Workplace

Original research

Randomised control trial of a proactive intervention

supporting recovery in relation to stress and irregular
work hours: effects on sleep, burn-­out, fatigue and
somatic symptoms
Anna Dahlgren  ‍ ‍,1 Philip Tucker  ‍ ‍,2,3 Majken Epstein,4 Petter Gustavsson,1
Marie Söderström5,6

► Additional supplemental ABSTRACT

material is published online Key message
Objectives  To examine if a proactive recovery
only. To view, please visit the
journal online (http://​dx.​doi.​intervention for newly graduated registered nurses (RNs)
What is already known about this subject?
org/​10.​1136/​oemed-​2021-​ could prevent the development of sleep problems, burn-­
⇒ Entering work life as a newly graduated
107789). out, fatigue or somatic symptoms.
registered nurse is stressful.
Methods  The study was a randomised control trial with
1
Department of Clinical ⇒ Newly graduated registered nurses have a high
parallel design. Newly graduated RNs with less than 12
Neuroscience, Karolinska prevalence of burn-­out.
Institute, Stockholm, Sweden months’ work experience were eligible to participate.
⇒ Recovery has been suggested as a key factor for
2
Department of Psychology, 461 RNs from 8 hospitals in Sweden were invited, of
Swansea University, Swansea,
preventing ill health due to stress.
which 207 signed up. These were randomised to either
UK intervention or control groups. After adjustments, 99
3
Department of Psychology, What are the new findings?
Stockholm University Stress RNs were included in the intervention group (mean ⇒ A proactive recovery intervention was shown to
Research Institute, Stockholm, age 27.5 years, 84.7% women) and 108 in the control be feasible in a working life context, promoting
Sweden group (mean age 27.0 years, 90.7% women). 82 RNs in beneficial strategies for sleep and recovery.
4
Liljeholmens akademiska the intervention group attended a group-­administered
vårdcentral/Academic Primary ⇒ Supporting recovery was associated with
Healthcare Center Liljeholmen,
recovery programme, involving three group sessions with positive results on health and well-­being.
Stockholm, Sweden 2 weeks between each session, focusing on proactive
5
Department of Psychology, strategies for sleep and recovery in relation to work How might this impact on policy or clinical
Karolinska Institutet Department stress and shift work. Effects on sleep, burn-­out, fatigue practice in the foreseeable future?
of Clinical Neuroscience, and somatic symptoms were measured by questionnaires
Stockholm, Sweden ⇒ Further development of methods for supporting
6
Stressmottagningen, at baseline, postintervention and at 6 months follow-­up. employees sleep and recovery is important.
Stockholm, Sweden Results  Preventive effect was seen on somatic ⇒ A proactive approach might be important for
symptoms for the intervention group. Also, the managing employee health.
Correspondence to intervention group showed less burn-­out and fatigue
Dr Anna Dahlgren, Department symptoms at postintervention. However, these latter
of Clinical Neuroscience,
effects did not persist at follow-­up. Participants used sleep deprivation can contribute to the development
Karolinska Institute, Stockholm,
Solna, Sweden; many of the strategies from the programme. of both somatic and psychological symptoms and ill
a​ nna.​dahlgren@​ki.​se Conclusions  A proactive, group-­administered recovery health, for example, burn-­out, depression, cardio-
programme could be helpful in strengthening recovery vascular disease, etc.5 6 While stress is a potential
Received 28 May 2021 and preventing negative health consequences for newly
Accepted 21 December 2021 cause of disturbed sleep,7 sleep deprivation can
Published Online First graduated RNs. itself be a stressor contributing to allostatic load.8
24 January 2022 Trial registration number  NCT04246736. According to the allostatic load theory, repeated or
prolonged stress exposure can have negative effects
on health. Stress reactions can also be sustained
after the actual stressor has subsided, through
INTRODUCTION perseverative cognition in the form of worries or
Work, and especially demanding work situations, rumination.9 Difficulties letting go of stressful
leads to effort expenditure and a need for recovery1 thoughts, together with high work demands and
that is signified by the manifestation of fatigue.2 insufficient sleep, have been shown to predict
Recovery is the process of psychophysiological clinical burn-­out.5 Hence, perseverative cognition
© Author(s) (or their unwinding after effort, in which mental and phys- could be one mechanism which, if sustained, may
employer(s)) 2022. Re-­use iological resources are replenished.3 According to lead to health problems.
permitted under CC BY. the effort-­recovery theory,1 recovery is crucial for Paradoxically, while situations with high work
Published by BMJ. preventing adverse health consequences due to demands featuring high stress levels increase the
To cite: Dahlgren A, stress exposure.4 need for recovery, those are also situations in which
Tucker P, Epstein M, There are multiple paths linking insufficient recovery is likely to be impaired, a phenomenon
et al. Occup Environ Med recovery with ill health. Sleep is essential for phys- referred to as the ‘recovery paradox’.10 Impaired
2022;79:460–468. iological and psychological recovery, and chronic recovery during stressful periods could be due to
460 Dahlgren A, et al. Occup Environ Med 2022;79:460–468. doi:10.1136/oemed-2021-107789
 Workplace
either sleep impairments, failure to detach from thoughts of There were seven primary outcomes, namely: two measures
work during free time, or lack of recovery behaviours such as of sleep problems (insomnia and sleep quality); a global measure
physical or social activities during leisure time. Work-­induced of burn-­out, along with two of its subindices, fatigue and cogni-
fatigue during free time, which is common during stressful work tive weariness; a measure of work-­induced fatigue during free
periods, may further limit the possibilities to engage in bene- time; and a measure of somatic symptoms. It was hypothesised
ficial recovery behaviours, and thus contribute to the recovery that there would be changes in the primary measures reflecting
paradox. improvements in well-­ being. In addition, a set of secondary
Sleep is not only affected by stress but is also regulated by outcomes were examined, focusing on factors that could help
