Content uploaded by Christos Lionis
Author content
All content in this area was uploaded by Christos Lionis
Content may be subject to copyright.
Content uploaded by Magdalena Esteva
Author content
All content in this area was uploaded by Magdalena Esteva
Content may be subject to copyright.
ÓThe Author 2008. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org.
doi:10.1093/fampra/cmn038 Family Practice Advance Access published on 11 July 2008
Burnout in European family doctors: the EGPRN study
Jean Karl Soler
a
, Hakan Yaman
b
, Magdalena Esteva
c
, Frank Dobbs
d
,
Radost Spiridonova Asenova
e
, Milica Katic
´
f
, Zlata Ozˇvac
ˇic
´
f
, Jean
Pierre Desgranges
g
, Alain Moreau
h
, Christos Lionis
i
,Pe
´ter Kota
´nyi
j
,
Francesco Carelli
k
, Pawel R.Nowak
l
, Zaida de Aguiar Sa
´Azeredo
m
,
Eva Marklund
n
, Dick Churchill
o
and Mehmet Ungan
p
(European
General Practice Research Network Burnout Study Group)
Soler JK, Yaman H, Esteva M, Dodds F, Spiridonova Asenova R, Katic´M,Oz
ˇvacˇic´ Z, Desgrange
JP, Moreau A, Lionis C, Kota´ nyi P, Carelli F, Nowak PR, de Aguiar Sa´ Azeredo Z, Marklund E,
Churchill D and Ungan M (European General Practice Research Network Burnout Study Group)
Burnout in European family doctors: the EGPRN study. Family Practice 2008; 25: 245–265.
Introduction. The aim of this study was to determine the prevalence of burnout, and of asso-
ciated factors, amongst family doctors (FDs) in European countries.
Methodology. A cross-sectional survey of FDs was conducted using a custom-designed and
validated questionnaire which incorporated the Maslach Burnout Inventory Human Services
Survey (MBI-HSS) as well as questions about demographic factors, working experience, health,
lifestyle and job satisfaction. MBI-HSS scores were analysed in the three dimensions of emo-
tional exhaustion (EE), depersonalization (DP) and personal accomplishment (PA).
Results. Almost 3500 questionnaires were distributed in 12 European countries, and 1393 were
returned to give a response rate of 41%. In terms of burnout, 43% of respondents scored high for
EE burnout, 35% for DP and 32% for PA, with 12% scoring high burnout in all three dimensions.
Just over one-third of doctors did not score high for burnout in any dimension. High burnout was
found to be strongly associated with several of the variables under study, especially those rela-
tive to respondents’ country of residence and European region, job satisfaction, intention to
change job, sick leave utilization, the (ab)use of alcohol, tobacco and psychotropic medication,
younger age and male sex.
Conclusions. Burnout seems to be a common problem in FDs across Europe and is associated
with personal and workload indicators, and especially job satisfaction, intention to change job
and the (ab)use of alcohol, tobacco and medication. The study questionnaire appears to be
a valid tool to measure burnout in FDs. Recommendations for employment conditions of FDs
and future research are made, and suggestions for improving the instrument are listed.
Keywords. Burnout, Europe, general practice, job satisfaction, job stress.
Introduction
Job-related ‘burn out’ or ‘burnout’ has been identified
as an occupational hazard for various professionals in-
volved in people-oriented services. Burnout is a syn-
drome, with reported symptoms including exhaustion,
frustration, anger, cynicism and a feeling of ineffec-
tiveness and/or failure. An important element of the
syndrome is a negative impact on job performance.
1,2
The predominant multidimensional model of burnout
defines it as a psychological response to chronic inter-
personal job stressors, characterized by overwhelming
Received 16 October 2006; Revised 5 June 2008; Accepted 6 June 2008.
a
Institute of Postgraduate Medicine and Primary Care, Faculty of Life and Health Sciences, University of Ulster, Coleraine BT52
1SA, UK (Maltese Co-ordinator),
b
Akdeniz University, Faculty of Medicine, Department of Family Medicine, Antalya, Turkey,
c
Research Unit, Majorca Primary Care Health District Department, Ib-salut, Spain (Spanish Co-ordinator),
d
Institute of Postgrad-
uate Medicine and Primary Care, Faculty of Life and Health Sciences, University of Ulster, Coleraine BT52 1SA, UK,
e
Plovdiv,
Bulgaria,
f
Zagreb, Croatia,
g
Brittany, France,
h
Lyons, France,
i
Crete, Greece,
j
Budapest, Hungary,
k
Milan, Italy,
l
Gdansk, Poland,
m
Portugal,
n
Sweden,
o
Nottingham, UK and
p
Turkey. Correspondence to Hakan Yaman, Faculty of Medicine, Department of Fam-
ily Medicine, Akdeniz University, 07059 Antalya, Turkey; Email: hakanyaman@akdeniz.edu.tr
245
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
exhaustion, feelings of cynicism and detachment from
the job.
1
Three dimensions of the syndrome are de-
scribed: emotional exhaustion (EE) is the depletion of
one’s emotional resources and reflects the basic stress
dimension of burnout; depersonalization (DP) usually
develops due to the effect of EE and exhibits features
of detachment and, eventually, dehumanization; and re-
duced personal accomplishment (PA) reflects reduced
feelings of competence and productivity at work, which
are linked to depression.
1,2
Factors related to burnout amongst professionals,
including doctors, include situational factors [organiza-
tional commitment and hierarchy, absence of job re-
sources (e.g. inadequate pay), overload, role conflict
and ambiguity, poor career progression and lack of
feedback] and individual characteristics (demographic
variables, personality characteristics, external locus of
control, job satisfaction, job withdrawal and lack of so-
cial support), with the effect of the situational factors
being stronger.
1,3,4
Although burnout has been described in health profes-
sionals
4–6
and has been reported to be common in family
doctors (FDs),
5,7–9
there are few published studies, and
there is an evident need for further research. This study
is an attempt to address this lacuna in the field of re-
search of burnout in FD populations in Europe.
The scale which has demonstrated the strongest psy-
chometric [Psychometrics: the field of study concerned
with the theory and technique of psychological mea-
surement, which includes the measurement of knowl-
edge, abilities, attitudes and personality traits. The
field is primarily concerned with the study of differ-
ences between individuals (http://en.wikipedia.org/
wiki/Psychometrics)] properties, has been shown to dis-
tinguish job-related neurasthenia from other mental
disorders
1
and continues to be used most widely by re-
searchers is the Maslach Burnout Inventory—Human
Services Survey (MBI-HSS) developed by Maslach and
Jackson in the early 1980s.
10
It comprises 22 seven-
point Likert-type questions on frequency of symptoms
(ranging from ‘0 = never’ to ‘6 = every day’). The three
dimensions are each measured by subscales: EE on
a subscale with nine items and a maximum score of 54,
DP on a five-item subscale with a maximum score of
30 and a decreased sense of PA (inverse scale, low
scores indicate high burnout) on a subscale with eight
items and a maximum score of 48.
10
The MBI-HSS
has demonstrated good external validity in FDs and
similar subscale correlations as in Maslach’s normative
sample.
11
The instrument of choice for this study of burnout is
consequently also a questionnaire. Thus, the MBI-HSS
was incorporated (unaltered) into a questionnaire de-
signed to measure other factors previously reported to
be associated with burnout in FDs.
8,12–16
The study in-
strument (Fig. 1) was based on a similar questionnaire
piloted previously in the population of interest.
17
This cross-sectional study was designed to address
two specific research questions.
Study research questions
1) What is the prevalence of burnout in European
FDs?
2) Which factors are associated with high levels of
burnout in European FDs?
Methodology
Study idea
This study was developed from an original idea to
study burnout in Turkish medical and education stu-
dents presented at a meeting of the World Organiza-
tion of Family Doctors (WONCA) European region
18
and was developed during various meetings of the Eu-
ropean General Practice Research Workshop
[EGPRW—since renamed ‘European General Prac-
tice Research Network’ (EGPRN)] from 1999 to
2002.
19,20
Twenty-four EGPRW members from 16 Eu-
ropean countries formed an interest group and worked
as a team to develop an instrument to collect data on
burnout and on factors associated with burnout.
Instrument
Questionnaire design process. A literature search was
performed to identify instruments and tools which mea-
sure burnout and to identify factors associated with
high levels of burnout. The questionnaire instrument
was developed on the basis of such literature. It was pi-
lot tested in 2000, the results being published in 2002.
17
Translation of the questionnaire instrument. In those
countries where the use of an English-language instru-
ment could potentially pose a language barrier, the
questionnaire was translated to the native language by
the key co-ordinating FD in that country, and the trans-
lation process was cross-checked by cross-translation in
most cases (see Table 1).
Questionnaire instrument structure. The final validated
questionnaire instrument (Fig. 1) comprised, in order:
a) a 25-item questionnaire including questions re-
garding age, gender, marital status, years since
qualification as a doctor, years in current work-
place, earning, workplace conditions (solo/
group, rural/urban setting), working conditions
(working hours per week, patients per week,
night shifts, weekends worked), intention of
changing job, sick leave utilization, sleep pat-
terns, alcohol consumption, smoking and psy-
choactive medication use;
b) a seven-point Likert-type job satisfaction ques-
tion and
c) the MBI-HSS.
