Skip to main content

Table 4 Relationships between well-being and patient safety

From: Integrating teamwork, clinician occupational well-being and patient safety – development of a conceptual framework based on a systematic review

Study Topic Primary topic Sample & setting Design & data collection methods Assessment of variables Analyses Findings Outcomes & effect sizes Quality scored
Arakawa et al., 2011 [98] Relationships between nurses’ work, health, and lifestyle characteristics and medical errors and incidents yes 6445 nurses, 99 hospitals, Japan Cross sectional self-report questionnaire Well-being: SF-36 scales mental health and vitalitya
Patient safety: Number of incidents and errors during the previous 6 months
Logistic regression No association between
1) mental health
2) vitality
and medical errors and incidents
1) NS
2) NS
9 (16)
Arimura et al., 2010 [99] Relationships between work characteristics, sleepiness, mental health state and self-reported medical errors yes 454 nurses, 2 general hospitals, Japan Cross sectional self- report questionnaire Well-being: GHQ-28a, daytime sleepiness (Epworth sleepiness scale)
Patient safety: medical errors during past month
Multiple logistic regression 1) Poorer mental health is associated with higher occurrence of medical errors
2) Daytime sleepiness is not associated with higher occurrence of medical errors
1) OR = 1.1, p < 0.05, 95 % CI 1.0 – 1.1
2) NS
(8 predictors altogether)
105 (16)
Chen et al., 2013 [114] Relationships between burnout, job satisfaction and medical malpractice yes 809 physicians, Taiwan Cross-sectional self-report questionnaire Well-being: MBIa
Patient safety: experiences of medical malpractice
Univariate logistic regression 1) Emotional exhaustion is associated with higher risk of medical malpractice, whereas
2) depersonalization and
3) personal accomplishment are associated with lower risk of medical malpractice
1) OR = 1.50, 95 % CI 0.68 –1.95
2) OR = 0.74, 95 % CI 0.40 –1.36
3) OR = 0.76, 95 % CI 0.07 –1.05
6 (16)
Cimiotti et al., 2012 [104] Relationships between nurse staffing, burnout, and hospital infections yes 7076 nurses, 161 hospitals, USA Cross-sectional self-report questionnaire,
hospital records
Well-being: MBIa
Patient safety: catheter-associated urinary tract & surgical site infections
Linear regression Burnout is positively associated
1) catheter-associated urinary tract and
2) surgical site infections
1) β = 0.82, p < .05
2) β = 1.56, p < .01
10.5 (16)
Fahrenkopf et al., 2008 [106] Relationships between depression, burnout, and medication errors yes 123 residents, 3 pediatric hospitals, USA Cross-sectional self-report questionnaire, record review Well-being: MBIa
Patient safety: medical errors (self-report & chart reviews)
Cluster adj. Poisson analysis,
Fisher’s exact test
1) Burnt out residents perceive their number of errors to be higher than residents who are not burnt out
2) Burnt out residents are more likely to attribute errors to sleep deprivation
3) No significant differences in error rates detected in chart reviews between both groups
1) Mhigh burnout = 2.3, Mlow burnout = 1.0, p = 0.002
2) 29 % vs. 10 %, p = 0.05
3) NS
[]b
8 (16)
Garrouste-Orgeas et al., 2015 [116] Relationships between medical errors, burnout, depression, and safety culture yes 1534 nurses, physicians, & other healthcare staff, 31 ICUs, France Cross-sectional self-report questionnaire,
hospital records and observations
Well-being: MBIa
Patient safety: Medical error
Negative binomial regression Burnout is not associated with medical error NS 10.5 (15)
Halbesleben et al., 2008 [22] Relationships between nurse burnout and patient safety perceptions/reporting behavior yes 148 nurses, 1 hospital, USA Cross sectional self- report questionnaire Well-being: Emotional Exhaustion and Depersonalizationa
Patient safety: AHRQ Patient Safety Culture Surveya & frequency of incident reports
Multiple linear regression 1) Emotional exhaustion and depersonalization predict patient safety dimensions
a) safety grade
b) safety perception
c) near-miss reporting frequency
2a) Emotional exhaustion and b) depersonalization do not predict patient safety dimension event reports
1a) βexhaustion = −0.40, p < 0.01,
βdepersonlization = −0.16, p < 0.05, R2 = 0.22, [f2 = 0.28]b,c
1b) βexhaustion = −0.84, p < 0.001, βdepersonlization = −0.26, p < 0.05, R2 = 0.36, [f2 = 0.56]b, c
1c) βexhaustion = −0.14, p < 0.05,
βdepersonlization = −0.36, p < 0.01, R2 = 0.18, [f2 = 0.22]b, c
2a) NS
2b) NS
13.5 (16)
Halbesleben & Rathert, 2008 [107] Relationship between physician burnout and patient satisfaction and patient recovery time after hospital discharge yes 178 patient and physician dyads, 1 hospital, USA Cross-sectional self- report questionnaire Well-being: MBIa, patients’ perception of physician depersonlization