homoeostatic and circadian factors. The homoeostatic regula- account for changes in the primary outcomes, namely: perceived
tion of sleep means that the neurophysiological drive for sleep stress; two remaining subscales of burn-­out (listlessness, tense)
increases with time awake.11 Circadian rhythms make sleep diffi- and dysfunctional attitudes about sleep.
cult during daytime, when melatonin is low and metabolism is
high. For shift workers this often means that they have to initiate METHODS
sleep at times that are biologically suboptimal. Disturbed sleep is Design
common among shift workers, and is one of the possible mech- A parallel randomised control trial was designed to include 100
anisms behind the increased risk of of both somatic and psycho- participants in each group (intervention and wait list control)
logical health problems among shift workers.6 to detect moderate effect sizes (Cohen’s d=0.5) resulting in a
In order to optimise employees’ health and work perfor- power of 0.94. Excel generator for random allocation to groups
mance, organisations should seek to minimise work stressors was used by the research team. Based on a previous feasibility
and promote work hours that enable sufficient sleep and study, adjustments to the process of random group allocation
recovery. On an individual level, organisations can encourage were made if many nurses from the same ward were initially
employees to adopt beneficial strategies for recovery. Sleep and allocated to one group.20 Adjustments were also made for partic-
sleep-­related outcomes can be improved by such interventions, ipants who were randomised to the intervention group but knew
with the most common being educational interventions that that they could not attend the group sessions. They were moved
focus on sleep hygiene and fatigue management.12 Cognitive– to the control group and replaced by a random participant from
behavioural therapy for insomnia (CBT-­I) has been shown to be the control group. Adjustments were made for 24 participants.
effective among adults in the general population.13 However, Masking was not applicable. After the follow-­up measure the
shift workers face more demanding challenges in managing sleep control group received the intervention.
in relation to irregular work hours. Group-­administered CBT-­I
for shift workers, including sleep hygiene, relaxation, cognitive
restructuring, etc, have shown improvements in sleep outcomes, Participants and data collection
although a follow-­up study did not show that CBT-­I was better RNs with less than 12 months’ work experience were eligible
than a sleep hygiene programme.14 15 to participate. Participants were recruited at eight Swedish
Few studies have examined interventions aimed at promoting hospitals within the induction programmes for newly graduated
recovery in forms other than sleep. Supporting recovery RNs at seven of the hospitals. One hospital did not have such a
behaviours in workers with high levels of stress symptoms was programme and so the RNs there were recruited via managers.
found to reduce stress and burn-­out, as well as depressive and The intervention was tested in ten subgroups with 5–13 partici-
anxiety symptoms.16 17 Recovery behaviours were defined as pants in each, between 2017 and 2018. All participants signed an
appetitive behaviours supporting psychophysiological detach- informed consent before entering the study and were thereafter
ment following exposure to stressors or effort expenditure. enrolled in the study by the research team.
Participants were encouraged to try various such behaviours in Digital questionnaires assessing the outcomes were sent to
different contexts for example, listening to music, engaging in participants by email about 1 month before entering the inter-
physical activity, etc. vention (baseline), 1 month after the intervention (postinter-
Entering working life is a period often characterised by high vention) and at 6 months after the intervention (follow-­up).
stress for registered nurses (RNs), described as a reality, or tran- Participants who had attended any of the group sessions received
sition, chock.18 Besides the high workload and the stress of being a short questionnaire, approximately 2 weeks after each session,
new in the professional role, many RNs also start working shifts, evaluating the use of recovery strategies from the programme.23
which is a risk factor for impaired sleep. New RNs may often As from the fourth subgroup, a global evaluation questionnaire
lack effective strategies for managing sleep and fatigue, and the was distributed after the intervention (in total 62 participants).
strategies used may sometimes be counterproductive.19 20 RNs
also have a high prevalence of burn-­out and somatic symptoms Intervention
early in their career.21 22 The intervention was a group-­administered proactive recovery
Given the challenges facing new RNs, actions are needed to programme focusing on enhancing beneficial strategies for sleep
protect the processes of recovery and thereby buffer the impact and recovery as a means of mitigating the impact of work stress
of their stressful work situation. The objective of the current and shift work.23 The programme was developed by MS (certi-
study was to examine whether a proactive intervention, a group-­ fied psychologist, PhD) and AD (PhD) and included three group
administered recovery programme focusing on promoting sessions (2,5 hours), with one session every second week (ie, 4
strategies for sleep and recovery, could mitigate the impact of weeks from the first session to the third), during work hours
work stress and shift work and thus prevent the development of at the hospitals. MS trained AD and ME (Bachelor of applied
sleep problems, burn-­out, fatigue and somatic symptoms among psychology) in delivering the recovery programme. Seven
new RNs. The intervention focused on three main themes: subgroups were led by MS together with AD and/or ME, three
(1) unwinding from stress; (2) promoting sleep according to subgroups were led by AD and ME.
homoeostatic and circadian factors; and (3) handling fatigue by The intervention was based on CBT and motivational inter-
increasing recovery behaviours. viewing techniques.13 16 20 24 The ‘sleep formula’—that is, the
Dahlgren A, et al. Occup Environ Med 2022;79:460–468. doi:10.1136/oemed-2021-107789 461
 Workplace