Family Practice—an international journal246
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
TABLE 1Details of the study process in each participating country: co-ordinator, translation and mailing procedure, sample selection and response rates
Country Name of
the co-ordinator
Translation from
English original
version
Cross-
translation
Method of
distribution
Target
population
of GPs
Method of sample
selection
Number of
questionnaires
sent out
Date of
mailshot
Responses
n(%)
Bulgaria Radost Spiridonova
Asenova
Yes No Post 5222 Random selection 250 May to September,
2003
69 (27.6)
Croatia Milica Katic
´and
Zlata Oz
ˇvac
ˇic
´
Yes Yes Post 2400 Stratified random selection 350 November, 2003 117 (33.4)
France Alain Moreau Yes No Post 3113 Partly from college list,
partly random selection
324 June to September,
2003
178 (54.9)
Greece Christos Lionis Yes Yes Not specified 1103 Random selection 85 Late 2003 to early
2004
45 (52.9)
Hungary Pe
´ter Kota
´nyi Yes Yes Post 7000 Random selection 299 February, 2003 87 (29.1)
Italy Francesco Carelli Yes No E-mail Non-random selection from
scientific society and continuing
medical education (CME) events
230 not specified 147 (63.9)
Malta Jean Karl Soler No N/A Post 260 Sent to entire population 260 March, 2003 129 (49.6)
Poland Pawel R. Nowak Yes No Post 5500 Random selection 282 May to September,
2003
150 (53.2)
Spain Magdalena Esteva Yes Yes Post 15 474 Stratified random selection 286 April, 2003 86 (30.1)
Sweden Eva Marklund No N/A E-mail and Fax 5000 Non-random snowball sample >250 not specified 109 (<43.6)
Turkey Mehmet Ungan Yes Yes E-mail 500 Non-random selection from
Turkish college e-mail distribution list
500 not specified 112 (22.4)
England (UK) Dick Churchill No N/A Post 30 000 Stratified random sample by
region of all the GPs in England
300 April to June, 2003 164 (54.7)
Total 3416 1393 (40.8)
N/A, not applicable.
249The EGPRN Burnout study
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
Data collection procedure
Sampling procedure. The country co-ordinators, one
lead FD in each of 12 participating countries who was
selected from the burnout study interest group, were
tasked to send the questionnaire to a representative
sample of their country’s FDs.
Sample size. The original target was a sample size of
160 completed questionnaires from each country, cal-
culated to allow the discrimination of a population dif-
ference of 10% or more in the mean MBI-HSS
burnout scores in any one dimension between two
countries to an a-value of 0.05 with a power (b)of
80%, based on the variability of the data obtained
from the pilot study. The inclusion criteria were: prac-
tising FDs working in private, or state employment.
Retired FDs and those working less than 50% full
time in family practice were excluded.
Ethical approval
International ethical approval. Many, but not all,
country co-ordinators did apply for and obtain ethical
approval in their country.
Data entry
Data coding. Each country co-ordinator coded the
data from the returned questionnaires into a custom-
designed Microsoft Excel
21
spreadsheet template, and
these were then imported into SPSS version 11
22
by HY.
Missing values. Up to one missing response per di-
mension of burnout in the MBI-HSS instrument was
replaced with the average score of the rest of that re-
spondent’s responses for that dimension (rounded to
an integer value). If more than two responses were
missing for any one dimension, the score for that di-
mension was replaced with a ‘missing value’ code rec-
ognized as such by SPSS.
Coding of burnout outcome variables. MBI-HSS
scores were output in the three dimensions of burnout
and were then transformed into dummy categorical
variables for high, average and low burnout in the di-
mensions of EE, DP and PA as recommended by Mas-
lach using the cutoff values applicable for doctors, as
listed below (Maslach C, personal communication,
July 6, 2004).
10
However, the burnout outcome varia-
bles were re-coded into high and not-high (average or
low burnout) for the statistical analyses.
EE: low burnout <13, average burnout 14–26,
high burnout >27 (The scoring guide actually recom-
mends that average scores for EE range from 19 to 26.
Scores in the range from 14 to 18 are thus difficult to
classify. For the purposes of the description of rates of
burnout found in this study, EE scores in the range of
14 to 18 were classified as average, to avoid unclassified
cases. However, all the statistical analyses performed
on the data set used the outcome variable of high as
against not high burnout in the three dimensions.)
DP: low burnout <5, average burnout 6–9, high
burnout >10.
PA: high burnout <33, average burnout 34–39,
low burnout >40 (inverse scale).
Statistical analysis
Validation of questionnaire instrument in each
country. The MBI-HSS section of the questionnaire
was internally validated by calculating a Cronbach’s
alpha coefficient for each dimension [for the nine
questions which scored for EE, the five for DP and
eight for PA in turn against the respective total for
each dimension], for each translation of the question-
naire used in the 12 different countries.
Respondents. Descriptive statistics have been used to
tabulate the characteristics of the respondents as
TABLE 2Cronbach’s alpha coefficient of the MBI-HSS instrument scores for each dimension of burnout in each of the versions/translations of the
questionnaire used in the 12 countries
Country Alpha
for EE
Worst
item-total
correlation
Alpha
with worst
item deleted
Alpha
for DP
Worst
item-total
correlation
Alpha
with worst
item deleted
Alpha
for PA
Worst
item-total
correlation
Alpha
with worst
item deleted
MBI-HSS
language
England 0.91 0.62 (Q6) 0.91 0.78 0.37 (Q22) 0.80 0.74 0.25 (Q4) 0.75 Original English version
Malta 0.89 0.44 (Q14) 0.90 0.75 0.29 (Q22) 0.77 0.76 0.25 (Q4) 0.77 Original English version
Sweden 0.90 0.44 (Q6) 0.90 0.68 0.29 (Q22) 0.68 0.69 0.12 (Q18) 0.75 Original English version
Bulgaria 0.90 0.16 (Q14) 0.92 0.69 0.26 (Q15) 0.71 0.67 0.09 (Q4) 0.69 Bulgarian one-way translation
Croatia 0.92 0.53 (Q14) 0.92 0.73 0.27 (Q15) 0.75 0.77 0.20 (Q4) 0.78 Croatian one-way translation
Italy 0.88 0.46 (Q20) 0.88 0.73 0.33 (Q11) 0.74 0.83 0.30 (Q12) 0.84 Italian one-way translation
France 0.88 0.45 (Q14) 0.89 0.69 0.37 (Q22) 0.67 0.85 0.49 (Q4) 0.84 French one-way translation
Poland 0.92 0.60 (Q3) 0.91 0.73 0.39 (Q15) 0.71 0.76 0.21 (Q4) 0.76 Polish one-way translation
Greece 0.90 0.45 (Q1) 0.90 0.91 0.67 (Q5) 0.92 0.87 –0.07 (Q4) 0.91 Greek, cross-translation
Hungary 0.86 0.25 (Q14) 0.88 0.73 0.31 (Q22) 0.76 0.80 0.40 (Q4) 0.79 Hungarian, cross-translation
Spain 0.86 0.41 (Q3) 0.86 0.75 0.40 (Q22) 0.75 0.78 0.19 (Q4) 0.79 Spain, cross-translation
Turkey 0.87 0.33 (Q16) 0.89 0.46 –0.70 (Q15) 0.58 0.80 0.31 (Q21) 0.81 Turkish, cross-translation
For each dimension and for each country, the Cronbach’s alpha for the scores from the relevant set of MBI-HSS questions against the total score is
tabulated, followed by the worst individual question Cronbach’s alpha (question number in parenthesis) and the corrected Cronbach’s alpha if that
one question were to be deleted. The lowest item-total correlation is highlighted.
