REVIEW
The Relationship Between Professional Burnout and Quality
and Safety in Healthcare: A Meta-Analysis
Michelle P. Salyers, Ph.D.
1,2
,KelseyA.Bonfils,M.S.
1,2
, Lauren Luther, M.S.
1,2
, Ruth L. Firmin, M.S.
1,2
,
Dominique A. White, M.S.
1,2
, Erin L. Adams, M.S.
1,2
, and Angela L. Rollins, Ph.D.
1,2,3
1
Department of Psychology, Indiana University-Purdue University Indianapolis, Indianapolis, IN, USA;
2
ACT Center of Indiana, Indianapolis, IN, USA;
3
VA HSR&D Center for Health Information and Communication, Richard L. Roudebush VAMC, Indianapolis, IN, USA.
BACKGROUND: Healthcare provider burnout is consid-
ered a factor in quality of care, yet little is known about the
consistency and magnitude of this relationship. This
meta-analysis examined relationships between provider
burnout (emotional exhaustion, depersonalization, and
reduced personal accomplishment) and the quality (per-
ceived quality, patient satisfaction) and safety of
healthcare.
METHODS: Publications were identified through targeted
literature searches in Ovid MEDLINE, PsycINFO, Web of
Science, CINAHL, and ProQuest Dissertations & Theses
through March of 2015. Two coders extracted data to
calculate effect sizes and potential moderators. We calcu-
lated Pearson’s r for all independent relationships be-
tween burnout and quality measures, using a random
effects model. Data were assessed for potential impact of
study rigor, outliers, and publication bias.
RESULTS: Eighty-two studies including 210,669
healthcare providers were included. Statistically signifi-
cant negative relationships emerged between burnout
and quality (r = −0.26, 95 % CI [−0.29, −0.23]) and safety
(r = −0.2 3, 95 % CI [−0.28, −0.17]). In both cases, the
negative relationship implied that greater burnout among
healthcare providers was associated with poorer-quality
healthcare and reduced safety for patients. Moderators
for the quality relationship included dimension of burn-
out, unit of analysis, and quality data source. Moderators
for the relationship between burnout and safety were
safety indicator type, population, and country. Rigor of
the study was not a significant moderator.
DISCUSSION: This is the first study to systematically,
quantitatively analyze the links between healthcare pro-
vider burnout and healthcare quality and safety across
disciplines. Provider burnout shows consistent negative
relationships with perceived quality (including patient
satisfaction), quality indicators, and perceptions of safety.
Though the effects are small to medium, the findings
highlight the importance of effective burnout interven-
tions for healthcare providers. Moderator analyses sug-
gest contextual factors to consider for future study.
KEY WORDS: burnout; quality of care; healthcare outcomes; meta-
analysis.
J Gen Intern Med 32(4):475–82
DOI: 10.1007/s11606-016-3886-9
© Society of General Internal Medicine 2016
INTRODUCTION
Professional burnout is characterized by high levels of emo-
tional exhaustion, cynical attitudes, and a diminished sense of
personal accomplishment at work.
1
A number of reviews have
highlighted the problematic nature of burnout for a variety of
healthcare professions, including medical residents,
2
nurses,
3
and mental health workers.
4, 5
Recent changes in healthcare
delivery have also raised concerns that provider burnout may
worsen if increased patient care and administrative demands
outpace resources.
6
Indeed, a recent national study found
significant increases in burnout among physicians compared
to the general population.
7
Burnout is associated with a number of problems, not only
for individual providers, but also for their employer organiza-
tions, patients, and the healthcare system as a whole. Workers
with burnout often experience physical health problems (e.g.,
insomnia, headaches, poor overall health), relationship prob-
lems, reduced job satisfaction, and increased mental health
problems (e.g., depression, anxiety, substance abuse).
8–15
Burn-
out has also been negatively associated with organizational
functioning, including excessive employee absences, tardiness,
frequent breaks, reduced job commitment, and in some studies,
poor job performance and increased turnover.
11, 13, 16, 17
Burnout can impact healthcare quality and safety in a num-
ber of ways. According to the job demands-resources model of
burnout,
18–21
job demands (e.g., interacting with patients with
intensive service needs, balancing competing priorities) re-
quire effort over time and can result in costs to the healthcare
provider (e.g., emotional exhaustion), particularly when re-
sources are low. As providers become exhausted, they with-
draw emotional energy from work, leading to depersonaliza-
tion. This conservation of resources
20
can also lead to pro-
viders spending less time with patients, and potentially be-
coming more directive than collaborative and patient-centered.