Table 1  Content of group sessions (I–III)

Session Content Strategies participants were encouraged to try
I.Unwinding from stress ►   The sleep formula ►   Observe behaviours in stressful work situation and
►   Stress factors and stress reactions at work reflect on alternatives
►   CBT-­model: Analysis of behaviour in stressful ►   Practice focusing on the present moment
work situations ►   Unwinding bedtime routine
►   Unwinding routines before bedtime ►   Body scan
►   Mindfulness, focus on the present moment
►   Body scan exercise
II. Promoting sleep according to homoeostatic and ►   Follow-­up from session I ►   Routine for leaving work
circadian factors ►   Routines for leaving work ►   Personal goal for supporting sleep related to the
►   Homoeostatic processes regulating sleep homoeostatic and circadian processes
►   Circadian processes regulating sleep ►   Evaluating work hours using the ArturNurse webtool
►   How work hours interact with sleep regulating
factors
 
III. Handling fatigue by increasing recovery behaviours ►   Follow-­up from session II ►   Practice recovery behaviours at work
►   Cognitive, physical and emotional fatigue ►   Engaging in activities boosting energy during free
►   Balance between activity and rest time
►   Short relaxation exercise ►   Practice short relaxation
►   Recovery behaviour on and off work
►   Activities boosting energy
 

CBT, cognitive–behavioural therapy.

influence of stress, homoeostatic and circadian factors on sleep— 15 or higher indicates clinical insomnia. A sleep quality index
was used as a pedagogical approach to summarise research-­based was calculated based on the mean of four items from the Karo-
knowledge about what regulates sleep. The sessions had three linska Sleep Questionnaire (KSQ)27 (Cronbach’s alpha=0.77)
main focuses: (1) unwinding from stress, including detach- rating the frequency of sleep problems (1 always—6 never).
ment from thoughts of work during free time; (2) supporting
sleep in relation to homoeostatic and circadian processes; and
(3) handling fatigue and increasing recovery behaviours (see
Burn-out, fatigue and cognitive weariness
Burn-­out symptoms during the last month were measured
table 1). Psychoeducative elements were interspersed with
with the Shirom-­ Melamed Burn-­ out Questionnaire (SMBQ)
group discussions and exercises. Participants were encouraged
consisting of 22 items (1 almost never—7 almost always).28 29 A
to reflect on their habitual behaviours connected to sleep and
global mean score was calculated (Cronbach’s a=0.95), and the
recovery and possible alternatives. Between sessions, the partic-
two indices: ‘fatigue’ (Cronbach’s a=0.89) and ‘cognitive weari-
ipants were encouraged to try strategies or behaviour changes
ness’ (Cronbach’s a=0.94).
of their choice, with the aim of enhancing sleep and recovery.
During the second and third sessions, participants reflected on
the experience of trying new strategies. All participants received Work-Induced fatigue
written material covering the content of each session, as well as Work-­induced fatigue during free time was measured with the
online access to an adapted version of a biomathematical model Work Interference with Personal Life index (WIPL) from the
(ArturNurse). ArturNurse evaluated fatigue risk levels based on Work Home Interference scale30 based on the mean of four items
their work schedules25 and provided suggestions of strategies (Cronbach’s alpha=0.90) measuring the extent to which work
from the programme on how to optimise sleep in relation to related fatigue affects free-­time (1 not at all—5 almost all the
different shifts. See online supplemental file 1 for more detail time). Scores of ≥3.5 indicates work-­home interference.31
about the intervention.

Somatic symptoms
Background measures
Somatic symptoms were measured with the Somatic Symptom
In the baseline questionnaire, participants reported gender
Scale-­8 (SSS8), which assesses the experience of eight somatic
(male, female, other), age (years), duration of working as a nurse
symptoms (eg, headache, stomach problems, back pain) during
(months), type of shift schedules, if they took any medication
the last 7 days (0 not at all—4 much). A sum score was calcu-
(yes/no), and frequency of the use of sleep medication, central
lated (Cronbach’s alpha=0.75). Scores 8–11 indicate a medium
stimulants, sedatives, opioid analgesics or other pain killers (1
somatic symptom burden, 12–15 indicate high and 16–32 indi-
never, 5 every day).
cate very high.32
PRIMARY OUTCOMES
Sleep SECONDARY OUTCOMES
Insomnia symptoms during the last month were measured with Perceived stress
the Insomnia Severity Index (ISI; 0 no problems—4 severe prob- Perceived stress during the last month was measured with the
lems).26 A sum score was calculated (Cronbach’s alpha=0.84), Perceived Stress Scale (PSS) consisting of 10 items (0 never—4