Family Practice—an international journal250
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
TABLE 3Respondent characteristics (n =1393) as measured by
the questionnaire instrument
Categorical
variables
Frequencies, n
(%, percentage valid)
a
Country
Bulgaria (Bu) 69 (5.0)
Croatia (Cr) 117 (8.4)
Hungary (Hu) 87 (6.2)
Italy (It) 147 (10.6)
France (Fr) 178 (12.8)
Malta (Mt) 129 (9.3)
Poland (po) 150 (10.8)
Spain (Sp) 86 (6.2)
Sweden (Sw) 109 (7.8)
Turkey (Tu) 112 (8.0)
England (UK) 164 (11.8)
Greece (Gr) 45 (3.2)
Total 1393 (100.0)
Gender
Male 758 (54.4, 54.6)
Female 630 (45.2, 45.4)
Missing 5 (0.4)
Marital status
Married 952 (68.3, 68.7)
Single 325 (23.3, 23.5)
Divorced–separated–
widowed
108 (7.8, 7.8)
Missing 8 (0.6)
Number of children less than
5 years old
0 1002 (71.9, 82.7)
1 170 (12.2, 14.0)
2 40 (2.9, 3.3)
Missing 181 (13.0)
Further qualifications
Yes 663 (47.6, 70.7)
No 275 (19.7, 29.3)
Missing 455 (32.7)
Type of work
State employed 717 (51.5, 51.9)
Private practice 607 (43.6, 43.9)
Education/academic 58 (4.2, 4.2)
Missing 11 (0.8)
All choices type of work
State employed only 571 (41.0, 41.3)
Private practice only 496 (35.6, 35.8)
Education/academic only 58 (4.2, 4.2)
State and private employment 77 (5.5, 5.6)
Private and education/
academic
49 (3.5, 3.5)
Private and in training 36 (2.6, 2.6)
State employed and in
training
32 (2.3, 2.3)
Other combinations not
otherwise specified
65 (4.7, 4.7)
Missing 9 (0.6)
Setting
Solo 715 (51.3, 53.1)
Group 631 (45.3, 46.9)
Missing 47 (3.4)
Type of practice
Rural 367 (26.3, 26.6)
Urban 643 (46.2, 46.6)
Mixed 371 (26.6, 26.9)
Missing 12 (0.9)
Hours of sleep
>8 hours 192 (13.8, 13.9)
<8 hours 1193 (85.6, 86.1)
TABLE 3Continued
Categorical
variables
Frequencies, n
(%, percentage valid)
a
Missing 8 (0.6)
Night visits
Yes 705 (50.6, 51.0)
No 676 (48.5, 49.0)
Missing 12 (0.9)
Weekend shifts
Yes 917 (65.8, 66.6)
No 460 (33.0, 33.4)
Missing 16 (1.1)
Sick leave last year
0 days 834 (59.9, 66.7)
1 or 2 days 130 (9.3, 10.4)
3 days or more 286 (20.5, 22.9)
Missing 143 (10.3)
Intention of changing job
Yes 450 (32.3, 32.4)
No 836 (60.0, 60.1)
Undecided 105 (7.5, 7.5)
Missing 2 (0.1)
Job satisfaction
1 (low) 74 (5.3, 5.5)
2 133 (9.5, 9.9)
3 307 (22.0, 22.9)
4 410 (29.4, 30.6)
5 293 (21.0, 21.9)
6 (high) 122 (8.8, 9.1)
Missing 54 (3.9)
Smoking
Yes 210 (14.4, 14.9)
No 1152 (82.7, 85.1)
Missing 40 (2.9)
Increasing smoking
Yes 64 (4.6, 33.2)
No 129 (9.3, 66.8)
Not applicable (non-smokers) 1152 (82.7)
Missing 48 (3.4)
Alcohol consumption
Yes 386 (27.7, 28.5)
No 969 (69.6, 71.5)
Missing 38 (2.7)
Increasing alcohol consumption
Yes 73 (5.2, 19.3)
No 305 (21.9, 80.7)
Not applicable (non-drinkers) 969 (69.6)
Missing 46 (3.3)
Psychotropic medication use
Yes 200 (14.4, 14.8)
No 1151 (82.6, 85.2)
Missing 42 (3.0)
North/South/East Europe
North (Sw, UK) 273 (19.6)
South (Cr, Gr, Tu, Sp,
Fr, It, Mt)
814 (58.4)
East (Hu, Bu, Po) 306 (22.0)
Continuous variables n(%) Mean SD
Age in years 1391 (99.9) 45.4 8.5
Number of children 1362 (97.8) 1.8 1.1
Years since graduation 1387 (99.6) 19.2 8.5
Years in current position 1365 (98.0) 12.2 9.0
Income difference/income
per capita
1313 (94.3) 0.09 1.1
Patients per week 1344 (96.5) 151 83
251The EGPRN Burnout study
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
measured by the questionnaire instrument, including
the MBI-HSS scores. Continuous variables which were
severely non-normal were re-coded as categorical
variables or transformed (income was re-coded as
a difference from average for that country, expressed
as a proportion of the Organization for Economic Co-
operation and Development (OECD) income per cap-
ita for that country).
Point prevalence of burnout—research question
1. Descriptive statistics have been used to present
the proportion of respondents who scored as high
burnout, average burnout or low burnout in each di-
mension, with the 95% confidence interval (CI).
Factors associated with high burnout—research
question 2. The associations between each of the
three principal yes/no outcome variables (high burn-
out present or not in each of the three dimensions of
EE, DP and PA) and each variable in the question-
naire were explored.
The statistical significance and strength of the asso-
ciations between the categorical variables and the
burnout outcome variables were analysed using Pear-
son’s chi-square test and Cramer’s V[Cramer’s Vis
a chi-square-based measure of the strength of associa-
tion, which can be used for nominal variables. Values
range from 0 (no association) to 1]. The statistical
analyses of the distributions of the continuous normal
variables within the two categories of the three burn-
out outcome variables (i.e. high burnout against not
high burnout in the three dimensions) were performed
using the independent samples t-test. Besides the one-
way analysis of associations between the independent
variables and the three burnout outcome variables as
above, a sensitivity analysis was performed to test the
associations in a two-way model with country as a con-
trol variable.
A multivariate analysis was subsequently performed
using SPSS complex samples logistic regression
analysis,
22
stratifying samples by country and including
as main effects all variables which reached an alevel
of at least 0.25 (P<0.25) in the one-way analysis, with
the same burnout outcome variables as the dependent
variables. The final model was the one which resulted
from a stepwise backward elimination process using
the maximum likelihood method. SPSS was used to
calculate odds ratios for all independent variables and
the 95% CI for the estimate and the a-value (cor-
rected for multiple comparisons using the sequential
Bonferroni method).
Results
Table 1 details the sample selection process and study
execution in the 12 countries. A total of 1393 com-
pleted and analysable questionnaires were returned
from more than 3416 sent (the reported number sent
out in Sweden was an estimate), giving a response rate
of approximately 41%.
Table 2 lists the Cronbach’s alpha coefficients of the
MBI-HSS instrument and its elements in each transla-
tion of the questionnaire. The worst correlation be-
tween a question and the total burnout score per
dimension is also tabulated, along with the Cronbach’s
alpha for the scale if that question were to be deleted.
The scores for all three dimensions in the various
translations range from good to excellent.
Table 3 gives the frequency distribution of the cate-
gorical variables and describes the distribution of con-
tinuous variables for the respondents. The 1393
respondents (758 males, 54.6%) had a mean age of
45.4 years (SD 8.5 years) and had graduated 19 years
previously to filling in the questionnaire (SD 8.5
years), worked 46 hours per week (SD 14 hours), saw
150 patients per week (SD 83 patients) and were
roughly evenly distributed amongst the 12 countries
(somewhat less respondents in Greece at 45, whilst
France and England were the only two countries that
achieved the target of 160 responses), with 58% of
respondents coming from the South of Europe.
Burnout point prevalence—research question 1
Table 4a lists the frequency distributions of respond-
ents by degree of burnout (high, average or low) in
the three dimensions (EE, DP, PA). Table 4b gives
the frequency distribution of respondents by presence
of high burnout scores in none (0), one or more of the
three dimensions (1, 2 or 3). For both tables, 95% CI
of the proportion is tabulated. In all, 43% of respond-
ents scored high for EE (95% CI = 40.5–45.6%),
35.3% for high DP (32.9–37.9%), 32.0% low for PA
(29.6–34.5%) and 12% of respondents (10.4–13.8%)
scored high for burnout in all three dimensions. Only
35.1% of doctors (32.6–37.7%) did not score high for
burnout in any dimension.
TABLE 3Continued
Continuous variables n(%) Mean SD
Hours per week 1360 (97.6) 45.9 14
Nights per month 1188 (85.3) 2.3 3.0
Weekends off per year 1119 (80.3) 32.2 17.4
Monthly earning (Euro) 1313 (94.3) 2100 3700
Sick leave days availed
of last year
1250 (89.7) 0 2
EE score 1392 (99.9) 24 16
DP score 1393 (100.0) 7 7
PA score 1391 (99.9) 37 11
For categorical variables, n, percentage of respondents and percent-
age of valid responses (ignoring missing values, in italics) are tabu-
lated; for continuous variables, n, mean and SD (or n, median and
IQR if the distribution is non-normal) are tabulated.
a
% valid when missing data excluded.
Family Practice—an international journal252
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
Figure 2 depicts the distribution of respondents by
percentage with a high burnout score in each of the
three dimensions and in all three dimensions by coun-
try and for all countries together (with error bars rep-
resenting 95% CI for proportions). Bulgarian, Italian
and English respondents demonstrated high propor-
tions of high EE burnout, Greek, Italian and English
respondents demonstrated high proportions of high
DP burnout, whilst Greek and Turkish respondents
demonstrated high proportions of high PA burnout.
Factors associated with high burnout—research
question 2
Tables 5a and 5b list the results of the analysis of asso-
ciation between the three dichotomous burnout out-
come variables (a categorical yes/no variable for high
burnout in each of the three dimensions of EE, DP
and PA) and the questionnaire categorical and contin-
uous variables, variable by variable, respectively. Also
listed are the results of the analysis of the association
between the categorical and continuous variables mea-
sured in the questionnaire and the three dichotomous
burnout outcome variables, all questionnaire re-
sponses from all countries being analysed together
(one way; except for the variable controlled for, i.e.
country). Finally, Tables 5a and 5b also list the results
of the analysis of association between the categorical
and continuous variables measured in the question-
naire and the three dichotomous burnout outcome
variables, variable by variable, controlling for country
(two way). The strongest associations include those
between the burnout outcome variables and country,
European region, job satisfaction and intention to
change job, the (ab)use of tobacco, alcohol and psy-
chotropic drugs, male sex, age, type of work and sick
leave utilization.
Table 6 lists the results of the logistic regression
analysis of the categorical and continuous variables
measured in the questionnaire controlling for country.
The odds ratios, 95% CI and P-values are presented.
The three models each correctly classify approxi-
mately three-quarters of all cases in the database. The
model for high EE burnout correctly classifies two-
thirds or respondents as having high burnout, whilst
the models for high DP and PA burnout perform con-
siderably better in predicting absence rather than pres-
ence of burnout. Again, the highest odds ratios were
found for job satisfaction and intention to change job,
the (ab)use of tobacco, alcohol and psychotropic
drugs, male sex, age, type of work and sick leave
utilization.
Discussion
In summary, this EGPRN study of burnout in FDs
from 12 European countries with a validated tool to
measure burnout achieved a response rate of 41%. In
all, 43% of respondents scored high for EE burnout,
35% scored high for DP burnout and 32% scored high
for PA burnout. Only 35% of respondents did not
score high for burnout in any dimension, whilst 21%
scored high for burnout in at least two dimensions
and 12% scored high for all three. There was a wide
variation in the proportions of respondents with high
burnout in the various countries. In the 12 countries,
between 15% and 68% of respondents scored high for
EE burnout, between 12% and 73% for DP burnout,
between 12% and 93% for PA burnout, and between
2% and 25% scored high for burnout in all three di-
mensions. FDs from Southern European countries
had significantly lower levels of EE burnout but higher
levels of PA burnout. After controlling for country,
low job satisfaction, expressed intention to change
job, (ab)use of alcohol, tobacco and psychotropic med-
ication, sick leave utilization, younger age, male sex
and type of work were associated with high burnout,
as previously reported.
