English

O R I G I N A L P A P E R

Knut W. Sørgaard Æ Peter Ryan Æ Robert Hill Æ Ian Dawson and the OSCAR group

Sources of stress and burnout in acute psychiatric care: inpatient vs. community staff

Received: 23 November 2006 / Accepted: 11 June 2007 / Published online: 13 August 2007

j Abstract Background Professionals who work alone or in small teams often provide services for people with serious mental health problems in com- munity settings. Stress is common in community teams and this may cause burnout and threaten the quality and stability of the services. This study com- pares levels of burnout and sources of stress among community and acute ward staff in six European centres. Methods A total of 6 acute ward (N = 204) and community staff (N = 209) in 5 different Euro- pean countries filled out the Maslach Burnout Inventory (MBI), the Mental Health Professional Scale (MHPSS) the Agervold Questionnaire for psy- chosocial work environment (QPWES) in addition to a comprehensive demographic questionnaire. Results In the univariate analyses, except for Emo- tional Exhaustion (MBI), there were no differences in burnout between the two groups of staff. Community teams reported more organisational problems, higher work demands, less contact with colleagues, but also better social relations and more control over their work. The ward staff was more satisfied with the or- ganisational structure and access to colleagues, but complained about lack of control over operating conditions at work. The multivariate analyses identi- fied four groups of staff: (1) a Control-dissatisfied and Contact satisfied group (N = 184) with 2/3 coming from the wards. (2) A Contact-satisfied and Work- demand dissatisfied group (N = 147) with 3=4 from the community staff. (3) A Control- and Contact

dissatisfied group (N = 47) with a majority from community teams, and (4) a Contact- and Work de- mand satisfied group (N = 37) with a majority from the wards. Conclusion Burnout as measured was not a serious problem among community and ward staff members, and did not differentiate between the two groups. Acute ward working implied lack of control but much contact with colleagues, whereas commu- nity work entailed more control but demanding work in terms of difficult task and hard-to-find-solutions.

j Key words stress – burnout – community care – acute wards

Introduction

Professionals who often work alone or in small teams are increasingly providing the care of people with serious mental health problems in community set- tings. Continuous structural changes in mental health systems, with accompanying changes in role requirements, exposes mental health professionals to new sources of stress. Studies show that community mental health staff experience considerable stress [6, 11, 20, 22] and that community work is more stressful than working in a ward [5, 23]. Concerns have been raised that such high levels of stress may cause de- moralisation and burnout and thus threaten the quality and stability of the community mental health services [21, 23].

Common stressors include workload, time man- agement, under-staffing, conflict with patients and relatives, difficult patients, job insecurity, role ambi- guity, shift work and organisational change [10, 22]. There is uncertainty about how widespread the problems of stress and burnout in community care are. Professions may be differently affected [10, 13, 21, 24] and amount and types of stress may vary locally, regionally and nationally. Stress and burnout hasSP

P E

2 2

8

K.W. Sørgaard, PhD (&) Æ I. Dawson Nordland Hospital Trust 8092 Bodø, Norway E-Mail: [email protected]

P. Ryan, PhD Middlesex University Archway Campus London, UK

R. Hill, PhD, PsychD South London and Maudsley NHS Trust Beckenham, Kent, UK

Soc Psychiatry Psychiatr Epidemiol (2007) 42:794–802 DOI 10.1007/s00127-007-0228-6

been studied less among community than general nurses and staff in acute psychiatric inpatient settings have received even less attention [14]. Stressors in community teams reflect the peculiarities of working in multi-professional non-institutional settings. Chief among these are: role conflicts; organisational prob- lems; lack of supervision; lack of resources; lack of opportunities for professional development, and security problems [4, 11, 12].

Much of the existing research on stressors and burnout in mental health is geographically unbal- anced and often based on small samples. With the exception of Priebe et al. [21] few studies have com- pared community teams in different countries. This makes generalisations uncertain and there is a need for large scale, international multi-centre studies. The present study compares sources of stress and levels of burnout among acute inpatient and community staff in 6 psychiatric centres in 5 European countries: 2 from Denmark and 1 from the following countries: Finland, Great Britain, Norway and Poland. In addi- tion to a descriptive overview of sources of stress and level of burnout among the community and acute ward staff in these centres, the main hypothesis guiding the study were that (1) community staff would have higher levels of burnout than acute ward staff, and (2) that there would be different sources of stress among the two groups reflecting different or- ganisational surroundings and responsibilities. The study was part of an EU-financed multi-centre study of stress, burnout and security problems in acute mental health care [19].

Subjects and method

j Study design

The overall design for the study was a simple longitudinal time series, with repeated measures at baseline, six and twelve months. The staff groups were assessed at baseline i.e. prior to receiving training specifically designed and developed to reduce levels of occupational stress, and to increase the efficacy of risk assess- ment. Staff were followed up at 6 and 12 months post interven- tion. This design was considered appropriate in that it would allow for the variability of a number of potentially unknown factors between sites. It would also overcome contamination issues that would otherwise be problematic in a randomised design.