Patient safety: recovery time: 1-item patient self-report
Path analysis,
Pearson’s correlation
1) Good overall model fit
2) Positive correlation between patient recovery time and a) depersonalization
b) but not emotional exhaustion
c) or personal accomplishment
3) Positive correlation between patients’ perception of physician depersonalization and recovery time
4) No correlation between physician emotional exhaustion and recovery time
1) GFI = 0.99, CFI = 1.00, NNFI = 1.02, AIC = −2.98, BIC = −8.45, RMSEA = 0.00
2a) r = 0.44, p < 0.05
2b) NS
2c) NS
3) r = 0.32, p < 0.05
4) NS
12 (16)
Hayashino et al., 2012 [108] Hope moderates relationship between distress and medical errors yes 836 physicians, Japan Longitudinal self-report questionnaire Well-being: MBIa (time 1)
Medical errors: self-report (time 2)
Poisson regression High scores in
1) emotional exhaustion
2) depersonalization
and low scores in
3) personal accomplishment
at time 1 are associated with medical errors at time 2
1) IRR = 2.34, p < 0.0001
2) IRR = 2.72, p < 0.0001
3) IRR = 0.62, p = 0.001
9.5 (16)
Hunziker et al., 2012 [109] Influence of self-reported, biochemical and physiological stress on cardio-pulmonary resuscitation (CPR) performance yes 28 residents, teaching hospital, Switzerland Self-report questionnaire, video observation of simulated resuscitation Well-being: Stress/overload index (self-report; blood cortisol, heart rate)
Patient safety: performance (time until CPR is started and hands-on time)
Multiple linear regression 1) Stress/overload is positively associated with
a) time to start CPR
b) but not hands-on-time during resuscitation
2) Heart rate is positively associated with
a) hands-on-time
b) and negatively with time to start CPR during resuscitation
3a) Cortisol level and
b) heart rate variability
do not predict
c) hands-on-time and
d) time to start CPR
4) The difference of
a) stress/overload
b) cortisol level
c) heart rate variability before to during resuscitation
do not predict
d) hands-on-time or
e) time to start CPR
5) The difference of heart rate before to during resuscitation predicts
a) hands-on-time and
b) time to start CPR
1a) β/B = 12.01, 95 % CI 0.65 – 23.36, p = 0.04
1b) NS
2a) β/B = 2.22, 95 % CI 0.53 – 3.92, p = 0.015
2b) β/B = −0.78, 95 % CI 1.44 to −0.11, p = 0.027
3 ac) NS
3ad) NS
3bc) NS
3bd) NS
4ad) NS
4ae) NS
4bd) NS
4be) NS
4 cd) NS
4ce) NS
5a) β/B = 2.73, 95 % CI 0.48 – 4.99, p = 0.022
5b) β/B = −1.12, 95 % CI −1.91 to −0.33, p = 0.01
(no information regarding standardization of coefficients)
12.5 (15)
Jones et al., 2012 [100] Effect of incident seriousness and work-based support on negative positive affect yes 171 nurses, 4 hospitals, UK Cross-sectional & longitudinal between & within-person design, diary study Well-being: Positive & Negative Affect Scale (PANAS) and mood diary entriesa
Patient safety: nurse-reported incidents
Random-effects multilevel model 1) Interaction of incident occurrence and seriousness leads to elevated negative affect during remainder of shift
2a) Incident occurrence
2b) but not incident seriousness
lead to reduced positive affect during remainder of shift
1) β = 0.07, z = 3.5, p < 0.005
2a) β = −2.39, z = 1.99, p < 0.05
2b) NS
13 (16)
Kirwan et al., 2013 [105] Relationships between working environment, burnout and patient safety no 1397 nurses, 108 wards, 30 hospitals, Ireland Cross-sectional self-report questionnaire Well-being: MBIa
Patient safety: one item from AHRQa, adverse events
Multilevel regression Emotional exhaustion on ward level does not predict
1) nurse-rated patient safety or
2) reporting of adverse events
1) NS
2) NS
12.5 (16)
Klein et al., 2010 [110] Relationship between burnout and self-reported quality of care yes 1311 surgeons, 489 hospitals, Germany Cross sectional self- report questionnaire Well-being: Copenhagen Burnout Inventory (CBIa)
Patient safety: Quality of care: frequency of diagnostic and therapeutic errors (Chirurgisches Qualitätssiegel survey CQS)
Multivariate logistic regression 1) Burnout is associated with
1a) lower quality of diagnosis/therapy
1b) more diagnostic errors
1c) more therapeutic errors among males
2) Unclear association of burnout with
2a) lower quality of diagnosis/therapy
2b) more diagnostic errors
2c) more therapeutic errors
among females
1a) OR = 1.71, 95 % CI 1.10 – 2.64
1b) OR = 1.94, 95 % CI 1.35 – 2.79
1c) OR = 2.56, 95 % CI 1.66 – 3.96
2a-c) contradictory information regarding significance in text and table
10.5 (16)
Maiden et al., 2011 [101] Relationship between moral distress, compassion fatigue, and causes of medication errors yes 205 nurses, ICU, USA Cross sectional self-report questionnaire, focus group Well-being: Moral distress scalea