Furthermore, burnout has been associated with cognitive im-
pairments, including attention deficits,
22
which can lead to
Electronic supplementary material The online version of this article
(doi:10.1007/s11606-016-3886-9) contains supplementary material,
which is available to authorized users.
Received March 31, 2016
Revised September 2, 2016
Accepted September 23, 2016
Published online October 26, 2016
475
JGIM
errors. Williams and colleagues
23
describe a cyclical model
whereby burnout negatively affects the quality of the patient
encounter, leading to dissatisfied patients, poor adherence, and
worse health outcomes, which can cause additional provider
burnout.
While links between burnout and quality of care have been
theorized since the term burnout was introduced,
24
research
empirically linking burnout to quality of care has varied wide-
ly across healthcare specialties and types of quality domains
(e.g., patient satisfaction, self-reported quality, errors). Some
studies have reviewed aspects of these relationships, but none
has attempted a comprehensive, quantitative review across
disciplines and domains. For example, the relationship be-
tween burnout and job performance was summarized,
25
but
this meta-analysis included only 16 studies, was not restricted
to healthcare settings, and included limited measures of per-
formance. Within healthcare, Lee and colleagues
26
studied
correlates of burnout, but the sample was restricted to physi-
cians, and few studies assessed quality. Another review
27
took
a broader approach, including a variety of healthcare profes-
sionals, but restricted studies to hospital settings, and because
it was a narrative review, the authors were unable to quantify
relationships or assess other contributing variables.
The objective of the current study was to systematically
review and quantify empirical studies linking healthcare pro-
vider burnout to quality and safety in order to better under-
stand the magnitude and consistency of these relationships.
We explored potential moderators to examine whether the
relationships would vary as a function of the aspect of burnout
or quality being studied. Given the importance of multidisci-
plinary teams in patient-centered care,
28, 29
we included a
variety of healthcare providers, and explored possible differ-
ences among provider types (e.g., nurses vs. physicians) and
settings (inpatient vs. outpatient). We hypothesized that there
would be negative relationships between each aspect of burn-
out (emotional exhaustion, depersonalization, and reduced
personal accomplishment) and quality and safety. We further
hypothesized that the effects would be largest for provider
perceptions, compared to more objective indicators of quality
(e.g., observation or medical records). Other relationships
were considered exploratory.
METHODS
Data Sources and Searches
We followed guidelines provided by the Preferred Reporting
Items for Systematic Reviews and Meta-Analyses
(PRISMA).
30
A systematic literature search of Ovid
MEDLINE, PsycINFO, Web of Science, CINAHL, and
ProQuest Dissertations & Theses was conducted to identify
all studies involving health professionals, burnout, and quality
of care through March 2015. The full electronic search string
used for PsycINFO included the following: ((DE
Boccupational str ess^)AND(SUBburnout^)) AND ((DE
Bquality of care^)OR(DEBquality of services^)OR(DE
Bsatisfaction^)OR(DEBclient satisfaction^)OR(DE
Bsafety^)OR(DEBperceived quality^)). Similar search strat-
egies were used for the other databases. All search strategies
and the coding protocol are available from the authors.
Study Selection
We included published and unpublished studies of any design
(e.g., cross-sectional surveys, intervention studies), as long as
empirical data was used to assess the relationship between
burnout and quality (including patient satisfaction) and/or safe-
ty; if these variables were assessed but the bivariate association
was not reported, the authors were contacted to gather addition-
al data for analyses. Attempts were made to contact 63 authors,
of whom 21 provided usable data, 6 responded that data did not
meet inclusion criteria or could not be obtained, 3 could not be
located, and 33 did not respond. Although review articles were
not included in our analyses, we examined reference sections of
review articles to identify primary studies for inclusion.
We retained articles that specifically examined burnout. The
Maslach
31
three-dimensional scale of burnout (emotional ex-
haustion, depersonalization or cynicism, and reduced personal
accomplishment) was used most often, although any study
measuring at least one dimens ion of burnout or a global
burnout score was included. We focused on healthcare pro-
viders and excluded studies of burnout in other occupations
(e.g., education, probation officers, vocational rehabilitation).
We categorized quality of care along two dimensions: per-
ceived quality (rating scales or items reflecting provider’s
perception, patient satisfaction) and safety (perceived safety,
adverse events, Bnear misses,^ medical errors). Included stud-
ies are briefly summarized in Table 1 of the supplemental
online material.