The numbers of participants per site were calculated by taking into account the longitudinal design of the study. For the study to have a Type 1 error rate of 5% (statistical significance) and Type 2 error rate of 80% (power) and treating each of the six settings as well as inpatient and community staff, as separate groups (i.e. 12 groups in total) to detect an effect-size of 0.5 (‘small’ to ‘medium’ effect size), and assuming that the measures to be used have a retest correlation of 60%, the sample size required per group was approximately 26. Taking into account a drop out rate of an esti- mated 35%, the final sample size was 35 subjects per group (i.e. 70 per setting, or 420 in total). Teams were randomly selected from each site, stratified by inpatient vs. community teams, until the approximate required sample size was reached. In some cases this resulted in all locally available teams entering the study. The sampling procedures for each site are summarised in Table 1.

Once the teams were selected for each centre as described above, the site researchers visited all teams included in the study, explained the study aims, policy with respect to confidentiality etc, handed each staff member an information leaflet about the study, and answered any questions. Questionnaires were left on the wards for self-completion and collected by the site researchers at an agreed timescale later. Staff who did not complete the instruments were followed up with a view to encourage completion. The total participation rate was 72 percent.

j Participating centres

The present study compared sources of stress and levels of burnout among acute inpatient and community staff in 6 psychiatric centres in 5 European countries, inpatient and community staff samples being drawn from each participating centre. There were two centres from Denmark and one each from Finland, Great Britain, Norway and Poland. Within centres, the inpatient and community staff were drawn from within the same catchment areas, thus ensuring that issues around demographic variations in patient populations were addressed. Catchment areas were randomly selected from within the total set of catchment areas available in the centres participating in the study. The main hypotheses guiding the study were that (1) community staff have higher levels of burnout than acute ward staff, and (2) that there are systematically different sources of stress among the two groups reflecting different or- ganisational surroundings and responsibilities related to for example control over one’s working conditions.

Descriptions of the centres are found in Table 2.

j Additional information about admission and treatment policies

(1) Aarhus (Denmark). The inpatient ward (16 beds) is open with a secluded section. It is not exclusively dedicated to acute patients. The Community Teams are considered the primary unit of the psychiatric services and have an inner city catchment area. Their diagnostic groups: long term psychosis, uni/bipolar affective dis- orders, borderline conditions. (2) Bodø (Norway). The acute ward (10 beds) serves 10 municipalities. Most patients are detained compulsorily. The Rehabilitation team (multi-disciplinary) was established to reduce the readmission rate of discharged patients.

Table 1 Numbers of participants per centre

Site Type of sample (inpatient) Type of sample (community) Inpatient (N) Community (N) Total (N)

Aarhus, Denmark Random sample of one team Census of both teams 33 20 53 Bodo, Norway Census of one team Census of 3 teams 27 34 61 Cambridge, UK Random sample of one team Census of both teams 37 50 87 Storstrom, Denmark Census of both teams Census of both teams 34 13 47 Tampere, Finland Random sample of two teams Random sample of seven teams 42 62 104 Warsaw, Poland Random sample of one team Census of one team 32 30 62 Total 205 209 414

795

T a

b le

2 D

em og

ra p

h ic

ch ar

ac te

ri st

ic s

of th

e p

ar ti

ci p

at in

g ce

n tr

es

Si te

/n u

m b

er of

p ar

ti ci

p an

ts P

op u

la ti

on R

eg io

n al

ch ar

ac te

ri st

ic s

N o.

of b

ed s

A g

e g

ro u

p p

at ie

n ts

Ca se

lo ad

to ta

l/ av

e- ra

g e

St af

f m

ea n

ag e

St af

f— ye

ar s

in m

en ta

l h

ea lt

h ,

av er

. Y

ea rs

cu rr

. jo

b R

ec ip

vi ol

en ce

% M

al es

%

A ar

h u

s 60

,0 00

R u

ra l,

sm al

l to

w n

s, 7

m u

n ic

ip al

it ie

s 16

18 –

69 43

6 6.

2 73

% 23

% In

p at

ie n

t (N

= 26

) In

n er

-c it

y ca

tc h

m en

t ar

ea A

ar h

u s

55 ,0

00 21

– 80

40 0/

20 –

25 44

12 5.

9 47

% 18

% Co

m m

u n

it y

te am

s (N

= 17

) 60

,0 00

B od

ø R

u ra

l, to

w n

, 11

m u

n ic

ip al

it ie

s In

p at

ie n

t (N

= 30

) 72

,0 00

10 18

– 65

38 4.

4 2,

1 90

% 40

% R

eh ab

ili ta

ti on

te am

(N =

8) 18

– 66

/7 44

16 2.

3 10

0% 33

% Ea

rl y

in te

rv en

ti on

te am

(N =

5) 18

– 23

25 /7

34 10

0. 6

10 0%

20 %

Co m

m u

n it

y St

af f

(N =

21 )

18 –

65 25

6/ 8

46 4.

3 2.

9 65

% 15

%

Ca m

b ri

d g

e 65

,0 00

a R

u ra

l vi

lla g

e, sm

al l

to w

s V

ar ie

sb 18

– 64

34 5.

0 0.

8 91

31 %

In p

at ie

n t

(N =

32 )

Co m

m u

n it

y te

am 1

(N =

17 )

17 –

65 25

0/ 25

– 30

42 6.

5 2.

