Académique Documents
Professionnel Documents
Culture Documents
396..404
Senior Nurse Consultant, Division of Psychiatry, 2Assistant Professor, Social and Forensic Psychiatry Program,
Division of Psychiatry, Department of Clinical Neuroscience, Karolinska Institutet, and 3Head of department of
Health Caring Sciences, Ersta Skndal University College, Stockholm, Sweden
Accessible summary
Correspondence:
A. Bjrkdahl
SLSO
Box 179 14
SE-118 95 Stockholm
Sweden
E-mail: anna.bjorkdahl@sll.se
Accepted for publication: 13 April 2012
doi: 10.1111/j.1365-2850.2012.01930.x
Abstract
Violence prevention and management is an important part of inpatient psychiatric
nursing and specific staff training is regarded essential. The training should be based on
primary, secondary and tertiary prevention. In Stockholm, Sweden, the Bergen model
is a staff-training programme that combines this preventive approach with the theoretical nursing framework of the City model that includes three staff factors: positive
appreciation of patients, emotional regulation and effective structure. We evaluated
this combination of the Bergen and City models on the violence prevention and
management climate in psychiatric inpatient wards. A 13-item questionnaire was
developed and distributed to patients and staff in 41 wards before the staff was trained
and subsequently to 19 of these wards after training. Data analyses included factor
analysis, Fishers exact test and MannWhitney U-test. The result showed that the staff
on trained wards had a more positive perception of four of the items and the patients
of one item. These items reflected causes of patient aggression, ward rules, the staffs
emotional regulation and early interventions. The findings suggest that a focus on three
levels of prevention within a theoretical nursing framework may promote a more
positive violence prevention and management climate on wards.
396
Introduction
Patient aggression and violent behaviour are well-known
phenomena within psychiatric inpatient wards. Many
studies have described its negative influence on the experience of safety and security for both patients and staff: the
risk of physical and psychological injury and the use of
restraint, seclusion and forced medication (Olofsson &
Jacobsson 2001, Bowers et al. 2006b, Richter & Whittington 2006, Jarrett et al. 2008, Stubbs et al. 2009). In the
present study, we therefore wanted to explore how a violence prevention and management programme used in
Sweden may have influenced the ward climate in psychiatric inpatient units.
Background
Violence prevention and management is considered to be
an important as well as challenging part of inpatient psychiatric nursing and specific staff training is regarded as
essential (International Labor Office et al. 2002, Farrell &
Cubit 2005, Beech & Leather 2006). The various theories
concerning causes for inpatient aggression and violence are
often grouped into three explanatory models: the internal,
the external and the situational/interactional model
(Nijman et al. 1999, Duxbury & Whittington 2005). Traditionally, many violence prevention and management
training programmes for staff have relied mostly on the
internal patient explanatory model to violence (Paterson
et al. 2010). By assuming that the major cause of patient
violence is related to symptoms of mental illness and other
individual patient characteristics, a reactive and controlling
approach to aggression and violence has often been
applied. As a consequence, the training has been focused on
self-defence as well as various control and restraint techniques (Duxbury 2002, Farrell & Cubit 2005, Beech &
Leather 2006, Paterson et al. 2010). However, during the
last two decades, an increasing amount of research has
supported the view that violent patient behaviour is often a
result of a complex interplay of different internal, external
and situational/interactional factors (Richter & Whittington 2006). The current international recommendations
regarding violence prevention and management training of
staff in health-care settings therefore state that this complexity must be taken into account and that a proactive
rather than reactive approach should dominate the training
(Krug et al. 2002, Council of Europe 2004, International
Council of Nurses et al. 2005). Moreover, it should be
based on preventive principles of public health including
three dimensions of prevention: primary, secondary and
tertiary prevention (International Labor Office et al. 2002,
Krug et al. 2002). In psychiatric inpatient care, the aim of
2012 Blackwell Publishing
A. Bjrkdahl et al.
Methods
We conducted a prospective non-randomized intervention
study with beforeafter intervention comparisons using an
independent measures design. The local research ethics
committee approved of the study. The study was conducted
in Stockholm, Sweden. In Sweden, employees working in
places where aggression and violence may be expected have
a legislated right to appropriate training provided by their
employer (Swedish Work Environment Authority 1993).
Within Swedish psychiatry, the arrangements have usually
been a matter for the local psychiatric clinics. Training
seems to have been provided by various sources such as
private training companies, self-defence or martial arts
sports clubs or by individual dedicated members of staff.
No common format for this type of staff training appears
to exist on the national level and the extent, content and
quality of such training programmes in Sweden has rarely
been systematically evaluated.
