a break of six months).
However, the
evidence is limited for each of those studies as they are the only papers
currently assessing the impact of such interventions.
80
Evidence Statement 9
All studies utilised questionnaire-based outcome measures.
There is evidence from one 1 study that written expression of emotions
can improve mental wellbeing in the short term further research
required
There is evidence from one 2 study that recreational music making has a
positive impact on mental wellbeing in the short term further research
required
There is evidence from one 3 study that eye movement desensitisation
has a positive impact on mental wellbeing in the short term further
research required.
1 Alford et al. 2005 Non-RCT 2/++
2. Wachi et al. 2007 RCT 1/++
3 Wilson et al. 2001 RCT 1/+
81
It is widely recognised that, in any individual, stress and poor mental wellbeing usually has a multifactorial causation. In a work context, advice from
the HSE (www.hse.gov.uk) indicates that both work and non-work factors can
be involved. Many of the studies which provided training to cope with stress
(3.2.2.1) or counselling and therapy (3.2.2.2) would not differentiate the
origins of stress and would be equally effective in countering the effects of
82
The convention in workplace health and safety, whatever the risk to health, is
that removal or reduction of the risk at source is preferred to interventions
which ameliorate its effect (e.g. Reynolds 1997).
One paper which did was that by Martin and Sanders (2003) who
demonstrated a significant benefit from a Positive Parenting scheme.
However, the study failed to demonstrate a significant effect on specific stress
parameters, although it did have a beneficial effect on child behaviour and
other measures (not examined in for this review).
interventions
in
addressing
stress,
anxiety
or
83
To the extent that they have been shown to have any significant benefit, a
number of intervention modalities such as the exercise and relaxation
interventions involving aerobic exercise, massage therapy and transcendental
meditation (3.2.2.3) or others not directly related to work such as Recreational
Music Making (3.2.2.5) are independent of the nature of the occupational
group or organisation and could therefore presumably be applied in any
context.
84
Evidence Statements 11
a There is evidence from two studies1,2 that aerobic exercise had a
positive impact on anxiety and other stress measures.
b There is currently insufficient3,4.5 research available to support the use
of relaxation training to improve mental wellbeing further research
required
c There is currently insufficient6,7,8 evidence to support the use of
massage therapy in promoting mental wellbeing further research
required
d There is evidence from one paper9 that transcendental meditation has
an impact on mental wellbeing in the longer term further research
required
e There is evidence from one 10 study that recreational music making
has a positive impact on mental wellbeing in the short term further
research required
1 Altchiler and Motta, 1994 RCT 1/+
2 Atlantis et al. 2004 RCT 1/+
3. de Lucio et al. 2000 RCT 1/+
4. Taniguchi et al. 2007 Non-RCT 2/++
5. Webb et al. 2000 Non-RCT 2/6. Field, 1997 RCT 1/++
7. Shulman & Jones, 1996 Non-RCT 2/++
8. McElligott et al. 2003 RCT 1/+
9. Sheppard et al. 1997 RCT 1/+
10 Wachi, 2007 RCT 1/++
On the basis that all employees in the UK are legally entitled to holiday
breaks, the study of Etzion (2003) demonstrating the short-term benefits of
85
86
Evidence Statements 12
There is evidence from one study1 that taking a vacation impacts
positively on burnout in the short term.
There is evidence from one study2 that changing the shift system from 7
day consecutive shifts to the 35 day Ottawa system can positively
impact on mental wellbeing
There is evidence from two small studies3,4 that Psychosocial
Intervention (PSI) training has a positive impact on burnout in the short
term.
1. Etzion, 2003 Non-RCT 2/+
2. Totterdell & Smith et al. 1992 Quasi-experimental 2/+
3. Doyle et al. 2007 RCT 1/++
4. Ewers et al. 2002 RCT 1/++
Although
not
usually
detailed
in
the
papers,
individually-orientated
Although not
reproduced here, the evidence statements given in these sections will assist
employers from different occupational groups and organisations.
87
3.6 What help do employers need to review and adapt working practices
and conditions to promote the mental wellbeing of employees?
What are the barriers and facilitators to implementation of any of the
above interventions for both employers and employees?
Summary statement
No papers were identified as part of this review which specifically addressed
the help which employers need to review and adapt working practices.
Similarly, no papers systematically explored the barriers and facilitators in
respect of implementing an effective intervention. However, some guidance
can be provided from the discussions presented, particularly where studies
either had no or only limited success in effecting beneficial change. This
might assist in helping employers and identifying barriers and facilitators. To
a large extent, this guidance echoes that provided in the UK by the HSE,
although very few of the papers emanated from UK researchers or describe
work carried out in the UK. For clarity, not all papers making a particular point
have been cited below. There is no particular significance to the order in
which points are listed.
88
Belief: Associated with that commitment is the belief that change will
happen. Mikkelson and Saksvik (1999) refer to the difficulties arising
when planned or proposed interventions are not introduced. In some
instances, as referred to by Landsbergis and Vivona-Vaughn (1995),
this arose because the interventions were limited to those which could
be introduced within the department. Lack of belief leads inevitably to
cynicism, disillusionment and a self-perpetuating cycle of decreasing
belief.
89
Evidence Statement 13
There is currently limited evidence1-9 to support the use of
organisational participatory interventions in the workplace to improve
mental wellbeing - further research is required.
1. Bourbonnais et al. 2006 Quasi-experimental 2/2. Dahl-Jorgensen et al. 2005 Quasi-experimental Non-RCT 2/+
3. Kawakami, 1997 Quasi-experimental 2/4. Landsbergis & Vivona-Vaughan, 1995 Quasi-experimental 2/+
5. Maes et al. 1998 Quasi-experimental 2/+
6. Mikkelsen & Saksvik, 1999 Non-RCT 2/+
7. Mikkelsen et al. 2000 Non-RCt 2/+
8. Mattila et al. 2006 Quasi-experimental 2/9. Munz et al. 2001 Quasi-experimental 2/10 Reynolds, 1997 Quasi-experimental 2/-
90
91
not always possible and that alternative approaches are often required, either
instead of or as well as organisational measures.
A number of the individual intervention studies were hampered by relatively
poor adherence to intervention regimes, even where formal attrition from the
study in terms of failure to complete all the evaluation measures was relatively
low. In most studies, this deficiency was not formally addressed and so the
implications from this for study outcome can only be surmised. It might be
assumed that better adherence would have resulted in more positive
outcomes but this cannot be stated with certainty. This is a limitation which
could be usefully addressed in further research, as there is an implicit
assumption in evaluating a particular intervention that those taking part have
actually received the intervention.
With the organisational interventions a similar problem applies. Many of the
authors comment adversely on the quality of implementation of interventions
with lack of management commitment frequently referred to as a specific
problem.
directing the intervention itself (quite correctly as it would have impaired their
ability to carry out an impartial evaluation) and the studies took the form of a
discovered experiment, rather than the individual interventions where the
researchers have had greater control over the intervention.
The studies
92
4. Evidence Tables
Table 2.Overview of Interventions Changing Working or Organisational Practice
Author and
Date
Bourbonnais
et al. 2006
Category of
Intervention
Organisational
Study
design and
research
type/
quality
Quasiexperiment
al (procohort)
2/-
Research
question
How does an
intervention
designed to
reduce adverse
psychological
factors in a
hospital affect
the
psychological
health of
workers in that
hospital?
Study
population,
setting,
country,
sample size
492 expt and
618 control at
baseline.
302 and 311 at
end.
Canada,
Hospital
nurses and
auxiliaries,
mainly female,
mixed ages
and
experience
Description of
intervention(s)
Introduction of
programme of
measure to
reduce adverse
psychosocial
factors
Length
of
followup
Mental
Wellbeing
outcome
variables
12
months
after
beginni
ng
Psychological
distress (PSI Psychologic
Symptoms
Index);
Personal and
client-related
burnout (CBI Copenhagen
Burnout
Inventory);
sleeping
problems
(NHP Nottingham
Health Profile)
Short term
findings (<1
year)
The comparison
of post
intervention
mean scores
between both
hospitals,
adjusting for preintervention
scores, proved
favourable to the
experimental
hospital; the
mean difference
was statistically
significant for
work related
burnout (p=0.03)
and borderline
for client related
burnout (p=0.08).
Psychological
distress and
sleeping
problems were
not sig.
Long term
findings (>1
year)
n/a
Comments,
confounder
s/ potential
sources of
bias
Most
planned
interventions
yet to be
implemented
at end so
good result.
Hospital with
good
expected
cooperation
selected so
results might
be biased by
this.
Analysis of
drop out
suggests no
survivor bias.
Subjects
were aware
of nature of
study and
status as
experimental
hospital.
Applica
bility to
the UK
DahlJorgensen et
al. 2005
Organisational
Quasiexperiment
al (NonRCT)
2/+
Kawakami
1997
Organisational
Quasiexperiment
al (procohort)
2/-
What is the
effect of an
organisational
intervention on
levels of workrelated stress
amongst
Japanese bluecollar workers
with pre-existing
high depressive
scores?
415/560
responded to
initial
questionnaire,
only 336 of
these were still
in same
employment at
post-test and
282
responded.
Breakdown of
response by
group is not
given
Shop workers
and municipal
workers in
Norway.
Mainly female,
approx 2550% part time,
12-25%
managers, av
age around
40yrs.
Japan,
Worksites with
mean
depression
scores higher
than mean
+1s.d. for
whole
company.
111 workers at
2 sites
matched for
age and mean
depression
score with 3
other site (183
workers).
Participatory
approach to
organisational
change. Different
specific
approach in two
workplaces
(shop workers
from a number of
shops in mall).
Not
clear,
about 6
months
post
interven
tion
SHC (Somatic
symptoms)
and burnout
(MBI)
With municipal
workers there
were no sig
changes in
expected
direction but
'emotional
exhaustion' was
sig increased in
intervention
group (p=0.05).
With mall
workers
depersonalisatio
n (p<0.05) and
somatic
symptoms
(p<0.05) sig
improved
compared to a
decrease in
controls.
n/a
Poor
implementati
on of
interventions
for a variety
of reasons
including
management
attitudes,
especially in
municipal
employer
and adverse
attitude to
study (again
municipal).
Authors
regard
changes as
Hawthorne
rather than
genuine
impact.
1 year
Workplace
participatory
intervention
aimed at
identifying &
reducing sources
of stress in
workplace.
1&2
years
Zungs SDS
score
(depression)
For intervention
group:
For
intervention
group:
ANCOVA
allowed for
pre-test
differences
in age or
outcome
scores.
Cannot rule
out
systematic
differences
between
worksite
groups.
Intervention
relied on
supervisor
support and
Sick leave
Blood
Pressure (2yr)
Progressive
reduction in
depressive
symptoms
across 2 years.
Total effect sig
(p=0.035)
No effect on BP.
sig gp x time
interaction for
sickness
(p=0.034)
Progressive
reduction in
depressive
symptoms
across 2
years. Total
effect sig
(p=0.035)
No effect on
BP.
94
Landsbergis &
VivonaVaughan 1995
Organisational
Quasiexperiment
al (nonRCT)
2/+
What is the
effect of an
occupational
stress
intervention in 2
municipal
departments on
levels of strain
and depression
in employees?
39 intervention
1; 10 control 1;
24 intervention
2; 26 control 2.
37, 23, 23 &
20 at follow up.
77 pairs (some
recruitment)
US municipal
employees in 4
paired
departments.
Mixed gender,
typically
around
30years av
age, mainly
caucasian.
Control 1 had
sig higher
proportion of
clerical, lower
paid, noncollege
educated
Data suggests
increase in year
1 (no separate
analysis)
Participatory
Action Research
(Organisational
Intervention) in
two different
departments
12
months
posttest
Depression
and sleeping
subscales
from Job
Content
Questionnaire
Neither
intervention had
any sig effect on
either outcome
measure
n/a
did not
involve
workers in
planning
stage. No
effect on
minority of
female
workers.
Other factors
such as
increased
work
demands on
intervention
sites during
follow-up.
This quasi
experimental
study used 2
pairs of
sections in 2
US municipal
departments
to mount 2
parallel
intervention
studies
(PAR). The
authors
express
doubts about
the quality of
the
intervention,
especially in
one of the
two depts
including
limited
worker
participation
95
Yes
employees
than
intervention 1
otherwise no
sig diffs.
Maes et al.
1998
Individual/Organ
isational
Quasiexperiment
al (NonRCT)
2/+
175 treatment,
171 control
Workers.
167, 157, 134
treatment at
T2-T4.
169, 157, 130
control at T2T4.
Dutch,
Household
goods
manufacturing.
1 intervention
site, 1 control
(random) plus
second control
site.
Extensive
battery of
interventions
including lifestyle
: (exercise,
healthy eating);
Stress, training
on social skills
and leadership,
plus
organisational
change - working
methods
3 years
SACL-90
(general
stress)
Absenteeism
No effect on
general stress
reactions;
Decrease of
8%
absenteeism
in treatment
group
compared to
5% on control
group.
(no stats
reported)
Extensive
longitudinal
study with
large battery
of measures,
mainly
focussed on
healthy living
and lifestyle
although
some
specifically
mental wellbeing.
Difficult to
determine
relative role
of different
aspects.
Treatment av
age 38.6yrs,
control 40.9
yrs (ns)
Treatment
26.1% female
sig more than
control
(12.2%)
Controls had
higher
educational
status.
96
Yes
Mattila et al.
2006
Organisational
Quasiexperiment
al (NonRCT)
2/-
253
intervention;
107 control 1;
165 control 2.
No details if
this is outset or
completion.
Finnish
municipal
workers.
Mainly men in
manual work,
typically about
44yrs old (av)
with most
leaving school
before
matriculation
Participative
work conference
but apparently
no actual
implementation
of change
Post
test 1,
immedi
ately at
conclusi
on of
interven
tion
period
(12
weeks)
and
posttest 2
12
months
after
onset of
study
Maslach
Burnout
Inventory general
survey
One general
stress
question
Participation had
no effect on
emotional
exhaustion or
perceived stress
n/a
Participation
in
'conference'
was
mandatory but many did
not take part.
No reasons
are given for
this but
these people
then formed
the first
control group
(so no
randomisatio
n) addition of
second
control group
offsets this.
Intervention
seems to
have been
about
identifying
need for
change and
establishing
nature of
those
changes but not
actually
implementin
g them - so it
is perhaps
not
surprising
that it had no
effect.
97
Yes
Mikkelsen &
Saksvik 1999
Organisational
Non-RCT
2/+
What impact
does a
participatory
organisational
intervention
have on job
stress and job
characteristics?
37 treatment
gp 1; 59
treatment gp 2;
31 control gp
1; 35 control
gp 2
33, 58, 29, 33
at post-test 1
16, 56, 23, 30
at post-test 1
Norway,
Postal
workers, Gp 1
30's gp 2 40's,
more female in
all
groups.differen
ces between
groups in
some
demographics,
allowed for in
analysis
2 intervention
groups in
different post
offices in 2
cities plus
control groups
from other
offices in same
city
Participation in
programme to
identify and
develop
workplace
changes. Initial 6
hour seminar
then Work subgroups met for 2
hours per week
over 9 weeks.
Post
test 1,
immedi
ately at
conclusi
on of
interven
tion
period
(12
weeks)
and
posttest 2
12
months
after
onset of
study
Cooper Job
Stress; UHI,
Work (Health
Inventory) and
STAI-T
plus
organisational
measures
Both
interventions had
no effect on job
stress health
inventory or Trait
anxiety.
n/a
Participatory
approach to
improving
work place
and work
organisation.
Some
isolated and
inconsistent
effects on
organisation
al
parameters
but key
outcomes
unaffected
by
interventions
Poor
adherence to
intervention
programme,
especially
intervention
1.
Change
limited to
local factors.
98
Mikkelsen et
al 2000
Organisational
Non-RCT
2/-
What effect
does a
participatory
organisational
intervention
have on job
stress in
community
health care
institutions?
Not clear.
Demog data
reported for
max of 45
treatment, 34
control
Supervisors
and
employees at
2 health care
institutions.
Predominantly
female, middle
aged
Participation in
programme to
identify and
develop
workplace
changes. Initial 6
hour seminar
then Work subgroups met for 2
hours per week
over 9 weeks
Intervention v
nointervention. 2
intervention
groups, results
merged
Munz et al.
2001
Stress
management
Quasiexperiment
al (procohort)
2/-
55 intervention
24 controls
USA,
Telecommunic
ations
customer
service/sales
reps. Very
limited
demographic
details. No sig
diff in years
worked for
company or
years in
Combination of
selfmanagement
and
organisational
stress
reduction
interventions.
3 X 4 hours
selfmanagement
skill
development
+ manual.
Participation
Post
test 1,
immedi
ately at
conclusi
on of
interven
tion
period
(12
weeks)
and
posttest 2
12
months
after
onset of
study
Post test
1,
immediat
ely at
conclusio
n of
interventi
on period
(12
weeks).
Cooper Job
Stress; UHI
(subjective
health)
plus
organisational
measures
Perceived
Stress Scale
(PSS);
Depression
(CES-D);
Positive and
Negative
Affect
Schedule
(PANAS)
n/a
Very similar
participatory
approach to
that in earlier
Mikkelson
paper but
appears to
have had
more
success
although loss
of follow up
is
disappointing
Yes
Two groups
from different
institutions
so changes
could be due
to other
factors.
n/a
Study took
place at time
of
considerable
change in
organisation.
Combination
of
participatory
and
individual
intervention not possible
to
differentiate
effects of
different
parts of
package.
Concerns
over the
initial
99
Yes
present job.
No age or
gender info.
Recruitment
not clear.
Approximately
150
employees in
each potential
pool. 55 took
part in the
intervention
with 24
controls. Not
stated how
many were
invited.
voluntary (all
55 took part).
4-6 work unit
employees in
participatory
programme to
improve
workplace.
affect (more
relaxation)
(p<0.05).
Intervention
group also
showed better
increase in
productivity and
lower
absenteeism.
selection.
Participation
in course
was
voluntary
and so selfselected
group
entered
intervention.
Controls
were from
different
work units
and so might
not have
been
comparable
although
ANCOVA
used to allow
for preintervention
differences.
Changes
restricted to
within work
unit.
Genuine
effect,
concern is
over extent
to which it
can be
generalised.
100
Reynolds 1997
Both
Quasiexperiment
al (procohort)
2/-
37 in Area A
(counselling)
76 in Area B
(organisational
) 43 in Area C
(control)
numbers
actually taking
part not stated
Area A received
individual
counselling,
Area B had
organisation
change, Area C
was control
12
months
after
interven
tion
period
GHQ-12 and
SCL-90R
(somatisation)
At 1 year sig
AreaXTime
interaction
(p<0.049). SCL90R decreased
in Area A and
increased in
B&C. According
to text same
effect was seen
with GHQ-12 but
figure doesn't
support this shows reduction
for B as well.
n/a
Three
management
programmes:
Mentor
programmes,
network
programmes and
Lecture group
versus no
intervention.
Questio
nnaire
followup at
one
year
Quality, work,
competence
tool (QWC)
used to
determine
individual well
being,
Improvement in
sickness
absence in
trainees cf
controls(p<0.05;
1.3 v 8.2d)
n/a
No
demographic
details given.
Workers in
council
housing
department
No reference
to method
used to select
departments
Von Vultee et
al. 2005
Organisational
Non-RCT
2/+
What is the
effect of
management
programmes on
physicians work
environment
and health?
52 in
intervention
group, 52 in
control group
42 & 42 at
follow up
25 in mentor
programme;
12 in network
programme; 5
in lecture
programme
No details of
programmes
Sickness
absence
According to
paper,
counselling
but not
organisation
al
intervention
worked in
reducing
physical and
psychologica
l symptoms.
Conflict in
data
reported
makes this
difficult to
interpret.
Also lack of
any
information
on
participation
rates and
demographic
comparisons
diminishes
value of
study. Fuller
paper cited
not located.
Lack of detail
on nature of
training
including
duration
makes
findings
difficult to
determine.
Trainees
could have
had time off
from clinical
duties or
101
Yes
Sweden,
Female
physicians
from 6 different
hospitals.
Allocated to
management
programmes
by managers
or HR.
Av age 47.5
yrs
could have
been
expected to
fit training
around such
duties.
Non random
allocation to
training.
No separate
analysis of
forms of
training.
102
Kawakami et
al. 2005
Category of
Intervention
Organisational
Study
design and
research
type/
quality
RCT
1/++
Research
question
What effect
does web-based
training for
supervisors on
supervisor
support have on
psychological
distress
amongst
subordinate
workers?
Study
population,
setting,
country,
sample size
Supervisors - 9
(intervention)
& 7 (controls);
Workers 100
(intervention)
& 90
(controls).
82 & 84 at
follow up
Description of
intervention(s)
Web-based
training of
supervisors on
work-site mental
health. 4 week
training period
based on
Guidelines for
Promoting
Mental Health
Care.
Length
of
followup
Mental
Wellbeing
outcome
variables
4
months
Brief Job
Stress
Questionnaire
(BJSQ)
subscales for
vigour,
anger/irritabilit
y, anxiety &
depression.
Short term
findings (<1
year)
No sig.
intervention
effect for any of
the 5 sub-scales
of psychological
distress.
Long term
findings (>1
year)
n/a
Japan,
Technicians
and clerks in
an IT company
Comments,
confounder
s/ potential
sources of
bias
Measures
suggested
that
supervisors
learned
messages
from training
but that this
was less-well
perceived by
subordinates
possibly due
to peak in
work
demands.
No measure
of knowledge
and
understandin
g of control
supervisors
at baseline.
Kawakami et
al. 2006
Organisational
RCT
1/++
What effect
does web-based
training for
supervisors on
supervisor
support have on
Supervisors 23
(intervention)
& 23
(controls);
Workers - 92
Web-based
training of
supervisors on
work-site mental
health. 4 week
training period
4
months
Brief Job
Stress
Questionnaire
(BJSQ) psychological
distress
No sig.
intervention
effect for total
psychological
distress.
n/a
More
females in
intervention
group
might be
affected by
male
supervisors.