0 10

0 35

% Co

m m

u n

it y

te am

B 2

(N =

28 )

60 ,0

00 17

– 65

28 0/

40 44

8. 0

2. 0

82 %

39 %

St or

st ro

m 80

,0 00

7 m

u n

ic ip

al it

ie s,

ru ra

l, to

w n

s an

d N

ae st

tv ed

t 12

18 –

69 41

6. 4

6 60

7% In

p at

ie n

t u

n it

1 (N

= 15

) 16

18 –

69 51

9. 0

10 53

5%

In p

at ie

n t

u n

it 2

(N =

19 )

7 m

u n

ic ip

al it

ie s,

ru ra

l, to

w n

s an

d N

ae st

tv ed

t Co

m m

u n

it y

te am

s A

(N =

6) 70

,0 00

8 m

u n

ic ip

al it

ie s,

ru ra

l, to

w n

s 25

– 69

22 0/

20 –

25 44

11 .5

5 30

10 0%

Co m

m u

n it

y te

am s

B (N

= 7)

30 –

40 ,0

00 3

m u

n ic

ip al

it ie

s 26

– 69

70 /1

5 48

16 .0

14 86

29 %

Ta m

p er

e U

n iv

h os

p it

al 34

m u

n ic

ip al

it ie

s, on

b ig

ci ty

(T am

p er

e) In

p at

ie n

t te

am s

11 0

(N =

24 )

45 0,

00 0

18 18

– 64

41 17

.3 16

.5 54

30 %

In p

at ie

n t

te am

13 0

(N =

19 )

16 18

– 64

41 16

.5 14

.0 90

42 %

Co m

m u

n it

y te

am 20

9 (N

= 9)

18 –

64 Co

m m

u n

it y

te am

21 1

(N =

10 )

18 –

25 Co

m m

u n

it y

te am

20 6

(N =

5) >

18 Co

m m

u n

it y

te am

21 3

(N =

23 )

> 18

Lo ca

l te

am s

· 3

(N =

3 ·

5) Co

m m

un it

y te

am s

av er

ag e

� 12

45 14

.2 6.

3 58

23 %

W ar

sa w

F1 W

ar d

1 (N

= 32

) 75

,0 00

In n

er ci

ty 35

10 0/

12 –

14 38

8. 9

6. 2

75 16

Co m

m u

n it

y te

am ZL

S 1

26 0,

00 0

> 18

R es

id en

ti al

d is

tr ic

t Co

m m

u n

it y

Te am

ZL S

2 56

0, 00

0 18

– 70

11 0–

14 0/

15 –

20 R

es id

en ti

al an

d su

b u

rb an

Co m

m u

n it

y Te

am ZL

D 3

> 18

25 0/

25 Co

m m

u n

it y

te am

s to

ta l

42 16

.5 9.

8 67

20

796

Diagnostic groups: psychosis and personality disorders. The early intervention team (multi-disciplinary) serves persons with first time psychosis and their carers. Both teams work on outreach principles. The community mental health staff comprises 22 employees in nine small municipalities who have the daily follow- up responsibility of patients in the local communities. (3) Cam- bridge (UK). The inpatient team serves patients with acute psy- chiatric illness or dual diagnoses. Many are detained compulsorily. Priority is given to establish/maintain links into the community. The community teams are multi-disciplinary and work with pa- tients who have serious and enduring mental health problems. One team had both community and inpatient responsibility for their patients. (4) Storstrom (Denmark). One inpatient unit with forensic patients (80% of the 12 available beds). Diagnoses are psychosis, depression, bipolar and forensic problems. The other ward is low security, patients ranging from depression to psychoses. Two community teams serve 11 municipalities. Their main aim is to prevent hospitalisation. Patients include long-term psychosis and bipolar disorders. (5) Tampere. One acute ward serves (mainly) first time admitted patients and other patients in crises or with security problems. Two of the community teams are day-centre based, one is specializing in treating younger patients. The other community teams are a crises intervention and home-visiting team working with mainly hard-to-engage clients. Three other commu- nity teams offer outpatient services for discharged and GP-referred patients. (6) Warsaw. The inpatient ward consists of two diagnos- tic-therapeutic units. Geographical area is inner city. Psychother- apy, occupational therapy and network meetings are provided. The second team addresses particularly difficult cases, 90% are referred after hospitalisation. Staff are available from 8.00 a.m. to 8.00 p.m. A third community team ‘‘A’’ has a semi-open catchment area. Patients with schizophrenia and long-term depression are in the majority. The Warsaw staff had the greatest number of weekly work hours (paid and unpaid) in the project (70 h for inpatient and 65 h. for community staff, in addition to1.5 h daily travelling to work.

The project was developed and administrated from Middlesex University in London.

In Table 3 demographic characteristics of ward and community staff is presented.

Community teams have more social workers, psychologists, doctors, but fewer nurses. They are older, work less shift work, are less exposed to violence from patients and are more seldom trained in dealing with such violence.