Intervention
The Bergen model is a non-commercial violence prevention
and management training programme for psychiatric inpatient staff. It originates from the Norwegian TERMA
training model that has been developed at the Haukeland
University Hospital, Department of Forensic Psychiatry in
Bergen. In Sweden, the TERMA model was first introduced at the psychiatric department of the Karolinska University Hospital, Huddinge in Stockholm. There, it was
subsequently adjusted to fit into Swedish general psychiatry
and renamed the Bergen model. In addition, the Bergen
model was adjusted to clarify a theoretical nursing framework, inspired by the public health approach (International
Labor Office et al. 2002) and strongly influenced by the
City model (Bowers 2002). The City model is a theoretical nursing-based framework that describes three staff
factors that are considered vital to reducing conflicts and
containment in psychiatric wards. These are: (i) positive
appreciation of patients, which refers to a psychiatric philosophy that promotes a psychological understanding of
difficult patient behaviour and a moral commitment to
values such as humanism and non-judgementalism; (ii) selfregulation of emotional responses, which includes awareness and control of feelings, especially fear and anger;
and (iii) effective structure of rules and routines, which
addresses teamwork skills, organizational support, clarity
of ward rules, early recognition of the interventions needed
and an organized manner in handling challenging situa398
Outcome measure
In order to evaluate the intervention, we sought a questionnaire that could be used for both patients and staff and that
adhered to the following premises: (i) each item would
relate to one or more of the three City model staff factors;
(ii) each item would be congruent to the content of the
Bergen model training programme and to the public health
approach; (iii) items should be observable by both staff and
patients; (iv) the number of items should be restricted,
making the questionnaire short and easy to use; and (v) the
items should be relevant to any type of psychiatric inpatient
ward. No such questionnaire was found in the literature.
We therefore formulated a number of questionnaire statements that were judged as specifically addressing all of
the stated premises and subsequently reduced them to a
13-item questionnaire. Of the 13 items, three were formulated as negative statements (item 4, 8 and 12) (DeVellis
2003). We called the questionnaire E13 (E being the first
letter in the Swedish word for questionnaire). For the
purpose of this paper, two professional translators conducted a translationcounter-translation process of the E13
from Swedish to English. The E13 was designed to collect
dichotomous data that would reflect the participants basic
agreement or disagreement to each of the questionnaire
statements. However, because participants may find it difficult to choose from only two response options, four levels
of agreement to the statements were included, from (1) not
at all, (2) unspecified, (3) unspecified, to (4) totally (Rossberg & Friis 2003). A fifth option, do not know, was also
available. Descriptive data included in the questionnaire
were for staff: sex, age category and occupation, and for
patients: sex and age category.
2012 Blackwell Publishing
Data collection
The data collection commenced in 2007. The E13 was
sent out to all participating wards (n = 41) 3 months
before the first wards were scheduled to start the training.
A research assistant was appointed on each participating
clinic to give information about the study and to distribute and collect questionnaires. The E13 was distributed to
all employed staff and to all patients that the staff assessed
as meeting the inclusion criteria. The criteria included the
ability to read and speak Swedish and the ability to understand the meaning of informed consent. Furthermore,
with respect to the mental and physical health of the
patient, the psychiatrist in charge was able to disapprove
of a patients participation. If possible, the E13 questionnaire was offered to the patients near discharge. An
enclosed letter described the purpose of the study and the
voluntary and anonymous nature of participation. No
coding or any other possibilities of identifying individual
patient or staff participants were made. Together with the
questionnaire, each participant received an unmarked
sealable envelope. A sealed box for collecting the questionnaires was placed on each ward. The data collection
continued for 1 month on each ward. Three to 6 months
after a ward had been trained, the same E13 questionnaire
was sent out again following the same procedures. In
December 2008, the data collection terminated. By then,
19 wards on six hospitals had finished their training and
completed the second round of the questionnaire (one
emergency and admission ward, 13 general wards, two
psychiatric intensive care units and three forensic wards).
Patient and staff turnover and the anonymity of the participants before and after the intervention meant that
the data were collected from independent samples. The
number of distributed questionnaires was not specifically
counted and response and exclusion rates were therefore
not known.
Data analysis
In order to analyse the interrelationship among the questionnaire items a factor analysis was performed, using
an exploratory principal component analysis including
varimax rotation and allowing for factors with an eigenvalue of >1.0 to emerge. In this analysis a three-factor
solution appeared. However, the three factors appeared
weak with a substantial cross-loading of >0.25 between
all factors on several items (Raubenheimer 2004). Furthermore, the second and third factor showed unsatisfactory internal consistency, Cronbachs a < 0.65 (Table 1).