Training
increased
knowledge
and attitude
of
supervisors
103
Applica
bility to
the UK
Yes
psychological
distress
amongst
subordinate
workers?
(intervention)
& 114
(controls)
81 & 108 at
follow up
based on
Guidelines for
Promoting
Mental Health
Care.
Japan,
Sales &
service
workers in an
office
machines
company
Logan, and
Ganster 2005
organisational
RCT
1/+
What effect
does training
Project
Managers have
on their
psychological
strain?
34 trainees: 33
controls
34 & 33 at 7
weeks post
intervention
23 & ?? at 17
weeks post
training
USA, Canada,
Mexico.
PMs from
trucking
company.
mainly male,
av. Age 36.8y,
managerial
Training to
increase
management
control,
organisational
change to
facilitate
increased
control. 10
hour training
session
17 weeks
postinterventi
on
Depression
(CES-D),
Anxiety
(Caplan)
No main effect
on well-being
outcomes
n/a
(which was
perceived by
employees)
but had no
effect on job
stressors.
Pre-existing
high levels of
support.
Limited
scope for
influence of
support on
sales staff
working out
of the office.
Improved
organisation
al control did
not enhance
well-being.
Trainees
were
directed to
attend
training on a
non-work
day.
Diffusion to
non-trainees
unlikely.
Some
suggestion
that training
raised
expectations
therefore
diluting any
beneficial
effect.
104
Yes
Takao et al.
2006
Organisational
RCT
1/++
What is the
effect of job
stress training
for supervisors
on psychological
distress and job
performance of
immediate
subordinates?
154
intervention,
101 control
134 & 92 on
completion
office and
manual
workers in
Japanese sake
brewery;
Intervention
group sig more
blue collar and
fewer years of
education.
Genders
reasonably
balanced
Supervisors
received training
which they were
then expected to
apply to
immediate
subordinates.
60 minute
lecture + 120
min active
listening training
3
months
post
interven
tion
Brief Job
Stress
Questionnaire
(vigour, anger,
fatigue,
anxiety,
depression,
somatic
symptoms)
No sig main
effects following
intervention.
Subgroup
analysis showed
younger male
white collar
workers to have
sig (p=0.012)
+ve effect of
intervention on
psychological
distress summing
subscales of
JSQ.
n/a
Interesting
study in that
intervention
was indirect training
supervisors.
Well
designed in
allowing for
possible
covariates as
subject
groups were
not balanced
in design
(depended
upon who
their
supervisor
was who had
been
randomly
allocated to
training or
not). Ratings
of 'supervisor
support' from
Job Content
Questionnair
e did not
show any
differences
although not
clear when
this was
measured
(no
intervention
change
reported).
105
Yes
Theorell 2001
Organisational
Non-RCT
2/++
(bloods)
2/+
(questionna
ires
223 treatment,
260 controls
workers at
Swedish
insurance
company
176 & 168
baseline blood
sampling
153 & 156 at
follow up
bloods
139 & 132 at
follow up
questionnaires
~60% female,
ages not given
psychosocial
training for
managers.
1 day at onset, 2
hours every 2
weeks for 6
months and 1
day at end.
1 year
Cortisol,
serum lipids,
serum GGT
(liver enzyme)
Swedish
version of
demandcontrol
questionnaire
No sig effect on
psychological
demands on
employees or
managers.
Sig reduction of
serum cortisol for
all (p=0.02) and
employees
(p=0.005) but not
managers in
intervention
group
Sig reduction of
serum GGT in all
(p=0.04) but not
employees or
managers
separately
n/a
Emergence
of subgroup
effect has
some
support from
other
aspects of
study although
could
possibly
mount
counterargument
Three other
management
programmes
were going
on at same
time as
intervention (
2 in control
groups)
which may
have
confounded
outcomes
though not
specifically
psychosocial in
nature. Also
fairly high
dropout rate
but unlikely
to explain all
of the
differences
106
yes
Bussing &
Glaser 1999
Category of
Intervention
Organisational
Study
design and
research
type/
quality
Quasiexperiment
al (procohort)
2/-
Research
question
What are the
effects of a
revised nursing
system on work
strain and
burnout
amongst nursing
staff?
Study
population,
setting,
country,
sample size
4 wards from
one hospital
(treatment)
and 13 wards
from another 2
hospitals
(control)
Description of
intervention(s)
holistic nursing
system
introduced in 4
wards in one
hospital
32 treatment,
75 control
Who remained
on same
wards
throughout.
German,
Nurses,
Length
of
followup
Mental
Wellbeing
outcome
variables
3 years;
1yr setup; 1 yr
transfer
phase;
1yr
testing
phase
T1 at
end of
set-up,
T2
towards
end of
testing
(i.e. ~2
years
later)
MBI-D
(German MBI)
(burnout)
Immedi
ately on
return
and 3
weeks
after
end of
vacatio
n
Burnout (not
MBI) with
physical and
mental
exhaustion
subscales.
Short term
findings (<1
year)
n/a
Long term
findings (>1
year)
Both burnout
subscales
increased
similarly and
significantly
(p=0.007) in
both groups
from T1 to T2.
This was
despite
measured
improvement
in
psychosocial
environment.
~90% female;
mean age 35
years, work
experience 15
years
Etzion 2003
Organisational
Non-RCT
2/+
What impact
does taking an
annual vacation
have on
perceived job
stress and
burnout
55 treatment
with 55 age,
gender and job
matched
controls
Israel,
industrial
employer,
mainly married
men, av age
~44yrs, mixed
jobs (white
collar)
taking a holiday
v not taking a
holiday
Burnout tended
to drop after
vacation and
stayed down
(p<0.10).
Job Stress fell
immediately after
but returned to
pre-vacation
levels (p<0.01)
n/a
Comments,
confounder
s/ potential
sources of
bias
Possible bias
due to
nurses
leaving any
of the wards
(initial
sample of
482). Also
clearly very
mobile
workforce
reducing
opportunities
for changes
to exert
influence.
Results
suggest any
experimental
effects
swamped by
other
influences.
No
randomisatio
n or
anonymity.
Subjective
scales.
Vacation
takers seen
as controlling
timing
might not
apply to non-
107
Applica
bility to
the UK
Yes
Yes
takers.
Totterdell &
Smith 1992
Quasiexperiment
al (NonRCT)
2/+
What is the
effect of a
change in
shiftwork rota
system on wellbeing and other
measures?
Not clear, as
initial
breakdown of
treatment and
control groups
not given. 150
questionnaires
sent out
initially.
31 treatment,
40 control at
completion.
UK police
officers
Treatment gp
sig older (34.2
v 29.5)no other
demo diffs
reported
Change in shift
system at 2
stations, cf 2
others in same
force
6
months
after
interven
tion
introduc
ed
GHQ-12
Change to
Ottawa shift
system was
associated with a
sig decrease in
GHQ-12 score
(p=0.001)
n/a
31/32 of
treatment
group had
used new
system for at
least 3 of the
last 6
months
Hospitals
and other
health care
establishmen
ts often work
a variety of
shift systems
to provide 24
hour cover
and cover for
peak
periods.
Recent work
on the
adverse
health
effects of
shift work
(cancer)
means this is
potentially a
very
important
finding of
particular
relevance
108
Yes
Table 5. Overview of Interventions for Support or Graining to Improve Skills or Job Role
Author and
Date
Doyle et al.
2007
Category of
Intervention
Individual/Organ
isational
Study
design and
research
type/
quality
RCT
1/++
Research
question
Can equipping
mental health
workers with the
skills to
integrate
psychosocial
interventions
(PSI) into
practice cut
levels of burnout
among staff?
Study
population,
setting,
country,
sample size
14 training and
12 control,
derived from
qualified staff
working at a
medium
secure unit in
the UK and
nominated to
attend training
by their work
group.
Description of
intervention(s)
PSI course, 3
hour sessions
weekly for 16
weeks during
work.
Length
of
followup
Mental
Wellbeing
outcome
variables
Not
stated,
believe
d to be
immedi
ately
post
course
(16
weeks)
MBI, 3
subscales for
personal
accomplishme
nt, emotional
exhaustion
and
depersonalisa
tion.
n/a
12
months
from
initial
implem
entation
but at
same
time as
last
element
sleep
disturbance
and perceived
stress
subscales of
SWQ
(Satisfaction
with Work
Questionnaire
)
No sig effects
although
perceived stress
improved in
intervention
group across
study
n/a
Short term
findings (<1
year)
Long term
findings (>1
year)
Organisational
Quasiexperiment
al (procohort)
2/+
What effect
does the
introduction of
new IT support
systems have
on
psychosomatic
health of mental
health care
workers?
No sig diffs.
27
intervention;
32 controls
falling to 17
and 16 at end.
Sweden,
Nurses
(licensed
practical
nurse) almost
all female av
age ~40yrs,
part-time
workers
IT measures to
improve
monitoring of
patients and to
provide relatives
with info.
Applica
bility to
the UK
Yes
Although
burnout not
affected PSI
was
apparently
used more
by training
group in their
work.
Av age 38yrs,
mainly female
nurses (3540% nonnursing)
Engstrom et al.
2005
Comments,
confounder
s/ potential
sources of
bias
Although
nominated
allocation to
groups was
random.
6 and 4
refused any
involvement
at baseline.
Allocation of
units to test
or control
was random.
Mainly parttime
workers.
Changes
added
109
Yes
responsibiliti
es as well as
making
some
aspects
easier.
Ewers et al.
2002
Individual/Organ
isational
RCT
1/++
Does training in
psychosocial
interventions
reduce burnout
rate in forensic
nurses?
10 + 10
stratified by
ward, gender
and day/night
duty.
20 day Training
in psychosocial
interventions
Immedi
ately
post 20
day
training
interven
tion
Maslach
Burnout
Inventory
subscales:
accomplishme
nt,
exhaustion,
depersonalisa
tion
Sig.
improvements
over controls:
accomplishment
(p=0.01),
exhaustion
(p=0.04)
depersonalisatio
n (p=0.01)
n/a
varied individual
and group
interventions.
Exercise,
empowerment,
IT training,
workshop on
development
20
months
Stress
symptoms &
vitality
n/a
sig. increase
in cognitive
stress reaction
at canteen A
(int: p<0.01);
B (int: p<0.05)
& C (cont:
p<0.01).
13 of initial
sample
declined to
take part.
Nielsen et al.
2006
Stress
management/
organisational
Quasiexperiment
al (procohort)
2/-
Mental health
care staff,
mainly nurses
varies:
Intervention
canteens: 27
and 17
completed
questionnaires
at both times;
control 11 and
16 did so.
Numbers at
pre-test
45,26,22,25
and at posttest 30, 26, 19,
28
sig increase in
vitality at
canteens A
(int: p<0.05);
B (int: p<0.05)
& D (cont:
p<0.01)
Limited time
for all
intervention
to have any
effect.
Small scale
study. PI
worked in
same unit.
Well-being
not main
focus of
paper.
Yes
Yes
Complex
array of
interventions
some
individual,
others
organisation
al. Different
interventions
at two sites.
Number of
criticisms by
110
canteen
workers at 4
hospitals/
elderly care
homes
Schaubroeck
et al. 1993
Organisational
Quasiexperiment
al (RCT)
1/+
What is the
effect of an
intervention
designed to
clarify individual
roles on
subjective
strain, physical
symptoms or
time lost through
illness?
authors of
process of
implementati
on of
interventions
Treatment group
received
supervisor role
clarification
(organisational
intervention)
T2 -10
months
after
pre-test
survey
(T1) T3
6
months
later.
Mental Health
Battery
(psychological
ill-health);
Somatic
Complaints
Index
No sig effect on
psych or phys illhealth although it
did reduce role
ambiguity and
supervisor
dissatisfaction
n/a
Many other
changes
extraneous
to project
(e.g. new
manger at
one site,
announceme
nt of
closures and
cuts at
another.
Lack of
clarity in
subject
groupings
makes
interpretation
unclear. The
authors
suggest that
there might
have been
some spread
of role
clarification
as a result of
the initial
meetings.
Also,
although
supervisors
agreed not to
apply
clarification
to those
employees in
control group
this would be
111
Yes
Shimizu et al.
2003
Stress
management/
Organisational
Non-RCT
2/+
What is the
effect of
communication
skills training on
burnout
amongst
Japanese
nurses?
19 treatment,
26 controls
12 & 14 at
completion
Allocated to
group by
supervisor
Japanese
female nurses
with low-tomoderate
communication
skills
Communication
training 2 x 2day sessions
5
months
J-MBI,
(Maslach
Burnout
Inventory)
significantly
greater increase
in personal
accomplishment
in treatment
group (2.6 vs 3.0 p); no
difference in
emotional
exhaustion or
depersonalisatio
n
n/a
difficult to
control for
and might
have diluted
any benefit.
Reduction of
role
ambiguity
was
nevertheless
seen as a
good step.
High dropout
rate and
nonrandomisatio
n of
participants
make results
difficult to
interpret.
Some
evidence of
change in
one aspect
of burnout
only
112
Yes
Aust et al.
1997
Category of
Intervention
Stress
management
Study
design and
research
type/
quality
Non-RCT
2/+
Bergdahl et al.
2005
Stress
management
RCT
1/+
Research
question
What effect
does a theorybased stress
management
approach have
on critical
coping
behaviour and
subjective
health
Can structured
affect-focussed
training reduce
heightened
levels of stress
and
psychological
symptoms such
as anxiety and
depression?
Study
population,
setting,
country,
sample size
26
(intervention)
28 (controls)
(22 & 24 at
end)
Length
of
followup
Mental
Wellbeing
outcome
variables
Stress
management
program (12
sessions)
1.5h 1 x per wk
after work
12
weeks
plus
addition
al follow
up
investig
ation
after 3
months
5weeks
after 7week
school
Positive and
negative
mood
questionnaire
based on
German
inventory of
QoL;
'Symptom'
questionnaire
based on
German
symptoms
inventory
PSQ
Perceived
Stress
Questionnaire
Description of
intervention(s)
Germany,
Inner city bus
drivers
Age 40 -60
(overall av
49.5y) min 5y
exp.
27 treatment,
23 control
(20 and 17 at
end)
Females
working in
areas of social
service, elderly
care and
education in
Swedish
municipality.
Randomly
selected from
top 50/120
high scores on
PSQ
1 x 2 hr session
per wk for 7 wks.
SCL-90
(Symptom
Check List)
and
GSI (Global
Severity Index
level of
psychological
symptoms)
Short term
findings (<1
year)
Long term
findings (>1
year)
No sig. change in
positive or
negative mood
scores or
Symptom scores
n/a
Treatment group
only - significant
decrease in PSQ
scores (0.46 to
0.37 p<0.01) and
GSI scores (0.72
to 0.57 p<0.05)
ANCOVA
showed
treatment/group
interaction for
GSI.
n/a
Comments,
confounder
s/ potential
sources of
bias
Volunteer
subjects.
Initially
poorer mood
scores in
intervention
group should
have
enhanced
chance of
effect.
Selection on
high PSQ
limits
generalisabili
ty.
Control
group did not
attend any
course so
results might
not be
related to
specific
course
content.
Lack of
volunteer
status of
subjects a
strength.
113
Applica
bility to
the UK
Yes
Yes
Stress
management
Non-RCT
2/-
Can either of 2
modes of
intervention
improve
psychological
well-being and
reduce strain?
45 IPP; 62
SMP; 84
controls at
outset.
20 IPP; 27
SMP; 70
controls at end
UK community
and hospitalbased health
workers, no
demographic
details given
but said not to
differ between
groups
Butterworth et
al. 2006
Stress
management
Non-RCT
2/-
What is the
impact of MIbased coaching
on the physical
and mental
health of
university-based
employees?
Initially 145
self-selected
into treatment
and 131 into
control groups.
112 treatment,
118 control at
end.
Subsidiary
case-control
study with 44
pairs matched
on propensity
score.
Health and
research
university
OHSU
Stress
Management
Programme
(SMP) aimed at
individual
behaviour and
coping;
Innovation
promotion
programme
(IPP) aimed at
organisational
improvements.
First
post
test
(T2) at
9
months
T3 - 15
months
from
outset
GHQ-12
Sig reduction in
GHQ T1 to T2
(p<0.01) in SMP.
Trend for
decrease in GHQ
T1 to T2 with IPP
but not sig (p not
cited).
Non-sig trend
GHQ increase
T2 to T3 in
SMP
suggesting
effect doesn't
last (p not
cited).
Decrease in
GHQ with IPP
sustained at
T3 but not sig
(p not cited).
Motivational
Interviewingbased health
coaching. 30
minute sessions,
minimum 1 initial
and 2 follow-up.
Actual number
determined by
participants
based on
(perceived) need
and interest
3
months
SF-12 giving
2 subscales:
PCS
(physical) &
MCS (mental)
Treatment group
improved on
PCS (1.69 pts,
p=0.035) and
MCS (4.4 pts,
p<0.0001).
Control group
didn't change sig.
Matched casecontrol study
showed similar
results (MCS
p=0.016) but not
sig. for PCS.
n/a
Occupational
groups
known to be
under higher
strain
targeted,
increasing
chances of
change.
Not all
subjects
were
volunteers,
some were
encouraged
to take part
by managers
no record
identifying
these to
allow
separate
analysis.
Volunteer
status and
apparently
differing
recruitment
methods
including self
allocation to
treatment or
control
seriously
undermines
results.
Derivation of
propensity
score
referenced
but not
explained.
May be other
variables not
114
Yes
Yes
(Oregon, USA)
workers;
average age ~
40; more men
in control
group (37%)
than treatment
(10%)
Cook et al.
2007
Stress
management
RCT
1/+
Craig 1996
Stress
management
Non-RCT
2/+
247 + 233
initially
15% & 13%
attrition
Office workers,
mainly white,
female, degree
educated,
salary >$50K
pa
143 initial
volunteers, 48
participants 41
completed
programme
(av age
41.5yrs); 28
attended
assessment
and 28
attended 2
year follow-up.
41 controls
drawn from
those of initial
143 unable to
attend course
matched for
gender (av age
40.2 yrs). 31
attended
stress
management,
diet and
physical
activity
3 month
test
period
followed
by posttest
on-line -vpaper
presentations
6 week, healthy
lifestyle
programme
focussing on
individual stress
management
and healthy
living. 1.5 hours
per week.
Testing
immedi
ately
postcourse
&2
years
later
Perceived
stress,
symptoms of
distress,
stress state of
change, brief
COPE
No diff between
2 groups on
ANCOVA but
print version
worked better
(p<0.01)
n/a
GHQ (form
not stated),
LAQ (Lifestyle
Appraisal
Questionnaire
) for perceived
stress (Part 2)
GHQ showed
2 year fall (ns)
controlled for
or other
factors
biasing
results.
Variability in
extent of
intervention
not
documented.
Examined
'dosage'
(access to
web
material) but
no sig effects
on stress
outcomes.
Selection to
treatment
group by
availability
and control
group by
unavailability
Very high
attrition rate,
especially at
2 year
follow-up.
Authors
attribute lack
of effects to
low statistical
power
115
Yes
Yes
assessment
and 17
attended 2
year follow-up.
Eriksen et al.
2002
Stress
management
RCT
1/+
Gardner 2005
Stress
management
Non-RCT
2/+
What effects do
Stress
Management
Training (SMT),
Physical
Exercise (PE) or
an Integrated
Health
Programme
(IHP) in a
workplace
setting have on
subjective
health
complaints?
What effects do
two forms of
stress
management
training have on
stress and
general health
(GHQ)?
Australia,
University
staff. Controls
77% support,
23% academic
participants
not given.
189 (SMT) +
165 (PE) +
162 (IHP) +
344 (Control)
(98, 114, 94,
166 by end).
Norwegian
postal workers;
~60% female.
Controls sig
less time in job
and fewer
hours per
week.
Mixture of blue
and white
collar jobs.
57 cognitive;
44 coping; 37
control
51, 38, 29 at
end of course
42, 37, 25 at
follow-up
UK NHS
employees,
Stress
Management
Training (2 hrs,
weekly for 12
weeks);
Physical
Education (1 hr,
twice weekly for
12 weeks);
Integrated
Health
Programme (2
hrs, weekly for
12 weeks);
Controls (no
stated treatment)
All during
working hours.
Pretest,
post 12
week
interven
tion, 1
year
followup
3, 3.5 hour
weekly sessions,
either cognitive
or behavioural
coping plus
homework.
3 month
followup
Order of courses
randomly
determined such
that nature of
course not
Cooper Job
Stress Q.
Subjective
Health
Checklist
(SHC)
At immediate
post training or
follow up:
n/a
No sig effect on
SHC (or sick
leave).
No sig effect of
interventions on
control for Job
Stress (Cooper).
Even when
adherence to
programme
(>50%) was
analysed
GHQ-12;
MHPSS
(Mental
Health
Professionals
Stress Scale)
Control group
less stressed
(MHPSS) at
outset (allowed
for in analysis).
No sig effects on
GHQ for whole
groups.
When only data
from those with
Poor
retention in
study, poor
adherence to
programmes
(even though
in work)
although this
was
analysed for
it possibly
indicates
poor attitude
to study.
Yes
Considered
generally
healthy
workforce.
n/a
Although
there were
problems
with
recruitment
and
allocation of
subjects to
study groups
these effects
were likely to
be minor (i.e.
116
Yes
mainly
'intellectual
disabilities
service';
mainly female
(82%); av age
37yrs
known to
subjects prior to
arrival.
GHQ 4 or more
were analysed
then sig
reduction
occurred and
continued postintervention.
(p<0.04).
GHQ scores fell
for all three at
post training.
Cognitive
continued to fall,
coping remained
down and control
returned to pretest levels.
MHPSS not used
at follow-up and
no data reported.
Horan 2002
Stress
management
RCT
1/-
66 in total, split
not given
USA, Mostly
married,
caucasian
females with
children, av
age 45yrs
"Chicken soup
for the soul at
work" workplace
story groups.
One meeting
(duration
apparently 1
hour) per week
for 11 weeks.
Immedi
ately
post
interven
tion
only
OSI-R; PMI
OSI_R - 13/14
scales not
significant,
interaction effect
on one scale
(p=0.002)
treatment means
reported but not
those for control
group.