Measures

The following self-report measures were used. The OSCAR Demographic Questionnaire—ODQ [27], the Maslach Burnout Inventory—MBI [16], the Mental Health Professional Stress Scale (MHPSS) [7] and the Psychosocial Work Environment and Stress Ques- tionnaire—PWSQ [1, 2]. The OSCAR demographic questionnaire comprises 105 items covering basic demographics (age, sex, martial status, number of children, other dependencies, education, work related questions and work related violence). The MBI was chosen as a reliable and valid indication of levels of burnout among mental health professionals amongst the sites. It is a 22-item measure intended to assess three aspects of the burnout syndrome: Emotional Exhaustion (EE 9 items), depersonalisation (DP 5 items) and personal accomplishment (PA 8 items). Each item is scored on a 7-point scale (None to Every day). Reliability and validity are good. The three- factor structure has proved invariant across different occupational groups, and the internal consistencies of

the subscales are satisfactory [25, 26]. The MBI is widely used in studies of mental health professionals (usually nursing staff).

To identify and measure the causes of stress per- tinent to the different professional groups the Mental Health Professionals Scale (MHPSS) was chosen [8]. The MHPSS is a 42-item discipline-neutral scale for identifying sources of stress in mental health profes- sionals. It is suitable for multi-disciplinary work, ad- dresses home-work conflicts and is psychometrically robust. It is grouped into seven scales: workload, client-related difficulties, organisational structure and processes, relationships and conflicts with other professionals, lack of resources, professional self- doubt and home-work conflict. It is scored on a 4- point scale (from Does not apply to me to Does apply to me). The scale has good reliability and validity [17] and there is some support for the original factor structure [17]. The social climate or environment is an important source of burnout related stress in a number of previous studies. The social climate was studied by using The Psychosocial Work Environ- ment and Stress Questionnaire (PWSQ) [1, 2]. The scale also has items on mental fatigue/burnout, psy- chological stress and psychosomatic fatigue. The majority of items are scored as Yes or No, and—in contrast to the MBI and the MHPSS—a small score signifies satisfaction/positive experiences. It has been thoroughly tested for reliability and validity (Rasch item analyses model) and consists of 10 scales (Job demands, Work load, Job control, Influence, Man- agement style, Role clarity, Social contact, Social cli- mate, Personal development via work and Work centrality) measuring psychosocial environment and 3 subscales covering mental burnout/fatigue, psy- chological stress and psychosomatic symptoms. The scale scores range from 0 to 5 and higher scores

Table 3 Demographic characteristics of the participants (Mann–Whitney, Pearson)

Variable Ward staff (N = 205)

Community staff (N = 209)

Age (sd) a

40.2 (10.8) 43.9 (8.8) Males 15.3% 16.3% Professional qualifications 72.9% 65.5% Professional quaificationsb

Nurses 46.2% 40.6% Doctors 9.2% 11.0% Psychologists 2.4% 7.6% Social workers 1.9% 8.6% Unskilled workers 23.0% 14.6% Other professionals 17.3% 17.6%

Time in unit (years) 7.2 (8.1) 5.5 (6.3) Time in mental health (years)c 11.93 (10.1) 13.25 (9.0) Shift work

d 52.2% 38.8%

Recipient of violence evere 77.9% 68.4% Training in dealing with violence

f 62.8% 35.7%

a Z—3.39, P = 0.001;

b all professionals: Pearson chi square 19.38, P = 0.002;

c Z—2.00, P = 0.045;

d Pearson chi square 7.31, P = 0.007;

e Z—4.76,

P = 0.0029; f Pearson chi square 29.74, P = 0.000

797

represent higher levels of dissatisfaction or stress. The national ethical committees relevant for each partici- pating centre approved the study.

j Statistics

Non-parametric univariate statistics (Mann–Whitney, Kruskal–Wallis) were used for studying univariate relations and Classification tree for multivariate analyses. The choice of Classification tree as the multivariate statistics was due to a marked deviation from the non-normality requirements of most of the explanatory variables. When there are many potential explanatory variables Classification trees can give a clear picture of the structure of the data and inter- action among the variables [3]. In the tree-analyses the variables from MBI, MHPSS, and QPWS were used, in addition a selections of variables from the ODQ (basic demographics, education, employment, and work related violence and training). The tree was estimated using the CRT routine in the SPSS. Data analyses were performed on the SPSS 14.00.

j Definitions: stress and burnout

Stress and burnout are well reviewed in Edwards et al. [10]. Stress can be defined as the psychological, physiological and behavioural response by an indi- vidual when they perceive a lack of equilibrium be- tween the demands placed upon them and their ability to meet those demands, which, over a period of time, leads to ill-health. Burnout is related to stress and it is most frequently linked to three factors: (a) emotional exhaustion, (b) depersonalisation related to the work environment, and (c) a sense of diminishing personal accomplishment. Burnout is a state of physical, emotional fatigue, and it is caused by a long- term commitment to demanding situations. It is de- scribed as a sense of helplessness and hopelessness, low energy level, chronic tiredness, fatigue, and a feeling of being trapped. Typical are also negative feelings for self, work and life. One of the conse- quences of these symptomatic effects can be a severe disruption or dislocation of the therapeutic relation- ship between the service provider and client [10, 16].