1
Table 7 summarizes the comparisons between burn-
out scores and rates reported previously in the litera-
ture and the data from this EGPRN study.
8,9,17,23–27
Some earlier studies did report lower rates of burnout,
but a similar number of recent studies did report
TABLE 4a Frequency and cumulative frequency distributions of respondents by degree of burnout (high, average and low) with 95% CI in each ofthe
three dimensions
Burnout EE
(n= 1392)
% (95% CI) Cumulative
%
DP
(n= 1393)
%
(95% CI)
Cumulative
%
PA
(n= 1391)
%
(95% CI)
Cumulative
%
High 599 43.0 (40.5–45.6) 43.0 492 35.3 (32.9–37.9) 35.3 445 32.0 (29.6–34.5) 32.0
Medium 557 40.0 (37.5–42.6) 83.0 379 27.2 (24.9–29.6) 62.5 396 28.5 (26.2–30.9) 60.5
Low 236 17.0 (15.1–19.0) 100.0 522 37.5 (35.0–40.0) 100.0 550 39.5 (37.0–42.1) 100.0
TABLE 4b Frequency and cumulative frequency distributions of
respondents by high burnout score in none (0) or any one, any two
or all three dimensions (1, 2 or 3) with 95% CI
High burnout n= 1390 % (95% CI)
No dimension 488 35.1 (32.6–37.7)
One dimension 438 31.5 (29.1–34.0)
Two dimensions 297 21.4 (19.3–23.6)
All three dimensions 167 12.0 (10.4–13.8)
253The EGPRN Burnout study
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
similar data. As expected, high burnout was more likely
with low job satisfaction and intention to change
job.
1,5,9,28
Additionally,
1,3,4,8,28
high levels of burnout
were found to be more likely with certain organiza-
tional factors (country of origin, as surrogate for health
care system, and type of work) and increased sick leave
utilization, and less so with high workload (patients per
week and hours per week) and other job stressors
(working nights and weekends). Personal factors such
as younger age, sex, marital status and number of chil-
dren were also linked with burnout, but male sex more
strongly so.
1,28
Academic work type was linked to
lower EE, but higher PA burnout, as previously re-
ported.
1
Low self-esteem has been previously reported
to be associated with burnout.
1
In this study, we ob-
served that burnout was more likely with increasing
smoking, increased use of alcohol and use of psychotro-
pic medication, which may be manifestations of low
self-esteem.
9
Other variables, such as income, were sur-
prisingly rather weakly linked with high burnout, whilst
others (non-academic type of work, years since gradua-
tion, not having further qualifications, increasing smok-
ing) seemed to be linked with high EE burnout, but
make high PA burnout less likely; however, such am-
biguous findings have been previously described in
burnout research in doctors, for example, by Deckard
et al.
4
Generally, the pattern of associated variables
appears similar to that reported by Goehring et al.
8
for
those variables which were included in both studies.
The questionnaire was constructed with reference to
the current literature at the time and included those
variables that had been reported to be associated with,
or to cause, burnout. Nonetheless, the possibility exists
that other variables may have an important role to
play. For example, FDs may be more likely to suffer
burnout if they perceive that they have poor control
of their place of work,
28
but this variable was unfortu-
nately not included in our questionnaire. The rather
low response rate in many countries and the conse-
quent failure to achieve the target sample size weaken
the power of this study to answer the principal re-
search questions, especially the second one using the
regression models. The survey was administered dif-
ferently in the 12 countries, and also not concurrently.
This may have had a variable influence on the non-
response rate, potentially introducing bias.
The validity of the questionnaire has been tested in
a pilot study, and the results have been published sep-
arately by Yaman and Soler.
17
In the pilot study, fac-
tor analysis confirmed that the three constructs of EE,
DP and PA are distinct and identified which items
loaded to which dimension. The results were consis-
tent with Maslach’s scoring key except for item 16.
10
The Cronbach’s alpha for EE, DP and PA in the pilot
study was all high, at 0.67, 0.66 and 0.70, respec-
tively.
17
The MBI-HSS section of the questionnaire
has again been validated in the main study using Cron-
bach’s alpha analysis (Table 2). The Cronbach’s alpha
coefficients for each dimension in each country were,
in fact, higher than those found in the pilot study, for
most countries (i.e. 0.9 for EE, 0.5 to 0.9 for DP and
0.7 to 0.9 for PA). In general, deleting questions would
not improve the internal consistency, with few excep-
tions. The validity and reliability of the questionnaire
appear to be high, based on the analyses and compari-
sons performed, and the good consistency of the re-
sults was obtained between countries and between the
pilot and main studies.
The sample size calculation, in retrospect, should
have been more precise since the calculated sample
FIGURE 2Distribution of the proportions of respondents by high burnout score in each of the three dimensions and in all three
dimensions in one respondent, by country and for all countries together (with error bars representing 95% CIs for proportions)
Family Practice—an international journal
254
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
TABLE 5a Associations between categorical variables in the questionnaire and the three burnout outcome variables (high burnout in each of the three dimensions) explored using Pearson
chi-square (or Somer’s d) and Cramer’s V test (or Spearman correlation) as appropriate
Categorical variables EE high burnout, n(%, 99% CI) DP high burnout, n(%, 99% CI) PA high burnout, n(%, 99% CI)
Country
Overall (95% CI) 599 (43.0, 40.5–45.6) 492 (35.3, 32.9–37.9) 445 (32.0, 29.6–34.5)
Bulgaria (Bu; 95% CI) 43 (62.3, 50.5–72.8) 21 (30.4, 20.8–42.1) 13 (18.8, 14.4–29.6)
Croatia (Cr; 95% CI) 49 (41.9, 33.3–50.9) 14 (12.0, 7.3–19.1) 16 (13.7, 8.6–21.1)
France (Fr; 95% CI) 60 (33.7, 27.2–40.9) 63 (35.4, 28.7–42.7) 49 (27.5, 21.5–34.5)
Greece (Gr; 95% CI) 14 (31.8, 20.0–46.6) 33 (73.3, 59.0–84.0) 41 (93.2, 81.8–97.7)
Hungary (Hu; 95% CI) 32 (36.8, 27.4–47.3) 31 (35.6, 26.4–46.1) 23 (26.4, 18.3–36.6)
Italy (It; 95% CI) 100 (68.0, 60.1–75.0) 81 (55.1, 47.0–62.9) 60 (40.8, 33.2–48.9)
Malta (Mt; 95% CI) 47 (36.4, 28.6–45.0) 40 (31.0, 23.7–39.4) 32 (24.8, 18.2–32.9)
Poland (Po; 95% CI) 72 (48.0, 40.2–55.9) 51 (34.0, 26.9–41.9) 45 (30.0, 23.2–37.8)
Spain (Sp; 95% CI) 26 (30.2, 21.5–40.6) 30 (34.9, 25.7–45.4) 22 (25.6, 17.5–35.7)
Sweden (Sw; 95% CI) 50 (45.9, 36.8–55.2) 38 (34.9, 26.6–44.2) 13 (11.9, 7.1–19.3)
Turkey (Tu; 95% CI) 17 (15.2, 9.7–23.0) 17 (15.2, 9.7–23.0) 77 (69.4, 60.3–77.2)
England (UK; 95% CI) 89 (54.3, 46.6–61.7) 73 (44.5, 37.1–52.2) 54 (32.9, 26.2–40.4)
P-value (one way) <0.0005** <0.0005** <0.0005**
Cramer’s V(one way) 0.283 0.280 0.383
Controlling for country (two way) Greater proportion in Bu,
It and UK; smaller in, Fr, Sp and Tu
Greater proportion in Greece,
It and UK; smaller in Croatia and Tu
Greater proportion in Greece, It and Tu;
smaller in Bu, Croatia and Sw
North/South/East European
North countries 139 (50.9, 43.2–58.6) 111 (40.7, 33.3–48.5) 67 (24.5, 18.5–31.8)
South countries 313 (38.5, 34.2–43.0) 278 (34.2, 30.0–38.6) 297 (36.6, 32.3–41.0)
East countries 147 (48.0, 40.8–55.4) 103 (33.7, 27.1–40.9) 81 (26.5, 20.5–33.4)
P-value <0.0005** 0.199 <0.0005**
Cramer’s V0.110 0.117
One-way effect Smaller proportion in South None statistically significant Greater proportion in South
Sex
Male 337 (44.5, 39.9–49.2) 316 (41.7, 37.2–46.4) 256 (33.8, 29.5–38.4)
Female 260 (41.3, 36.3–46.4) 175 (27.8, 23.4–32.6) 185 (29.4, 25.0–34.3)
P-value 0.224 <0.00005** 0.080
Cramer’s V0.145
One-way effect None statistically significant Greater proportion in males None statistically significant
Controlling for country
(significant two-way effects)
It greater if male Bu, Tu and UK, as above None statistically significant
Marital status
Married 434 (45.6, 41.5–49.8) 343 (36.0, 32.1–40.1) 271 (28.5, 24.9–32.4)
Single 103 (31.7, 25.5–38.7) 104 (32.0, 25.7–39.0) 136 (42.0, 35.1–49.1)
Divorced–separated 58 (53.7, 41.5–65.5) 42 (38.9, 27.8–51.3) 37 (34.3, 23.7–46.6)
P-value <0.0005** 0.304 <0.0005**
Cramer’s V0.134 0.121
One-way effect Smaller proportion in singles None statistically significant Greater proportion in singles
Controlling for country
(significant two-way effects)
Tu, as above Tu greater in divorced–separated Fr, Hu and Tu, as above
Children under 5
0 427 (42.7, 38.7–46.7) 336 (33.5, 29.8–37.5) 292 (29.1, 25.6–33.0)
1 64 (37.6, 28.7–47.5) 52 (30.6, 22.3–40.3) 65 (38.5, 29.4–48.4)
2 15 (37.5, 20.9–57.7) 13 (32.5, 17.1–52.9) 11 (27.5, 13.5–47.9)
255The EGPRN Burnout study
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
TABLE 5a Continued
Categorical variables EE high burnout, n(%, 99% CI) DP high burnout, n(%, 99% CI) PA high burnout, n(%, 99% CI)
P-value 0.404 0.750 0.047*
Cramer’s V0.071
One-way effect None statistically significant None statistically significant Greater proportion if only one child