Of the 24
subscales within
the PMI, 3
subscales, 2
within the Mental
Wellbeing
category, had
significant
effects: (State of
Mind (p=0.04) &
Confidence Level
n/a
only 3
people
allocated to
treatment
who should
have been
wait listed).
Data
analyses are
reported to
explore likely
impact of
these to
support
findings.
GHQ
outcomes
seen as
clinically
significant in
size of effect
for both
interventions
Poor control
group.
Treatment
group had a
meeting per
week away
from job control group
didn't.
Those
without time
to attend
didn't attend;
Very poor
documentati
on of noncompleters
and nonattenders
117
Yes
(p=0.02))
Jones &
Johnston 2000
Stress
management
RCT
1/++
What is the
effect of a stress
management
intervention
designed to
reduce affective
distress in
student nurses
who have
previously
reported
significant
distress?
40
(intervention)
& 39 (controls)
3
months
& 18
months
3 month
attrition stated
as 6% & 7%
(2.4 & 2.7
subjects)
GHQ, STAI,
Beck
Depression
Inventory
(BDI), Beck &
Srivatava
Stress
Inventory.
Objective
performance
measures
(sickness
absence and
general
absence)
UK,
'Distressed'
student nurses
with score of 4
or more in
GHQ-30 20
weeks prior to
intervention.
85% female.
In treatment
group:
Significant fall in
GHQ-30
(p<0.0005)
STAI-T sig lower
(p<0.0005)
BDI sig lower
(p<0.0005)
STAI-S sig lower
(p<0.0005)
No sig effect on
objective
measures
Higher GHQ
scores in nonparticipant
group
Lindquist &
Cooper 1999
Stress
management
RCT
1/++
Can a battery of
training and
counselling
measures
reduce levels of
stress amongst
office workers?
52 treatment
group, 52
control group
STAI-T
remained
lower
(p<0.0005)
BDI remained
lower
(p<0.0005)
STAI-S
difference
reduced but
still sig lower
(p=0.002)
No effect on
objective
measures.
Stress
awareness,
lifestyle and
coping.
100% retention
Australia,
Government
Office workers,
55% female,
educational
and other
GHQ-30
remained sig
lower
(p<0.0005)
4 x weekly
workshops plus
45 minute
personal
counselling
session at end.
Not
followe
d up as
controls
given
treatme
nt after
first
posttreatme
nt
evaluati
Subscales of
OSI for
perceived
stress,
home/work
interface and
physical
health, plus
physiol (BP)
No significant
effects of
treatment on any
outcomes
Apparent
concerns
about
confidentialit
y of
responses
Study on
those with
existing
distress.
Authors
acknowledge
that effects
might be
attributable
to other,
unintended
aspects of
course such
as increased
attention,
diversion
from other
cares,
increased
socialisation,
etc rather
than the
course
content per
se.
Battery of 4
workshops
aimed at
stress in
particular but
also adverse
lifestyle
factors
(smoking,
drinking etc)
had no
significant
118
Yes
demographic
factors not
given
Lucini 2007
Stress
management
Quasi
experiment
al (NonRCT)
2/-
What effect
does a stress
management
programme
have on
psychological
profiles and
autonomic
nervous system
regulation?
26 treatment,
25 controls in
training
programme
White-collar
workers,
similar ages
and BMIs
between 2
groups, more
females in
treatment
group
on at 8
weeks.
Not
followe
d up
beyond
end of
year of
study.
impact
although
perceived
job stress
showed
trend
(p=0.06)
Overall stress,
tiredness
perception
and stresssymptoms
measured
using nonstandard
instruments
used (and
published) by
same group
previously.
Plus
physiological
measures of
blood
pressure and
heart rate (RR
Interval)
variability
Levels of stress,
tiredness and
somatic
symptoms fell in
treatment group
(authors state
this is sig but no
stats reported)
although
absolute levels
still higher in 2 of
3 than with sham
group. In
contrast levels of
all three rose in
sham group.
n/a
Effects at
post training
follow-up
(after control
group had
received
training)
suggests
longer term
benefit but
lack of
control group
makes this
uncertain.
Intervention
not main
focus of
study which
was
cardiovascul
ar health.
Intervention
poorly
designed
with subjects
volunteering
to treatment
or sham
group.
Treatment
group
appears sig
worse than
sham at
outset
although no
stats
119
Yes
reported.
Sham
treatment not
comparable
to main
treatment.
McCraty et al.
2003
Stress
management
RCT
1/++
Mino et al..
2006
Stress
management
RCT
1/+
What effect
does a
workplacebased stress
management
programme
have on blood
pressure and
emotional health
of hypertensive
employees?
21
(intervention)
& 17 (controls)
18 & 14 on
completion
USA,
Hypertensive
employees of
a global IT
company
28 (treatment)
& 30 (controls)
Positive
emotion-focused
stress
management
programme
(Inner Quality
Management)
3
months
after
training
1 x 8 hour + 2 X
4 hours over 2
week period
21 & 30 at
follow-up
Cognitive
behaviour
therapy stress
management
programme.
Japan,
2 hour
3
months
BP; Personal
and
Organizational
Quality
Assessment
(POQA); Brief
Symptom
Inventory
(BSI)
GHQ-30,
Centre for
Epidemiologic
Studies for
Depression
(CES-D),
Health Status
BP 10.6mmHg
systolic reduction
sig larger than
controls
(P=0.05);
n/a
Improvements in
individual
wellbeing and
organizational
effectiveness at
3m follow-up
including
Positive Outlook
(p<0.01); Stress
Symptoms
(p<0.05) from the
POQA
Reductions in
depression
(p<0.05) and
phobic anxiety
symptoms
(p<0.05) from the
BSI and the
Global Severity
Index derived
from the BSI
(p<0.05)
CES-D scores
reduced in SM
group, difference
1.1 (p=0.003),
No effect on
GHQ-30
Positive
emotionfocused
stress
management
programmes
can have
health and
wellbeing
benefits for
workers with
hypertension
Yes
Small scale
study,
controls just
wait-listed.
Involvement
in training
might have
improved
adherence to
other
treatment
elements
(e.g.
medication)
n/a
No
differences
in outcome
measures
amongst
those lost to
follow up.
120
Yes
Manufacturing
Company,
male manual
workers, av
age 38yrs.
All volunteers.
Munz et al.
2001
Stress
management
Quasiexperiment
al (procohort)
2/-
55 intervention
24 controls
USA,
Telecommunic
ations
customer
service/sales
reps. Very
limited
demographic
details. No sig
diff in years
worked for
company or
years in
present job.
No age or
gender info.
Recruitment
not clear.
Approximately
150
employees in
each potential
pool. 55 took
part with 24
controls. Not
stated how
many were
behaviouralcognitive training
+
2 hour muscle
relaxation
training plus
advice to
continue
individually.
Followed by
email based
advice and
counselling as
required.
Combination of
selfmanagement
and
organisational
stress reduction
interventions.
3 X 4 hours selfmanagement
skill
development +
manual.
Participation
voluntary (all 55
took part).
4-6 work unit
employees in
participatory
programme to
improve
workplace.
Questionnaire
Poor
compliance
with stress
self
management
advice and
reluctance to
use email
counselling.
Post
test 1,
immedi
ately at
conclusi
on of
interven
tion
period
(12
weeks).
Perceived
Stress Scale
(PSS);
Depression
(CES-D);
Positive and
Negative
Affect
Schedule
(PANAS)
Intervention
group showed
sig less stress
post-intervention
(p<0.05); and
less depression
(p<0.05); less
negative effect
(negative
arousal)
(p<0.05); less
tiredness
(p<0.05); more
positive energy
(p<0.05); less
low negative
affect (more
relaxation)
(p<0.05).
Intervention
group also
showed better
increase in
productivity and
lower
absenteeism.
n/a
Combination
of
participatory
and
individual
intervention not possible
to
differentiate
effects of
different
parts of
package.
Concerns
over the
initial
selection.
Participation
in course
was
voluntary
and so selfselected
group
entered
intervention.
Controls
were from
different
work units
and so might
121
Yes
invited.
Pelletier 1998
RCT
1/+
3 X 27
subjects (full
intervention,
no telephone
component,
control).
21, 20 & 25
completed
study.
County
employees,
secretarial and
lower middle
management,
mix of
ethnicity, race
and gender
(details not
given)
4 written training
modules on
reducing stress
at work and
home, book by
first author,
stress reduction
audiotape, stress
card to assess
daily stress level,
written
personalised
assessment.
4 Telephone
follow-up to
assist in skills
development.
Material
s every
6
weeks
across
a year.
Follow
up at
one
year
from
onset.
Survey
covering self
report
physical and
psychological
health,
perceived
stress, selfreport stressrelated
absenteeism,
psychological
health. Plus
Job Strain
Survey with
subscales
covering
somatisation,
depression
and anxiety.
No differences at
baseline.
Gp 1, sig
reduction in
perceived work
stress (p<0.01)
No changes in
perceived health
n/a
not have
been
comparable
although
ANCOVA
used to allow
for preintervention
differences.
Changes
restricted to
within work
unit.
Genuine
effect,
concern is
over extent
to which it
can be
generalised.
No
differences
in pre-test
scores for
drop outs.
Limited data
on
demographic
variables.
Gp 1, sig
decrease in
somatisation
(p<0.05). No
change in other
dimensions
122
Yes
Rahe et al.
2002
Stress
management
RCT
1/+
What is the
effect of a
workplace
stress
management
programme on
illness and
health services
utilisation?
171 full
intervention vs
166 partial
intervention vs
164 waiting list
control.
Overall attrition
32% (343
remaining), no
group numbers
given.
USA,
computer
industry and
local
government
employees,
typically 40s,
50:50 gender.
Stress
management
program that
used
personalized
feedback and
small group
education.
Full intervention
included face-toface feedback
and 6
supplementary
small group
sessions. In
computer
industry these
were 60 mins in
lunch period (av
attendance 3
sessions), in
local gov 90
mins during work
(av attendance 4
sessions). Both
were every other
week for 3
months. Partial
intervention had
mail feedback
and no additional
contact.
16
months
Stress Coping
Inventory
(SCI), StateTrait Anxiety
Inventory
(STAI Trait
Form Y-2),
Quarterly
Health
Questionnaire
(QHRQ),
Physician
visits from
medical
records
n/a
All computer
industry
groups
showed
decreases in
stress and
anxiety over
time. One
measure (neg.
responses to
stress)
showed a
group x time
interaction
with full
intervention
more change
than partial
which was
more change
than control
(p=0.012).
Limited
support for
expectation
that
intervention
would have
beneficial
effects and
that full
intervention
would be
more
beneficial.
Main
impression is
that
extraneous
factors
affecting all
groups were
dominant.
City gov.
employees
also showed
sig decrease
across all
measures with
no sig
interactions.
Lower selfreported
illness in
expected
directions with
that for city
employees
approaching
sig (p=0.068)
Fewer
physician
visits in year,
123
Yes
Sheppard et
al. 1997
Stress
management
RCT
1/+
Shimazu et al.
2003
Stress
management
Non-RCT
2/-
22 in both
groups.
12 in each
group
17 and 15 at
completion
Mean age
50.5yrs; mixed
ethnic, 85%
female
(although all
dropouts were
female)
8 & 8 on
completion
Mainly female,
Japanese
teachers,
mostly married
av age 44yrs
Transcendental
Meditation (TM)
or conventional
Stress
Management
(including
physical
relaxation).
Same hands-on
time for both and
monitoring of
uptake for 3
months
Post
formal
interven
tion (3
months)
and 3
year
followup
STAI and
IPAT
Depression
scale
After 3 months
Trait Anxiety
(p=0.05) and
Depression
(p=0.025) were
sig lower in TM
group. Text says
State Anxiety as
well (p=0.03)
although this is
not marked in the
table.
Post
test one
week
after
final
session
Brief Job
Stress
Questionnaire
(vigour, anger,
fatigue,
anxiety,
depression,
somatic
symptoms)
No sig effect on
any stress
response scales
(p>0.05)
in line with
expected
changes. Sig
for city
employees
(p=0.04)
After 3 years
State Anxiety
(p=0.025)
Trait Anxiety
(p=0.05) and
Depression
(p=0.01) were
all sig lower in
TM group
n/a
Appears to
be a well
conducted
study.
Seems to
have
balanced
face-to-face
time with two
treatments
and also
monitored
adherence,
at least for
first 3
months. No
formal 'at
home' task
for relaxation
group unlike
TM. High
level of
adherence to
both
suggests
strong
motivation.
Study
severely
limited by
small scale.
Additionally,
although
initial
recruitment
was intended
to be
voluntary,
124
yes
Yes
both gps, av
21yrs in job
both gps
Shimazu et al.
2005
Stress
management
Non-RCT
2/++
109
intervention;
113 controls
sent first
questionnaire.
94 & 104
complete data
at follow-up
Mainly male,
Japanese
office workers,
av age 42-44.
Controls
slightly older
and more of
them male (not
sig).
web-based
training on stress
awareness and
coping. 1 month
learning period
1 week
after
end of
learning
period,
followup after
a
further
5
weeks.
Brief Job
Stress
Questionnaire
(vigour, anger,
fatigue,
anxiety,
depression,
somatic
symptoms)
No sig main
effect on any
stress subscales.
Males showed
trend to sig effect
(p=0.069) on
physical stress,
younger showed
trend to sig effect
(p=0.093) on
psych distress,
those expressing
an interest in
stress showed
sig +ve effect
(p=0.029) on
psych distress.
n/a
64% had
been
instructed to
attend by
their
manager and
12% for
'other
reasons' with
only 24%
volunteers.
Interesting
subset
analyses
despite lack
of main
effects.
Authors
report signs
of spill-over
to controls
working in
same
department.
Short term
before
follow-up
intervention
seen as too
short.
Implications
of use of
employee
number for
randomisatio
n not
apparent.
Not stated
whether
course was
completed in
own or work
time.
125
Yes
Shimazu et al.
2006
Stress
management
Non-RCT
2/++
Walach et al.
2007
Stress
management
Non-RCT
2/+
149
(treatment),
151 (controls)
12 treatment,
17 controls
telephone call
centre workers
in Germany.
Volunteers
from 185.
CBT, stress
awareness and
coping
8
weeks
after
course,
BJSQ
'psychological
distress'
n/a
Mindfulnessbased Stress
Reduction.
8
weeks
(post
course)
and 4
months
(follow
up)
General
complaints,
tension and
tiredness
subscales of
Freiburg
Complaint List
There were no
sig changes on
complaints
n/a
12 on first
course
designated
treatment, 17
on second,
control.
Nature of
control
unclear.
8, weekly
evening classes
of 2.5 hrs per
class, 1 X 6 hour
mindfulness day,
practice at least
5 days per week
at least 20 mins
per day.
Authors
suggest
adverse
effect could
either be due
to increased
awareness
of stress
issues or to
stress
created by
additional
workload
due to need
to implement
training.
Suggest
longer period
of evaluation
would show
benefit.
Stress score
actually rose
in treatment
group but fell
in control
group
Pilot study
on small
groups of
volunteers.
Treatment
group older
than
controls, and
the 2 groups
also reported
different
motivations.
Overall some
evidence of
a positive
effect though
high time
commitment
126
Yes
11 & 16 on
completion
Willingness to
practice
condition of
entry.
6 & 10 female
treatment sig
older than
control (41.3 &
33.7 yrs)
36% & 44%
some chronic
disease (not
specified)
out of work
hours for
participants.
No effect on
well-being
outcomes.
Lack of
clarity on
how second
group
(control)
were treated
(possibly
wait listed
but not
stated).
Assignment
to 2 courses
not
explained.
127
Category of
Intervention
Stress
management
Study
design and
research
type/
quality
RCT
1/+
Grime 2004
Stress
management
RCT
1/+
Research
question
How do ACT
(psychotherapy)
and IPP
(Innovation
Promotion
programme)
interventions
affect both
mental health
and workrelated
outcomes?
What is the
effect of an 8
week
computerised
Cognitive
Behavioural
Therapy (CBT)
programme on
emotional
distress in
employees with
recent stressrelated
absenteeism?
Study
population,
setting,
country,
sample size
30 treatment
gps 1 & 2; 30
controls
Length
of
followup
Mental
Wellbeing
outcome
variables
Acceptance and
Commitment
Therapy (ACT);
Innovation
Promotion
Programme
(IPP) both 3x0.5
day sessions at
weeks 1, 2 and
14.
27
weeks
in total
(13
after
end of
interven
tion)
GHQ-12 &
BDI
(depression)
24 (treatment)
24 (controls). 8
cases failed to
complete
treatment
course.
CBT &
conventional -vconventional
(whatever care
they were
receiving).
end,
1m, 3m
& 6m
HADS
(Anxiety &
Depression)
14 & 19 at
follow-up.
CBT 8 weekly
sessions
(duration not
given) presented
on PC at Occ.
Health Centre.
Volunteer
Media workers
av age 36.4y,
mainly
graduates,
balanced for
gender
Description of
intervention(s)
Short term
findings (<1
year)
ACT sig reduced
GHQ score cf
other treat and
control
(p<0.0001)
Long term
findings (>1
year)
n/a
Comments,
confounder
s/ potential
sources of
bias
High initial
GHQ scores
indicate
strong scope
for change.
Applica
bility to
the UK
Yes
ACT (p<0.0001)
and IPP
(p<0.006) both
sig reduced BDI
cf control.
With CBT v
control:
Depression
lower at end
(p=0.028),
Anxiety
(p=0.021), &
Depression
(p=0.040) lower
at +1m.
No effect at +3 &
+6 months
although
adjusted values
remained below
end of treatment
scores.
n/a
Low
attendance
and
adherence to
training a
problem
(might have
been due to
need to
attend clinic).
Some
problems
with
employers
allowing time
for access
also
reported.
Also
preferences
seen for
conventional
treatment as
being more
tailored and
128
Yes
Subjects
already under
treatment
Reynolds 1997
Both
Quasiexperiment
al (procohort)
2/-
37 in Area A
(counselling)
76 in Area B
(organisational
) 43 in Area C
(control)
numbers
actually taking
part not stated
No
demographic
details given.
Workers in
council
housing
department
No reference
to method
used to select
departments
more faceto-face
Area A received
individual
counselling,
Area B had
organisation
change, Area C
was control
12
months
after
interven
tion
period
GHQ-12 and
SCL-90R
(somatisation)
At 1 year sig
AreaXTime
interaction
(p<0.049). SCL90R decreased
in Area A and
increased in
B&C. According
to text same
effect was seen
with GHQ-12 but
figure doesn't
support this shows reduction
for B as well.
n/a
According to
paper,
counselling
but not
organisation
al
intervention
worked in
reducing
physical and
psychologica
l symptoms.
Conflict in
data
reported
makes this
difficult to
interpret.
Also lack of
any
information
on
participation
rates and
demographic
comparisons
diminishes
value of
study. Fuller
paper cited
not located.
129
Yes
Altchiler and
Motta1994
Category of
Intervention
Stress
management
Study
design and
research
type/
quality
RCT
1/+
Research
question
Do different
forms of
exercise
(aerobic and
nonaerobic)
have any effect
on state and
trait anxiety,
absenteeism,
job satisfaction
and resting
heart rate?
Study
population,
setting,
country,
sample size
43/90 to
completion (23
aerobic, 20
nonaerobic).
Mainly white
(84%), female
(88%), US,
workers with
disabled
children &
adults.
Av. age 33.48y
(Aerobic);
30.40y
(nonaerobic)
Description of
intervention(s)
Low-impact
aerobics or
nonaerobic
exercises, 30
mins, 3 times per
week for 8
weeks
Length
of
followup
Pre-test
1; 4
weeks
to pretest 2;
posttest 1 at
end of 8
weeks
of
exercisi
ng;
posttest 2, 2
weeks
later
Mental
Wellbeing
outcome
variables
STAI-S &
STAI-T
Short term
findings (<1
year)
Collapsed across
groups,
immediate postexercise
sessions had sig
(p=0.005)
reduction in
STAI-S. Effect
was due to
aerobic not
nonaerobic
sessions.
Combined post
test scores,
aerobic sig lower
STAI-S than
nonaerobic
(p=0.018) (effect
size =0.22).
Effect mainly at
post test 1.
STAI-T sig
reduced across
study for aerobic
(p=0.018) but not
nonaerobic
(effect size =
0.60).
Further analysis
showed only
previously nonexercising
members of
aerobics group
had sig red in
STAI-T (p=0.016)
Long term
findings (>1
year)
n/a
Comments,
confounder
s/ potential
sources of
bias
Reasonable
adherence to
programme
(2.3/3 per
wk).
No nonintervention
but
differential
effect
suggests not
due to other
uncontrolled
feature of
intervention
(e.g. social
benefit of
sessions).
Limitation of
effect on
STAI-T to
only previous
nonexercisers
suggests
subjects
might be
anxious
about lack of
exercise.
130
Applica
bility to
the UK
Yes
Atlantis et al.
2004
Stress
management
RCT
36 (treatment)
37 (controls)
(20 & 24
completed).
Stratified for
gender &
normal and
high DASS
scores
1/+
Australia,
Casino shift
work
employees
Av Age
treatment gp
30y; controls
33y.
24 week
combined
aerobic (mod to
high int 20 mins
3 days per wk)
and weighttraining (light to
mod at least 2
per wk) exercise.
Plus behaviour
modification
interventions
(health
education
seminars and
health
counselling).
24
weeks
(6m)
SF-36 and
DASS
questionnaire
s
Mental Health
(p=0.005 effect
size 0.68);
Vitality (p<0.001
es=1.54);
General Health
(p=0.009
es=0.44);
Bodily Pain
(p=0.005
es=0.62);
Physical
Functioning
(p=0.004
es=0.93);
Stress (p=0.036
es=-0.56);
Depression
(p=0.048 es= 0.16)
n/a
immedi
ately
post 10
min
interven
tion
STAI-S &
POMS
(depression
and vigour)
No significant
changes
between groups,
all of which
showed pre-post
test improvement
in all measures.
n/a
Field 1997
Stress
management
RCT
1/++
100 subjects
(64% female)
in 5 groups
(gp. Nos not
given but
assumed 20)
USA, Hospital
employees
(?nurses)
Massage (10
mins either at
workplace or in
separate room),
Music,
Relaxation (10
mins in separate
room), Social
support (10 mins
talking)
Supervised
exercise
sessions
probably
enhanced
adherence to
programme
and certainly
ensured
good
compliance
during
sessions.