Results

Univariate analyses. Table 4 shows the results of the univariate analyses

The community teams experienced more emo- tional exhaustion, were more dissatisfied with the organisational structure and process, had better social relationships, lower work demands, had more control over their work, were more dissatisfied with the management style and with their collegial contacts. On the MBI the sample means (SD) was EE 17.1 Ta

b le

4 U

n iv

ar ia

te d

if fe

re n

ce s

b et

w ee

n in

p at

ie n

t an

d co

m m

u n

it y

st af

f. W

it h

th e

ex ce

p ti

on of

th e

M B

I, on

ly si

g n

if ic

an t

d if

fe re

n ce

s ar

e sh

ow n

(M an

n –

W h

it n

ey )

V ar

ia b

le d

ef in

it io

n s

In p

at ie

n t

Co m

m u

n it

y

M B

I Em

ot io

n al

ex h

au st

io n

Fe el

in g

s of

b ei

n g

em ot

io n

al ly

ov er

ex te

n d

ed an

d ex

h au

st ed

15 .8

(9 .7

4) 18

.3 1

(1 0.

5) D

ep er

so n

al is

at io

n U

n fe

el in

g an

d im

p er

so n

al re

sp on

se s

to w

ar d

s re

ci p

ie n

ts of

se rv

ic es

4. 9

(4 .7

) 4.

5 (4

.4 )

P er

so n

al ac

co m

p lis

h m

en t

Fe el

in g

s of

co m

p et

en ce

an d

ac h

ie ve

m en

t in

on e’

s w

or k

36 .3

(8 .1

) 36

.4 (7

.4 )

M H

P SS

O rg

an is

at io

n al

st ru

ct u

re an

d p

ro ce

ss b

Su p

p or

t fr

om m

an ag

er ,

re la

ti on

s w

it h

lin e

m an

ag er

, co

m m

u n

ic at

io n

an d

in fo

rm at

io n

fl ow

, co

n fl

ic t

re so

lu ti

on ,

or g

. st

ru ct

u re

an d

p ol

ic ie

s 0.

9 (0

.7 )

1. 1

(. 7)

A g

er vo

ld So

ci al

en vi

ro n

m en

t Cl

iq u

es ,

co n

fl ic

ts ,

p er

so n

al re

la ti

on s,

q u

ar re

lli n

g ,

u n

it y

0. 96

(1 .4

) 0.

62 (1

.1 )

W or

k d

em an

d s

Co n

ce n

tr at

io n

, re

ca lls

, ta

sk d

if fi

cu lt

y, ac

ce ss

ib le

so lu

ti on

s, 3.

09 (1

.0 )

3. 52

(0 .8

2) Co

n tr

ol of

yo u

r w

or ke

d In

fl u

en ce

, d

ec id

e w

or k

rh yt

h m

, in

d ep

en d

en t

d ec

is io

n s,

p la

n n

in g

, re

sp on

si b

ili ty

1. 44

(0 .7

5) 0.

75 (1

.1 1)

M an

ag em

en t

st yl

e P

re fe

re n

ti al

tr ea

tm en

t, n

eg at

iv e

fe ed

b ac

k, co

n fu

se d

le ad

er sh

ip ,

in fo

rm at

io n

0. 95

(1 .3

) 1.

31 (1

.4 )

Co n

ta ct

w it

h co

lle ag

u es

g Co

lla b

or at

io n

an d

p ro

b le

m so

lv in

g ,

cl os

e co

n ta

ct d

u ri

n g

w or

k, ta

lk in

g w

it h

co lle

ag u

es ,

q u

al it

y of

re la

ti on

s, ch

at ti

n g

0. 46

(0 .8

5) 1.

11 (1

.2 2)

a Z—

2. 51

, P

= .0

12 ;

b Z—

2. 17

, P

= 0.

00 5;

c Z —

2. 08

. P

= 0.

03 7;

d Z—

4. 67

. P

= 0.

00 0;

e Z—

6. 24

, P

= 0.

00 0;

f Z —

3. 75

, P

= 0.

00 1;

g Z—

6. 05

P <

0. 00

1

798

(10.2), DP 4.6 (4.7) and PA 35.6 (7.9). These can be compared with the MBI manual categories (17) EE: High ‡ 27 Average 17–26 and Low £ 16; DP:‡13, 7–12 and £ 6, respectively, and PA:‡39, ‡32–38 and £ 31, respectively.

Figure 1 shows the results of classification tree- analysis. The method was CRT, the minimum number of cases in the parent nodes was 30 and in the child nodes 16.

In this almost equal sized sample of ward (N = 204) and community staff (N = 209) the vari- able that primarily distinguished between the ward and the community staff was Control (influence on the work situation, the pace, important decisions, planning and having adequate responsibilities). The scores on this variable produced two groups: a high- score group (N = 229) of persons with negative experiences on the control indicators and a low-score/ satisfied group (N = 184). The high score/control- dissatisfied group had a majority of ward staff (62.9%). A second, less important variable—Con- tact—dichotomised this control-dissatisfied group into one comparatively smaller contact-dissatisfied one (N = 47) consisting mainly of community team members (72.3%) and a much bigger contact-satisfied group (N = 182) of mainly (72.0%) ward staff. Con-

tact referred to cooperation during the performance of work, how the work situation was arranged to make contact possible, the possibility of talking together during the work performance, the quality of social relations and the availability of colleagues to chat with. If on the figure we follow the Control-satisfied group (N = 184), this group was also dichotomised by a second variable—Work demands—into a large satisfied group (N = 147) of mainly community team members (74.1%), and much smaller (N = 37) Work- demand dissatisfied group with a majority of ward members (59.5%). The ‘‘Work demands’’ variable referred to concentration, lots to bear in mind, diffi- cult tasks and hard-to-find solutions. Summing up we might say that working in acute wards implied lack of control but much contact with colleagues, whereas community work implied more control but demand- ing work in terms of difficult task, hard-to-find- solutions etc. Thus four different groups were iden- tified. Group 1 consisted of Control-dissatisfied and Contact satisfied staff (N = 182) with a majority (72%) of ward staff. Group 2 comprised Control sat- isfied and Work demand dissatisfied employees (N = 147) with a majority of community staff (74.1%), Group 3 consisted of Control- and Contact dissatisfied staff (N = 47) with a majority from com-