Therefore we assumed that the 13 items may be viewed as
399
A. Bjrkdahl et al.
Table 1
Summary of exploratory principal component analysis for the E13 questionnaire
Rotated factor loadings
Item
0.717
0.692
0.680
0.614
0.579
0.527
0.091
0.370
0.282
0.282
-0.069
0.224
0.404
4.558
35.2
0.729
0.092
0.232
0.064
0.408
0.427
0.478
0.736
0.560
0.525
0.407
0.294
0.184
-0.441
1.193
9.17
0.624
0.136
-0.007
0.170
0.148
0.093
0.182
0.047
0.292
0.197
0.264
0.721
0.649
0.557
1.027
7.90
0.400
The table shows the three-factor solution that was rejected in favour of a one-dimensional solution (Cronbachs a = 0.828).
Results
Staff
A total of 854 staff questionnaires were collected from 41
wards before the training started and 260 staff questionnaires were subsequently collected from the 19 wards that
had been trained. Descriptive data for staff and patients are
presented in Table 2. In the analysis of the differences in the
total questionnaire sum scores, the perception was significantly more positive among staff on the trained wards as
compared with the wards that had yet not been trained,
MannWhitney P = 0.045. The results of the separate item
analysis showed that staff working on wards that had been
trained according to the Bergen model differed significantly
in their perception of violence prevention and management
climate regarding four of the 13 statements compared with
staff working on wards that had not yet been trained
(Table 3). The differences all corresponded to a more positive perception on the trained wards. The areas that were
perceived significantly more positive concerned ward rules
(statement 6), the emotional regulation of staff members in
challenging situations (statement 9), the staffs interest in
possible causes for patient aggression (statement 10) and
the staffs readiness to intervene at an early stage of patient
aggression (statement 11). In addition, just above the level
of significance was statement 2 concerning good relationships between patients and staff (P = 0.06) and statement
3 that related to the staffs ability to co-operate when
approaching aggressive patients (P = 0.058).
2012 Blackwell Publishing
Table 2
Descriptive data of staff and patients
Staff
Wards (n)
Total responses (n)
Responses/ward (n) (mean, range)
Female/male (%)
Age (%)
25
2640
>40
Occupation (%)
Nursing assistant
Registered nurse
Physician
Paramedic
Other
Patients
Before training
After training
Before training
After training
41
854
21 (256)
60/40
19
260
14 (419)
58/42
41
297
8 (118)
45/55
19
156
8 (120)
49/51
4
31
65
2
30
68
63
28
4
2
3
67
28
2
1
2
14
34
52
12
26
62
Table 3
E13 scale content: staff ratings before and after training, Fishers exact test P-value
Item
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
Agree
Disagree
Agree
Disagree
Effect size
(odds ratio)
88.4
94.6
91.3
42.8
91.8
76.9
87.4
65.5
11.6
5.4
8.7
57.2
8.2
21.9
12.6
34.5
89.6
97.6
95.0
39.0
94.6
87.3
91.0
68.9
10.4
2.4
5.0
61.0
5.4
12.7
9.0
31.1
0.729
0.060
0.058
0.332
0.169
0.001*
0.144
0.348
1.12
2.31
1.85
0.86
1.56
1.97
1.45
1.17
(750)
(800)
(772)
(350)
(779)
(655)
(746)
(538)
(98)
(46)
(74)
(468)
(70)
(187)
(108)
(284)
(215)
(241)
(231)
(93)
(226)
(214)
(221)
(162)
(25)
(6)
(12)
(145)
(13)
(31)
(22)
(73)
88.6 (749)
83.8 (707)
80.3 (653)
11.4 (96)
16.2 (137)
19.7 (160)
94.5 (227)
89.4 (211)
89.5 (214)
5.5 (13)
10.6 (25)
10.5 (25)
0.007*
0.031*
0.001*
2.24
1.64
2.10
24.7 (201)
85.3 (719)
75.3 (612)
14.7 (124)
20.6 (49)
87.8 (208)
79.4 (189)
12.2 (29)
0.195
0.399
0.79
1.24
*P < 0.05.
Patients
Discussion
The City model (Bowers 2002) is an important and influential theoretical nursing framework of the Bergen model.
Thus, the Bergen model may serve as one example of how
the results of Bowers and colleagues intensive and longterm psychiatric nursing research on violence prevention
and management is used and implemented in clinical practice in Sweden. By using the City model as the foundation
of the construction of the E13, we believe that this questionnaire may be useful to evaluate the City model within
clinical practice in general and not only in the context of
the Bergen model.