Correlation
between
baseline and
changes ie
those worse
pre-study
improved
more use
of
stratification
therefore
enhanced
positive
overall
effect.
Concerns
over low
initial
recruitment
and high
drop out
rate.
Could be
Hawthorne
effect or
general
relaxation
effect of 10
mins away
from work.
No follow-up.
131
Yes
Yes
group sessions
on anxiety,
depression and
vigour?
Hinman et al.
1997
Stress
management
RCT
1/++
de Lucio et al.
2000
Stress
management
RCT
1/+
McElligott et
al. 2003
Stress
management
RCT
1/+
Can a
computerised
exercise
programme
reduce stress
levels in office
workers who
use video
display
terminals?
What effect
does training in
relaxation,
cognitive
restructuring
and some
communication
skills have on
communication
skills and stateanxiety?
What effect
does touch
therapy have on
relaxation and
anxiety of
24 (treatment),
26 (controls) (3
excluded from
treatment
group due to
complete noncompliance)
USA, female
office workers,
30's
29 (treatment)
& 32 (controls)
(30 declined).
(20 and 31
STAI post
intervention)
Spain, female
(?) nurses
stratified by
area of work,
shift and
professional
category.
12 (treatment),
8 (controls)
12 & 3 on
completion
Computerdirected exercise
break, 2 per day
15 mins per
break. In
addition to
normal breaks.
Exercise period
could be put on
hold for work
reasons.
Self-reported
compliance
ranged from 3.8100%
(av=39.5%)
Emotional selfcontrol and
communication
training
programme
At end
of 8
week
progra
mme
4 dimensions
of stress on
PSQ:
vocational,
psychological,
interpersonal,
physical
No sig effect
between
treatment and
control groups or
between
compliers and
non-compliers in
treatment group
n/a
Immedi
ately
followin
g
interven
tion.
State-TraitAnxiety
Inventory
(STAI-S)
No significant
differences in
STAI scores
n/a
Before
and
after
each
treatme
Anxiety on
Visual
Analogue
Scale + BP,
heart rate,
5hr weekly
sessions over 5
weeks outside
working hours,
plus homework.
Touch therapy
compared with
laying on of
hands. 45 mins
sessions (plus
No sig diffs in
any parameters.
n/a
Subjective
questionnair
es might
load results
(e.g. I feel
nervous I
feel at ease)
Small
sample size,
small
differences
between
nonsymptomatic
groups, low
compliance,
remained at
computer for
exercise.
Selfconsciousne
ss of
participants.
Text unclear
nature and
extent of
drop outs.
High STAI-S
scores
before and
after (84%
higher than
median for
Spanish
adult
women)
Touch
therapy
(physical
therapy) had
no sig effect
132
Yes
Yes
Yes
nurses?
Mainly (75%)
female, 30's,
degree
educated
Sheppard et
al. 1997
Stress
management
RCT
1/+
Shulman &
Jones 1996
Stress
management
RCT
1/++
22 in both
groups.
What is the
effect of
massage
therapy on
anxiety?
18
intervention,
16 control
17 and 15 at
completion
Mean age
50.5yrs; mixed
ethnic, 85%
female
(although all
dropouts were
female)
18 & 15 at
end.
Office workers
15 mins for
tests)
nt
session
only
pulse oximetry
and
respiration
rate.
Transcendental
Meditation (TM)
or conventional
Stress
Management
(including
physical
relaxation).
Same hands-on
time for both and
monitoring of
uptake for 3
months
Post
formal
interven
tion (3
months)
and 3
year
followup
STAI and
IPAT
Depression
scale
After 3 months
Trait Anxiety
(p=0.05) and
Depression
(p=0.025) were
sig lower in TM
group. Text says
State Anxiety as
well (p=0.03)
although this is
not marked in the
table.
After 3 years
State Anxiety
(p=0.025)
Trait Anxiety
(p=0.05) and
Depression
(p=0.01) were
all sig lower in
TM group
15 minute at
chair massage, 1
per week for 6
weeks.
12
weeks
from
start, 34
weeks
from
end of
STAI, state
and trait
n/a
Controls had 15
minute break.
in small
scale poorly
sustained
study with
only 3/8
controls
completing
and no follow
up.
Appears to
be a well
conducted
study.
Seems to
have
balanced
face-to-face
time with two
treatments
and also
monitored
adherence,
at least for
first 3
months. No
formal 'at
home' task
for relaxation
group unlike
TM. High
level of
adherence to
both
suggests
strong
motivation.
Subjects had
higher than
Norm values
for STAI
(both)
suggesting
highly
anxious
133
yes
Yes
in US
company,
more females,
av age 40yrs.
Taniguchi et
al. 2007
Stress
management
Non-RCT
2/+
38 in treatment
group, 41 in
control
100% retention
Japan,
Female health
care workers
massag
e period
1 hour lecture
and 10 mins
relaxation
training versus
same lecture
Immedi
ately
post
training
test (weeks 3
and 6 of
treatment
period). For
state anxiety, all
three sets were
sig diff for
massage group
(fell sig and then
rose sig to
intermediate
value in delayed
post-test. Sig diff
between gps at
post-test (during
intervention). No
diff in trait
anxiety between
gps but general
trend to reduce
across expt.
Salivary IgA,
Iceberg profile
of mood
states
Significant
increase in S-IgA
in intervention
group (p=0.03)
however, after
adjustment for
age effect was
marginal
(p=0.09)
n/a
group explained by
authors in
terms of
ongoing
downsizing.
Use of break
balanced out
effect of
massage
break to
some extent
although a
break for
'non-work'
might not be
as distracting
as massage
which might
enhance any
difference.
Not known if
massage
allowed
conversation
for example.
Post-test
questionnair
es believed
to be
completed
shortly after
massage but
limited
persistence
interesting.
Very short
training
period and
no
assessment
of
persistence.
Subjects not
randomly
134
Yes
No significant
difference in
Iceberg scores
(mood)
Van Rhenen,
2005
Stress
management
RCT
1/+
Initial sample
396, 59
cognitive, 71
physical into
trial (266 drop
outs)
36 & 39 at
completion
Dutch
telecommunica
tions workers
(apparently
white collar),
mean age
44.2yrs, 90%
men, av work
exp. 21.1yrs.
Selected on
basis of
'Distress
Score' in top
decile for 700
staff. No
breakdown by
groups
reported.
4 x 1 hour
training sessions
over a 8 week
period (2wk,
2wk, 4wk
breaks). Either
physical therapy
(exercises,
relaxation) or
cognitive
therapy.
6
months
after
end of
progra
mme
Psychological
complaints
(4DSQ);
Burnout
(UBOS Dutch MBIGS); Fatigue
Checklist
Individual
Strength (CIS)
At short-term
immediately post
training, both
groups showed a
sig decline in
psychological
complaints but
no diff between
groups. Both
groups show a
decrease in
burnout
subscales but no
diff between
interventions
(although trend
to one for
exhaustion
subscale with
phy clearer than
cog). Both
groups showed a
decrease in
fatigue scores
but again no
intervention
effect. After 6
months the
reduction in
psychological
n/a
allocated but
were
assigned to
groups by
manager to
one of 2
sessions a
month apart.
No
assessment
of possible
++confounde
rs such as
exercise
levels.
No
differential
effect
between the
two
interventions
was
demonstrate
d although
both had a
positive
benefit which
lasted
through to
follow-up.
Welldesigned
study but
lack of
absolute
control group
makes
extrapolation
difficult - was
the benefit
just from the
breaks from
work or a
genuine
effect of the
135
Yes
Webb et al..
2000
Stress
management
Non-RCT
2/-
What is the
effect on blood
pressure and
personal strain
of a progressive
relaxation
intervention for
AfricanAmerican
women?
48 enrolled, 43
completed. No
group numbers
given
Demographics
only reported
for whole
sample.
Av age,
33.5yrs
AfricanAmerican
women
selected on
basis of high
degree of
hypertension.
Baseline data
not analysed
separately
Progressive
muscle
relaxation
compared with
being 'instructed
to take time out
(30mins) each
day. No record
of compliance.
8
weeks
from
entry
Personal
Stress
Questionnaire
(PSQ)
complaints
remained stable.
Burnout was
reduced in phys
group and
exhaustion
raised in cog
group but
interaction term
still not sig.
Effect of either
intervention on
fatigue remained,
again with no sig
interaction effect.
Experimental
group (size not
known) showed
sig reduction in
'interpersonal
strain' (p=0.02);
and 'physical
strain' (p=0.01).
Both groups
showed sig
reduction in
psychological
strain
(p=0.0001).
No sig change in
'vocational strain'
although data
suggests trend
for control group
to reduce levels.
n/a
interactions?
Complex
analysis
increases
risk of
chance
significant
effects but
persistence
and similarity
of effects
between two
groups
makes this
unlikely.
Concerns
over
recruitment
(allocated to
treatment or
control on
basis of
course
attended).
Poor
reporting of
group
numbers and
dynamics.
Poor
adherence to
treatment
intervention
Study
appears to
show
positive
value from
relaxation.
Subjects
were all
136
Yes
recruited as
having a
family history
of
hypertension
and selfreported
stress.
137
Hasson et al.
2005
Category of
Intervention
Stress
management
Study
design and
research
type/
quality
RCT
1/++
Research
question
What are the
effects on
mental and
physical wellbeing and
stress-related
biological
markers of a
web-based
health
promotion tool?
Study
population,
setting,
country,
sample size
129
(treatment) &
174 (controls)
drawn from
volunteers.
121 & 156 at
end. Analysed
on intention to
participate
basis
Description of
intervention(s)
Web-based tool
for health
promotion and
stress
management.
Control group
had access to
website but not
lifestyle training
& chat
elements.
Length
of
followup
Mental
Wellbeing
outcome
variables
6
months
Physiological
markers Cardiovascula
r and lifestyle;
Stress related;
Recovery
related;
Immune
markers and
neuropeptides
; Self-rated
health and
stress
questionnaire
(nonstandard)
Sweden, IT
and media
company
employees.
Slight excess
of males.
Mixed ages.
No analyses of
pre-test
demographics
reported but
ANCOVA
applied.
Nielsen et al.
2006
Stress
management/
organisational
Quasiexperiment
al (procohort)
2/-
varies:
Intervention
canteens: 27
and 17
completed
varied individual
and group
interventions.
Exercise,
empowerment,
IT training,
20
months
Stress
symptoms &
vitality
Short term
findings (<1
year)
Treatment group
Improved
significantly in
terms of selfperceived ability
to manage stress
(p=0.001), sleep
quality (p=0.04),
mental energy
(p=0.002),
concentration
ability (p=0.038),
& social support
(p=0.049).
Levels of DHEAS decreased
significantly
(p=0.04), levels
of neuropeptide
(NPY) increased
significantly
(p=0.02) CGA
(associated with
catecholamine
activity)
decreased
(p=0.01), levels
of immune
marker TNFa
decreased
(p<0.016).
n/a
Long term
findings (>1
year)
n/a
Comments,
confounder
s/ potential
sources of
bias
No
knowledge of
total
potential
participants
(estimated
80% take-up
overall).
Applica
bility to
the UK
Yes
Web tool
only differed
in interactive
training and
chat
elements.
sig. increase
in cognitive
stress reaction
at canteen A
(int: p<0.01);
B (int: p<0.05)
Well-being
not main
focus of
paper.
Yes
Complex
138
being?
questionnaires
at both times;
control 11 and
16 did so.
workshop on
development
& C (cont:
p<0.01).
sig increase in
vitality at
canteens A
(int: p<0.05);
B (int: p<0.05)
& D (cont:
p<0.01)
Numbers at
pre-test
45,26,22,25
and at posttest 30, 26, 19,
28
canteen
workers at 4
hospitals/
elderly care
homes
Peters &
Carlson 1999
Stress
management
RCT
1/+
24 (treatment),
26 (controls)
21 & 19
respectively
completed
USA,
Maintenance
workers, 60%
male,
asian/pacific
islander
extraction,
middle-aged,
Health
Promotion
initiative
described by
author as having
strong emphasis
on stress
management.
Exact nature
unclear, seems
to be:
1 hour HRA
feedback
session,
Followed by 8
10
week
interven
tion
period
then 3
months
Many
measures,
STPI relevant
STPI subscales
measure anxiety,
anger, curiosity
and depression.
Curiosity scale
was higher in
treatment group
(p<0.05)
n/a
array of
interventions
some
individual,
others
organisation
al. Different
interventions
at two sites.
Number of
criticisms by
authors of
process of
implementati
on of
interventions
Many other
changes
extraneous
to project
(e.g. new
manger at
one site,
announceme
nt of
closures and
cuts at
another.
Comprehens
ive Health
Promotion
initiative of
which mental
well-being
only a small
aspect of
outcome
measuremen
t (although
stated as a
key part of
study). Few
effects other
139
Yes
46% obese
weeks of 1X45
minute large
group training
session, and
1X60 minutes
small group
training session
than 'I am
curious'??
Very limited
applicability
140
Alford et al.
2005
Category of
Intervention
Stress
management
Study
design and
research
type/
quality
Non-RCT
2/++
Research
question
Does written
expression of
emotions by
employees in
stressful
occupations
help to minimise
stress
reactions?
Study
population,
setting,
country,
sample size
34 treatment
31 controls
31 treatment,
30 controls on
completion
child protective
service
officers,
Description of
intervention(s)
Length
of
followup
Mental
Wellbeing
outcome
variables
Short term
findings (<1
year)
Long term
findings (>1
year)
write about
recent stresses,
emotions,
related thoughts
and plans in
journal for 15-20
minutes per day,
3 consecutive
days
interven
tion
start of
week 2,
postinterven
tion
questio
nnaire
end of
week 2
GHQ-12;
PANAS
(Positive &
Negative
Affect
Schedule);
JIG (Job in
General)
scale
significantly
greater change
among treated
for (i) GHQ 12.03
to 8.10 vs 12.30
to 12.10
(p=0.003) and (ii)
JIG 42.97 to
45.26 vs 41.60 to
39.97 (p=0.002)
n/a
Written advice
tailored to
individual
(physical activity,
nutrition,
breakfast,
alcohol,
relaxation, etc.)
Included pre-test
GHQ-12 score.
1 year
follow
up
GHQ-12,
Blood
pressure,
cholesterol,
triglycerides &
sick leave in
year.
No sig effect on
any of these
outcome
measures.
n/a
Australia.
85% women.
Comments,
confounder
s/ potential
sources of
bias
Consider
possibility
that time
spent writing
about
anything
could have
been
beneficial;
mainly
females; use
of volunteers
Applica
bility to
the UK
Yes
Average age
35.2 years
Kawakami
1999
Stress
management
RCT
1/+
What effects
does a mailed
advice leaflet on
stress reduction
have on
psychological
distress, blood
pressure, serum
lipids and sick
leave amongst
Japanese
manufacturing
employees?
Japan,
Workers with
GHQ-12 score
of 3 or more.
113 initially
selected in 2
groups, others
excluded at
baseline
reduced
numbers to 91
(intervention)
and 88
(control).
81 & 77 at 1
year follow-up
(health checks
on 48 & 45)
Both groups
reduced
GHQ-12
scores,
attributed to
natural
regression.
Could have
been
crossover
between
individuals in
different
groups at
same plant.
BP
measures
limited by
141
Yes
reduced
group size.
Martin &
Sanders 2003
Stress
management
RCT
1/+
Nhiwatiwa
2003
Stress
management
Non-RCT
2/++
What effect
does a Positive
Parenting
training
programme
have on a
number of
measures
including workrelated stress?
Can a brief
intervention
reduce
symptoms of
distress
following
assault?
23 (treatment)
& 22
(controls).
16 & 11 at
completion
Work-place 3P
(Positive
Parenting
Program) over 8
weeks.
6
months
(4
months
post
training)
DepressionAnxiety-Stress
Scale 21
(DASS 21),
Work Stress
Measure
Australia,
General and
academic staff
in a
metropolitan
university.
Needed to
have a child
with
behavioural
problems and
be
experiencing a
significant
degree of
distress.
Not explicitly
stated 45/90
initially agreed
to take part.
prob 20
treatment, 20
control
UK,
nurses at
medium
secure
Given booklet on
effects of trauma
and coping
mechanisms
3
months
GHQ-28;
Impact of
events scale
(IES)
No effect of
training on work
stress (scale
referred to not
given) although
post-training
levels fell in
treatment group
and rose in
control group.
treatment group
had higher levels
of distress (IES)
post-intervention
(change = +2,
compared to -6
in controls;
p<0.03). No
difference for
GHQ-28
although mean
scores went up
in treatment and
n/a
n/a
Training had
beneficial
effect on
child
behaviour
and other
measures
not of
concern here
but these did
not manifest
themselves
in terms of
stress
although
positive
trend.
Analyses of
those
dropping out
suggests no
difference to
those
remaining
but high rate
of concern.
No account
taken of
previous
number of
assaults or
of severity of
assault
which might
have
influenced
outcomes.
Poor
142
Yes
Yes
hospitals who
had been
assaulted
within the past
month.
Approx equal
males and
females, mean
age not given
69% from 1
hospital.
Wachi 2007
Stress
management
RCT
(crossover)
1/++
What is the
effect of
recreational
music making
on natural killer
cell activity,
cytokines and
mood states on
employees?
20 (treatment)
and 20
(controls).
20 & 19 in
phase 2 (RMM
crossover)
Electric
company,
Japan
down in control
groups.
Recreational
Music Making
(RMM) versus
leisurely reading,
1 X 3 hour
intervention
Crosso
ver
design
with 6
months
betwee
n
phases.
Mood states
questionnaire
(POMS),
measures of
natural killer
(NK) cell
activity
Improvements in
NK cell activity,
and mood states.
In one phase
(phase 2)
anger/hostility
was sig lower in
RMM than
control, post
intervention (p
not stated).
Consistent
decrease in total
mood (TMD) for
both RMM &
control. Sig diffs
in change
between groups
for P2 (p=0.019)
& P1+P2
(p=0.012).
Plots of pre-post
NK cell activity
showed sig diff in
slopes for RMM
& control (Phase
1, p=0.05, Phase
2, p=0.019).
reporting of
sampling
and
numbers.
n/a
No analysis
of immediate
case-control
comparisons
Data
analysed
after second
phase with
assumption
of no carryover
between
phases.
Small
sample size
and results
not
consistent
between
phases.
Study funded
by Yamaha
carried out
with Yamaha
employees
possible
conflict.
143
Yes
Wilson et al.
2001
Stress
management
RCT
1/+
33 to EMDR,
29 to standard
training
EMDR versus
stress
management
USA,
Police officers
3 X 2 hour
EMDR sessions
1:1 basis, 6
different
therapists
Volunteers
from 531 in
Department
Av age, 36.8
yrs
79% male,
75%
Caucasian.
No
demographics
of groups)
6 X 1 hour (video
+ workbook)
sessions for SM
Either could be
undertaken
during work
hours
Post
test, six
month
followup and
exit
intervie
w
SUDS, State
Trait Anger
Inventory, Job
Stress survey
scale, Police
Stress
inventory,
Symptom
Check List
(SCL), marital
adjustment
scale, PSDS,
Coping
Response
Inventory
SUDS found
level of stress
lower in EMDR
(p<0.05),
Significant
reduction in trait
anger in EMDR
(rose in SM)
(p<0.05)
State anger fell
in EMDR & rose
in SM (p<0.05)
Job Stress
Survey scores
lower for EMDR
than SM (p<0.05)
n/a
Positive
findings in
favour of
EMDR.
Although
total contact
was similar,
face to face
contact of
EMDR
possible
issue,
regardless of
nature of
contact
No effects on
PSI or SCL-90.
144
Yes
145
146
147
148
149
150
Totterdell P, Smith L. (1992). Ten-hour days and eight-hour nights: Can the
Ottawa shift system reduce the problems of shiftwork? Work & Stress; 6: 139152.
van Rhenen RW. (2005). The effect of a cognitive and a physical stressreducing programme on psychological complaints 721. International Archives
of Occupational and Environmental Health; 78: 139-148.
von Vultee PJ, Arnetz B, von Vultee PJ, Arnetz B. (2004). The impact of
management programs on physicians' work environment and health. A
prospective, controlled study comparing different interventions. Journal of
Health Organization & Management; 18: 25-37.
Wachi MK. (2007). Recreational music-making modulates natural killer cell
activity, cytokines, and mood states in corporate employees. Medical Science
Monitor; 13: 57-70.
Walach H, Nord E, Zier C, etz-Waschkowski B, Kersig S, Schupbach H.
(2007). Mindfulness-based stress reduction as a method for personnel
development: A pilot evaluation. International Journal of Stress Management;
14: 188-198.
Webb MS, Smith KA,Yarandi H. (2000). A progressive relaxation intervention
at the worksite for African-American women 978. Journal of National Black
Nurses' Association : JNBNA; 11: 1-6.
Wilson SA, Tinker RH, Becker LA, Logan CR. (2001). Stress management
with law enforcement personnel: A controlled outcome study of EMDR versus
a traditional stress management program. International Journal of Stress
Management; 8: 179-200.
151
6. Bibliography
Department of Health (1999) National service framework for mental health:
modern standards and service models. London: Department of Health
Coast D and Max C (2005) Healthy work: productive workplaces. London: The
London Health Commission.
Dunham J (2001) Stress in the workplace. Past, present and future. London:
Whurr Publishers
Health and Safety Executive (2006) Self-reported work-related illness in
2004/2005: results from the labour force survey. London: Health and Safety
Executive
Jones JR, Huxtable CS, Hodgson JT. (2006) Self-reported work-related illness
in 2004/05: Results from the Labour Force Survey. Sudbury: HSE Books.