Node 0

Category % n

49 .4 204Ward 50 .6 209Com Teams

Total 100 .0 413

Control Improvement=0.047

Teams

Node 1

Category % n

32 .6 60Ward 67 .4 124Com Teams

Total 44 .6 184

Work demands Improvement=0.016

<= 0.50

Node 2

Category % n

62 .9 144Ward 37 .1 85Com Teams

Total 55 .4 229

Contact Improvement=0.036

> 0.50

Node 3

Category % n

59 .5 22Ward 40 .5 15Com Teams

Total 9.0 37

<= 2.34

Node 4

Category % n

25 .9 38Ward 74 .1 109Com Teams

Total 35 .6 147

> 2.34

Node 5

Category % n

72 .0 131Ward 28 .0 51Com Teams

Total 44 .1 182

<= 1.62

Node 6

Category % n

27 .7 13Ward 72 .3 34Com Teams

Total 11 .4 47

> 1.62

Ward Com Teams

Fig. 1 The structure of the most important sources of stress characterising ward and community staff. Tree analysis

799

munity teams (72.3%), and Group 4 (N = 36) was made up of Control and Work demand satisfied persons with a majority (59.5%) from the wards. The percent of correct classifications was 76.1% for Ward and 68.9% for community teams, overall 72.5%. Normalised relative variable importance to the model showed that the Control variable was 100%, the Contact variable 75.4% and the Work Role variable 34.9%. It is of interest that none of the burnout (MBI), sources of stress variables (MHPSS) nor demographic variables (ODQ) appeared as predictors. Control, Contact and Work demands all stem from the PWSQ.

Discussion

Assumptions about different sources of stress and levels of burnout among community mental health staff compared with staff from acute mental health wards motivated the study. The analyses identified four groups of staff. There were to main groups: the first was a Control-dissatisfied and Contact satisfied group (44.1% of the total sample) with a majority (72.0%) coming from the wards. A second group comprised Control satisfied and Work demand dis- satisfied employees (35.6% of the total sample) mainly (74%) consisting of community staff. The two smaller groups were a Control- and Contact dissatisfied staff (11% of the total sample) with a majority (72.3%) from the community teams, and a group (9.0% of the total sample) made up of Control and Work demand sat- isfied persons with a majority (59.5%) from the wards. A short summing up of the results would be that the Ward staff were characterised by lack of control, but also ample opportunities for social contact at work, whereas the Community staff was much more in control over their work situation but complained about high work demands. None of the burnout (MBI) nor the work-stress variables (MHPSS), nor the demographic variables discriminated between acute ward and community staff in the multivariate analyses.

Research especially from the UK has found high levels of stress and burnout in community teams [11, 20, 22]. The univariate differences in burnout between acute ward and community staff measured on the MBI were in the predicted direction (community staff experiencing more) on Emotional Exhaustion only. The differences in the two other MBI dimensions were nonsignificant. Univariate comparisons between inpatient and community staff showed that the com- munity staff complained more about organisational aspects of their work: lack of support from and problems with managers, deficient communication and information flow, preferential treatment, feed- back only on wrongdoings and confused leadership. Community staff were also less satisfied with their work demands in terms of concentration, having lots to grasp, difficult tasks and solutions that were often hard to find. All this reflects a job where colleagues

are not always easily accessible and converges with a more negative score on the contact-with-colleagues variable: less close collaboration, less talking and chatting, and being more alone in finding solutions to problems. On the positive side the social environment at work suffered less from cliques, conflicts, negative personal relations and quarrelling. Community staff also reported more control over their work: more influence on planning and decision making, adequate responsibility and reasonable work rhythm. These results are more differentiated than Prosser et al.’s [23] who found that community work was more stressful than working in hospitals.

The MBI has been the main measure of burnout in most of the studies that have focused on stress and burnout in community care. Compared with the MBI manual norms [16] the total study mean for Emo- tional Exhaustion in the present study was on the low end of the average level, Depersonalisation was low and Personal Accomplishment average. This study does not therefore indicate a population of staff with higher normative scores on any of the three MBI factors. This is not because the services were pri- marily rural. Three of the sites were based within cities (Warsaw, Cambridge and Aarhus), although only one (Warsaw) is a large city. Kilfedder et al.’s [15] in their study of a random sample of 501 nurses from one Scottish Trust also found average to low MBI mean scores, but the proportion of persons exceeding the MBI norms in the present study were lower (Emotional exhaustion: 10.2% vs. 21.6%; De- personalisation: 5.3% vs. 7.1% and Personal accom- plishment 22.2% vs. 33.1%).