462 Dahlgren A, et al. Occup Environ Med 2022;79:460–468. doi:10.1136/oemed-2021-107789

 Workplace
very often). A global mean score was calculated (Cronbach’s RESULTS
alpha=0.88).33 Figure 1 shows the participant flow chart, showing how the
final sample was arrived at. Of 461 invited new RNs, 207 (45%)
signed up for the study.
Tension and listlessness
The indices ‘listlessness’ (Cronbach’s a=0.82) and ‘tense’ (Cron-
Baseline data
bach’s a=0.73) from the SMBQ were calculated.28
The intervention and control group consisted of 84.7% and
90.7% women, respectively. The average age in the intervention
Dysfunctional beliefs and attitudes about sleep group was 27.5±5.3 and 27.0±5.1 in the control group. The
Dysfunctional beliefs and attitudes about sleep were measured average time of employment was 2.8±2.1 months in the inter-
vention group and 3.3±2.7 months in the control group. Most
through the Dysfunctional Beliefs and Attitudes about Sleep
participants (73%) had a rotating morning and evening shift
scale (DBAS-­10).34 In the original version, the degree of agree-
schedule, and almost one fifth (19%) had a rotating morning,
ment with 10 statements is measured on a Visual Analogue Scale
evening and night shift schedule. No significant differences were
between 0 and 100. However, due to technical problems, data
observed between the two groups at baseline for any of the back-
from the first three subgroups (in total 46 participants) were
ground variables or any of the outcome measures at baseline (see
excluded from the analyses, while as from the fourth sub-­group
online supplemental file 2).
a ten point scale (0 do not agree—10 do fully agree; Cronbach’s
alpha=0.80) was used.
Sleep, burn-out, fatigue and somatic symptoms
Results relating to the primary outcomes are shown in table 2
STATISTICAL METHODS and figure 2. Insomnia symptoms (ISI) and sleep quality (KSQ)
Longitudinal analysis of mean response profiles, 35 with showed no significant group by time interaction.
time coded as a categorical variable (in order to account Symptoms of burn-­out (SMBQ) showed significant group by
for possible non-­linear relationships), was performed using time interactions for both the global score and for the indices
the mixed model procedure in IBM SPSS Statistics V.26. ‘fatigue’ and ‘cognitive weariness’. Post hoc analysis showed
Maximum likelihood was used to estimate the parameters the same general pattern for all three outcomes; the interven-
(using all available data) under the assumption that incom- tion group reported less symptoms postintervention (small to
plete data were missing at random. A significant group-­by-­ moderate effect sizes), but did not differ from the control group
time interaction was interpreted as reflecting differential at follow-­up.
patterns of change between the groups over time. Calcula- Ratings of work-­ induced fatigue (WIPL) during free time
tions of effect sizes based on group mean differences postin- showed a significant group by time interaction, where the inter-
tervention and at follow-­u p were calculated on model-­based vention group reported less fatigue postintervention (small effect
estimated means and SD where a Cohen’s d around 0.5 was size), but not at follow-­up.
considered as moderate and around 0.2 as small. Ratings of somatic symptoms were relatively stable over time
in the intervention group, whereas the control group reported
increased somatic symptoms (SSS8). This was reflected in the
significant group by time interaction and in significant differ-
ences in the post hoc tests postintervention and at follow-­up,
with higher somatic symptoms observed in the control group
(small to moderate effect sizes).

Perceived stress, tension, listlessness, DBAS

Results relating to the secondary outcomes are shown in table 3
and figure 2. No significant group by time interactions were
found for the ratings of perceived stress (PSS), for either of the
SMBQ indices ‘tense’ or ‘listlessness’, or for beliefs and attitudes
about sleep (DBAS).

COMPLIANCE AND PROGRAMME EVALUATION

Unwinding bedtime routines were used by 95% of those
who attended any of the group sessions (N=82), routines for
leaving work by 87%, relaxation exercise by 86%, activities
promoting recuperation by 75% and body scan meditation by
74% (response rates 91%–94%). Recovery behaviours during
work and free time were used by 82% and 80%, respectively,
and the short relaxation exercise by 70%, whereas the webtool
ArturNurse was used by 21% (response rates 60%–78%). Strat-
egies related to homoeostatic or circadian processes were used
by 78% (response rate 45%). All respondents (100%) reported
that they would recommend the programme to others, and 98%
rated the programme as good or very good. The majority, 90%,
reported that they would use the strategies in the future, and 8%
Figure 1  Participant flow chart. that they might do so (79% response rate) .
Dahlgren A, et al. Occup Environ Med 2022;79:460–468. doi:10.1136/oemed-2021-107789 463
464
Workplace

Table 2  Estimated group means and tests of effects, taken from a multilevel analysis of the primary outcome measures using an intention to treat approach
Cohen’s d between
Estimated group means Tests of effects groups
Intervention Control Group Time Group * Time
Follow-­
Mean SE 95% CI Mean SE 95% CI F P value df F P value df F P value Post up