Siegrist J and Marmot M (2004) Health inequalities and the psychosocial
environment two scientific challenges. Social Science Medicine 58: 14631473.
World Health Organization (2004) Promoting mental health: concepts,
emerging evidence, practice: summary report. Geneva: World Health
Organization
World Health Organization (in press) (2007) Shift-work, painting and firefighting. Geneva WHO (IARC Monographs on the Evaluation of Carcinogenic
Risks to Humans Vol. no. 98)
152
154
Population
Adult*
Work*
Employ*
Labour force
Personnel
Vocational
Professional
Organisation*
Industr*
Settings
Work*
Occupation*
Organisation*
Work Characteristics
Management Style
Transformational
Transactional
Leader exchange
Supervision style
Supervisor style
Labour
Interventions
Intervention
Job design
Team work
Strategy
Prevention
Control
Stress Management
Relaxation
CBT (Cognitive Behaviour Therapy)
ACT (Acceptance Commitment Therapy)
Coaching
Development
Education
Awareness
Information Counselling
Occupational Health
Rehabilitation
Employee Assistance Programm*
Change Management
Anti-bullying polic*
Anti-discrimination polic*
Anti-violence polic*
155
Anti-harassment polic*
Attendance polic*
Counselling
Sickness absence polic*
Outcomes
Consult*
Involve*
Prevent*
Reduct*
Barriers
Facilitators
Decreas*
Increas*
Best practice
Self-esteem
Control over workload
Health
Stress
Anxiety
Depression
Absenteeism
Presenteeism
Costs
Benefits
Health inequalities
Staff retention
Performance
Productivity
Hours (working)
156
Appendix B Checklists
157
1.1
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.2
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.3
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.4
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.5
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.6
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.7
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.8
1.9
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.1
0
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
158
2.2
2.3
2.4
3.1
3.2
159
160
Guideline topic:
1.1
Not addressed
Not reported
Not applicable
SELECTION OF SUBJECTS
1.2
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.3
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.4
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.5
1.6
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
ASSESSMENT
1.7
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.8
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.9
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
161
1.10
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.11
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.12
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
CONFOUNDING
1.13
STATISTICAL ANALYSIS
1.14
2.2
2.3
SECTION 3: DESCRIPTION OF THE STUDY (Note: The following information is required for evidence
tables to facilitate cross-study comparisons. Please complete all sections for which information is
available).
PLEASE PRINT CLEARLY
3.1
3.3
3.4
162
3.6
3.8
3.9
163
Guideline topic:
1.1
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
SELECTION OF SUBJECTS
1.2
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.3
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.4
Cases:
Controls:
1.5
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.6
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.7
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
ASSESSMENT
1.8
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
1.9
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
CONFOUNDING
164
1.10
Well covered
Adequately addressed
Poorly addressed
Not addressed
Not reported
Not applicable
STATISTICAL ANALYSIS
1.11
2.2
2.3
3.3
3.4
3.5
3.6
3.7
165
3.9
166
Intervention
Outcome
Study
Measures
Design
Bourbonnais et al.
Introduction
of
Psychological distress
2006
programme
of
(PSI
measure
to
reduce
Psychologic
Symptoms
Quality
Effective
Applic-
Intervent
ability to
ion?
UK?
Y/N
Y/N
Quasiexperimental
(pro-cohort)
2/-
Quasiexperimental
(Non-RCT)
2/+
Quasiexperimental
(pro-cohort)
2/-
Quasiexperimental
(non-RCT)
2/+
Quasiexperimental
(Non-RCT)
2/+
Index);
adverse psychosocial
factors
sleeping
(NHP
Nottingham
Health
Profile)
Dahl-Jorgensen
et
al. 2005
SHC
Participatory
approach
to
(Somatic
symptoms)
and
organisational
burnout
change.
Burnout Inventory)
Different
(Maslach
specific approach in
two workplaces (shop
workers
from
number of shops in
mall).
Kawakami, 1997
Landsbergis
&
1 year Workplace
participatory
intervention aimed at
identifying & reducing
sources of stress in
workplace.
Participatory
Depression
Action
Vivona-Vaughan
Research
in
1995
different
municipal
two
Sick leave
Blood Pressure (2yr)
sleeping
from
and
subscales
Job
Content
departments
Questionnaire
Extensive battery of
interventions including
lifestyle : (exercise,
healthy eating);
SACL-90 (general
stress)
Absenteeism
167
Stress, training on
social skills and
leadership, plus
organisational change
working methods
Mikkelsen, & Saksvik
Participation
1999
programme to identify
and
in
Non-RCT
2/+
Non-RCT
2/-
Quasi-
2/-
2/-
2/-
2/+
Quality
Effective
Applic-
Intervent
ability to UK
UHI, STAI-T
develop
workplace changes
Mikkelsen,
et
al.
2000
Participation
in
programme to identify
and
develop
workplace changes
Mattiilla et al. 2006
Participative
work
conference
GHQ-12
and
SCL-
90R
experimental
(non-RCT)
Combination of selfmanagement
organisational
GHQ-30, CES-D
Quasiexperimental
and
stress
reduction
interventions
Reynolds, 1997
Individual counselling
Quasi-
organisation
experiement
(pro-cohort)
versus
group
von
Vultee
et
al.
2004
Quality
management
competence
programmes
versus
work
Non-RCT
(QWC)
tool
no intervention
Intervention
Outcome
Study
Measures
Design
ion?
Y/N
Y/N
Kawakami
et
al.
2005
Web-based training of
Brief
Job
supervisors impact on
Questionnaire
Stress
RCT
1/++
Stress
RCT
1/++
subordinate workers
Kawakami
2006
et
al.
Brief
supervisors impact on
Questionnaire
Job
subordinate workers
168
Training
2004
managers
project
on
their
RCT
1/+
RCT
1/++
2/++ (for
Effective
Applic-
Intervent
ability to UK
(Caplan)
psychological strain
Takao et al. 2006
Brief
supervisors impact on
Questionnaire
Job
Stress
subordinate workers
Theorell, 2001
Psychosocial training
for
and GGT.
managers
and
Non-RCT
physiolog
Swedish
impact on subordinate
version
of
workers
Demands/Control
ical
the
measure
s
Questionnaire
2/+ for
question
naire
Intervention
Outcome
Study
Measures
Design
Quality
ion?
Y/N
Y/N
Bussing & Glaser,
Introduction
of
1999
holistic
nursing
Etzion, 2003
Quasi
Maslach
experimental pro
Burnout
2/-
system
Inventory
cohort
Non-RCT
2/+
annual vacation
job stress
Quasi-
2/+
Quality
Effective
Applic-
Intervent
ability to UK
The
1992
impact
on
GHQ-12
experimental
Intervention
Outcome
Study
Measures
Design
ion?
Y/N
Y/N
Doyle et al. 2007
Maslach
Psychosocial
Intervention
Training
Burnout
RCT
1/++
Inventory
169
The
impact
of
the
introduction of new IT
support
system
2/+
1/++
2/-
RCT
1/+
Non-RCT
2/+
Quality
Effective
Applic-
Intervent
ability to UK
Quasi-
Questionnaires
experimental
pro-cohort
to
monitor patients
Ewers et al. 2002
Training
in
Maslach
Burnout
RCT
Inventory
psychosocial
interventions
(PSI)
Multiple
health
promotion intervention
Quasi-
vitality
experimental
pro-cohort
Schaubroeck et al.
Individual
1993
clarification
role
and
its
impact on subjective
strain,
Somatic
Complaints Index
physical
Communication skills
Maslach
Inventory
Burnout
on burnout
Intervention
Outcome
Study
Measures
Design
ion?
Y/N
Y/N
Aust et al. 1997
Stress
management
programme
of
12
Non-RCT
2/+
RCT
1/+
Non-RCT
2/-
PSS
sessions
Bergdahl et al. 2005
Structured
Affect-
SCL-90,
Focussed
Training
Severity
and
its
stress,
impact
anxiety
on
and
Global
Index
Perceived
and
Stress
Questionnaire
depression
Bunce & West 1996
Stress
management
GHQ-12
programme aimed at
individual
and
170
Innovation promotion
programme aimed at
organisational
improvements
Butterworth
et
al.
2006
Motivational
interviewing
SF-12
Non-RCT
2/-
COPE
RCT
1/+
Non-RCT
2/+
RCT
1/+
Non-RCT
2/+
based
health coaching
Cook et al. 2007
Web
based
health
promotion programme
and stress
Craig, 1996
Healthy
lifestyle
programme
on
physiological
and
psychological health
Eriksen et al. 2002
Stress
management
Cooper
Job
Stress
training
or
Questionnaire
and
exercise
physical
or
an
Subjective
heatlh
checklist
Cognitive behavioural
GHQ-12,
stress
Health
integrated
Health
programme
Gardner, 2005
Horan, 2002
management
Mental
Professionals
or behavioural coping
Stress Scale
RCT
1/-
RCT
1/++
RCT
1/++
2/-
1/++
Stress
management
2000
intervention designed
Depression Inventory
to reduce distress
Lindquist & Cooper,
1999
counselling measures
Lucini, 2007
Non-standard
stress
Quasi-
cognitive restructuring
measures
used.
experimental
and
Physiological
its
impact
on
OSI
non-RCT
measures
included
nervous system
Positive-emotion
BP,
focussed
Organizational Quality
stress
Personal
and
management
Assessment,
programme
Symptom Inventory
RCT
Brief
171
Cognitive
behaviour
therapy
stress
GHQ-30,
RCT
1/+
Quasi-
2/-
RCT
1/+
RCT
1/+
RCT
1/+
Stress
Non-RCT
2/-
Stress
Non-RCT
2/++
Stress
Non-RCT
2/++
Y adverse
CES-D,
health
status
questionnaire
management
programme
Munz et al. 2001
Workplace
stress
CES-D,
Perceived
management
experimental
programme including
and
pro-cohort
personal
schedule
and
negative
affect
organisational
elements
Pelletier, 1998
Stress
management
intervention delivered
self
and
report
physical
psychological
health
Rahe et al. 2002
management
programme including
personalised
Quarterly
feedback
Questionnaire
Stress
and small
Health
group sessions
Sheppard
et
al.
1997
STAI
Transcendental
Meditation
versus
conventional
stress
management
and
IPAT
depression scale
on
anxiety
and
depression
Shimazu et al. 2003
Brief
pact
Questionnaire
of
stress
Job
management
programme
on
teachers
(Shimazu
et
al.
2005)
Web
based
psychoeducaiton
on
Brief
Job
Questionnaire
self-efficacy, problem
solving behaviour and
stress
Shimazu et al. 2006
Brief
management
Questionnaire
Job
effect
programme on coping
skills,
psychological
Mindfulness
based
Subscales
of
the
Non-RCT
2/-
172
stress reduction
Intervention
Outcome
Study
Measures
Design
Quality
Effective
Applic-
Intervent
ability to UK
ion?
Y/N
Y/N
Bond & Bunce 2000
Acceptance
and
Commitment Therapy
and
GHQ-12
and
Beck
RCT
1/+
RCT
1/+
Quasi-
2/-
Quality
Effective
Applic-
Intervent
ability to UK
Depression Inventory
Innovation
Ptromotion
Programmes
Grime, 2004
Computerised
HADS
Cognitive Behavioural
and anxiety)
(depression
Therapy programme
Reynolds, 1997
Individual
and
GHQ-12
and
SCL-
90R (somatisation)
organisational
interventions
experimental
pro-cohort
for
psychological
wellbeing
Intervention
Outcome
Study
Measures
Design
ion?
Y/N
Y/N
Altchiler and Motta
1994
and
RCT
1/+
RCT
1/+
anaerobic
anxiety,
absenteeism,
job
satisfaction
and
resting HR
Atlantis et al. 2004
Exercise intervention
including
weight
aerobic,
training
and
behaviour
modification
173
interventions
Field, 1997
RCT
1/++
Personal
RCT
1/++
STAI
RCT
1/+
Anxiety on a visual-
RCT
1/+
therapy on relaxation
and anxiety
HR,
RCT
1/+
STAI
Non-RCT
2/++
Salivary
Non-RCT
2/++
RCT
1/+
Non-RCT
2/-
Brief
massage
therapy,
music
muscle
group
sessions
Hinman et al. 1997
The
impact
of
Stress
computerised
Questionnaire
exercise
including
programme
subscales
on stress levels in
for
vocational,
VDU workers
psychological,
interpersonal
and
physical
de Lucio et al. 2000
Relaxation, cognitive
restructuring
and
communications skills
training
McElligott
et
al.
2003
pulse
oximetry
et
al.
1997
STAI
Transcendental
meditation
conventional
and
and
IPAT
depression scale
stress
management
Shulman & Jones,
1996
Taniguchi
et
al.
2007
The
impact
of
relaxation training on
immunoglobulin
salivary
Iceberg
immunoglobulin
and
profile
and
mood states
and
4DSQ,
of
mood state
Van Rhenen, 2005
Cognitive
Physical
stress
Burnout
Maslach
Inventory,
reduction programme
Fatigue Checklist
Personal
relaxation on BP and
Questionnaire
Stress
personal strain
174
Health Promotion
Author
Intervention
Outcome
Study
Measures
Design
Quality
Effective
Applic-
Intervent
ability to UK
ion?
Y/N
Y/N
Hasson et al. 2005
Physiological markers
based
including
health
promotion
tool
on
RCT
1/++
2/-
1/+
Quality
Effective
Applic-
Intervent
ability to UK
cardiovascular,
wellbeing
neuropeptides;
rated
selfstress
questionnaire
Nielsen et al.2006
Quasi-
promotion intervention
vitality
experimental
on
health
pro-cohort
and
wellbeing
Peters
&
Carlson
1999
Effects of multimodal
worksite
STPI
RCT
Outcome
Study
Measures
Design
stress
management
programme
Others
Author
Intervention
ion?
Y/N
Y/N
Alford et al. 2005
Kawakami, 1999
Written expression of
emotions
General Scale
GHQ-12,
on stress reduction
cholesterol,
trigycerides
BP,
Non-RCT
2/++
RCT
1/+
RCT
1/+
and
sickness absence
Martin & Sanders
The
2003
positive
impact
of
parenting
programme on work
related stress
175
Nhiwatiwa, 2003
Given
booklet
on
trauma
and
coping
mechanisms
post
GHQ-28,
Impact
of
Non-RCT
2/++
RCT
1/++
RCT
1/+
events scale
assault
Wachi, 2007
Recreational
music
POMS, measures of
natural
activity
killer
cell
State
Trait
Eye
movement
SUDS,
and
anger
inventory,
reprocessing (EMDR)
Police
stress
on
desensitisation
job
stress,
inventory,
symptom
subjective
distress
checklist,
marital
and anger
176
Author and
Date
(Ref ID)
Category of
Intervention
Study
design and
research
type/
quality
Research
question
Does written
expression of
emotions by
employees in
stressful
occupations
help to minimise
stress
reactions?
Study
population,
setting,
country,
sample size
34 treatment
31 controls
31 treatment,
30 controls on
completion
child protective
service
officers,
Australia.
85% women.
Non-RCT
Alford et al.
2005
Stress
management
Short term
findings (<1
year)
Long term
findings (>1
year)
write about
recent
stresses,
emotions,
related
thoughts and
plans in journal
for 15-20
minutes per
day, 3
consecutive
days
interventi
on start of
week 2,
postinterventi
on
questionn
aire end
of week 2
GHQ-12;
PANAS
(Positive &
Negative
Affect
Schedule);
JIG (Job in
General)
scale
significantly
greater change
among treated
for (i) GHQ 12.03
to 8.10 vs 12.30
to 12.10
(p=0.003) and (ii)
JIG 42.97 to
45.26 vs 41.60 to
39.97 (p=0.002)
n/a
Low-impact
aerobics or
nonaerobic
exercises, 30
mins, 3 times
per week for 8
weeks
Pre-test
1; 4
weeks to
pre-test 2;
post-test
1 at end
of 8
weeks of
exercising
; post-test
2, 2
weeks
later
STAI-S &
STAI-T
Collapsed across
groups,
immediate postexercise
sessions had sig
(p=0.005)
reduction in
STAI-S. Effect
was due to
aerobic not
nonaerobic
sessions.
Combined post
test scores,
aerobic sig lower
STAI-S than
nonaerobic
(p=0.018) (effect
n/a
Comments,
confounder
s/ potential
sources of
bias
Consider
possibility
that time
spent writing
about
anything
could have
been
beneficial;
mainly
females; use
of volunteers
Applica
bility to
the UK
Yes
2/++
RCT
Stress
management
Length of
follow-up
Mental
Wellbeing
outcome
variables
Average age
35.2 years
Do different
forms of
exercise
(aerobic and
nonaerobic)
have any effect
on state and
trait anxiety,
absenteeism,
job satisfaction
and resting
heart rate?
Altchiler and
Motta 1994
Description of
intervention(s
)
1/+
43/90 to
completion (23
aerobic, 20
nonaerobic).
Mainly white
(84%), female
(88%), US,
workers with
disabled
children &
adults.
Av. age 33.48y
(Aerobic);
30.40y
(nonaerobic)
Reasonable
adherence to
programme
(2.3/3 per
wk).
No nonintervention
but
differential
effect
suggests not
due to other
uncontrolled
feature of
intervention
(e.g. social
benefit of
sessions).
Yes
177
36 (treatment)
37 (controls)
(20 & 24
completed).
Stratified for
gender &
normal and
high DASS
scores
Australia,
Casino shift
work
employees
Av Ages
Treatment
group 30y;
controls 33y.
24 week
combined
aerobic (mod
to high int 20
mins 3 days
per wk) and
weight-training
(light to mod at
least 2 per wk)
exercise. Plus
behaviour
modification
interventions
(health
education
seminars and
health
counselling).
Out of work
time.
Wait list
controls had
minimum
intervention
RCT
Atlantis et al.
2004
Stress
management
1/+
24 weeks
(6m)
SF-36 and
DASS
questionnaire
s
size =0.22).
Effect mainly at
post test 1.
STAI-T sig
reduced across
study for aerobic
(p=0.018) but not
nonaerobic
(effect size =
0.60).
Further analysis
showed only
previously nonexercising
members of
aerobics group
had sig red in
STAI-T (p=0.016)
Mental Health
(p=0.005 effect
size 0.68);
Vitality (p<0.001
es=1.54);
General Health
(p=0.009
es=0.44);
Bodily Pain
(p=0.005
es=0.62);
Physical
Functioning
(p=0.004
es=0.93);
Stress (p=0.036
es=-0.56);
Depression
(p=0.048 es= 0.16)
Limitation of
effect on
STAI-T to
only previous
nonexercisers
suggests
subjects
might be
anxious
about lack of
exercise.
n/a
Supervised
exercise
sessions
probably
enhanced
adherence to
programme
and certainly
ensured
good
compliance
during
sessions.
Correlation
between
baseline and
changes ie
those worse
pre-study
improved
more use
of
stratification
therefore
enhanced
positive
overall
effect.
Concerns
over low
initial
Yes
178
What effect
does a theorybased stress
management
approach have
on critical
coping
behaviour and
subjective
health
26 (treatment)
28 (controls)
(22 & 24 at
end)
Can structured
affect-focussed
training reduce
heightened
levels of stress
and
psychological
symptoms such
as anxiety and
depression?
27 treatment,
23 control
(20 and 17 at
end)
Non-RCT
Aust et al.
1997
Stress
management
Germany,
Inner city bus
drivers
Stress
management
program (12
sessions)
1.5h 1 x per
wk after work
12 weeks
plus
additional
follow up
investigati
on after 3
months
5weeks
after 7week
school
Age 40 -60
(overall av
49.5y) min 5y
exp.
2/+
Females
working in
areas of social
service, elderly
care and
education in
Swedish
municipality.
Positive and
negative
mood
questionnaire
based on
German
inventory of
QoL;
'Symptom'
questionnaire
based on
German
symptoms
inventory
PSQ
Perceived
Stress
Questionnaire
SCL-90
(Symptom
Check List)
and
GSI (Global
Severity Index
level of
psychological
symptoms)
1 x 2 hr
session per wk
for 7 wks.
No sig. change in
positive or
negative mood
scores or
Symptom scores
n/a
Treatment group
only - significant
decrease in PSQ
scores (0.46 to
0.37 p<0.01) and
GSI scores (0.72
to 0.57 p<0.05)
ANCOVA
showed
treatment/group
interaction for
GSI.
n/a
RCT
Stress
management
1/+
RCT
Bond & Bunce
2000
Stress
management
1/+
How do ACT
(psychotherapy)
and IPP
(Innovation
Promotion
programme)
interventions
affect both
mental health
30 treatment
groups 1 & 2;
30 controls
Volunteer
Media workers
av age 36.4y,
mainly
graduates,
Acceptance
and
Commitment
Therapy
(ACT);
Innovation
Promotion
Programme
(IPP) both
27 weeks
in total
(13 after
end of
interventi
on)
GHQ-12 &
BDI
(depression)
Yes
Selection on
high PSQ
limits
generalisabili
ty.
Control
group did not
attend any
course so
results might
not be
related to
specific
course
content.
Randomly
selected from
top 50/120
high scores on
PSQ
Bergdahl et al.
2005
recruitment
and high
drop out
rate.
Volunteer
subjects.
Initially
poorer mood
scores in
intervention
group should
have
enhanced
chance of
effect.
n/a
Lack of
volunteer
status of
subjects a
strength.
High initial
GHQ scores
indicate
strong scope
for change.
Yes
ACT (p<0.0001)
and IPP
(p<0.006) both
179
and workrelated
outcomes?
balanced for
gender
3x0.5 day
sessions at
weeks 1, 2 and
14.
How does an
intervention
designed to
reduce adverse
psychological
factors in a
hospital affect
the
psychological
health of
workers in that
hospital?
Introduction of
programme of
measure to
reduce
adverse
psychosocial
factors
12
months
after
beginning
Canada,
Hospital
nurses and
auxiliaries,
mainly female,
mixed ages
and
experience
Psychological
distress (PSI Psychologic
Symptoms
Index);
Personal and
client-related
burnout (CBI Copenhagen
Burnout
Inventory);
sleeping
problems
(NHP Nottingham
Health Profile)
Quasiexperiment
al (procohort)
Bourbonnais
et al. 2006
Organisational
Non-RCT
Stress
management
n/a
2/-
45 IPP; 62
SMP; 84
controls at
outset.