Only 5 persons (1.2%) had scores that exceeded the MBI norms on all three dimensions (‡27, ‡ 13 and £ 31, respectively) compared with Kilfedder et al.’s

2.0%. A reasonable conclusion is that burnout as de- fined in the Maslach tradition, was not a wide-spread problem in the present multi-centre sample although it affected a minority of employees. The differences compared with Kilfedder et al. may be due to national or regional variations in organisation, work load, responsibilities etc. In a comparative study, Nolan et al. [18] found that English nurses encountered more violence in their work than Swedish colleagues did. They also reported lower individual well-being, lower levels of self-esteem and a higher workload.

Violent and disruptive patients are a major source of stress in mental health staff [11, 14, 18, 19]. In the present study 78 percent of the staff had been recipients of violence from patients, slightly (not significantly) more often among inpatient than community staff. An important difference related to security was that almost twice as many ward staff had received training in dealing with such violence than community staff (63% of the ward and 36% of the community staff). The Danish centres had had a much higher percentage of staff who had previously received training in the management of violence (78%, 85% of the ward and

800

64% of the community staff) than Bodø (33%; 50% vs. 18%), Cambridge (36%; 51% vs. 24%), Tampere (56%; 77% vs. 44%) and Warsaw (26%; 25% vs. 27%).

Most of the explanatory variables had distributions that made them unfit for parametric multivariate analyses, and accordingly a classification tree was chosen for multivariate comparisons. The variable that above all differentiated between community and ward staff was Control (influence, being able to affect work rhythm, taking independent decisions, contrib- ute to planning and having adequate responsibility). Community staff reported that they, to a significantly higher degree, could influence the policy of their organisation, the daily routines at work, the work rhythm and that they, more than acute ward staff, could take independent and responsible decisions. Acute wards are probably the most squeezed part of today’s mental health services with limited possibili- ties to restrict admissions and select patients, and it is not easy to see how the high patient turnover, sick patients and the legal responsibilities of these wards can effectively be combined with high general staff influence with regard to management and policy.

On the positive side acute wards may, to a larger extent than the community teams, give room for contact with colleagues, as was the case in the present study. The Contact variable dichotomised the Control- dissatisfied and mainly wards based group into a contact-satisfied and a Contact-dissatisfied group. About 3=4 of the contact-satisfied group were inpa- tient staff and about the same proportion of contact- dissatisfied staff were community staff. As will be remembered the contact dimension measured in this study covered the opportunities for contact in carrying out the job, how the work situation is arranged to enable social contact, talking to colleagues during the work, the quality of social relations and the general access to colleagues. Going back to the control satis- fied group, community staff was in majority (67%). The majority of those satisfied with the control aspect of their work came from the community staff. This was group was dichotomised by the Work-demand vari- able. Community staff were in general far more dis- satisfied with their Work demands than the ward staff. They complained about having much to bear in mind, having to concentrate a lot, task difficulty and dealing with many hard-to-find solutions in carrying out the work. With regard to the hypothesis: In the multi- variate analyses (i) we found significant differences between the ward and community staff with regard to social climate (as measured by the PWSQ), (ii) but no differences in levels of burnout and stress.

Conclusions

With the exceptions of emotional exhaustion there were no differences in burnout or aspects of burnout between acute ward and community staff. There were

significant differences related to sources of stress: Community teams experiences more organisational problems, higher work demands, less contact with colleagues, but also better social relations and more control over their work. The ward staff were more satisfied with the organisational structure and had easy access to colleagues during the work, but they also reported a worse social environment and a lack of control.

j Limitations of study

A major limitation lies in the fact that the countries chosen for inclusion in the study were purposive: these countries came together as a bid consortium and agreed to collaborate in the submission of an EU Framework 5 bid in which they were successful. However, it was not possible to ensure at this stage that these countries in any sense were randomly drawn from all EU countries, or that therefore their services are in any sense necessarily representative of the EU as a whole. However, as the section on study design makes clear steps were taken within this lim- itation to ensure that the services chosen within the participating centres were randomly selected. It could therefore be claimed that the study’s services were representative of the total set of services within the participating centres. By the same token it is not possible to claim that these centres were necessarily representative of national Danish, British or Finish services etc as a whole and this therefore is a further limitation. Thus it is difficult to generalise as to what degree any of the centres were representative of typ- ical working conditions operative in their own countries on a wider basis.

A further limitation is that selection criteria and exclusion criteria were different on the different sites, both for inpatient and community teams. For example in Storstrom (Denmark), one inpatient unit worked with forensic patients, which was not the case with inpatient settings in the other participating centres. Given that patient characteristics therefore varied between sites, it is difficult to know the degree to which differences in stress and burnout were simply an expression of different levels of ‘difficulty’ in the patient populations. Another limitation was the mat- ter of qualitative semi-structured questionnaire translations and their revalidation in new languages. Whilst the MBI already was translated reliably into all the languages pertinent to this study, this was not the case for the ODQ, PWSQ or the MHPSS. While it was anticipated that back translation would occur during the course of the project, the complexity and cost of the task could not be born by the project’s funding. However, all of the key researchers were English speakers as a second or third language and in one site, Bodo (Norway), the researcher was a native English speaker. While this linguistic expertise was crucial to the project’s success, possible nuances within some of

801

the questionnaires may not have been picked up upon. While there is no reason to believe that there would be systematic differences between community and inpatient teams in this regard, the validity of some questionnaire items must remain open. Con- sideration must also be given to the sample sizes achieved for each site and team type. While overall the numbers compared between inpatient and com- munity services were equal, there was some imbalance within some of the sites.