ISI 2.88 0.09 185.39 6.73 0.00 148.41 2.57 0.08 0.30 0.39
Baseline 10.65 0.51 9.64 to 11.66 10.74 0.49 9.78 to 11.71
Post 8.83 0.54 7.76 to 9.90 10.15 0.52 9.12 to 11.19
Follow-­up 8.83 0.59 7.66 to 10.00 10.58 0.54 9.51 to 11.65
KSQ 1.02 0.31 196.14 11.25 0.00 153.04 0.75 0.47 −0.22 −0.17
Baseline 4.07 0.10 3.87 to 4.27 4.03 0.10 3.84 to 4.22
Post 4.42 0.10 4.22 to 4.62 4.23 0.10 4.04 to 4.43
Follow-­up 4.34 0.11 4.11 to 4.56 4.20 0.10 4.00 to 4.40
SMBQ Global 1.45 0.23 190.01 0.17 0.84 149.74 4.03 0.02 0.36 0.19
Baseline 3.74 0.12 3.51 to 3.97 3.71 0.11 3.49 to 3.93
Post 3.54 0.12 3.30 to 3.77 3.89 0.11 3.67 to 4.12
Follow-­up 3.66 0.12 3.43 to 3.90 3.84 0.11 3.63 to 4.06
SMBQ Fatigue 3.90 0.05 187.18 1.05 0.35 147.17 3.87 0.02 0.45 0.30
Baseline 3.73 0.13 3.48 to 3.98 3.79 0.12 3.55 to 4.02
Post 3.62 0.14 3.35 to 3.88 4.14 0.13 3.89 to 4.40
Follow-­up 3.64 0.14 3.37 to 3.91 3.96 0.13 3.71 to 4.21
SMBQ Cognitive 0.10 0.76 189.40 0.43 0.65 147.13 3.53 0.03 0.24 0.03
Baseline 3.47 0.15 3.17 to 3.77 3.33 0.14 3.04 to 3.61
Post 3.16 0.16 2.85 to 3.47 3.47 0.15 3.18 to 3.77
Follow-­up 3.37 0.16 3.06 to 3.69 3.37 0.15 3.08 to 3.66
WIPL Fatigue 0.31 0.58 184.30 0.44 0.65 149.17 5.37 0.01 0.33 0.02
Baseline 3.43 0.09 3.25 to 3.62 3.32 0.09 3.14 to 3.49
Post 3.20 0.10 2.99 to 3.40 3.48 0.10 3.29 to 3.68
Follow-­up 3.30 0.11 3.08 to 3.51 3.32 0.10 3.13 to 3.52
SSS8 4.48 0.04 176.33 16.30 0.00 132.72 3.81 0.03 0.37 0.49
Baseline 10.82 0.61 9.61 to 12.03 11.25 0.58 10.11 to 12.39
Post 11.73 0.69 10.36 to 13.09 13.67 0.65 12.38 to 14.95
Follow-­up 11.96 0.73 10.51 to 13.41 14.69 0.68 13.36 to 16.02
Factor labels in tests of effects: Group=Intervention vs Control, Time=Baseline vs Post vs Follow-­up, Group*time=interaction term. Df (df) for Time and Group*time are identical.
ISI, Insomnia Severity Index, 0–28 severe problems; KSQ, Karolinska Sleep Questionnaire, 1 always— 6 never; Mean, modelled mean values; SMBQ, Shirom-­Melamed Burn-­out Questionnaire, 1 almost never—7 almost always; SSS8, Somatic Symptom Scale-­8, 0–32 very high somatic symptom burden;
WIPL, Work Interference with Personal Life, 1 not at all—5 almost all the time.

Dahlgren A, et al. Occup Environ Med 2022;79:460–468. doi:10.1136/oemed-2021-107789

 Workplace

Figure 2  Mean values and SEs in intervention and control group at baseline, post and follow-­up measures. DBAS, Dysfunctional Beliefs and Attitudes
about Sleep, 0 do not agree–10 do fully agree; ISI, Insomnia Severity Index, 0–28 severe problems; KSQ, Karolinska Sleep Questionnaire, 1 always–6 never;
PSS, Perceived Stress Scale, 0 never–40 very often; SSS8, Somatic Symptom Scale-­8, 0–32 very high somatic symptom burden; SMBQ, Shirom-­Melamed
Burn-­out Questionnaire, 1 almost never–7 almost always; WIPL, Work Interference with Personal Life, 1 not at all–5 almost all the time.

DISCUSSION showed stable ratings for these symptoms, while the control
This study examined whether a proactive intervention for group showed increased somatic symptoms over time. Further,
newly graduated RNs, supporting strategies for the enhance- promising effects were seen on burn-­out measures and on work-­
ment of sleep and recovery in relation to work stress and shift induced fatigue during free time at postintervention. However,
work, could prevent negative development of sleep problems, these latter effects did not persist at follow-­up 6 months later.
burn-­out, fatigue and somatic symptoms. The results indicated a The intervention group showed lower global burn-­out scores
preventive effect on somatic symptoms, as the intervention group compared with the control group, as well as lower scores on
Dahlgren A, et al. Occup Environ Med 2022;79:460–468. doi:10.1136/oemed-2021-107789 465
466
Workplace