Controlling for country
(significant two-way effects)
It greater if none It greater if none; Tu if 2 It greater if 1; Sp if 2
Further qualifications
Yes 299 (45.1, 40.2–50.1) 241 (36.3, 31.7–41.3) 208 (31.4, 26.9–36.2)
No 151 (54.9, 47.1–62.4) 93 (33.8, 26.9–41.5) 64 (23.3, 17.4–30.4)
P-value 0.006** 0.461 0.013*
Cramer’s V0.089 0.081
One-way effect Greater proportion if no None statistically significant Greater proportion if yes
Controlling for country
(significant two-way effects)
None statistically significant None statistically significant Po greater if no
Type of work (first choice)
Public 292 (40.8, 36.1–45.6) 269 (37.5, 33.0–42.3) 250 (34.9, 30.5–39.6)
Private 289 (47.6, 42.4–52.8) 212 (34.9, 30.1–40.1) 149 (24.6, 20.4–29.4)
Academic 13 (22.4, 11.6–38.9) 8 (13.8, 5.8–29.2) 40 (69.0, 52.0–82.0)
P-value <0.0005** 0.001** <0.0005**
Cramer’s V0.110 0.098 0.199
One-way effect Smaller proportion if academic Smaller proportion if academic Smaller proportion if private; greater
proportion if academic
Controlling for country
(significant two-way effects)
It greater if private It greater if private; Mt and Tu if public It and Mt greater if public
Type of work all choices
State employed only 240 (42.1, 36.9–47.5) 216 (37.8, 32.8–43.2) 211 (37.0, 32.0–42.4)
Private only 249 (50.2, 44.4–56.0) 179 (36.1, 30.7–41.8) 131 (26.5, 21.7–31.9)
Academic only 13 (22.4, 11.6–38.9) 8 (13.8, 5.8–29.2) 40 (69.0, 52.0–82.0)
Private and state employed 35 (45.5, 31.8–59.9) 26 (33.8, 21.7–48.5) 22 (28.6, 17.4–43.1)
Private and academic 13 (26.5, 13.8–44.9) 14 (28.6, 15.3–45.7) 4 (8.2, 2.5–23.9)
Private and in training 16 (44.4, 25.7–65.0) 7 (19.4, 7.9–40.6) 8 (22.2, 9.6–43.5)
State employed and in train 5 (15.6, 5.4–37.7) 14 (43.8, 24.2–65.4) 6 (18.8, 7.1–41.2)
Others 24 (36.9, 23.4–52.9) 27 (41.5, 27.3–57.4) 19 (29.2, 17.2–45.2)
P-value <0.0005** 0.005** <0.0005**
Cramer’s V0.164 0.122 0.221
One-way effect Greater proportion if private only;
smaller proportion if academic only
or state employed and in training
Smaller proportion if academic only Greater proportion if public only or
academic only; smaller proportion if
private only or private and academic
Controlling for country
(significant two-way effects)
It greater if private only It greater if private only; Tu if public only;
Mt if works both public and private
It greater if public only; Mt if public only
or both private and public
Setting
Solo 312 (43.7, 39.0–48.5) 238 (33.3, 28.9–38.0) 222 (31.1, 26.8–35.7)
Group 271 (42.9, 38.0–48.1) 236 (37.4, 32.6–42.5) 188 (29.8, 25.3–34.7)
P-value 0.782 0.115 0.606
One-way effect None statistically significant None statistically significant None statistically significant
Controlling for country
(significant two-way effects)
It greater if solo Tu greater if group None
Family Practice—an international journal256
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
TABLE 5a Continued
Categorical variables EE high burnout, n(%, 99% CI) DP high burnout, n(%, 99% CI) PA high burnout, n(%, 99% CI)
Type of practice
Rural 176 (48.0, 41.3–54.7) 126 (34.3, 28.3–41.0) 117 (32.0, 26.0–38.5)
Urban 258 (40.1, 35.3–45.2) 240 (37.3, 32.6–42.4) 208 (32.4, 27.8–37.3)
Mixed 157 (42.4, 36.0–49.1) 124 (33.4, 27.4–40.0) 115 (31.0, 25.2–37.5)
P-value 0.053 0.396 0.899
One-way effect None statistically significant None statistically significant None statistically significant
Controlling for country
(significant two-way effects)
Bu and It greater if rural It greater if rural; Sw smaller if urban It, Sp and Tu greater if rural;
Gr smaller if rural
Hours of sleep
>8 71 (37.0, 28.6–46.3) 57 (29.7, 22.0–38.8) 61 (31.8, 23.8–40.9)
<8 527 (44.2, 40.5–47.9) 433 (36.3, 32.8–40.0) 380 (31.9, 28.5–35.5)
P-value 0.060 0.076 0.970
One-way effect None statistically significant None statistically significant None statistically significant
Controlling for country
(significant two-way effects)
None statistically significant Tu greater if >8 None statistically significant
Night visits
Yes 305 (43.3, 38.5–48.1) 265 (37.6, 33.0–42.4) 216 (30.7, 26.4–35.3)
No 291 (43.0, 38.2–48.0) 219 (32.4, 27.9–37.2) 219 (32.4, 27.9–37.2)
P-value 0.936 0.043* 0.493
Cramer’s V0.054
One-way effect None statistically significant Greater proportion if yes None statistically significant
Controlling for country
(significant two-way effects)
Tu greater if yes; UK if no None None statistically significant
Weekend shifts
Yes 396 (43.2, 39.0–47.4) 340 (37.1, 33.1–41.3) 263 (28.7, 25.0–32.7)
No 199 (43.3, 37.4–49.3) 141 (30.7, 25.4–36.4) 172 (37.4, 31.8–43.4)
P-value 0.978 0.018* 0.001**
Cramer’s V0.064 0.088
One-way effect None statistically significant Greater proportion if yes Greater proportion if no
Controlling for country
(significant two-way effects)
Tu greater if yes Tu, as above It and Sp greater if no; Tu if yes
Sick leave
0 days 316 (37.9, 33.7–42.3) 261 (31.3, 27.35–35.6) 249 (29.9, 26.0–34.1)
1–2 days 68 (52.3, 41.2–63.2) 54 (41.5, 31.1–52.8) 37 (28.5, 19.5–39.5)
3 days or more 143 (50.2, 42.6–57.7) 114 (39.9, 32.7–47.5) 111 (38.9, 31.8–46.6)
P-value <0.0005** 0.006** 0.012*
Cramer’s V0.124 0.091 0.084
One-way effect Smaller proportion if 0 days Smaller proportion if 0 days Greater proportion if >3 days
Controlling for country
(significant two-way effects)
It greater if 0; Fr, Tu, UK smaller if 0 It greater if 0 Fr and It, as above
Changing job
Yes 298 (66.4, 60.45–71.8) 212 (47.1, 41.1–53.2) 190 (42.3, 36.5–48.4)
No 248 (29.7, 25.8–33.9) 244 (29.2, 25.3–33.4) 222 (26.6, 22.8–30.7)
Undecided 51 (48.6, 36.5–60.9) 35 (33.3, 22.8–45.9) 33 (31.4, 21.1–43.9)
P-value <0.0005** <0.0005** <0.0005**
Cramer’s V0.341 0.172 0.155
257The EGPRN Burnout study
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
TABLE 5a Continued
Categorical variables EE high burnout, n(%, 99% CI) DP high burnout, n(%, 99% CI) PA high burnout, n(%, 99% CI)
One-way effect Greater proportion if yes; smaller
proportion if no
Greater proportion if yes; smaller
proportion if no
Greater proportion if yes, smaller
proportion if no
Controlling for country
(significant two-way effects)
All except Gr, as above Bu, Fr, Hu, It, Mt, Sw and UK,
as above
Fr, Hu, Tu and UK, as above
Satisfaction
High (5–6) 78 (18.8, 14.4–24.2) 97 (23.4, 18.5–29.1) 66 (15.9, 11.8–21.1)
Moderate (3–4) 359 (50.1, 45.3–54.9) 277 (38.6, 34.1–43.4) 255 (35.7, 31.2–40.4)
Low (0–2) 127 (61.4, 52.4–69.6) 98 (47.3, 38.6–56.2) 98 (47.3, 38.6–56.2)
P-value <0.0005** <0.0005** <0.0005**
Cramer’s V0.326 0.178 0.240
One-way effect Greater proportion if low or moderate,
smaller proportion if high
Greater proportion if low, smaller
proportion if high
Greater proportion if low, smaller
proportion if high
Controlling for country
(significant two-way effects)
All, as above Fr, It, Mt, Po, Sw, Tu and UK,
as above
Hu, It, Mt, Po, Sw, Tu and UK,
as above
Smoking
Yes 90 (44.8, 36.0–53.8) 76 (37.8, 29.5–46.9) 79 (39.3, 30.9–48.4)
No 497 (43.1, 39.4–46.9) 387 (33.6, 30.1–37.3) 332 (28.8, 25.5–32.4)
P-value 0.666 0.245 0.003**
Cramer’s V0.081
One-way effect None statistically significant None statistically significant Greater proportion if yes
Controlling for country
(significant two-way effects)
None statistically significant None statistically significant It, as above
Increasing smoking
Yes 41 (64.1, 48.0–77.5) 31 (48.4, 33.2–63.9) 22 (34.4, 21.2–50.5)
No 45 (34.9, 25.0–46.2) 41 (31.8, 22.3–43.0) 50 (38.8, 28.5–50.1)
P-value 0.001** 0.043* 0.029*
Cramer’s V0.106 0.068 0.063
One-way effect Greater proportion if yes Greater proportion if yes Greater proportion if no
Controlling for country
(significant two-way effects)
It and Sp, as above None statistically significant It and Tu, as above
Alcohol
Yes 196 (50.8, 44.3–57.3) 166 (43.0, 36.7–49.6) 111 (28.8, 23.2–35.0)
No 394 (40.7, 36.7–44.8) 299 (30.9, 27.2–34.8) 300 (31.0, 27.3–34.9)
P-value 0.001** <0.0005** 0.419
Cramer’s V0.092 0.115
One-way effect Greater proportion if yes Greater proportion if yes None statistically significant
Controlling for country
(significant two-way effects)
Hu and Mt greater if no; It if yes Tu, as above None statistically significant
Increasing alcohol
Yes 58 (79.5, 65.1–88.9) 46 (63.0, 47.9–75.9) 23 (31.5, 19.5–46.6)
No 133 (43.6, 36.5–51.0) 114 (37.4, 30.6–44.7) 82 (26.9, 20.9–33.9)
P-value <0.0005** <0.0005** 0.380
Cramer’s V0.176 0.157
One-way effect Greater proportion if yes Greater proportion if yes None statistically significant
Controlling for country
(significant two-way effects)
Hu, Fr, It, Mt, Tu and UK, as above Fr as above None statistically significant
Family Practice—an international journal258
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
would only have achieved enough power to measure
the projected differences if the rate of burnout was con-
siderably lower. Additionally, the effect of non-response
was not fully considered. The responses from the vari-
ous European countries were pooled together in the
one-way analyses of association; if the variability of
burnout rates within the countries is less than that be-
tween countries, then an a-value of 0.05 may be too
large to exclude associations due to chance.