20 IPP; 27
SMP; 70
controls at end
UK community
and hospitalbased health
Stress
Management
Programme
(SMP) aimed
at individual
behaviour and
coping;
Innovation
promotion
programme
(IPP) aimed at
organisational
First post
test (T2)
at 9
months
T3 - 15
months
from
outset
GHQ-12
Sig reduction in
GHQ T1 to T2
(p<0.01) in SMP.
Non-sig trend
GHQ increase
T2 to T3 in
SMP
suggesting
effect doesn't
last (p not
cited).
Trend for
decrease in GHQ
T1 to T2 with IPP
but not sig (p not
cited).
Most
planned
interventions
yet to be
implemented
at end so
good result.
Yes
Hospital with
good
expected
cooperation
selected so
results might
be biased by
this.
Analysis of
drop out
suggests no
survivor bias.
2/Can either of 2
modes of
intervention
improve
psychological
well-being and
reduce strain?
The comparison
of post
intervention
mean scores
between both
hospitals,
adjusting for preintervention
scores, proved
favourable to the
experimental
hospital; the
mean difference
was statistically
significant for
work related
burnout (p=0.03)
and borderline
for client related
burnout (p=0.08).
Psychological
distress and
sleeping
problems were
not sig.
Decrease in
GHQ with IPP
sustained at
Subjects
were aware
of nature of
study and
status as
experimental
hospital.
Occupational
groups
known to be
under higher
strain
targeted,
increasing
chances of
change.
Yes
Not all
subjects
180
workers, no
demographic
details given
but said not to
differ between
groups
improvements.
4 wards from
one hospital
(treatment)
and 13 wards
from another 2
hospitals
(control)
holistic nursing
system
introduced in 4
wards in one
hospital
32 treatment,
75 control
Who remained
on same
wards
throughout.
German,
Nurses,
Quasiexperiment
al (procohort)
Bussing &
Glaser 1999
Organisational
MBI-D
(German MBI)
(burnout)
n/a
Both burnout
subscales
increased
similarly and
significantly
(p=0.007) in
both groups
from T1 to T2.
This was
despite
measured
improvement
in
psychosocial
environment.
2/-
Non-RCT
Stress
management
3 years;
1yr setup; 1 yr
transfer
phase;
1yr
testing
phase
T1 at end
of set-up,
T2
towards
end of
testing
(i.e. ~2
years
later)
~90% female;
mean age 35
years, work
experience 15
years
What is the
impact of MIbased coaching
on the physical
and mental
health of
university-based
employees?
Butterworth et
al. 2006
2/-
Initially 145
self-selected
into treatment
and 131 into
control groups.
112 treatment,
118 control at
end.
Subsidiary
case-control
study with 44
Motivational
Interviewingbased health
coaching. 30
minute
sessions,
minimum 1
initial and 2
follow-up.
Actual number
determined by
participants
based on
3 months
SF-12 giving
2 subscales:
PCS
(physical) &
MCS (mental)
Treatment group
improved on
PCS (1.69 pts,
p=0.035) and
MCS (4.4 pts,
p<0.0001).
Control group
didn't change sig.
Matched casecontrol study
showed similar
results (MCS
n/a
were
volunteers,
some were
encouraged
to take part
by managers
no record
identifying
these to
allow
separate
analysis.
Possible bias
due to
nurses
leaving any
of the wards
(initial
sample of
482). Also
clearly very
mobile
workforce
reducing
opportunities
for changes
to exert
influence.
Results
suggest any
experimental
effects
swamped by
other
influences.
Volunteer
status and
apparently
differing
recruitment
methods
including self
allocation to
treatment or
control
seriously
undermines
results.
Yes
Yes
181
pairs matched
on propensity
score.
(perceived)
need and
interest
Health and
research
university
OHSU
(Oregon, USA)
workers;
average age ~
40; more men
in control
group (37%)
than treatment
(10%)
RCT
Cook et al.
2007
Stress
management
1/+
What is the
effect of a
healthy lifestyle
programme on
physical and
psychological
health?
Non-RCT
Craig 1996
Stress
management
2/+
247 + 233
initially
15% & 13%
attrition
Office workers,
mainly white,
female, degree
educated,
salary >$50K
pa
143 initial
volunteers, 48
participants 41
completed
programme
(av age
41.5yrs); 28
attended
assessment
and 28
attended 2
year follow-up.
41 controls
drawn from
those of initial
143 unable to
attend course
matched for
stress
management,
diet and
physical
activity
3 month
test
period
followed
by posttest
Perceived
stress,
symptoms of
distress,
stress state of
change, brief
COPE
No diff between
2 groups on
ANCOVA but
print version
worked better
(p<0.01)
n/a
Testing
immediat
ely postcourse &
2 years
later
GHQ (form
not stated),
LAQ (Lifestyle
Appraisal
Questionnaire
) for perceived
stress (Part 2)
GHQ showed
2 year fall (ns)
on-line -vpaper
presentations
6 week,
healthy
lifestyle
programme
focussing on
individual
stress
management
and healthy
living. 1.5
hours per
week.
Derivation of
propensity
score
referenced
but not
explained.
May be other
variables not
controlled for
or other
factors
biasing
results.
Variability in
extent of
intervention
not
documented.
Examined
'dosage'
(access to
web
material) but
no sig effects
on stress
outcomes.
Selection to
treatment
group by
availability
and control
group by
unavailability
Yes
Yes
Very high
attrition rate,
especially at
2 year
follow-up.
Authors
attribute lack
of effects to
low statistical
power
182
Can equipping
mental health
workers with the
skills to
integrate
psychosocial
interventions
(PSI) into
practice cut
levels of burnout
among staff?
Individual/Organ
isational
1/++
Quasiexperiment
al (procohort)
Engstrom et al
2005
Organisational
2/+
PSI course, 3
hour sessions
weekly for 16
weeks during
work.
Not
stated,
believed
to be
immediat
ely post
course
(16
weeks)
MBI, 3
subscales for
personal
accomplishme
nt, emotional
exhaustion
and
depersonalisa
tion.
n/a
What effect
does the
introduction of
new IT support
systems have
on
psychosomatic
health of mental
health care
workers?
No sig diffs.
27
intervention;
32 controls
falling to 17
and 16 at end.
Sweden,
Nurses
(licensed
practical
nurse) almost
all female av
age ~40yrs,
IT measures to
improve
monitoring of
patients and to
provide
relatives with
info.
12
months
from
initial
implemen
tation but
at same
time as
last
element
sleep
disturbance
and perceived
stress
subscales of
SWQ
(Satisfaction
with Work
Questionnaire
)
No sig effects
although
perceived stress
improved in
intervention
group across
study
n/a
Although
nominated
allocation to
groups was
random.
Yes
Although
burnout not
affected PSI
as
apparently
used more
by training
group in their
work.
Av age 38yrs,
mainly female
nurses (3540% nonnursing)
RCT
Doyle et al.
2007
Australia,
University
staff. Controls
77% support,
23% academic
treatment
group details
not given.
14 training and
12 control,
derived from
qualified staff
working at a
medium
secure unit in
the UK and
nominated to
attend training
by their work
group.
6 and 4
refused any
involvement
at baseline.
Yes
Allocation of
units to test
or control
was random.
Mainly parttime
workers.
183
part-time
workers
What effects do
Stress
Management
Training (SMT),
Physical
Exercise (PE) or
an Integrated
Health
Programme
(IHP) in a
workplace
setting have on
subjective
health
complaints?
RCT
Eriksen et al
2002
Stress
management
1/+
189 (SMT) +
165 (PE) +
162 (IHP) +
344 (Control)
(98, 114, 94,
166 by end).
Norwegian
postal workers;
~60% female.
Controls sig
less time in job
and fewer
hours per
week.
Mixture of blue
and white
collar jobs.
Changes
added
responsibiliti
es as well as
making
some
aspects
easier.
Stress
Management
Training (2 hrs,
weekly for 12
weeks);
Physical
Education (1
hr, twice
weekly for 12
weeks);
Integrated
Health
Programme (2
hrs, weekly for
12 weeks);
Controls (no
stated
treatment) All
during working
hours.
Pre-test,
post 12
week
interventi
on, 1 year
follow-up
Cooper Job
Stress Q.
Subjective
Health
Checklist
(SHC)
At immediate
post training or
follow up:
No sig effect on
SHC (or sick
leave).
No sig effect of
interventions cf
control for Job
Stress (Cooper).
Even when
adherence to
programme
(>50%) was
analysed
n/a
Limited time
for all
intervention
to have any
effect.
Poor
retention in
study, poor
adherence to
programmes
(even though
in work)
although this
was
analysed for
it possibly
indicates
poor attitude
to study.
Yes
Considered
generally
healthy
workforce.
184
What impact
does taking an
annual vacation
have on
perceived job
stress and
burnout
Non-RCT
Etzion 2003
Organisational
55 treatment
with 55 age,
gender and job
matched
controls
taking a
holiday v not
taking a
holiday
Israel,
industrial
employer,
mainly married
men, av age
~44yrs, mixed
jobs (white
collar)
Immediat
ely on
return
and 3
weeks
after end
of
vacation
10 + 10
stratified by
ward, gender
and day/night
duty.
RCT
1/++
What are the
immediate
effects of: brief
massage
therapy, music
relaxation with
visual imagery,
muscle
relaxation, and
social support
group sessions
on anxiety,
depression and
vigour?
RCT
Field 1997
Stress
management
n/a
No
randomisatio
n or
anonymity.
Yes
Subjective
scales.
Vacation
takers seen
as controlling
timing
might not
apply to nontakers.
20 day
Training in
psychosocial
interventions
Immediat
ely post
20 day
training
interventi
on
Maslach
Burnout
Inventory
subscales:
accomplishme
nt,
exhaustion,
depersonalisa
tion
Sig.
improvements
over controls:
accomplishment
(p=0.01),
exhaustion
(p=0.04)
depersonalisatio
n (p=0.01)
n/a
Small scale
study. PI
worked in
same unit.
Yes
Massage (10
mins either at
workplace or in
separate
room), Music,
Relaxation (10
mins in
separate
room), Social
support (10
mins talking)
immediat
ely post
10 min
interventi
on
STAI-S &
POMS
(depression
and vigour)
No significant
changes
between groups,
all of which
showed pre-post
test improvement
in all measures.
n/a
Could be
Hawthorne
effect or
general
relaxation
effect of 10
mins away
from work.
Yes
13 of initial
sample
declined to
take part.
Individual/Organ
isational
Burnout tended
to drop after
vacation and
stayed down
(p<0.10).
2/+
Does training in
psychosocial
interventions
reduce burnout
rate in forensic
nurses?
Ewers et al.
2002
Burnout (not
MBI) with
physical and
mental
exhaustion
subscales.
1/++
Mental health
care staff,
mainly nurses
100 subjects
(64% female)
in 5 groups
(gp. Nos not
given but
assumed 20)
USA, Hospital
employees
(?nurses)
No follow-up.
Subjective
questionnair
es might
load results
(e.g. I feel
nervous I
feel at ease)
185
What effect
does training in
relaxation,
cognitive
restructuring
and some
communication
skills have on
communication
skills and stateanxiety?
RCT
de Lucio 2000
Stress
management
1/+
What effects do
two forms of
stress
management
training have on
stress and
general health
(GHQ)?
29 (treatment)
& 32 (controls)
(30 declined).
(20 and 31
STAI post
intervention)
Spain, female
(?) nurses
stratified by
area of work,
shift and
professional
category.
57 cognitive;
44 coping; 37
control
51, 38, 29 at
end of course
42, 37, 25 at
follow-up
UK NHS
employees,
mainly
'intellectual
disabilities
service';
mainly female
(82%); av age
37yrs
Immediat
ely
following
interventi
on.
State-TraitAnxiety
Inventory
(STAI-S)
No significant
differences in
STAI scores
5hr weekly
sessions over
5 weeks
outside
working hours,
plus
homework.
3, 3.5 hour
weekly
sessions,
either cognitive
or behavioural
coping plus
homework.
Order of
courses
randomly
determined
such that
nature of
course not
known to
subjects prior
to arrival.
3 month
follow-up
GHQ-12;
MHPSS
(Mental
Health
Professionals
Stress Scale)
Control group
less stressed
(MHPSS) at
outset (allowed
for in analysis).
No sig effects on
GHQ for whole
groups.
When only data
from those with
GHQ 4 or more
were analysed
then sig
reduction
occurred and
continued postintervention.
(p<0.04).
GHQ scores fell
for all three at
post training.
Cog continued to
fall, coping
remained down
and control
returned to pretest levels.
Non-RCT
Gardner 2005
Stress
management
2/+
n/a
n/a
Text unclear
nature and
extent of
drop outs.
High STAI-S
scores
before and
after (84%
higher than
median for
Spanish
adult
women)
Although
there were
problems
with
recruitment
and
allocation of
subjects to
study groups
these effects
were likely to
be minor (i.e.
only 3
people
allocated to
treatment
who should
have been
wait listed).
Data
analyses are
reported to
explore likely
impact of
these to
support
findings.
GHQ
outcomes
seen as
clinically
significant in
size of effect
for both
interventions
Yes
Yes
186
What is the
effect of an 8
week
computerised
Cognitive
Behavioural
Therapy (CBT)
programme on
emotional
distress in
employees with
recent stressrelated
absenteeism?
RCT
Grime 2004
Stress
management
1/+
What are the
effects on
mental and
physical wellbeing and
stress-related
biological
markers of a
web-based
health
promotion tool?
RCT
Hasson et al.
2005
Stress
management
1/++
24 (treatment)
24 (controls). 8
failed to
complete all
treatment.
14 & 19 at
follow-up.
UK, NHS and
Local Authority
workers, 10 or
more days
absence due
to stress,
anxiety or
depression in
6 months.
GHQ>4.
Recruitment
via a number
of channels
(direct and
indirect)
Subjects
already under
treatment
keep in
review?
129
(treatment) &
174 (controls)
drawn from
volunteers.
121 & 156 at
end. Analysed
on intention to
participate
basis
Sweden, IT
and media
company
employees.
Slight excess
of males.
CBT &
conventional v- conventional
(whatever
care they were
receiving).
end, 1m,
3m & 6m
HADS
(Anxiety &
Depression)
With CBT v
control:
Depression
lower at end
(p=0.028),
n/a
Anxiety
(p=0.021), &
Depression
(p=0.040) lower
at +1m.
CBT 8 weekly
sessions
(duration not
given)
presented on
PC at Occ.
Health Centre.
No effect at +3 &
+6 months
although
adjusted values
remained below
end of treatment
scores.
Low
attendance
and
adherence to
training a
problem
(might have
been due to
need to
attend clinic).
Some
problems
with
employers
allowing time
for access
also
reported.
Yes
Also
preferences
seen for
conventional
treatment as
being more
tailored and
more faceto-face
Web-based
tool for health
promotion and
stress
management.
Control group
had access to
website but not
lifestyle
training &
chat
elements.
6 months
Physiological
markers Cardiovascula
r and lifestyle;
Stress related;
Recovery
related;
Immune
markers and
neuropeptides
; Self-rated
health and
stress
questionnaire
(nonstandard)
Treatment group
Improved
significantly in
terms of selfperceived ability
to manage stress
(p=0.001), sleep
quality (p=0.04),
mental energy
(p=0.002),
concentration
ability (p=0.038),
& social support
(p=0.049).
n/a
No
knowledge of
total
potential
participants
(estimated
80% take-up
overall).
Yes
Web tool
only differed
in interactive
training and
chat
elements.
187
Mixed ages.
No analyses of
pre-test
demographics
reported but
ANCOVA
applied.
Can a
computerised
exercise
programme
reduce stress
levels in office
workers who
use video
display
terminals?
24 (treatment),
26 (controls) (3
excluded due
to complete
noncompliance
with treatment)
USA, female
office workers,
30's
RCT
Hinman et al.
1997
Stress
management
1/++
What are the
effects of
organisational
interventions on
employee
health?
DahlJorgensen et
al. 2005
Organisational
Quasiexperiment
al (NonRCT)
415/560
responded to
initial
questionnaire,
only 336 of
these were still
in same
employment at
post-test and
282
responded.
Breakdown of
response by
group is not
given
2/+
Shop workers
Computerdirected
exercise
break, 2 per
day 15 mins
per break. In
addition to
normal breaks.
Exercise
period could
be put on
hold for work
reasons.
Self-reported
compliance
ranged from
3.8-100%
(av=39.5%)
Participatory
approach to
organisational
change.
Different
specific
approach in
two
workplaces
(shop workers
from a number
of shops in
mall).
At end of
8 week
program
me
4 dimensions
of stress on
PSQ:
vocational,
psychological,
interpersonal,
physical
Not clear,
about 6
months
post
interventi
on
SHC (Somatic
symptoms)
and burnout
(MBI)
of neuropeptide
(NPY) increased
significantly
(p=0.02) CGA
(associated with
catecholamine
activity)
decreased
(p=0.01), levels
of immune
marker TNFa
decreased
(p<0.016).
No sig effect
between
treatment and
control groups or
between
compliers and
non-compliers in
treatment group
With municipal
workers there
were no sig
changes in
expected
direction but
'emotional
exhaustion' was
sig increased in
intervention
group (p=0.05).
With mall
workers
depersonalisatio
n (p<0.05) and
somatic
symptoms
n/a
Small
sample size,
small
differences
between
nonsymptomatic
groups, low
compliance,
remained at
computer for
exercise.
Selfconsciousne
ss of
participants.
Yes
n/a
Poor
implementati
on of
interventions
for a variety
of reasons
including
management
attitudes,
especially in
municipal
employer
and adverse
attitude to
study (again
municipal).
Authors
Yes
188
and municipal
workers in
Norway.
Mainly female,
approx 2550% part time,
12-25%
managers, av
age around
40yrs.
66 in total, split
not given
USA, Mostly
married,
caucasian
females with
children, av
age 45yrs
(p<0.05) sig
improved
compared to a
decrease in
controls.
"Chicken soup
for the soul at
work"
workplace
story groups.
Immediat
ely post
interventi
on only
OSI-R; PMI
One meeting
(duration
apparently 1
hour) per week
for 11 weeks.
OSI_R - 13/14
scales not
significant,
interaction effect
on one scale
(p=0.002)
treatment means
reported but not
those for control
group.
regard
changes as
Hawthorne
rather than
genuine
impact.
n/a
Very poor
documentati
on of noncompleters
and nonattenders
RCT
Horan 2002
Jones &
Johnstson
2000
1/-
RCT
Stress
management
1/++
What is the
effect of a stress
management
intervention
designed to
reduce affective
distress in
student nurses
who have
previously
reported
40 (treatment)
& 39 (controls)
3 months
& 18
months
3 month
attrition stated
as 6% & 7%
(2.4 & 2.7
subjects)
GHQ, STAI,
Beck
Depression
Inventory
(BDI), Beck &
Srivatava
Stress
Inventory.
UK,
'Distressed'
Objective
performance
In treatment
group:
Significant fall in
GHQ-30
(p<0.0005)
Yes?
Those
without time
to attend
didn't attend;
Of the 24
subscales within
the PMI, 3
subscales, 2
within the Mental
Wellbeing
category, had
significant
effects: (State of
Mind (p=0.04) &
Confidence Level
(p=0.02))
Stress
management
Poor control
group.
Treatment
group had a
meeting per
week away
from job control group
didn't.
GHQ-30
remained sig
lower
(p<0.0005)
STAI-T
remained
lower
(p<0.0005)
BDI remained
Apparent
concerns
about
confidentialit
y of
responses
Study on
those with
existing
distress.
Yes
Authors
acknowledge
that effects
might be
attributable
to other,
189
significant
distress?
student nurses
with score of 4
or more in
GHQ-30 20
weeks prior to
intervention.
85% female.
measures
(sickness
absence and
general
absence)
(p<0.0005)
STAI-S sig lower
(p<0.0005)
No sig effect on
objective
measures
Higher GHQ
scores in nonparticipant
group
What is the
effect of an
organisational
intervention on
levels of workrelated stress
amongst
Japanese bluecollar workers
with pre-existing
high depressive
scores?
Quasiexperiment
al (procohort)
Kawakami
1997
Organisational
2/-
Japan,
Worksites with
mean
depression
scores higher
than mean
+1s.d. for
whole
company.
111 workers at
2 sites
matched for
age and mean
depression
score with 3
other site (183
workers).
Int sites av age
33y, 76%
male.
Control av age
35y, 56%
male.
110 & 175
actually took
part and 79 &
108 remained
at follow-up.
lower
(p<0.0005)
STAI-S
difference
reduced but
still sig lower
(p=0.002)
No effect on
objective
measures.
1 year
Workplace
participatory
intervention
aimed at
identifying &
reducing
sources of
stress in
workplace.
1&2
years
Zungs SDS
score
(depression)
Sick leave
Blood
Pressure (2yr)
For intervention
group:
Progressive
reduction in
depressive
symptoms
across 2 years.
Total effect Sig
(p=0.035)
No effect on BP.
Sig gp x time
interaction for
sickness
(p=0.034)
Data suggests
increase in year
1 (no separate
analysis)
For
intervention
group:
Progressive
reduction in
depressive
symptoms
across 2
years. Total
effect Sig
(p=0.035)
No effect on
BP.
unintended
aspects of
course such
as increased
attention,
diversion
from other
cares,
increased
socialisation,
etc rather
than the
course
content per
se.
ANCOVA
allowed for
pre-test
differences
in age or
outcome
scores.
Cannot rule
out
systematic
differences
between
worksite
groups.
Intervention
relied on
supervisor
support and
did not
involve
workers in
planning
stage. No
effect on
minority of
female
workers.
Other factors
such as
increased
work
demands on
intervention
sites during
Yes
190
follow-up.
What effects
does a mailed
advice leaflet on
stress reduction
have on
psychological
distress, blood
pressure, serum
lipids and sick
leave amongst
Japanese
manufacturing
employees?