Data Extraction and Quality Assessment
Articles were coded independently by a pair of coders (from a
group of six coders comprising a clinical psychologist and five
doctoral students). To maintain consistency and ensure reli-
ability, coders met to review and come to consensus for each
independent sample. We extracted information on burnout
type and measure(s) used, quality and safety indicators, pro-
vider type (nurses, physicians, interdisciplinary), setting (out-
patient, inpatient, or mixed inpatient/outpatient), and country
(coded by region: North America, South America, Europe,
Asia, Australia, Africa). Where available, we coded provider
characteristics (age, gender, experience/length of time in the
field) and patient characteristics (age, gender). We extracted
information on potential methods-related moderators includ-
ing study year, unit of analysis (individual, dyad, service unit,
hospital/organization), and quality or safety data source (pro-
vider, patient, observer, medical records).
We rated the quality of each study to account for bias in
individual studies (see Table 2 of the supplemental online
material). Because quality rubrics commonly recommended
476 Salyers et al.: Burnout, Quality of Care, and Safety JGIM
for meta-analyses
32, 33
include items not relevant for correla-
tional designs (e.g., blinding, allocation of intervention), we
created items based on common sources of bias in observa-
tional studies.
34, 35
We tested and refined the initial rating
system on several studies before rating the full sample. Two
raters independently coded each study; disagreements were
resolved through discussion. Measures of central tendency
highlighted the presence of eight items as a potentially valu-
able cutpoint (mean = 8.12, median and mode = 8). Following
other meta-analyses that examined subgroups based on quality
ratings,
36–38
we used quality rating as a moderator, examining
the effect sizes for those with high quality (8 or a bove)
compared to effect sizes of studies scoring below 8.
Data Synthesis and Analysis
We extracted effect size information at the level of burnout and
quality (or safety) relationship. All associations were first con-
verted into Pearson’s correlations; Fisher’s Z-transformation was
conducted to adjust for the non-normal distribution of Pearson’s
r. When a study reported multiple measures of the same con-
struct, we averaged the effect sizes and weighted them by
sample size in order to maintain statistical independence.
39
We
calculated an overall relationship, with one effect size per inde-
pendent sample, to describe the relationships between burnout
and perceived quality and safety. We conducted separate meta-
analyses to examine the relationships aggregated at the level of
predictor (burnout type) and aggregated at the level of the quality
indicator (perceived quality and safety). We conducted modera-
tor analyses for perceived quality and safety .
We used a random effects model to calculate the mean effect
sizes using Comprehensive Meta-Analysis (CMA) software,
version 2.
40
At the aggregate level, Z-scores and confidence
intervals were examined to determine the statistical signifi-
cance of each association. The strength of the mean effect sizes
were interpreted in light of Cohen’s
41
recommendation for
correlations, where 0.10 is small, 0.30 is medium, and 0.50
is large. We conducted one-study-removed sensitivity analy-
ses to determine whether any single sample unduly influenced
the results (indicating a potential outlier); because the point
estimate of the mean effect size did not change substantially
upon removal of any study, we performed the remainder of the
analyses with the full sample.
We examined heterogeneity with the Q-statistic and the I
2
index; a significant Q-statistic informs whether moderation
may be present, and the I
2
index informs the extent of the
heterogeneity, ranging from 0 to 100 %, with higher values
indicating greater heterogeneity.
42–44
Although I
2
is of value
in determining the need for moderation analyses, it does not
speak to the source of heterogeneity or dispersion of effects.
45
We used I
2
values of 25 % or more as a cutoff to examine the
presence of moderators, as this suggests that between-study
variability in effect size is greater than expected by chance.
43
To document dispersion of effects, we report 95 % confidence
intervals for each effect size.
We tested study-level moderators for both quality and safe-
ty, including year, type of report, provider type and setting,
region, and quality of study. Additional moderators for burn-
out and perceived quality included burnout type, quality
source, and unit of analysis. Additional moderator analysis
for safety compared perceived safety (e.g., questionnaires)
versus events (e.g., reported adverse events, near misses).
For categorical moderators, we used an analysis of variance
(ANOVA) analog. To test continuous moderator variables, we
conducted random effects meta-regressions using unrestricted
maximum likelihood estimation. Because meta-regressions
use list-wise deletion, each moderator was examined indepen-
dently to maximize the number of studies included in the
analysis. Continuous moderators were considered significant
if beta weights were significant and I
2
decreased. We
interpreted statistical tests at p < 0.05. All moderator analyses
were conducted in CMA, version 2.
40
Finally, we assessed the potential influence of publication
bias by examining funnel plots and testing for asymmetry
using Egger’s
46
regression approach. Although Egger’stest
may be prone to bias in low-power situations, our sample size
was well beyond the recommended minimum of ten sam-
ples.
47
In addition, Failsafe N was not appropriate because of
the high level of study heterogeneity and the random effects
model used.