Table 3  Estimated group means and tests of effects, taken from a multilevel analysis of the secondary outcome measures using an intention to treat approach
Estimated group means Tests of effects Cohen’s d between groups
Intervention Control Group Time Group * Time
Mean SE 95% CI Mean SE 95% CI F P value df F P value df F P value Post Follow-­up
PSS 0.08 0.78 175.63 0.52 0.60 144.83 2.24 0.11 0.22 0.01
Baseline 18.34 0.75 16.87 to 19.81 17.77 0.66 16.48 to 19.07
Post 16.95 0.70 15.56 to 18.34 18.23 0.66 16.93 to 19.53
Follow-­up 17.64 0.78 16.10 to 19.18 17.63 0.70 16.25 to 19.01
SMBQ Tense 0.05 0.82 191.85 2.78 0.07 148.94 0.89 0.41 0.03 0.14
Baseline 3.83 0.13 3.58 to 4.08 3.75 0.12 3.51 to 3.99
Post 3.66 0.14 3.38 to 3.94 3.69 0.14 3.42 to 3.96
Follow-­up 3.82 0.15 3.54 to 4.11 3.98 0.13 3.72 to 4.24
SMBQ Listlessness 1.99 0.16 190.88 0.59 0.56 150.50 2.89 0.06 0.39 0.15
Baseline 4.06 0.13 3.81 to 4.31 4.11 0.12 3.88 to 4.35
Post 3.81 0.12 3.57 to 4.05 4.21 0.12 3.98 to 4.45
Follow-­up 3.99 0.14 3.72 to 4.27 4.16 0.13 3.91 to 4.41
DBAS 1.32 0.25 153.20 1.15 0.32 129.95 2.94 0.06 0.26 0.32
Baseline 5.34 0.20 4.94 to 5.74 5.27 0.19 4.89 to 5.65
Post 4.91 0.23 4.46 to 5.36 5.36 0.21 4.94 to 5.79
Follow-­up 5.00 0.23 4.54 to 5.45 5.51 0.21 5.10 to 5.93
Factor labels in tests of effects: Group=Intervention vs Control, Time=Baseline vs Post vs Follow-­up. Df (df) for Time and Group*time are identical.
DBAS, Dysfunctional Beliefs and Attitudes about Sleep, 0 do not agree—10 do fully agree; Mean, modelled mean values; PSS, Perceived Stress Scale, 0 never—40 very often; SMBQ, Shirom-­Melamed Burn-­out Questionnaire.

Dahlgren A, et al. Occup Environ Med 2022;79:460–468. doi:10.1136/oemed-2021-107789

 Workplace
the indices ‘fatigue’ and ‘cognitive weariness’ postinterven- boosted by the participants receiving written materials after each
tion. However, the effects on these burn-­out measures did not session.
persist at follow-­up 6 months later. It remains to be determined The programme included a wide range of strategies aimed
whether a booster session or if changes to the programme could at enhancing both sleep and other forms of recovery, possibly
contribute to a longer-­lasting effect. making it easier for participants to find strategies to apply. On
Another promising finding was that the intervention group the other hand, the intervention’s broad approach limits the
reported less work-­ induced fatigue during free time, postin- possibility to explain specific mechanisms behind the results.
tervention. Previous research indicates that nurses’ work-­home Some limitations are worth noting. The sampling may have
balance suffers at the start of their career.36 It is possible that the been biased by self-­selection into the study, towards nurses with
intervention helped the nurses to detach from work stress during a high motivation to participate. Mandatory participation might
free time, thereby enabling them to achieve a better quality of have produced different results. Moreover, we cannot draw any
recovery. An improved work-­ home balance suggests that the conclusions as to whether the recovery programme would be
recovery programme may strengthen important preconditions feasible or effective in other occupational groups, or for partic-
for a sustainable working life and counteract the so-­ called ipants with more extreme workloads, or with clinically signif-
’recovery paradox’.10 icant sleep or burn-­out problems. Notably, only 37% attended
There was no significant interaction effect in the analysis all three group sessions, highlighting the need to develop
of sleep quality or insomnia symptoms. However, the results approaches to increase attendance at group sessions. Another
showed trends towards interaction, suggesting less insomnia limitation was the variation in response rates regarding the use
symptoms and decreased DBAS in the intervention group, thus of strategies within the programme. A deeper understanding of
following the same pattern as for the other outcome variables. how the different strategies have been used will be examined in
A possible explanation for the lack of significant effects on sleep analysis of follow-­up interviews with participants and reported
quality might be that work requirements (eg, timing of shifts) in future publications.
constrain the extent to which sleep can be altered. More fine-­ The results merit further evaluation of the recovery
grained analysis, such as day-­to-­day comparisons, might capture programme as a part of induction programmes for new RNs.
a more nuanced picture of sleep quality or other sleep parame- Future studies should examine the feasibility of implementing
ters. It is also important to note that the recovery programme the recovery programme in nursing education, or whether it
was not consistent with a regular CBT-­protocol for insomnia. could be adapted for nurses who are further into their career
Rather, a preventive approach was taken with participants being or for other professional groups. Enabling shift workers to
included regardless of whether insomnia was present. This study cope with their demanding work hours makes strong economic
is therefore not comparable to other therapeutic intervention sense, as it may help reduce turnover and absenteeism rates, to
studies14 15 37 in which participants were included on the basis the mutual benefit of employees and employers.40 Nevertheless,
of sleep problems. The proactive approach of the intervention organisations still have a responsibility to provide healthy work
may also partly explain why the effect sizes were only small to environments and work schedules that enable sufficient recu-
peration on and between shifts, in order to promote sustainable
moderate.
work conditions.
Previous studies have reported impaired self-­ rated health
To conclude, a short, proactive, group-­administered recovery
among new RNs during the transition from education into
programme was helpful in strengthening recovery for newly
working life.38 Therefore, the present finding that somatic
graduated RNs, by way of preventing somatic symptoms,
symptoms did not increase over time for nurses in the interven-
and reducing burn-­ out symptoms and work-­ induced fatigue,
tion group, but did so for the control group, is important, indi-
suggesting recovery as a key factor in the prevention of negative
cating a preventive effect of the recovery programme on somatic
health consequences of work stress.
symptoms.
While the intervention was effective in reducing fatigue and Acknowledgements  Many thanks to all the newly graduated RNs that
preventing somatic symptoms, it had no significant effect on participated in the ’Bädda för Kvalitet’ study and to all dedicated persons within HR
the secondary outcomes of perceived stress (PSS), listlessness or and the introduction programmes that made this study possible.
tension (SMBQ indices). This may imply that both groups reacted Contributors  AD is the guarantor and acquired funding, designed the study,
similarly to the challenges they face as new nurses. Notably, the developed and delivered the intervention, collected data, monitored data collection,
intent of the recovery programme was not to decrease stress cleaned and analysed data, drafted and revised the paper. MS acquired funding,
designed the study, developed and delivered the intervention, drafted and revised
reactions per se, but to improve the quality of recovery and the paper. ME delivered the intervention, collected data, monitored data collection,
increase the use of recovery behaviours—in line with the theo- cleaned and analysed data, drafted and revised the paper. PT drafted and revised the
retical perspective that stress is not necessarily harmful as long as paper. PG advised on statistical methods, drafted and revised the paper
there is sufficient recovery.4 Funding  This study was funded by AFA Försäkring (150024).
The broad approach of the recovery programme, targeting Competing interests  None declared.
factors regulating sleep and recovery (unwinding from stress,
Patient consent for publication  Not applicable.
supporting sleep according to homoeostatic and circadian
factors, increasing recovery behaviours), may have helped coun- Ethics approval  This study was approved by Ethics Review Board in Stockholm,
Sweden ID: 2016/1395-­31/2.
teract fatigue development. Fatigue is a signal of the need for
recovery and so fatigue should decrease when recuperation is Provenance and peer review  Not commissioned; externally peer reviewed.
strengthened.2 39 Our results point to the value of a holistic Data availability statement  No data are available. Data obtained for the study
will not be accessible to others.
approach to recovery.
Major strengths of the intervention were that it was short, Supplemental material  This content has been supplied by the author(s). It
has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have
proactive and proved feasible in a working life context. Despite
been peer-­reviewed. Any opinions or recommendations discussed are solely those
that only 37% attended all three sessions, the programme of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and
achieved high measured compliance. Compliance may have been responsibility arising from any reliance placed on the content. Where the content