j Acknowledgement The study was made possible by a grant from the EU (QLG4-CT-2001-01476)

References

1. Agervold M (1998) Measuring and changing the psychosocial work environment: theoretical background for the development of The Psychosocial Work Environment and Stress Question- naire (in Danish) 1998. The Danish Work Environment Fund, Copenhagen

2. Agervold M (1998) The Psychosocial Work Environment and Stress Questionnaire (PWSQ) (in Danish). The Danish Work Environment Fund, Copenhagen

3. Breiman L, Friedman JH, Olshen RA, Stone CJ (1988) Classi- fication and regression trees. Wadsworth, California

4. Burnard P, Edwards D, Fothergill A, Hannigan B, Coyle D (2000) Community mental health nurses in Wales: self-reported stressors and coping strategies. J Psychiatric Ment Health Nurs 7:523–528

5. Butterworth T, Carson J, Jeacock J, et al. (1999) Stress, coping burnout and job satisfaction in British nurses; Findings from the Clinical Supervision Evaluation Project. Stress Med 15:27–33

6. Carson J, Maal S, Roche S, et al. (1999) Burnout in mental health nurses: Much ado about nothing? Stress Med 15:127–134

7. Cushway D, Tyler PA, Nolan P (1996) Development of a stress scale for mental health professionals. Br J Clin Psychol 35:279– 295

8. Cushway D, Tyler PA, Nolan P (1996) Development of a stress scale for mental health professionals. Br J Clin Psychol 35:279– 295

9. Dunn LA, Ritter SA. Stress in mental health nursing: a review of the literature. In Carson J, Fagin K, Ritter S (eds) Stress and Coping in Mental Health, Chapman & Hall, London, pp 29–45

10. Edwards D, Burnard P, Coyle D, Foterghill A, Hannigan B (2000) Stress and burnout in community mental health nursing: a review of the literature. J Psychiatr Ment Health Nurs 7:7–14

11. Edwards E, Burnard P, Coyle D, et al. (2001) A stepwise mul- tivariate analysis of factors that contribute to stress for mental health nurses working in the community. J Adv Nurs 36:805– 813

12. Galloway J (2002) Personal safety when visiting patients in the community. Adv Psychiatr Treat 8:214–222

13. Hopkinson PC, Carson J, Brown D, Fagin L, Bartlett H, Leary J (1998) occupational stress and community mental health nursing: what CPNs really said. J Adv Nurs 27:707–712

14. Jenkins R, Elliott P (2004) Stressors, burnout and social sup- port: nurses in acute mental health settings. J Adv Nurs 48:622– 631

15. Kilfedder CJ, Power KG, Wells TJ (2001) Burnout in psychiatric nursing. J Adv Nurs 34:383–396

16. Maslach C, Jackson SE, Leiter MP. Maslach Burnout Inventory manual 3rd edn. Consult. Psychol Press, Palo Alto, CA

17. Mehrotra S, Rao K, Subbakrishna DK (2000) Factor structure of the Mental Health Professionals Stess Scale (MHPSS) among clinical psychologists in India. Int J Soc Psychiatry 46:142–150

18. Nolan P, Dallender J, Soares J, et al. (1999) Violence in mental health care: the experience of mental health nurses and psy- chiatrists. J Adv Nurs 30:934–941

19. OSCAR—Occupational Stress with mental health Clients in Acute Response. Middlesex University, Archway Campus, Archway, London

20. Onyett S, Pillenger T, Muijen M (1997) Job satisfaction and burnout among members of community mental health teams. J Ment Health 6:55–66

21. Priebe S, Kafhoury WKH, Hoffmann K, Powell RA (2005) Morale and job perception of community mental health pro- fessionals in Berlin and London. Soc Psychiatry Psychiatr Ep- idemiol 40:223–232

22. Prosser D, Johnson S, Kuipers E, et al. (1996) Perceived sources of work stress and satisfaction among hospital and community mental health staff and their relationships to mental health, burnout and job satisfaction. J Psychosom Res 43:51–59

23. Prosser D, Johnson S, Kuipers E, Dunn G, Szmukler G, Reid Y, Bebbington P, Thornicroft G (1999) Mental health, burnout and job satisfaction in a longitudinal study of mental health staff. Soc Psychiatry Psychiatr Epidemiol 34:295–300

24. Reid Y, Johnson S, Morant N, Kuipers R, Szmukler G, Thorn- icroft T, Bebbington P, Prosser D (1999) Explanations for stress and satisfaction in mental health professionals: a qualitative study. Soc Psychiatry Psychiatr Epidemiol 34:301–308

25. Richardsen AM, Martinussen M (2004) The Maslach Burnout Inventory: Factorial validity and consistency across occupa- tional groups in Norway. J Occupat Org Psychol 77:377–384

26. Schutte N, Toppinen S, Kalimo R, Schaufeli W (2000) The factorial validity of the Maslach Burnout Inventory—Generaly Survey (MBI-GS) across occupational groups and nations. J Occupat Organizational Psychology 73:53–66

27. The Demographic Questionnaire—OSCAR (2002) Middlesex University, Archway Campus, London

802

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.