Stress
management
RCT
Organisational
1 year
follow up
GHQ-12,
Blood
pressure,
cholesterol,
triglycerides &
sick leave in
year.
No sig effect on
any of these
outcome
measures.
n/a
1/+
What effect
does web-based
training for
supervisors on
supervisor
support have on
psychological
distress
amongst
subordinate
workers?
Kawakami et
al. 2005
Written advice
tailored to
individual
(physical
activity,
nutrition,
breakfast,
alcohol,
relaxation,
etc.) Included
pre-test GHQ12 score.
81 & 77 at 1
year follow-up
(health checks
on 48 & 45)
RCT
Kawakami
1999
Japan,
Workers with
GHQ-12 score
of 3 or more.
113 initially
selected in 2
groups, others
excluded at
baseline
reduced
numbers to 91
(intervention)
and 88
(control).
1/++
Supervisors - 9
(treatment) & 7
(controls);
Workers - 100
(treatment) &
90 (controls).
82 & 84 at
follow up
Japan,
Technicians
and clerks in
an IT company
Web-based
training of
supervisors on
work-site
mental health.
4 week training
period based
on Guidelines
for Promoting
Mental Health
Care.
4 months
Brief Job
Stress
Questionnaire
(BJSQ)
subscales for
vigour,
anger/irritabilit
y, anxiety &
depression.
No sig.
intervention
effect for any of
the 5 sub-scales
of psychological
distress.
n/a
Both groups
reduced
GHQ-12
scores,
attributed to
natural
regression.
Could have
been
crossover
between
individuals in
different
groups at
same plant.
BP
measures
limited by
reduced
group size.
Measures
suggested
that
supervisors
learned
messages
from training
but that this
was less-well
perceived by
subordinates
possibly due
to peak in
work
demands.
Yes
Yes
No measure
of knowledge
and
understandin
191
g of control
supervisors
at baseline.
What effect
does web-based
training for
supervisors on
supervisor
support have on
psychological
distress
amongst
subordinate
workers?
Supervisors 23 (treatment)
& 23
(controls);
Workers - 92
(treatment) &
114 controls
81 & 108 at
follow up
Web-based
training of
supervisors on
work-site
mental health.
4 week training
period based
on Guidelines
for Promoting
Mental Health
Care.
4 months
Brief Job
Stress
Questionnaire
(BJSQ) psychological
distress
No sig.
intervention
effect for total
psychological
distress.
n/a
Participatory
Action
Research
(Organisationa
l Intervention)
in two different
departments
12
months
post-test
Depression
and sleeping
subscales
from Job
Content
Questionnaire
Neither
intervention had
any sig effect on
either outcome
measure
n/a
Japan,
Sales &
service
workers in an
office
machines
company
RCT
Kawakami et
al. 2006
Landsbergis &
VivonaVaughan 1995
Organisational
1/++
Quasiexperiment
al (nonRCT)
Organisational
2/+
What is the
effect of an
occupational
stress
intervention in 2
municipal
departments on
levels of strain
and depression
in employees?
39 intervention
1; 10 control 1;
24 intervention
2; 26 control 2.
37, 23, 23 &
20 at follow up.
77 pairs (some
recruitment)
US municipal
employees in 4
paired
departments.
More
females in
intervention
group
might be
affected by
male
supervisors.
Training
increased
knowledge
and attitude
of
supervisors
(which was
perceived by
employees)
but had no
effect on job
stressors.
Pre-existing
high levels of
support.
Limited
scope for
influence of
support on
sales staff
working out
of the office.
This quasi
experimental
study used 2
pairs of
sections in 2
US municipal
departments
to mount 2
parallel
intervention
studies
(PAR). The
authors
express
doubts about
Yes
192
Can a battery of
training and
counselling
measures
reduce levels of
stress amongst
office workers?
RCT
Lindquist &
Cooper 1999
Stress
management
1/++
Mixed gender,
typically
around
30years av
age, mainly
caucasian.
Control 1 had
sig higher
proportion of
clerical, lower
paid, noncollege
educated
employees
than
intervention 1
otherwise no
sig diffs.
52 treatment
group, 52
control group
the quality of
the
intervention,
especially in
one of the
two depts
including
limited
worker
participation
Stress
awareness,
lifestyle and
coping.
100% retention
Australia,
Government
Office workers,
55% female,
educational
and other
demographic
factors not
given
4 x weekly
workshops
plus 45 minute
personal
counselling
session at end.
Not
followed
up as
controls
given
treatment
after first
posttreatment
evaluatio
n at 8
weeks.
Subscales of
OSI for
perceived
stress,
home/work
interface and
physical
health, plus
physiol (BP)
No significant
effects of
treatment on any
outcomes
Battery of 4
workshops
aimed at
stress in
particular but
also adverse
lifestyle
factors
(smoking,
drinking etc)
had no
significant
impact
although
perceived
job stress
showed
trend
(p=0.06)
Effects at
post training
follow-up
(after control
group had
received
training)
suggests
longer term
benefit but
lack of
193
control group
makes this
uncertain.
What effect
does a stress
management
programme
have on
psychological
profiles and
autonomic
nervous system
regulation?
26 treatment,
25 controls in
training
programme
White-collar
workers,
similar ages
and BMIs
between 2
groups, more
females in
treatment
group
1 hour per
week for 1
year, mental
relaxation and
cognitive
restructuring in
small
workshops
during lunch
breaks.
Not
followed
up
beyond
end of
year of
study.
Overall stress,
tiredness
perception
and stresssymptoms
measured
using nonstandard
instruments
used (and
published) by
same group
previously.
Plus
physiological
measures of
blood
pressure and
heart rate (RR
Interval)
variability
3 years
SACL-90
(general
stress)
Sham group
had yearly
meeting and
monthly
emails.
Levels of stress,
tiredness and
somatic
symptoms fell in
treatment group
(authors state
this is sig but no
stats reported)
although
absolute levels
still higher in 2 of
3 than with sham
group. In
contrast levels of
all three rose in
sham group.
n/a
Quasi
experiment
al (NonRCT)
Lucini 2007
Stress
management
2/-
Quasiexperiment
al (NonRCT)
Maes et al.
1998
Individual/Organ
isational
2/+
175 treatment,
171 control
Workers.
167, 157, 134
treatment at
T2-T4.
169, 157, 130
control at T2T4.
Dutch,
Household
goods
Extensive
battery of
interventions
including
lifestyle :
(exercise,
healthy
eating); Stress,
training on
social skills
and
leadership,
plus
organisational
change -
Absenteeism
No effect on
general stress
reactions;
Decrease of
8%
absenteeism
in treatment
group
compared to
5% on control
group.
(no stats
reported)
Intervention
not main
focus of
study which
was
cardiovascul
ar health.
Intervention
poorly
designed
with subjects
volunteering
to treatment
or sham
group.
Treatment
group
appears sig
worse than
sham at
outset
although no
stats
reported.
Sham
treatment not
comparable
to main
treatment.
Extensive
longitudinal
study with
large battery
of measures,
mainly
focussed on
healthy living
and lifestyle
although
some
specifically
mental wellbeing.
Yes
Yes
194
manufacturing.
working
methods
Difficult to
determine
relative role
of different
aspects.
1 intervention
site, 1 control
(random) plus
second control
site.
Treatment av
age 38.6yrs,
control 40.9
yrs (ns)
What effect
does a Positive
Parenting
training
programme
have on a
number of
measures
including workrelated stress?
RCT
Martin &
Sanders 2003
Stress
management
1/+
Treatment
26.1% female
sig more than
control
(12.2%)
Controls had
higher
educational
status.
23 (treatment)
& 22
(controls).
16 & 11 at
completion
Australia,
General and
academic staff
in a
metropolitan
university.
Needed to
have a child
with
behavioural
problems and
be
experiencing a
significant
degree of
distress.
Work-place
3P (Positive
Parenting
Program) over
8 weeks.
6 months
(4 months
post
training)
DepressionAnxiety-Stress
Scale 21
(DASS 21),
Work Stress
Measure,
No effect of
training on work
stress (scale
referred to not
given) although
post-training
levels fell in
treatment group
and rose in
control group.
n/a
Training had
beneficial
effect on
child
behaviour
and other
measures
not of
concern here
but these did
not manifest
themselves
in terms of
stress
although
positive
trend.
Yes
Analyses of
those
dropping out
suggests no
difference to
those
remaining
but high rate
of concern.
195
What effect
does training
Project
Managers have
on their
psychological
strain?
34 (treatment):
33 (controls)
34 & 33 at 7
weeks post
intervention
23 & ?? at 17
weeks post
training
Training to
increase
management
control,
organisational
change to
facilitate
increased
control. 10
hour training
session
17 weeks
postinterventi
on
Depression
(CES-D),
Anxiety
(Caplan)
No main effect
on well-being
outcomes
n/a
Diffusion to
non-trainees
unlikely.
RCT
organisational
1/+
What effect
does a
workplacebased stress
management
programme
have on blood
pressure and
emotional health
of hypertensive
employees?
RCT
McCraty et al.
2003
Stress
management
1/++
21 (treatment)
& 17 (controls)
18 & 14 on
completion
USA,
Hypertensive
employees of
a global IT
company
Yes
Trainees
were
directed to
attend
training on a
non-work
day.
USA, Canada,
Mexico.
PMs from
trucking
company.
mainly male,
av. Age 36.8y,
managerial
Logan &
Ganster, 2004
Improved
organisation
al control did
not enhance
well-being.
Positive
emotionfocused stress
management
programme
(Inner Quality
Management)
1 x 8 hour + 2
X 4 hours over
2 week period
3 months
after
training
BP; Personal
and
Organizational
Quality
Assessment
(POQA); Brief
Symptom
Inventory
(BSI)
BP 10.6mmHg
systolic reduction
sig larger than
controls
(P=0.05);
Improvements in
individual
wellbeing and
organizational
effectiveness at
3m followup
including
Positive Outlook
(p<0.01); Stress
Symptoms
(p<0.05) from the
POQA
Reductions in
depression
(p<0.05) and
n/a
Some
suggestion
that training
raised
expectations
therefore
diluting any
beneficial
effect.
Positive
emotionfocused
stress
management
programmes
can have
health and
wellbeing
benefits for
workers with
hypertension
Yes
Small scale
study,
controls just
wait-listed.
Involvement
in training
might have
improved
196
What effects
does touch
therapy have on
relaxation and
anxiety of
nurses?
12 (treatment),
8 (controls)
12 & 3 on
completion
Mainly (75%)
female, 30's,
degree
educated
Touch therapy
compared with
laying on of
hands. 45
mins sessions
(plus 15 mins
for tests)
Before
and after
each
treatment
session
only
Anxiety on
Visual
Analogue
Scale + BP,
heart rate,
pulse oximetry
and
respiration
rate.
Participation in
programme to
identify and
develop
workplace
changes. Initial
6 hour seminar
then Work
sub-groups
met for 2 hours
per week over
9 weeks.
Post test
1,
immediat
ely at
conclusio
n of
interventi
on period
(12
weeks)
and posttest 2 12
months
after
onset of
study
Cooper Job
Stress; UHI,
Work (Health
Inventory) and
STAI-T
plus
organisational
measures
phobic anxiety
symptoms
(p<0.05) from the
BSI and the
Global Severity
Index derived
from the BSI
(p<0.05)
No sig diffs in
any parameters.
adherence to
other
treatment
elements
(e.g.
medication)
n/a
RCT
McElligott et
al. 2003
Stress
management
1/+
What impact
does a
participatory
organisational
intervention
have on job
stress and job
characteristics?
37 treatment
gp 1; 59
treatment gp 2;
31 control gp
1; 35 control
gp 2
33, 58, 29, 33
at post-test 1
16, 56, 23, 30
at post-test 1
Non-RCT
Mikkelsen &
Saksvik 1999
Organisational
Norway,
Postal
workers, Gp 1
30's gp 2 40's,
more female in
all
groups.differen
ces between
groups in
some
demographics,
allowed for in
analysis
Both
interventions had
no effect on job
stress health
inventory or Trait
anxiety.
n/a
Touch
therapy
(physical
therapy) had
no sig effect
in small
scale poorly
sustained
study with
only 3/8
controls
completing
and no follow
up.
Participatory
approach to
improving
work place
and work
organisation.
Some
isolated and
inconsistent
effects on
organisation
al
parameters
but key
outcomes
unaffected
by
interventions
Yes
Yes
Poor
adherence to
intervention
programme,
especially
intervention
1.
2/+
197
What effect
does a
participatory
organisational
intervention
have on job
stress in
community
health care
institutions?
2 intervention
groups in
different post
offices in 2
cities plus
control groups
from other
offices in same
city
Not clear.
Demog data
reported for
max of 45
treatment, 34
control
Supervisors
and
employees at
2 health care
institutions.
Predominantly
female, middle
aged
Change
limited to
local factors.
Participation in
programme to
identify and
develop
workplace
changes. Initial
6 hour seminar
then Work
sub-groups
met for 2 hours
per week over
9 weeks
Post test
1,
immediat
ely at
conclusio
n of
interventi
on period
(12
weeks)
and posttest 2 12
months
after
onset of
study
Cooper Job
Stress; UHI
(subjective
health)
plus
organisational
measures
Intervention v
nointervention. 2
intervention
groups, results
merged
Organisational
RCT
Stress
management
1/+
28 (treatment)
& 30 (controls)
21 & 30 at
follow-up
Cognitive
behaviour
therapy stress
management
programme.
Japan,
Manufacturing
Company,
2 hour
behaviouralcognitive
3 months
GHQ-30,
Centre for
Epidemiologic
Studies for
Depression
(CES-D),
Health Status
Questionnaire
CES-D scores
reduced in SM
group, difference
1.1 (p=0.003),
No effect on
GHQ-30
Very similar
participatory
approach to
that in earlier
Mikkelson
paper (4637)
but appears
to have had
more
success
although loss
of follow up
is
disappointing
Yes
Treatment
and control
groups from
different
institutions
so changes
could be due
to other
factors.
Mino et al.
2006
n/a
There were no
significant effects
of the
intervention on
subjective health
and anxiety (text
says this but no
ref to
measurement of
anxiety
elsewhere)
Non-RCT
Mikkelsen et
al. 2000
n/a
Study took
place at time
of
considerable
change in
organisation.
No
differences
in outcome
measures
amongst
those lost to
follow up.
Yes
Poor
198
male manual
workers, av
age 38yrs.
All volunteers.
Quasiexperiment
al (procohort)
Munz et al.
2001
Stress
management
2/-
55 intervention
24 controls
USA,
Telecommunic
ations
customer
service/sales
reps. Very
limited
demographic
details. No sig
diff in years
worked for
company or
years in
present job.
No age or
gender info.
Recruitment
not clear.
Approximately
150
employees in
each potential
pool. 55 took
part in the
treatment with
24 controls.
Not stated how
many were
invited.
training +
2 hour muscle
relaxation
training plus
advice to
continue
individually.
Followed by
email based
advice and
counselling as
required.
Combination of
selfmanagement
and
organisational
stress
reduction
interventions.
3 X 4 hours
selfmanagement
skill
development
+ manual.
Participation
voluntary (all
55 took part).
4-6 work unit
employees in
participatory
programme to
improve
workplace.
compliance
with stress
self
management
advice and
reluctance to
use email
counselling.
Post test
1,
immediat
ely at
conclusio
n of
interventi
on period
(12
weeks).
Perceived
Stress Scale
(PSS);
Depression
(CES-D);
Positive and
Negative
Affect
Schedule
(PANAS)
Intervention
group showed
sig less stress
post-intervention
(p<0.05); and
less depression
(p<0.05); less
negative effect
(negative
arousal)
(p<0.05); less
tiredness
(p<0.05); more
positive energy
(p<0.05); less
low negative
affect (more
relaxation)
(p<0.05).
Intervention
group also
showed better
increase in
productivity and
lower
absenteeism.
n/a
Combination
of
participatory
and
individual
intervention not possible
to
differentiate
effects of
different
parts of
package.
Concerns
over the
initial
selection.
Participation
in course
was
voluntary
and so selfselected
group
entered
intervention.
Controls
were from
different
work units
and so might
not have
been
comparable
although
ANCOVA
used to allow
Yes
199
Can a brief
intervention
reduce
symptoms of
distress
following
assault?
Nielsen et al.
2006
Stress
management
Stress
management/
organisational
Given booklet
on effects of
trauma and
coping
mechanisms
3 months
GHQ-28;
Impact of
events scale
(IES)
prob 20
treatment, 20
control
UK,
nurses at
medium
secure
hospitals who
had been
assaulted
within the past
month.
Approx equal
males and
females, mean
age not given
69% from 1
hospital.
Non-RCT
Nhiwatiwa
2003
Not explicitly
stated 45/90
initially agreed
to take part.
treatment group
had higher levels
of distress (IES)
post-intervention
(change = +2,
compared to -6
in controls;
p<0.03). No
difference for
GHQ-28
although mean
scores went up
in treatment and
down in control
groups.
n/a
n/a
sig. increase
in cognitive
stress reaction
at canteen A
(int: p<0.01);
B (int: p<0.05)
& C (cont:
p<0.01).
for preintervention
differences.
Changes
restricted to
within work
unit.
Genuine
effect,
concern is
over extent
to which it
can be
generalised.
No account
taken of
previous
number of
assaults or
of severity of
assault
which might
have
influenced
outcomes.
Yes
Poor
reporting of
sampling
and
numbers.
2/++
Quasiexperiment
al (procohort)
2/-
varies:
I-ntervention
canteens: 27
and 17
completed
questionnaires
at both times;
control 11 and
varied
individual and
group
interventions.
Exercise,
empowerment,
IT training,
workshop on
development
20
months
Stress
symptoms &
vitality
Well-being
not main
focus of
paper.
Yes
Complex
array of
interventions
some
200
16 did so.
sig increase in
vitality at
canteens A
(int: p<0.05);
B (int: p<0.05)
& D (cont:
p<0.01)
Numbers at
pre-test
45,26,22,25
and at posttest 30, 26, 19,
28
canteen
workers at 4
hospitals/
elderly care
homes
Quasiexperiment
al (NonRCT)
Mattila et al.
2006
Organisational
2/-
253
intervention;
107 control 1;
165 control 2.
No details if
this is outset or
completion.
Finnish
municipal
workers.
Mainly men in
manual work,
typically about
44yrs old (av)
with most
leaving school
before
matriculation
Participative
work
conference but
apparently no
actual
implementatio
n of change
Post test
1,
immediat
ely at
conclusio
n of
interventi
on period
(12
weeks)
and posttest 2 12
months
after
onset of
study
Maslach
Burnout
Inventory general
survey
One general
stress
question
Participation had
no effect on
emotional
exhaustion or
perceived stress
n/a
individual,
others
organisation
al. Different
interventions
at two sites.
Number of
criticisms by
authors of
process of
implementati
on of
interventions
Many other
changes
extraneous
to project
(e.g. new
manger at
one site,
announceme
nt of
closures and
cuts at
another.
Participation
in
'conference'
was
mandatory but many did
not take part.
No reasons
are given for
this but
theses
people then
formed the
first control
group (so no
randomisatio
n) addition of
second
control group
offsets this.
Intervention
seems to
Yes
201
RCT
1/+
Pelletier 1998
Stress
management
1/+
3 X 27
subjects (full
intervention,
no telephone
component,
control).
21, 20 & 25
completed
study.
County
employees,
secretarial and
lower middle
management,
mix of
ethnicity, race
and gender
(details not
given)
24 (treatment),
26 (controls)
21 and 19
respectively
completed
USA,
Maintenance
workers, 60%
4 written
training
modules on
reducing
stress at work
and home,
book by first
author, stress
reduction
audiotape,
stress card to
assess daily
stress level,
written
personalised
assessment.
4 Telephone
follow-up to
assist in skills
development.
Health
Promotion
initiative
described by
author as
having strong
emphasis on
stress
management.
Exact nature
Materials
every 6
weeks
across a
year.
Follow up
at one
year from
onset.
10 week
interventi
on period
then 3
months
Survey
covering self
report
physical and
psychological
health,
perceived
stress, selfreport stressrelated
absenteeism,
psychological
health. Plus
Job Strain
Survey with
subscales
covering
somatisation,
depression
and anxiety.
No differences at
baseline.
Many
measures,
STPI relevant
STPI subscales
measure anxiety,
anger, curiosity
and depression.
Curiosity scale
was higher in
treatment group
(p<0.05)
n/a
Gp 1, sig
reduction in
perceived work
stress (p<0.01)
have been
about
identifying
need for
change and
establishing
nature of
those
changes but not
actually
implementin
g them - so it
is perhaps
not
surprising
that it had no
effect.
No
differences
in pre-test
scores for
drop outs.
Yes
Limited data
on
demographic
variables.
No changes in
perceived health
Gp 1, sig
decrease in
somatisation
(p<0.05). No
change in other
dimensions
n/a
Comprehens
ive Health
Promotion
initiative of
which mental
well-being
only a small
aspect of
outcome
measuremen
Yes
202
male,
asian/pacific
islander
extraction,
middle-aged,
46% obese
What is the
effect of a
workplace
stress
management
programme on
illness and
health services
utilisation?
171 full
intervention vs
166 partial
intervention vs
164 waiting list
control.
Overall attrition
32% (343
remaining), no
group numbers
given.
USA,
computer
industry and
local
government
employees,
typically 40s,
50:50 gender.
RCT
Rahe et al.
2002
Stress
management
1/+
unclear,
seems to be:
1 hour HRA
feedback
session,
Followed by 8
weeks of 1X45
minute large
group training
session, and
1X60 minutes
small group
training
session
Stress
management
program that
used
personalized
feedback and
small group
education.
Full
intervention
included faceto-face
feedback and
6
supplementary
small group
sessions. In
computer
industry these
were 60 mins
in lunch period
(av attendance
3 sessions), in
local gov 90
mins during
work (av
attendance 4
sessions).
Both were
every other
week for 3
months. Partial
intervention
had mail
t (although
stated as a
key part of
study). Few
effects other
than 'I am
curious'??