Dahlgren A, et al. Occup Environ Med 2022;79:460–468. doi:10.1136/oemed-2021-107789 467

 Workplace
includes any translated material, BMJ does not warrant the accuracy and reliability 18 Kramer M, Brewer BB, Maguire P. Impact of healthy work environments on new
of the translations (including but not limited to local regulations, clinical guidelines, graduate nurses’ environmental reality shock. West J Nurs Res 2013;35:348–83.
terminology, drug names and drug dosages), and is not responsible for any error 19 Epstein M, Söderström M, Jirwe M, et al. Sleep and fatigue in newly graduated
and/or omissions arising from translation and adaptation or otherwise. nurses-­experiences and strategies for handling shiftwork. J Clin Nurs
2020;29:184–94.
Open access  This is an open access article distributed in accordance with the
Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits 20 Frögéli E, Rudman A, Ljótsson B, et al. Preventing stress-­related ill health among
others to copy, redistribute, remix, transform and build upon this work for any newly registered nurses by supporting engagement in proactive behaviors:
purpose, provided the original work is properly cited, a link to the licence is given, development and feasibility testing of a behavior change intervention. Pilot Feasibility
and indication of whether changes were made. See: https://creativecommons.org/​ Stud 2018;4:28.
licenses/by/4.0/. 21 Rudman A, Arborelius L, Dahlgren A, et al. Consequences of early career nurse
burnout: a prospective long-­term follow-­up on cognitive functions, depressive
ORCID iDs symptoms, and insomnia. EClinicalMedicine 2020;27:100565.
Anna Dahlgren http://orcid.org/0000-0001-8252-3961 22 Lövgren M, Gustavsson P, Melin B, et al. Neck/shoulder and back pain in
Philip Tucker http://orcid.org/0000-0002-8105-0901 new graduate nurses: a growth mixture modeling analysis. Int J Nurs Stud
2014;51:625–39.
23 Söderström M, Epstein M, Dahlgren A. Återhämtningsprogrammet Bädda för
REFERENCES Kvalitet: Innehåll, upplägg och utvärdering. Stockholm: Karolinska Institutet,
1 Meijman TF, Mulder G. Psychological aspects of workload. In: Drenth PJD, Thierry H, de Institutionen för klinisk neurovetenskap, 2020. https://ki.se/media/88984/​
Wolff J, eds. Handbook of work and organizational psychology: work psychology. 2. download
ed. Hove, England: Psychology Press/Erlbaum (UK) Taylor & Francis, 1998: 5–33. 24 Hettema J, Steele J, Miller WR. Motivational interviewing. Annu Rev Clin Psychol
2 Sluiter JK, de Croon EM, Meijman TF, et al. Need for recovery from work related 2005;1:91–111.
fatigue and its role in the development and prediction of subjective health complaints. 25 Ingre M, Van Leeuwen W, Klemets T, et al. Validating and extending the three process
Occup Environ Med 2003;60 Suppl 1:62i–70. model of alertness in airline operations. PLoS One 2014;9:e108679.
3 Zijlstra FRH, Cropley M, Rydstedt LW. From recovery to regulation: an attempt to 26 Bastien CH, Vallières A, Morin CM. Validation of the insomnia severity index as an
reconceptualize ’recovery from work’. Stress Health 2014;30:244–52. outcome measure for insomnia research. Sleep Med 2001;2:297–307.
4 Geurts SAE, Sonnentag S. Recovery as an explanatory mechanism in the relation
27 Åkerstedt T, Knutsson A, Westerholm P, et al. Sleep disturbances, work stress and work
between acute stress reactions and chronic health impairment. Scand J Work Environ
hours: a cross-­sectional study. J Psychosom Res 2002;53:741–8.
Health 2006;32:482–92.
28 Melamed S, Ugarten U, Shirom A, et al. Chronic burnout, somatic arousal and
5 Söderström M, Jeding K, Ekstedt M, et al. Insufficient sleep predicts clinical burnout. J
elevated salivary cortisol levels. J Psychosom Res 1999;46:591–8.
Occup Health Psychol 2012;17:175–83.
29 Grossi G, Perski A, Evengård B, et al. Physiological correlates of burnout among
6 Kecklund G, Axelsson J. Health consequences of shift work and insufficient sleep. BMJ
women. J Psychosom Res 2003;55:309–16.
2016;355:i5210.
30 Fisher GG, Bulger CA, Smith CS. Beyond work and family: a measure of work/nonwork
7 Partinen M. Sleep disorders and stress. J Psychosom Res 1994;38 Suppl 1:89–91.