29
Multicollinearity is present when the independent
variables in a multiple regression equation are highly
correlated, and this causes a lack of precision of the re-
gression coefficient estimates (here expressed as odds
ratios in Table 6). During the logistic regression analysis,
interfactor correlation was examined, and in fact, there
was only one correlation greater than 0.6, that between
age and years since graduation (0.9, data not tabulated).
This is the first reported study investigating the
prevalence of burnout in an European FD workforce,
designed to investigate the factors associated with high
burnout. The limitations of this study include the fact
that it is cross-sectional, that it has not been conducted
concurrently in all countries, that the cultural and lin-
guistic equivalence of the concept of burnout and the
MBI-HSS instrument itself have not been fully investi-
gated in Europe and that the study involved FDs in
various European countries and working in different
health care systems without measuring the complexity
of this environment. The response rate was moderate,
but it is quite possible that non-respondents might
have scored differently to respondents to the MBI-
HSS. However, similar response rates are common in
anonymous questionnaire studies. The burnout scores
found appeared comparable or high with respect to
earlier studies. However, the Italian respondents in
this study, which response rate was the highest in this
study, scored very high for burnout, suggesting that
non-responders may also have high levels of burnout.
Conclusions
Burnout seems to be a common problem in FDs
across Europe, with high levels apparently affecting
two-thirds of respondents in this study. In all, 41% of
respondents reported high levels of EE, 35% DP and
32% low feelings of PA. There is considerable varia-
tion between countries, with doctors from Southern
European countries reporting lower rates of EE but
also lower feelings of PA.
High burnout was found to be more likely in associ-
ation with several of the variables under study, espe-
cially those relative to respondents’ country of
residence and European region, job satisfaction, inten-
tion to change job, sick leave utilization, the (ab)use of
alcohol, tobacco and psychotropic medication, youn-
ger age and male sex.
TABLE 5a Continued
Categorical variables EE high burnout, n(%, 99% CI) DP high burnout, n(%, 99% CI) PA high burnout, n(%, 99% CI)
Psychotropic drugs
Yes 130 (65.0, 55.9–73.1) 89 (44.5, 35.8–53.6) 76 (38.0, 29.7–47.1)
No 457 (39.7, 36.1–43.5) 373 (32.4, 29.0–36.1) 331 (28.8, 25.5–32.3)
P-value <0.0005** 0.001** 0.009**
Cramer’s V0.181 0.091 0.071
Controlling for country
(significant two-way effects)
Greater proportion if yes; Fr, It, Po, Sw,
Tu and UK, as above
Greater proportion if yes; Fr,
as above
Greater proportion if yes Tu,
as above
N, proportion (%) of respondents with high burnout and 99% CI (to correct for multiple comparisons; except for the control variable ‘country’) in brackets, P-value for the statistical significance of
a one-way association, Cramer’s Vfor significant one-way associations, text description of statistically significant one-way trends and text description of associations found to be statistically significant
(P<0.05) when controlling for country (where applicable) are tabulated.
*Significant to an a-value of 0.05.
**Significant to an a-value of 0.01. Recommended minimum alevel for multiple comparisons.
259The EGPRN Burnout study
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
TABLE 5b The associations between continuous variables and the three burnout outcome variables (high burnout in each of the three dimensions)
explored using the independent samples t-test, one-way ANOVA or two-way ANOVA F-tests, as appropriate
Continuous variables EE burnout mean (99% CI) DP burnout mean (99% CI) PA burnout mean (99% CI)
Age
BO high 46.1 (45.3–46.9) 45.4 (44.5–46.4) 44.7 (43.7–45.8)
BO not high 44.9 (44.1–45.8) 45.4 (44.7–46.2) 45.8 (45.1–46.5)
P-value 0.011* 0.953 0.030*
One-way effect Greater mean with high BO None statistically significant Smaller mean with high BO
Controlling for country (two way)
P-value for interaction <0.0005** 0.002** n.s.
P-value for main effect n.s. 0.001** n.s.
Effect Varies Mostly smaller mean with high BO
Number of children
BO high 1.9 (1.8–2.0) 1.9 (1.7–2.0) 1.6 (1.5–1.8)
BO not high 1.8 (1.7–1.9) 1.8 (1.7–1.9) 1.9 (1.8–2.0)
P-value 0.106 0.495 <0.0005**
One-way effect None statistically significant None statistically significant Smaller mean with high BO
Controlling for country (two way)
P-value for interaction n.s. n.s. n.s.
P-value for main effect n.s. n.s. n.s.
Years since graduation
BO high 19.9 (19.0–20.7) 19.1 (18.2–20.1) 18.7 (17.7–19.8)
BO not high 18.8 (18.0–19.6) 19.3 (18.6–20.1) 19.5 (18.8–20.2)
P-value 0.016* 0.681 0.108
One-way effect Greater mean with high BO None statistically significant None statistically significant
Controlling for country (two way)
P-value for interaction n.s. 0.002** n.s.
P-value for main effect n.s. 0.001** n.s.
Effect Mostly smaller mean with high BO
Years in current job
BO high 12.6 (11.7–13.5) 12.8 (11.8–13.8) 11.7 (10.6–12.9)
BO not high 11.9 (11.1–12.8) 11.9 (11.1–12.7) 12.5 (11.7–13.2)
P-value 0.182 0.087 0.164
One-way effect None statistically significant None statistically significant None statistically significant
Controlling for country (two way)
P-value for interaction n.s. n.s. n.s.
P-value for main effect n.s. n.s. n.s.
Patients per week
BO high 160 (151–169) 150 (141–159) 148 (138–159)
BO not high 146 (138–154) 153 (145–160) 153 (146–160)
P-value 0.003* 0.605 0.330
One-way effect Greater mean with high BO None statistically significant None statistically significant
Controlling for country (two way)
P-value for interaction 0.0005** 0.0005** n.s.
P-value for main effect 0.0005** 0.017* n.s.
Effect Mostly greater mean with high BO Varies
Hours per week
BO high 47.5 (45.9–49.0) 47.2 (45.7–48.7) 44.7 (42.9–46.5)
BO not high 44.8 (43.5–46.0) 45.2 (44.0–46.5) 46.5 (45.3–47.7)
P-value 0.001** 0.013* 0.029*
One-way effect Greater mean with high BO Greater mean with high BO Smaller mean with high BO
Controlling for country (two way)
P-value for interaction 0.009** n.s. n.s.
P-value for main effect 0.0005** n.s. n.s.
Effect Mostly greater mean with high BO
Nights per month
BO high 2.4 (2.0–2.8) 2.8 (2.4–3.2) 2.6 (2.1–3.1)
BO not high 2.2 (1.9–2.5) 2.0 (1.7–2.3) 2.1 (1.9–2.4)
P-value 0.191 <0.0005** 0.019*
One-way effect None statistically significant Greater mean with high BO Greater mean with high BO
Controlling for country (two way)
P-value for interaction n.s. 0.004** n.s.
P-value for main effect n.s. 0.001** n.s.
Effect Mostly greater mean with high BO
Weekends off per annum
BO high 33 (31–35) 32.3 (30.2–34.4) 31.3 (28.9–33.9)
BO not high 32 (30–33) 32.1 (30.3–33.8) 32.5 (31.0–34.0)
P-value 0.165 0.804 0.293
Family Practice—an international journal260
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
Future research is needed to explore the problem in
depth, develop models to describe the phenomenon
and to identify causative factors and effective inter-
vention strategies. Job satisfaction is an important ele-
ment in such research, and it should be prioritized by
EGPRN and WONCA Europe as an action point for
research and intervention.