The one statement that was more positively perceived by
the patients on trained wards as well as by the staff was
statement 10: The staff try to understand why a patient is
acting aggressively. Interestingly, this item reflects some
fundamental aspects of nursing, such as relating to and
401
A. Bjrkdahl et al.
Table 4
E13 scale content: patient ratings before and after training, Fishers exact test P-value
Item
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
Agree
Disagree
Agree
Disagree
Effect size
(odds ratio)
83.8
83.6
86.8
31.8
79.4
75.3
84.0
61.5
16.2
16.4
13.2
68.2
20.6
24.7
16.0
38.5
87.2
86.4
88.6
35.2
81.3
72.0
85.6
64.2
12.8
13.6
11.4
64.8
18.7
28.0
14.1
35.8
0.400
0.487
0.731
0.556
0.686
0.484
0.681
0.706
1.32
1.24
1.19
1.17
1.13
0.84
1.16
1.12
(249)
(245)
(177)
(78)
(185)
(217)
(247)
(112)
(48)
(48)
(27)
(167)
(48)
(71)
(47)
(70)
(130)
(127)
(109)
(43)
(109)
(103)
(134)
(68)
(19)
(20)
(14)
(79)
(25)
(40)
(22)
(38)
78.8 (175)
70.5 (146)
71.5 (138)
21.2 (47)
29.5 (61)
28.5 (55)
86.2 (119)
82.6 (95)
76.1 (89)
13.8 (19)
17.4 (20)
23.9 (28)
0.093
0.022*
0.428
1.68
1.98
1.27
38.6 (76)
86.4 (178)
61.4 (121)
13.6 (28)
35.3 (41)
88.0 (110)
64.7 (75)
12.0 (15)
0.629
0.738
0.87
1.15
*P < 0.05.
Methodological considerations
The methodological strength of this study includes the
participation of patients. The importance of including the
opinions of patients in the evaluation of effects of violence
prevention and management staff training has been emphasized in previous research (National Institute for Clinical
Excellence 2006). Nonetheless, literature reviews show that
this is still uncommon (Richter et al. 2006, Johnson 2010).
However, the result of this study should be interpreted with
caution and there are several methodological issues to be
addressed. The lack of control groups is a limitation of the
study design, which gives rise to uncertainty regarding to
what extent the observed changes were due to the staff
training or other confounding variables (Johnson 2010). It
should also be noted that the E13 scale was developed
without having the item pool reviewed by experts or prior
testing of the items on a development sample. Because the
psychometric testing was made directly on the research
sample, the robustness of the E13 scale should be further
established by a repeated factor analysis the next time the
scale is used. The interpretation of a factor analysis ultimately comes down to a subjective evaluation of what
solution appears to be the most meaningful in the light of
theory (DeVellis 2003). Our interpretation was in favour of
the E13 being a one-dimensional scale. In further testing,
this interpretation may be supported or challenged.
2012 Blackwell Publishing
Conclusion
The findings support to some extent that the Bergen training model has a positive influence on the violence prevention and management climate on the wards from the
perspectives of both patients and staff. The findings also
show that the combination of the public health approach to
violence prevention and the City model may be a promising example of a new integrated theoretical framework
that could become a valuable contribution to the development of staff training. Considering the heavy reliance on
the use of incident rates as the outcome variable of staff
training, the inclusion of measurements such as the E13
scale could present a more nuanced picture of the effects of
training, especially because it includes ratings made by
patients. Furthermore, the use of more complex evaluation
variables is also congruent with the current view on multiple causes for inpatient violence. In addition, it may
provide valuable feedback to staff and management in the
efforts to establish a ward climate characterized by an
atmosphere of safety and security.
Acknowledgments
This study was supported by funding from AFA Insurance
(a non-profit organization owned by Swedens labour
market parties). We want to thank the patients and nurses
who took the time to participate in this study. We are also
grateful to the European Violence in Psychiatry Research
Group and to Charlotte Pollak for valuable discussions and
assistance in the research process.
403
A. Bjrkdahl et al.
References
Nursing 9, 325337.
832834.
Olofsson B. & Jacobsson L. (2001) A plea for
respect: involuntarily hospitalized psychiatric
14, 810815.
Paterson B., Ryan D. & McComish S. (2009)
London.
204.
162167.
Raubenheimer J. (2004) An item selection proce-
Services
Framework
International
(2002)
Geneva.
Strasbourg.
Violence
Menaces
in
the
Working
Oaks, CA.
Stockholm.
Nursing, London.
404