interference and enhancement. J Occup Health Psychol 2009;14:441–56.
8 McEwen BS. Sleep deprivation as a neurobiologic and physiologic stressor: allostasis
31 Leineweber C, Kecklund G, Lindfors P, et al. Change in Work-­Time control and Work-­
and allostatic load. Metabolism 2006;55:S20–3.
Home interference among Swedish working men and women: findings from the
9 Brosschot JF, Verkuil B, Thayer JF. Generalized Unsafety theory of stress: unsafe
environments and conditions, and the default stress response. Int J Environ Res Public SLOSH cohort study. Int J Behav Med 2016;23:670–8.
Health 2018;15:464. 32 Gierk B, Kohlmann S, Kroenke K, et al. The somatic symptom scale-­8 (SSS-­8): a brief
10 Sonnentag S. The recovery paradox: Portraying the complex interplay between job measure of somatic symptom burden. JAMA Intern Med 2014;174:399–407.
stressors, lack of recovery, and poor well-­being. Res Organ Behav 2018;38:169–85. 33 Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health
11 Tononi G, Cirelli C. Sleep and the price of plasticity: from synaptic and cellular Soc Behav 1983;24:385–96.
homeostasis to memory consolidation and integration. Neuron 2014;81:12–34. 34 Insomniacs’ attributions: psychometric properties of the Dysfunctional Beliefs and
12 Redeker NS, Caruso CC, Hashmi SD, et al. Workplace interventions to promote sleep Attitudes about Sleep Scale and the Sleep Disturbance Questionnaire.. J Psychosom
health and an alert, healthy workforce. J Clin Sleep Med 2019;15:649–57. Res 2000:141–8.
13 Trauer JM, Qian MY, Doyle JS, et al. Cognitive behavioral therapy for chronic insomnia: 35 Fitzmaurice GM, Laird NM, Ware JH. Applied longitudinal analysis. Hoboken, NJ:
a systematic review and meta-­analysis. Ann Intern Med 2015;163:191–204. Wiley, 2004.
14 Järnefelt H, Sallinen M, Luukkonen R, et al. Cognitive behavioral therapy for chronic 36 Boamah SA, Read EA, Spence Laschinger HK. Factors influencing new graduate nurse
insomnia in occupational health services: analyses of outcomes up to 24 months post-­ burnout development, job satisfaction and patient care quality: a time-­lagged study. J
treatment. Behav Res Ther 2014;56:16–21. Adv Nurs 2017;73:1182–95.
15 Järnefelt H, Härmä M, Sallinen M, et al. Cognitive behavioural therapy interventions 37 Schiller H, Söderström M, Lekander M, et al. A randomized controlled intervention
for insomnia among shift workers: RCT in an occupational health setting. Int Arch of workplace-­based group cognitive behavioral therapy for insomnia. Int Arch Occup
Occup Environ Health 2020;93:535–50. Environ Health 2018;91:413–24.
16 Almén N, Lisspers J, Öst L-­G, et al. Behavioral stress recovery management 38 Hasson D, Lindfors P, Gustavsson P. Trends in self-­rated health among nurses: a 4-­year
intervention for people with high levels of perceived stress: a randomized controlled longitudinal study on the transition from nursing education to working life. J Prof Nurs
trial. Int J Stress Manag 2020;27:183–94. 2010;26:54–60.
17 Lisspers J, Almén N, Sundin Örjan, . The Effects of a Recovery-­Focused Program 39 Sonnentag S, Venz L, Casper A. Advances in recovery research: what have we learned?
for Stress Management in Women<br/>—An Exploratory Study. Health what should be done next? J Occup Health Psychol 2017;22:365–80.
2014;06:2825–36. 40 Monk TH, Folkard S. Making shiftwork Tolerable. London: Taylor & Francis, 1992.

468 Dahlgren A, et al. Occup Environ Med 2022;79:460–468. doi:10.1136/oemed-2021-107789

 © 2022 Author(s) (or their employer(s)) 2022. Re-use permitted under CC
BY. Published by BMJ. https://creativecommons.org/licenses/by/4.0/This is
an open access article distributed in accordance with the Creative Commons
Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy,
redistribute, remix, transform and build upon this work for any purpose,
provided the original work is properly cited, a link to the licence is given, and
indication of whether changes were made. See:
https://creativecommons.org/licenses/by/4.0/. Notwithstanding the ProQuest
Terms and Conditions, you may use this content in accordance with the
terms of the License.