Future research
Recent research work on burnout aims to develop new
theoretical frameworks that explicitly integrate both
individual and situational factors, using a model of
job–person fit. Maslach and Leiter
30
address the chal-
lenge by formulating a model that focuses on the de-
gree of match or mismatch between the person and
six domains of the job environment, namely workload,
control, reward, community, fairness and values. Re-
search has indicated that the greater the mismatch,
the greater the potential for burnout.
Future research into the phenomenon should ad-
dress these factors when studying burnout in FDs, and
the focus should be on positive rather than negative
states, dealing with job engagement and satisfaction
and not just job stress.
1
In this regard, the strong rela-
tionships found in this study between low job satisfac-
tion and burnout support the notion of focusing
future research on improving job satisfaction rather
than addressing burnout directly.
Surprisingly, little research has been conducted into
interventions for burnout. Although research indicates
that it is the organizational attributes that seem to have
stronger associations with burnout, most interventions
have in the past ironically been centered on changing
individuals.
1
Various intervention strategies have been
studied, some focusing on prevention of burnout and
others on treatment when it has already occurred, and
results have been varied.
1
This is another important
area where levels of knowledge should improve. A con-
trolled trial of organizational interventions for FDs to
improve job satisfaction should be considered.
Acknowledgements
We would like to thank Prof. Dr Henk Lamberts and
Dr John Beasley for their invaluable advice on the
analysis of data and presentation of the results. Author
JKS is very grateful to Prof. Isabel Stabile for her in-
valuable assistance in obtaining ethical approval for this
study in Malta. Author HY has been supported by The
Akdeniz University Research Foundation. Distribu-
tion of work: JKS led the project, developed the re-
search idea and study methodology, presented the
research protocol and design at EGPRN meetings, re-
cruited the country co-ordinators, designed the ques-
tionnaire, co-ordinated the pilot and main studies,
provided data, analysed the data and wrote the manu-
script. HY originated and developed the research idea,
designed the questionnaire, analysed the data for the
pilot and main studies and participated in writing the
manuscript. ME provided data, participated in the de-
sign and translation of the questionnaire, analysed the
data and participated in writing the manuscript. FD
developed the research idea and methodology, advised
on data analysis and participated in writing the manu-
script. In the EGPRN Burnout Study Group, DC
TABLE 5b Continued
Continuous variables EE burnout mean (99% CI) DP burnout mean (99% CI) PA burnout mean (99% CI)
One-way effect None statistically significant None statistically significant None statistically significant
Controlling for country (two way)
P-value for interaction 0.0005** n.s. 0.0005**
P-value for main effect n.s. n.s. n.s.
Effect Varies Varies
Income—median income/IPC
BO high 13.5% (2.1–24.8%) 13.1% (1.4–24.8%) 10.7% (0–25.7%)
BO not high 5.2% (0–12.7%) 6.3% (0–15.9%) 8.1% (0–16.4%)
P-value 0.155 0.262 0.666
One-way effect None statistically significant None statistically significant None statistically significant
Controlling for country (two way)
P-value for interaction 0.0005** n.s. 0.010*
P-value for main effect 0.002** 0.04* n.s.
Effect Varies Greater mean with high BO Varies
n.s., not significant; ANOVA, analysis of variance. The mean, 99% CI (to correct for multiple comparisons) and P-value of the test of difference in
means for the distribution of each variable in each of the categories of the three main outcome variables are tabulated. The two-way ANOVA F-test
with Bonferroni correction analysis is summarized textually as such: P-value for the interaction effect of country and burnout category if significant
(a<0.05), P-value for the main effect of burnout high/not high if significant (a<0.05) and a text description of the observed graphical trend for
different countries.
*Significant to an a-value of 0.05.
**Significant to an a-value of 0.01. Recommended minimum alevel for multiple comparisons.
261The EGPRN Burnout study
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
TABLE 6Logistic regression analysis, controlling for country
Variable EE high burnout,
odds ratio
(95% CI)
DP high burnout,
odds ratio
(95% CI)
PA high burnout,
odds ratio
(95% CI)
Intercept –4.32 –2.99 –0.19
P-value <0.0005** <0.0005** <0.0005**
Sex
Female 1.00 reference 1.00 reference 1.00 reference
Male 1.05 (1.01–1.10) 1.83 (1.77–1.89) 1.71 (1.63–1.79)
P-value 0.031* <0.0005** <0.0005**
Marital status
Single 1.00 reference 1.00 reference
Married 1.08 (1.04–1.12) 0.79 (0.75–0.83)
Divorced–separated 1.10 (0.99–1.22) 0.97 (0.85–1.10)
P-value 0.179 0.003**
Children under 5
0 1.00 reference
1 0.86 (0.81–0.92)
2 0.72 (0.56–0.92)
P-value 0.019*
Further qualifications
No 1.00 reference 1.00 reference
Yes 0.60 (0.58–0.62) 1.58 (1.52–1.64)
P-value <0.0005** <0.0005**
Type of work (first choice)
Public 1.00 reference 1.00 reference 1.00 reference
Private 1.51 (1.43–1.59) 1.51 (1.49–1.63) 0.84 (0.79–0.89)
Academic 0.07 (0.07–0.08) 0.43 (0.41–0.44) 1.94 (1.82–2.08)
P-value <0.0005** <0.0005** <0.0005**
Setting
Solo 1.00 reference
Group 1.34 (1.28–1.41)
P-value <0.0005**
Type of practice
Urban 1.00 reference
Rural 1.19 (1.12–1.27)
Mixed 0.68 (0.64–0.73)
P-value <0.0005**
Hours of sleep
<8 1.00 reference 1.00 reference
>8 1.15 (1.08–1.22) 1.02 (0.96–1.08)
P-value <0.0005** 0.501
Night visits
No 1.00 reference
Yes 0.85 (0.81–0.90)
P-value <0.0005**
Weekend shifts
No 1.00 reference 1.00 reference
Yes 1.33 (1.27–1.40) 1.01 (0.95–1.08)
P-value <0.0005** 0.697
Sick leave
0 days 1.00 reference 1.00 reference 1.00 reference
1–2 days 2.20 (2.06–2.35) 1.54 (1.48–1.60) 1.44 (1.35–1.55)
>3 days 1.02 (0.96–1.08) 1.21 (1.17–1.26) 1.50 (1.42–1.60)
P-value <0.0005** <0.0005** <0.0005**
Changing job
No 1.00 reference 1.00 reference 1.00 reference
Yes 3.36 (3.14–3.59) 1.70 (1.63–1.77) 1.08 (1.00–1.16)
Undecided 1.39 (1.31–1.47) 1.03 (0.99–1.08) 1.16 (1.08–1.24)
P-value <0.0005** <0.0005** <0.0005**
Satisfaction
High (5–6) 1.00 reference 1.00 reference 1.00 reference
Low (0–2) 22.8 (21.1–24.7) 3.66 (3.48–3.86) 5.08 (4.54–5.67)
Moderate (3–4) 3.85 (3.63–4.10) 2.34 (2.22–2.47) 3.89 (3.47–4.29)
P-value <0.0005** <0.0005** <0.0005**
Increasing smoking
No 1.00 reference 1.00 reference 1.00 reference
Yes 2.89 (2.61–3.20) 2.10 (1.95–2.27) 0.56 (0.49–0.64)
Family Practice—an international journal262
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
TABLE 6Continued
Variable EE high burnout,
odds ratio
(95% CI)
DP high burnout,
odds ratio
(95% CI)
PA high burnout,
odds ratio
(95% CI)
Non-smoker
a
1.45 (1.37–1.54) 1.00 (0.94–1.06) 0.50 (0.46–0.54)
P-value <0.0005** <0.0005** <0.0005**
Increasing alcohol
No 1.00 reference 1.00 reference
Yes 2.94 (2.27–3.81) 2.09 (1.88–2.32)
Non-drinker
a
0.81 (0.75–0.88) 0.88 (0.83–0.93)
P-value <0.0005** <0.0005**
Psychotropic drugs
No 1.00 reference 1.00 reference 1.00 reference
Yes 3.10 (2.83–3.39) 1.35 (1.28–1.42) 1.41 (1.30–1.52)
P-value <0.0005** <0.0005** <0.0005**
Age
Per 10 years 0.59 (0.54–0.66) 0.84 (0.76–0.93)
P-value <0.0005** 0.001**
Number of children
Per child 0.96 (0.93–0.99) 0.91 (0.88–0.94)
P-value 0.002** <0.0005**
Years since graduation
Per 10 years 2.34 (2.13–2.68) 0.89 (0.80–1.00)
P-value <0.0005** 0.048*
Years in current job
Per 10 years 0.73 (0.70–0.75) 1.07 (1.05–1.10) 1.24 (1.20–1.29)
P-value <0.0005** <0.0005** <0.0005**
Patients per week
Per 10 patients 1.02 (1.02–1.02)
P-value <0.0005**
Hours per week
Per 5 hours 1.05 (1.04–1.06) 0.97 (0.97–0.98) 0.92 (0.91–0.93)
P-value <0.0005** <0.0005** <0.0005**
Nights per month
Per night 1.01 (1.00–1.02) 1.07 (1.06–1.07) 0.99 (0.98–1.00)
P-value 0.011* <0.0005** 0.003**
Weekends off per annum
Per 12 weekends 1.03 (1.02–1.05)
P-value <0.0005**
Income—median income/
IPC
Per 10% change 1.00 (1.00–1.01)
P-value 0.066
Sample
Cases (n) 681 1040 693
Excluded (n) 712 353 700
Model characteristics
Pseudo –2 log likelihood 713.8 1194.0 734.8
Nagelkerke pseudo R
2
0.397 0.174 0.181
Classification
Correctly classified
high BO (%)
66.4 30.6 23.1
Correctly classified not
high BO (%)
78.0 90.5 94.9
Correctly classified
overall (%)
72.6 70.9 75.4
The three burnout outcome variables (high burnout in each of the three dimensions) as the dependent variable in each model, with selected ques-
tionnaire variables (a<0.25 in one-way analysis) as independent variables. Odds ratios, 95% CI and P-values (corrected for multiple comparisons
using the sequential Bonferroni method) sample and model characteristics and classification.
a
Information from another variable (tobacco or alcohol consumption) re-coded into this variable.