39
RESULTS
The search yielded 1674 citations, resulting in 102 studies
with 82 unique samples of healthcare providers (see Fig. 1
for PRISMA flow diagram
30
). A summary description of the
included studies appears i n Table 1, and a detailed table
showing each study including individual effect sizes is pre-
sented in Table 1 of the supplemental online material. A total
of 210,669 healthcare providers were included, from 32 coun-
tries on 6 continents. The majority of studies measured at least
the emotional exhaustion component of burnout; about half
studied depersonalization and reduced personal accomplish-
ment. Some (19.5 %) included a total or global measure of
burnout. Most assessed perceived quality, and about half
assessed safety. Most studies took place in a setting that
included both inpatient and outpatient care. Nurses were the
most frequent target populations.
The meta-analysis of the relationship between burnout and
perceived quality including 63 independent samples resulted
in a significant negative relationship (r = −0.26), with 95 % CI
ranging from −0.29 to −0.23. The Q-statistic of the overall
effect was significant, with a large amount of heterogeneity
(I
2
= 93 %). A second meta-analysis was conducted between
burnout and safety, including 40 independent samples, which
also yielded a significant negative relationship (r = −0.23) and
95 % CI ranging from −0.28 to −0.17. The burnout–safety
relationship also demonstrated high levels of heterogeneity
(I
2
= 97 %). Forest plots for the meta-analyses of perceived
477Salyers et al.: Burnout, Quality of Care, and SafetyJGIM
quality and safety are shown in Tables 3 and 4 of the supple-
mental online material.
Moderator analyses for the relationship between burnout
and perceived quality are shown in Table 2. Three categorical
moderators were significant: type of burnout, unit of analysis,
and source of quality rating. The relationship between burnout
and perceived quality was strongest for emotional exhaustion
(r = −0.27) or overall burnout (r = −0.25), whereas effects for
depersonalization (r = −0.21) and reduced personal accom-
plishment (r = −0.20) were weaker but still significant. Effect
sizes were significantly stronger when examining individuals
(r = −0.27) compared to service units (r = −0.12). Effect sizes
were significantly stronger for provider report (r = −0.28)
compared with patient reports of quality (i.e., patient satisfac-
tion, r = −0.17). No continuous moderators were significant
predictors.
Moderators for the safety meta-analyses are shown in
Table 3. None of the continuous moderators were significant,
but three categorical moderators were identified: safety indi-
cator, study population, and the country in which the study
was conducted. Though both types of safety indicators were
statistically significant, burnout had a stronger relationship
with perceptions of safety (r = − 0.28) than events
(r = −0.16). In terms of discipline, the strongest relationship
wasfornurses(r = −0.27), followed by interdisciplinary sam-
ples (r = −0.24) and physicians (r = −0.15). Regarding coun-
try, effect sizes were stronger in studies from Eu rope
Fig. 1 PRISMA flow chart of article identification and exclusion.
Table 1 Summary of Study Characteristics Across Independent
Samples (k = 82)
Sample characteristics k %
Median year (range) 2010 (1982 to 2015)
Mean sample size (SD) 2569.1 (8021.0)
Median sample size (range) 454 (30 to 68,724)
Country grouping of study
Africa 1 1.2
Asia 5 6.1
Australia 2 2.4
Europe 31 37.8
North America 42 51.2
South America 1 1.2
Burnout type measured*
Emotional exhaustion 68 82.9
Depersonalization 40 48.8
Reduced personal accomplishment 37 45.1
Total/Global 16 19.5
Quality indicator*
Quality 63 76.8
Safety 40 8.8
Setting
Outpatient 15 18.3
Inpatient 21 25.6
Mixed 46 56.1
Population
Nurses 43 52.4
Physicians 21 25.6
Interdisciplinary sample 18 22.0
Mean provider age (k = 57) 39.1 (6.0)
Sex (mean % female; k = 60) 69.9 (25.0)
Education (mean % doctoral; k = 39) 53.6 (48.9)
Mean years in the field (k = 41) 11.0 (5.7)
*Percentages add to greater than 100 %, as some samples used more
than one burnout predictor and/or quality indicator
478 Salyers et al.: Burnout, Quality of Care, and Safety JGIM
(r = −0.36) than in those from North America (r = −0.18).
Discipline, however, was confounded with location. Most
European studies with safety outcomes consisted of nursing
samples (78 %), whereas only 26 % of North American studies
focused on nursing samples. Follow-up analyses to parse the
overlap in these results indicated that in North American
studies, nurses (r = −0.22) and physicians (r = −0.15) did not
differ in relationships between burnout and safety (Q = 0.59,
df =1,p = 0.44). There were not enough studies of physicians
within the European sample to conduct a parallel analysis.