Very limited
applicability
16
months
Stress Coping
Inventory
(SCI), StateTrait Anxiety
Inventory
(STAI Trait
Form Y-2),
Quarterly
Health
Questionnaire
(QHRQ),
Physician
visits from
medical
records
n/a
All computer
industry
groups
showed
decreases in
stress and
anxiety over
time. One
measure (neg.
responses to
stress)
showed a
group x time
interaction
with full
intervention
more change
than partial
which was
more change
than control
(p=0.012).
Limited
support for
expectation
that
intervention
would have
beneficial
effects and
that full
intervention
would be
more
beneficial.
Main
impression is
that
extraneous
factors
affecting all
groups were
dominant.
Yes
City gov.
employees
also showed
sig decrease
across all
measures with
no sig
interactions.
Lower selfreported
illness in
203
feedback and
no additional
contact.
37 in Area A
(counselling)
76 in Area B
(organisational
) 43 in Area C
(control)
numbers
actually taking
part not stated
No
demographic
details given.
Workers in
council
housing
department
No reference
to method
used to select
departments
Quasiexperiment
al (procohort)
Reynolds 1997
Both
2/-
Area A
received
individual
counselling,
Area B had
organisation
change, Area
C was control
expected
directions with
that for city
employees
approaching
sig (p=0.068)
12
months
after
interventi
on period
GHQ-12 and
SCL-90R
(somatisation)
At 1 year sig
AreaXTime
interaction
(p<0.049). SCL90R decreased
in Area A and
increased in
B&C. According
to text same
effect was seen
with GHQ-12 but
figure doesn't
support this shows reduction
for B as well.
Fewer
physician
visits in year,
in line with
expected
changes. Sig
for city
employees
(p=0.04)
n/a
According to
paper,
counselling
but not
organisation
al
intervention
worked in
reducing
physical and
psychologica
l symptoms.
Conflict in
data
reported
makes this
difficult to
interpret.
Also lack of
any
information
on
participation
rates and
demographic
comparisons
diminishes
value of
study. Fuller
paper cited
not located.
Yes
204
What is the
effect of an
intervention
designed to
clarify individual
roles on
subjective
strain, physical
symptoms or
time lost through
illness?
22 in both
groups.
Treatment
group received
supervisor role
clarification
(organisational
intervention)
T2 -10
months
after pretest
survey
(T1) T3 6
months
later.
Mental Health
Battery
(psychological
ill-health);
Somatic
Complaints
Index
No sig effect on
psych or phys illhealth although it
did reduce role
ambiguity and
supervisor
dissatisfaction
n/a
Transcendenta
l Meditation
(TM) or
conventional
Stress
Management
(including
physical
relaxation).
Same hands-
Post
formal
interventi
on (3
months)
and 3
year
follow-up
STAI and
IPAT
Depression
scale
After 3 months
Trait Anxiety
(p=0.05) and
Depression
(p=0.025) were
sig lower in TM
group. Text says
State Anxiety as
well (p=0.03)
although this is
After 3 years
State Anxiety
(p=0.025)
Trait Anxiety
(p=0.05) and
Depression
(p=0.01) were
all sig lower in
TM group
57% male,
median age
41yrs
Quasiexperiment
al (RCT)
Schaubroeck
et al. 1993
Organisational
1/+
RCT
Sheppard et
al.1997
Stress
management
1/+
17 and 15 at
completion
Mean age
50.5yrs; mixed
ethnic, 85%
female
Lack of
clarity in
subject
groupings
makes
interpretation
unclear. The
authors
suggest that
there might
have been
some spread
of role
clarification
as a result of
the initial
meetings.
Also,
although
supervisors
agreed not to
apply
clarification
to those
employees in
control group
this would be
difficult to
control for
and might
have diluted
any benefit.
Reduction of
role
ambiguity
was
nevertheless
seen as a
good step.
Appears to
be a well
conducted
study.
Seems to
have
balanced
face-to-face
time with two
treatments
Yes
yes
205
on anxiety and
depression?
(although all
dropouts were
female)
on time for
both and
monitoring of
uptake for 3
months
12 in each
group
CBT and
muscle
relaxation.
Included
coping with
unruly pupils. 5
x 2-hour
sessions after
work, 2-4
weeks
between
sessions
Post test
one week
after final
session
Brief Job
Stress
Questionnaire
(vigour, anger,
fatigue,
anxiety,
depression,
somatic
symptoms)
No sig effect on
any stress
response scales
(p>0.05)
n/a
109
intervention;
113 controls
sent first
questionnaire.
web-based
training on
stress
awareness
and coping. 1
month learning
period
1 week
after end
of
learning
period,
follow-up
after a
further 5
weeks.
Brief Job
Stress
Questionnaire
(vigour, anger,
fatigue,
anxiety,
depression,
somatic
symptoms)
No sig main
effect on any
stress subscales.
Males showed
trend to sig effect
(p=0.069) on
physical stress,
younger showed
trend to sig effect
(p=0.093) on
n/a
8 & 8 on
completion
Mainly female,
Japanese
teachers,
mostly married
av age 44yrs
both gps, av
21yrs in job
both gps
Non-RCT
Shimazu et al.
2003
Stress
management
2/-
Non-RCT
Shimazu et al.
2005
Stress
management
2/++
94 & 104
complete data
at follow-up
and also
monitored
adherence,
at least for
first 3
months. No
formal 'at
home' task
for relaxation
group unlike
TM. High
level of
adherence to
both
suggests
strong
motivation.
Study
severely
limited by
small scale.
Additionally,
although
initial
recruitment
was intended
to be
voluntary,
64% had
been
instructed to
attend by
their
manager and
12% for
'other
reasons' with
only 24%
volunteers.
Interesting
subset
analyses
despite lack
of main
effects.
Authors
report signs
of spill-over
to controls
Yes
Yes
206
Non-RCT
Shimazu et al.
2006
Stress
management
2/++
job satisfaction?
Mainly male,
Japanese
office workers,
av age 42-44.
Controls
slightly older
and more of
them male (not
sig).
149
(treatment),
151 (controls)
144 & 143 at
completion
Japan,
Predominantly
males,
managers Av
ages 36
(treatment), 37
(controls)
psych distress,
those expressing
an interest in
stress showed
sig +ve effect
(p=0.029) on
psych distress.
CBT, stress
awareness
and coping
8 weeks
after
course,
BJSQ
'psychological
distress'
n/a
working in
same
department.
Short term
before
follow-up
intervention
seen as too
short.
Implications
of use of
employee
number for
randomisatio
n not
apparent.
Not stated
whether
course was
completed in
own or work
time.
Authors
suggest
adverse
effect could
either be due
to increased
awareness
of stress
issues or to
stress
created by
additional
workload
due to need
to implement
training.
Suggest
longer period
of evaluation
would show
benefit.
Stress score
actually rose
in treatment
group but fell
in control
group
Yes
207
What is the
effect of
communication
skills training on
burnout
amongst
Japanese
nurses?
Shimizu et al .
2003
Stress
management/
Organisational
19 treatment,
26 controls
J-MBI,
(Maslach
Burnout
Inventory)
significantly
greater increase
in personal
accomplishment
in treatment
group (2.6 vs 3.0 p); no
difference in
emotional
exhaustion or
depersonalisatio
n
n/a
High dropout
rate and
nonrandomisatio
n of
participants
make results
difficult to
interpret.
Some
evidence of
change in
one aspect
of burnout
only
Yes
15 minute at
chair
massage, 1
per week for 6
weeks.
12 weeks
from start,
3-4
weeks
from end
of
massage
period
STAI, state
and trait
n/a
Subjects had
higher than
Norm values
for STAI
(both)
suggesting
highly
anxious
group explained by
authors in
terms of
ongoing
downsizing.
Use of break
balanced out
effect of
massage
break to
some extent
although a
break for
'non-work'
might not be
as distracting
as massage
which might
enhance any
difference.
Not known if
massage
allowed
Yes
12 & 14 at
completion
Allocated to
group by
supervisor
Non-RCT
2/+
18
intervention,
16 control
18 & 15 at
end.
Office workers
in US
company,
more females,
av age 40yrs.
RCT
Stress
management
5 months
Japanese
female nurses
with low-tomoderate
communication
skills
What is the
effect of
massage
therapy on
anxiety?
Shulman &
Jones 1996
Communicatio
n training 2 x
2-day sessions
1/++
Controls had
15 minute
break.
208
What is the
effect of job
stress training
for supervisors
on psychological
distress and job
performance of
immediate
subordinates?
RCT
Takao et al.
2006
Organisational
1/++
154
intervention,
101 control
134 & 92 on
completion
office and
manual
workers in
Japanese sake
brewery;
Intervention
group sig more
blue collar and
fewer years of
education.
Genders
reasonably
balanced
Supervisors
received
training which
they were then
expected to
apply to
immediate
subordinates.
60 minute
lecture + 120
min active
listening
training
3 months
post
interventi
on
Brief Job
Stress
Questionnaire
(vigour, anger,
fatigue,
anxiety,
depression,
somatic
symptoms)
No sig main
effects following
intervention.
Subgroup
analysis showed
younger male
white collar
workers to have
sig (p=0.012)
+ve effect of
intervention on
psychological
distress summing
subscales of
JSQ.
n/a
conversation
for example.
Post-test
questionnair
es believed
to be
completed
shortly after
massage but
limited
persistence
interesting.
Interesting
study in that
intervention
was indirect training
supervisors.
Well
designed in
allowing for
possible
covariates as
subject
groups were
not balanced
in design
(depended
upon who
their
supervisor
was who had
been
randomly
allocated to
training or
not). Ratings
of 'supervisor
support' from
Job Content
Questionnair
e did not
show any
differences
although not
clear when
this was
measured
(no
Yes
209
38 in treatment
group, 41 in
control
223 treatment,
260 controls
workers at
Swedish
insurance
company
100% retention
1 hour lecture
and 10 mins
relaxation
training versus
same lecture
Immediat
ely post
training
Salivary IgA,
Iceberg profile
of mood
states
Significant
increase in S-IgA
in intervention
group (p=0.03)
however, after
adjustment for
age effect was
marginal
(p=0.09)
No significant
difference in
Iceberg scores
(mood)
n/a
psychosocial
training for
managers.
1 year
Cortisol,
serum lipids,
serum GGT
(liver enzyme)
Swedish
version of
demandcontrol
No sig effect on
psychological
demands on
employees or
managers.
n/a
Japan,
Female health
care workers
Non-RCT
Taniguchi et
al. 2007
Stress
management
2/+
Non-RCT
Theorell 2001
Organisational
2/++
(bloods)
2/+
(questionna
ires
1 day at onset,
2 hours every
2 weeks for 6
months and 1
Sig reduction of
serum cortisol for
intervention
change
reported).
Emergence
of subgroup
effect has
some
support from
other
aspects of
study although
could
possibly
mount
counterargument
Very short
training
period and
no
assessment
of
persistence.
Subjects not
randomly
allocated but
were
assigned to
groups by
manager to
one of 2
sessions a
month apart.
No
assessment
of possible
++confounde
rs such as
exercise
levels.
Three other
management
programmes
were going
on at same
time as
intervention (
2 in control
Yes
yes
210
stress markers
of subordinate
employees?
baseline blood
sampling
day at end.
questionnaire
Sig reduction of
serum GGT in all
(p=0.04) but not
employees or
managers
separately
~60% female,
ages not given
What is the
effect of a
change in
shiftwork rota
system on wellbeing and other
measures?
Not clear, as
initial
breakdown of
treatment and
control groups
not given. 150
questionnaires
sent out
initially.
31 treatment,
40 control at
completion.
UK police
officers
Treatment gp
sig older (34.2
v 29.5)no other
demo diffs
reported
Quasiexperiment
al (NonRCT)
Totterdell &
Smith 1992
2/+
Change in shift
system at 2
stations, cf 2
others in same
force
6 months
after
interventi
on
introduce
d
GHQ-12
Change to
Ottawa shift
system was
associated with a
sig decrease in
GHQ-12 score
(p=0.001)
n/a
groups)
which may
have
confounded
outcomes
though not
specifically
psychosocial in
nature. Also
fairly high
dropout rate
but unlikely
to explain all
of the
differences
31/32 of
treatment
group had
used new
system for at
least 3 of the
last 6
months
Yes
Hospitals
and other
health care
establishmen
ts often work
a variety of
shift systems
to provide 24
hour cover
and cover for
peak
periods.
Recent work
on the
adverse
health
effects of
shift work
(cancer)
means this is
potentially a
very
important
finding of
211
particular
relevance
Initial sample
396, 59
cognitive, 71
physical into
trial (266 drop
outs)
36 & 39 at
completion
Dutch
telecommunica
tions workers
(apparently
white collar),
mean age
44.2yrs, 90%
men, av work
exp. 21.1yrs.
Selected on
basis of
'Distress
Score' in top
decile for 700
staff. No
breakdown by
groups
reported.
RCT
Van Rhenen,
2005
Stress
management
1/+
4 x 1 hour
training
sessions over
a 8 week
period (2wk,
2wk, 4wk
breaks). Either
physical
therapy
(exercises,
relaxation) or
cognitive
therapy.
6 months
after end
of
program
me
Psychological
complaints
(4DSQ);
Burnout
(UBOS Dutch MBIGS); Fatigue
Checklist
Individual
Strength (CIS)
At short-term
immediately post
training, both
groups showed a
sig decline in
psychological
complaints but
no diff between
groups. Both
groups show a
decrease in
burnout
subscales but no
diff between
interventions
(although trend
to one for
exhaustion
subscale with
phy clearer than
cog). Both
groups showed a
decrease in
fatigue scores
but again no
intervention
effect. After 6
months the
reduction in
psychological
complaints
remained stable.
Burnout was
reduced in phys
group and
exhaustion
raised in cog
group but
interaction term
still not sig.
Effect of either
n/a
No
differential
effect
between the
two
interventions
was
demonstrate
d although
both had a
positive
benefit which
lasted
through to
follow-up.
Welldesigned
study but
lack of
absolute
control group
makes
extrapolation
difficult - was
the benefit
just from the
breaks from
work or a
genuine
effect of the
interactions?
Complex
analysis
increases
risk of
chance
significant
effects but
persistence
and similarity
of effects
Yes
212
intervention on
fatigue remained,
again with no sig
interaction effect.
What is the
effect of
management
programmes on
physicians work
environment
and health?
52 in
intervention
group, 52 in
control group
42 & 42 at
follow up
25 in mentor
programme;
12 in network
programme; 5
in lecture
programme
Non-RCT
Von Vultee et
alI. 2005
Organisational
2/+
Sweden,
Female
physicians
from 6 different
hospitals.
Allocated to
management
programmes
by managers
or HR.
Av age 47.5
yrs
Three
management
programmes:
Mentor
programmes,
network
programmes
and Lecture
group versus
no
intervention.
No details of
programmes
Question
naire
follow-up
at one
year
Quality, work,
competence
tool (QWC)
used to
determine
individual well
being,
Sickness
absence
Improvement in
sickness
absence in
trainees cf
controls(p<0.05;
1.3 v 8.2d)
Apart from that
no significant
differences
between the
groups
between two
groups
makes this
unlikely.
n/a
Lack of detail
on nature of
training
including
duration
makes
findings
difficult to
determine.
Trainees
could have
had time off
from clinical
duties or
could have
been
expected to
fit training
around such
duties.
Yes
Non random
allocation to
training.
No separate
analysis of
forms of
training.
213
What is the
effect of
recreational
music making
on natural killer
cell activity,
cytokines and
mood states on
employees?
20 (treatment)
and 20
(controls).
20 & 19 in
phase 2 (RMM
crossover)
Recreational
Music Making
(RMM) versus
leisurely
reading, 1 X 3
hour
intervention
Crossove
r design
with 6
months
between
phases.
Mood states
questionnaire
(POMS),
measures of
natural killer
(NK) cell
activity
Electric
company,
Japan,
Improvements in
NK cell activity,
and mood states.
n/a
In one phase
(phase 2)
anger/hostility
was sig lower in
RMM than
control, post
intervention (p
not stated).
Wachi 2007
Stress
management
Study funded
by Yamaha
carried out
with Yamaha
employees
possible
conflict.
1/++
Non-RCT
Stress
management
Small
sample size
and results
not
consistent
between
phases.
Plots of pre-post
NK cell activity
showed sig diff in
slopes for RMM
& control (Phase
1, p=0.05, Phase
2, p=0.019).
Walach et al.
2007
Yes
Data
analysed
after second
phase with
assumption
of no carryover
between
phases.
Consistent
decrease in total
mood (TMD) for
both RMM &
control. Sig diffs
in change
between groups
for P2 (p=0.019)
& P1+P2
(p=0.012).
RCT
(crossover)
No analysis
of immediate
case-control
comparisons
2/+
12 treatment,
17 controls
telephone call
centre workers
in Germany.
Volunteers
from 185.
12 on first
course
designated
treatment, 17
on second,
control.
Nature of
control
Mindfulnessbased Stress
Reduction.
8, weekly
evening
classes of 2.5
hrs per class,
1 X 6 hour
mindfulness
day, practice
at least 5 days
per week at
least 20 mins
per day.
8 weeks
(post
course)
and 4
months
(follow
up)
General
complaints,
tension and
tiredness
subscales of
Freiburg
Complaint List
There were no
sig changes on
complaints
n/a
Pilot study
on small
groups of
volunteers.
Treatment
group older
than
controls, and
the 2 groups
also reported
different
motivations.
Overall some
evidence of
a positive
effect though
214
unclear.
high time
commitment
out of work
hours for
participants.
11 & 16 on
completion
Willingness to
practice
condition of
entry.
No effect on
well-being
outcomes.
6 & 10 female
treatment sig
older than
control (41.3 &
33.7 yrs)
36% & 44%
some chronic
disease (not
specified)
What is the
effect on blood
pressure and
personal strain
of a progressive
relaxation
intervention for
AfricanAmerican
women?
48 enrolled, 43
completed. No
group numbers
given
Demographics
only reported
for whole
sample.
Av age,
33.5yrs
AfricanAmerican
women
selected on
basis of high
degree of
hypertension.
Non-RCT
Webb et al.
2000
Stress
management
2/-
Baseline data
not analysed
separately
Progressive
muscle
relaxation cf
being
'instructed to
take time out
(30mins) each
day. No record
of compliance.
8 weeks
from entry
Personal
Stress
Questionnaire
(PSQ)
Experimental
group (size not
known) showed
sig reduction in
'interpersonal
strain' (p=0.02);
and 'physical
strain' (p=0.01).
Both groups
showed sig
reduction in
psychological
strain
(p=0.0001).
No sig change in
'vocational strain'
although data
suggests trend
for control group
to reduce levels.
n/a
Lack of
clarity on
how second
group
(control)
were treated
(possibly
wait listed
but not
stated).
Assignment
to 2 courses
not
explained.
Concerns
over
recruitment
(allocated to
treatment or
control on
basis of
course
attended).
Yes
Poor
reporting of
group
numbers and
dynamics,.
Poor
adherence to
treatment
intervention
Study
appears to
show
positive
215
value from
relaxation.
RCT
Wilson et al.
2001
Stress
management
1/+
33 to EMDR,
29 to standard
training
EMDR versus
stress
management
USA,
Police officers
3 X 2 hour
EMDR
sessions 1:1
basis, 6
different
therapists
Volunteers
from 531 in
Department
Av age, 36.8
yrs
79% male,
75%
Caucasian.
No
demographics
of groups)
6 X 1 hour
(video +
workbook)
sessions for
SM
Either could be
undertaken
during work
hours
Post test,
six month
follow-up
and exit
interview
SUDS, State
Trait Anger
Inventory, Job
Stress survey
scale, Police
Stress
inventory,
Symptom
Check List
(SCL), marital
adjustment
scale, PSDS,
Coping
Response
Inventory
SUDS found
level of stress
lower in EMDR
(p<0.05),
Significant
reduction in trait
anger in EMDR
(rose in SM)
(p<0.05)
State anger fell
in EMDR & rose
in SM (p<0.05)
Job Stress
Survey scores
lower for EMDR
than SM (p<0.05)
n/a
Subjects
were all
recruited as
having a
family history
of
hypertension
and selfreported
stress.
Positive
findings in
favour of
EMDR.
Although
total contact
was similar,
face to face
contact of
EMDR
possible
issue,
regardless of
nature of
contact
Yes
No effects on
PSI or SCL-90.
216
Recruitment unclear
No control group
No control group
No control group
No control group
Evaluation of a learning
model, no measures of
mental wellbeing
217
No control group
No control group
Excluded as description of
intervention and
implementation, no follow
up
Excluded as not an
intervention
218
No control group
No control group
No control group
No control group
219
Risk management
approach in hospital setting
- effect on stress
No control group
No control group
Non-occupational group
220
Cross-sectional study
No control group
No control group
No control group
No control group
Not an intervention
221
No control group
No control group
222
No control group
No control group
No control group
223
Military Population
No control group
224
Validated outcome
measures not used
Validated outcome
measures not used
No control group
No control group
Kitchener BA, Jorm AF, Kitchener BA, Jorm AF. (2004). Mental
health first aid training in a workplace setting: a randomized
controlled trial [ISRCTN13249129]. BMC Psychiatry; 4: 23.
225
Validated outcome
measures not used
No control group
Validated outcome
measures not used
Lo Sasso AT, Lindrooth RC, Lurie IZ, Lyons JS, Lo Sasso AT,
Lindrooth RC, Lurie IZ, Lyons JS. (2006). Expanded mental
health benefits and outpatient depression treatment intensity.
Medical Care; 44: 366-372.
No control group
226
Non-occupational sample
No control group
Description of intervention
and baseline measures but
no follow-up
227
Medical Intervention
Not a workplace
intervention
No control group
No control group
228
Non-occupational sample
No control group
Non-occupational sample
Schuster RJ, Weber ML, Schuster RJ, Weber ML. (2003). Noise
in the ambulatory health care setting. How loud is too loud?
Journal of Ambulatory Care Management; 26: 243-249.
Cross-sectional study
229
Not a workplace
intervention
No control group
230
No pre-intervention
measures made
No control group
231
Non-occupational sample
Non-occupational sample
232