*Significant to an a-value of 0.05.
**Significant to an a-value of 0.01.
263The EGPRN Burnout study
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
TABLE 7Descriptive analysis of previously published studies of burnout in FDs or primary care doctors compared with the EGPRN study
Population Authors and year Burnout rates Comparable EGPRN study data Comparison Limitations of comparison
Switzerland, Swiss
primary care doctors
Goehring et al.
8
published 2005
High EE burnout in 19%;
high DP burnout in 22%;
high PA burnout in 16%
High EE burnout in 43.0%
(95% CI = 40.5–45.6%),
high DP burnout in 35.3%
(32.9–37.9%), high PA
burnout in 32.0% (29.6–34.5%).
Lower rates of burnout
reported
Only a third of the
respondents in this study
were FDs. There was no
Swiss arm in the EGPRN
study. Comparisons of Swiss
data are against EGPRN
global study rates.
France, French FDs Cathebras et al.
23
published 2004
5% scored high in all
three dimensions
10% of French respondents
scored high in all three
dimensions (95% CI 6–15%)
Lower rate of burnout
reported
Spain, Spanish
primary care doctors
Prieto Albino et al.
24
published 2002
66% scored high in at
least one dimension
55% of Spanish respondents
scored high in at least one
dimension (95% CI 44–65%)
Comparable rate of
burnout reported
Not all respondents were FDs
Spain, Spanish FDs
and paediatricians
Esteva et al.
25
published 2006
High EE burnout in 53%;
high DP burnout in 47%;
high PA burnout in 33%
30% of Spanish respondents
scored high for EE burnout
(95% CI 22–40%); 35% for
DP (26–45%); 26% for PA
(18–36%)
Higher rate of EE burnout and
DP burnout and comparable
rate of PA burnout reported
Not all respondents were FDs
Italy, Italian FDs Grassi and Magnani
26
published 2000
High EE burnout in 32%;
high DP burnout in 27%;
high PA burnout in 13%
68% of Italian respondents
scored high for EE burnout
(95% CI 60–75%); 55%
for DP (47–63%); 41% for
PA (33–49%)
Lower rates of burnout reported
Canada, Canadian FDs Thommasen et al.
9
published 2001
Moderate to high EE
burnout in 80%; moderate
to high DP burnout in 61%;
moderate to high PA
burnout in 44%
83.0% of all respondents
scored moderate to high for
EE burnout (95% CI
81.0–84.9%); 62.5% for
DP (60.0–65.0%); 60.5%
for PA burnout (57.9–63.0%)
EE and DP rates comparable,
lower PA burnout reported
Comparison of European
and Canadian data
Britain, British FDs Kirwan and Armstrong
27
published 1995
Mean score of 26.1 for EE;
9.8 for DP; 36.2 for PA
Mean scores for English
respondents were 27.2 for
EE; 9.5 for DP; 36.3 for PA
Comparable scores reported Study in mid-1990s
Europe, European FDs Yaman and Soler
17
published 2002
Mean score of 25.1 for EE;
7.3 for DP; 34.5 for PA
Mean scores for respondents
were 25.2 for EE; 8.2 for
DP; 36.7 for PA
Comparable scores reported
Family Practice—an international journal264
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
provided data, participated in the design of the ques-
tionnaire and participated in writing the manuscript;
CL provided data, participated in the design and
translation of the questionnaire and participated in
writing the manuscript; RSA, Z de ASA, FC, JPD,
MK, PK, ZO, AM, EM, PRN and MU provided data
and participated in the design and translation of the
questionnaire.
Declaration
Funding: During the course of the main study, finan-
cial support to total Euro 8000 was requested and re-
ceived jointly from European General Practice
Research Network and WONCA Europe for the costs
incurred by country co-ordinators, and to support the
protected time of the study co-ordinators and authors.
Ethical approval: Obtained in most countries individu-
ally, and at the University of Malta by the first author
for the whole study.
Conflicts of interest: None.
References
1
Maslach C, Schaufeli WB, Leiter MP. Job burnout. Annu Rev Psy-
chol. 2001; 52: 397–422.
2
Maslach C, Goldberg J. Prevention of burnout: new perspectives.
Appl Prev Psychol. 1998; 7: 63–74.
3
Beemsterboer J, Baum BH. ‘‘Burnout’’ definitions and health care
management. Soc Work Health Care. 1984; 10: 96–109.
4
Deckard GJ, Metarko M, Field D. Physician burnout: an examina-
tion of personal, professional and organisational relationships.
Med Care. 1994; 32: 745–754.
5
Deckard GJ, Hicks LL, Hamory BH. The occurrence and distribu-
tion of burnout amongst infectious diseases physicians. J Infect
Dis. 1992; 165: 224–228.
6
Gundersen L. Physician burnout. Ann Intern Med. 2001; 135:
145–148.
7
Chambers R, Wall D, Campell I. Stress, coping mechanisms and
job satisfaction in general practitioner registrars. Br J Gen
Pract. 1996; 46: 343–348.
8
Goehring C, Bouvier Gallacchi M, Ku
¨nzi B, Bovier P. Psycholog-
ical and professional characteristics of burnout in Swiss pri-
mary care practitioners: a cross-sectional survey. Swiss Med
Wkly. 2005; 135: 101–108.
9
Thommasen HV, Lavanchy M, Connelly I, Berkowitz J, Grzybow-
ski S. Mental health, job satisfaction and intention to relocate.
Can Fam Physician 2001; B47: 737–744.
10
Maslach C, Jackson SE. Maslach Burnout Inventory. 2nd edn. Palo
Alto, CA: Consulting Psychologists Press, 1986.
11
Rafferty JP, Lemkau JP, Purdy RR, Rudisill JR. Validity of the
Maslach Burnout Inventory for family practice physicians.
J Clin Psychol. 1986; 42: 488–493.
12
Howie JGR, Hopton JL, Heaney DJ. Attitudes to medical care,
the organization of work, and stress among general practi-
tioners. Br J Gen Pract. 1992; 42: 181–185.
13
McManus IC, Winder BC, Gordon D. The causal links between
stress and burnout in a longitudinal study of UK doctors. Lan-
cet 2002; 359: 2089–2090.
14
Smith R. Why are doctors so unhappy? BMJ 2001; 322: 1073–1074.
15
Rowe MM. Hardiness, stress, temperament, coping and burnout in
health professionals. Am J Health Behav. 1997; 21: 163–171.
16
Yaman H, Ungan M. Burnout in young doctors. [Turkish]. Tu
¨rk
Psikoloji Derg. 2002; 17 (49): 37–44.
17
Yaman H, Soler JK. The job-related burnout questionnaire in fam-
ily practice: a multinational pilot study. Aust Fam Physician.
2002; 31: 1055–1056.
18
European Society of General Practice and Family Medicine
(Wonca Europe). http://www.woncaeurope.org/ (accessed on
January, 2007).
19
European General Practice Research Network. http://www.egprn.
org. (accessed on January 2007).
20
Yaman H, Ungan M, Soler JK. Frequency of job-related burnout
in family physicians working in family practice: proposal for
a multinational study of European countries. Paper Presented
at the EGPRW Meeting in October 2001 in Gdynia, Poland.
www.egprn.org/gdyn2t.htm.
21
Microsoft Corporation. Excel for Windows Version 6. Redmond,
WA: Microsoft Corporation. 2000.
22
SPSS Inc. SPSS for Windows Version 11 and Version 13. Chicago,
IL: SPSS Inc, 2004.
23
Cathebras P, Begon A, Laporte S, Bois C, Truchot D. Burn out
among French general practitioners. Presse Med. 2004; 33:
1569–1574.
24
Prieto Albino L, Robles Aguero E, Salazar Martinez LM, Daniel
Vega E. Burnout in primary care doctors of the province of Ca-
ceres. Aten Primaria. 2002; 29: 294–302.
25
Esteva M, Larraz C, Jimenez F. La salud mental en los me
`dicos de
familia: efectos de la satisfaccio
`n y el estre
`s en el trabajo. Rev
Clin Esp. 2006; 206: 77–83.
26
Grassi L, Magnani K. Psychiatric morbidity and burnout in the
medical profession: an Italian study of general practitioners
and hospital physicians. Psychother Psychosom. 2000; 69:
329–334.
27
Kirwan M, Armstrong D. Investigation of burnout in a sample of
British general practitioners. Br J Gen Pract. 1995; 45: 259–260.
28
Linzer M, Visser MRM, Oort FJ, et al. Predicting and preventing
physician burnout: results from the United States and the Neth-
erlands. Am J Med. 2001; 111: 170–175.
29
Altman DG, Machin D, Bryant TN, Gardner MJ. Statistics with
Confidence. 2nd edn. London: BMJ Books, 2000.
30
Maslach C, Leiter MP. The EGPRN burnout study. San Francisco,
CA: Jossey-Bass 1997.
265The EGPRN Burnout study
at University of Crete on April 23, 2013http://fampra.oxfordjournals.org/Downloaded from