We created a funnel plot, displaying overall study effect as a
function of precision (calculated as 1/SE; see Figures 1 and 2
in supplemental online material). Egger’s test of the intercept
was not significant for either perceived quality (t =1.13,p =
0.26) or safety (t =0.72,p = 0.48). As both meta-analyses had
substantially more than the recommended minimum number
of ten samples,
47, 48
the lack of significance suggests that
publication bias did not influence these findings.
DISCUSSION
Summarizing all available empirical literature on
healthcare provider burnout, we found small to
medium-sized relationships between burnout and both
decreased quality of care and decreased safety. For
perceived quality, the effect size of r = −0.26 translates
into approximately 7 % of variance accounted for by
provider burnout. For safety, the effect size of r = − 0.23
translates into approximately 5 % of the variance in
safety being attributable to provider burnout. These re-
lationships were robust to potential publication bias and
ratings of study rigor, which increases confidence in the
findings. Given the increasing rates of burnout, particu-
larly among physicians,
7
these findings could have im-
portant ramifications.
Of the burnout components, emotional exhaustion had the
strongest relationship with quality, followed by depersonali-
zation and reduced personal accomplishment. These findings
parallel a meta-analysis of burnout and objective job perfor-
mance in other service industries.
25
Together, these findings
suggest that emotional exhaustion may be the most critical
element of burnout to address. Earlier conceptualizations of
burnout also posit a primary role for emotional
exhaustion—that it may be the driving element that leads to
other aspects of burnout.
49
In terms of quality, burnout had a medium-sized rela-
tionship with lower perceived (provider-reported) quality
and a weaker, but still significant, relationship with
reduced patient satisfaction. This is important, given
the increasing role of satisfaction as a benchmark for
performance evaluations. Medicare payments are now
Table 2 Moderator Analyses of the Relationship Between Professional Burnout and Quality
Sample characteristics k* r 95 % CI Q
b
p Q
w
I
2
Burnout type 7.79 0.05
Emotional exhaustion 54 −0.27 −0.30 −0.24 648.13 92 %
Depersonalization 30 −0.21 −0.26 −0.17 551.96 95 %
Reduced pers. acc. 29 −0.20 −0.25 −0.15 484.68 94 %
Total/Global 11 −0.25 −0.32 −0.17 92.42 89 %
Unit of analysis 4.65 0.03
Individual 50 −0.27 −0.30 −0.24 854.69 94 %
Service unit 11 −0.12 −0.25 −0.01 8.32 0 %
Quality indicator source 4.67 0.03
Provider 49 −0.28 −0.31 −0.24 1148.33 96 %
Patient (satisfaction) 14 −0.17 −0.26 −0.07 33.34 61 %
Study-level moderators
Study quality rating 0.15 0.56
Lower-quality studies 29 −0.25 −0.29 −0.22 572.9 95 %
High-quality studies 34 −0.27 −0.31 −0.22 248.7 87 %
Population 3.35 0.19
Nurses 38 −0.27 −0.30 −0.23 755.01 95 %
Physicians 13 −0.27 −0.34 −0.20 71.77 83 %
Interdisciplinary sample 12 −0.18 −0.27 −0.09 27.79 60 %
Setting 1.65 0.44
Ou
tpatient 13 −0.21 −0.29 −0.13 92.57 87 %
Inpatient 13 −0.26 −0.34 −0.18 11.53 0 %
Mixed 37 −0.27 −0.30 −0.23 787.17 95 %
Country grouping 0.14 0.93
North America 30 −0.26 −0.31 −0.21 500.48 94 %
Europe 27 −0.26 −0.30 −0.21 217.00 88 %
Other 6 −0.28 −0.36 −0.18 79.57 94 %
Continuous study level moderators k B 95 % CI Z p Q
w
I
2
Year published 63 0 −0.01 0.01 0.24 0.81 58.52 0 %
Provider age 42 −0.01 −0.02 0 −1.82 0.07 39.20 0 %
Provider sex (percent female) 45 0 0 0 0.46 0.65 41.11 0 %
Time in the field 32 0 −0.01 −0.01 −0.20 0.84 30.55 1 %
*k = number of studies; r = correlation; 95 % CI = 95 % confidence interval; Q
b
= Q-statistic comparing heterogeneity between groups; Q
w
=Q-
statistic showing heterogeneity within groups; I
2
= degree of heterogeneity within the category; B = beta weight in regression for continuous
moderators; Z = test of significance of the predictor
479Salyers et al.: Burnout, Quality of Care, and SafetyJGIM