Compassion fatigue: Professional Quality of Life Scalea
Patient safety: Medication Administration Error Surveya
Pearson’s correlation 1) Positive correlation between moral distress and
a) transcription related medication errors and
b) physician communication related medication errors
c) but not with medication packaging
d) pharmacy processes
2) Compassion fatigue is positively correlated with
a) transcription related medication errors
but not with medication error due to
b) physician communication
c) medication packaging
d) pharmacy processes
1a) r = 0.20, p = 0.05
1b) r = 0.24, p = 0.01
1c) NS
1d) NS
2a) r = 0.15, p = 0.05
2b) NS
2c) NS
2d) NS
9 (16)
Merlani et al., 2011 [13] Relationships between hospital, patient, and clinician characteristics and burnout/stress yes 3052 physicians, nurses, and nurse-assistants, 74 ICUs, Switzerland Cross-sectional self-report questionnaire, record review Well-being: MBIa, 1 stress item
Patient safety: mortality rates and length of stay (unit records)
Multivariate logistic regression 1) Mortality is associated with higher level of burnout
2) Length of stay is not associated with burnout
1) OR = 1.060, p = 0.04, 95 % CI 1.003 – 1.120
2) NS
12.5 (16)
Prins et al., 2009 [97] Relationships between self-reported errors, burnout, and engagement yes 2115 residents, The Netherlands Cross-sectional self- report questionnaire Well-being: Utrecht Burnout Scale (UBOS)a, Utrecht Work Engagement Scale (UWES)a
Patient safety: medical errors
Pearson’s correlation 1) Errors due to wrong actions/inexperience
a) are positively correlated with emotional exhaustion
b) depersonalization
c) and negatively correlated with personal accomplishment
2) Errors due to wrong actions/inexperience are not correlated with
d) vigor
e) dedication
f) absorption
3) Errors due to lack of time
a) are positively correlated with emotional exhaustion
b) depersonalization
c) and negatively correlated with personal accomplishment
4) Errors due to lack of time are negatively correlated with
a) vigor
b) dedication
c) absorption
1a) r = 0.20, p < 0.001
1b) r = 0.29, p < 0.001
1c) r = −0.05, p < 0.001
2a) NS
2b) NS
2c) NS
3a) r = 0.43, p < 0.001
3b) r = 0.42, p < 0.001
3c) r = −0.08, p < 0.001
4a) r = −0.23, p < 0.001
4b) r = −0.24, p < 0.001
4c) r = −0.11, p < 0.001
10.5 (16)
Ramanujam et al., 2008 [102] Relationship between nurses’ work characteristics, burnout, and patient safety yes 430 nurses, 2 hospitals, USA Cross sectional self- report questionnaire Well-being: Not described, although it can be deducted from the paper that the MBIa was used
Patient safety: nurses’ safety perception
Path analysis 1) Unsatisfactory initial model fit statistics, final model statistics not reported
2) Positive association between depersonalization and perceived patient safety
3) No association between emotional exhaustion and perceived patient safety
1) χ2 = 1100.60, df = 455, χ2/df = 2.419, CFI = 0.876, RMSEA = 0.058
2) β = 0.189, p < 0.001
3) NS
8.5 (16)
Shanafelt et al., 2002 [112] Prevalence of burnout in medical residents and the relationship to self-reported patient care practices yes 115 internal medicine residents, USA Cross sectional self- report questionnaire Well-being: MBIa
Patient safety: self-developed patient care practices measure
Stepwise logistic regression 1) Overall burnout score is associated with higher levels of
a) monthly
b) weekly suboptimal patient care practices
2) Depersonalization is associated with higher levels of
a) monthly
b) weekly suboptimal patient care practices
3) No associations between
a) emotional exhaustion
b) personal accomplishment and
c) monthly
d) weekly suboptimal patient care practices
1a) OR = 8.3, p < 0.001, 95 % CI 2.6 – 26.5
1b) OR = 4.0, p = 0.036, 95 % CI 1.1 – 14.2
2a) OR = 5.8, p < 0.001, 95 % CI 2.2 – 15.4
2b) OR = 2.8, p = 0.041, 95 % CI 1.1 – 7.7
3 ac) NS
3ad) NS
3bc) NS
3bd) NS
10.5 (16)
Shanafelt et al., 2010 [111] Relationship between burnout, quality of life, depression and perceived major medical errors yes 7905 surgeons, USA Cross sectional self- report questionnaire Well-being: MBIa
Patient safety: medical errors
Logistic regression 1a) Emotional exhaustion and
b) depersonalization
are associated with higher odds of reporting an error
2) Personal accomplishment is associated with lower odds of reporting an error
1a) OR = 1.048, p < 0.0001, 95 % CI 1.042 – 1.055
1b) OR = 1.109, p < 0.0001, 95 % CI 1.096 – 1.122
2) OR = 0.965, p < 0.0001, 95 % CI 0.955 – 0.975
9 (16)
Squires et al., 2010 [103] Relationships between nurse leadership, work environment, safety climate, and nurse and patient outcomes no 600 acute care nurses, USA Cross sectional self- report questionnaire Well-being: Emotional Exhaustiona
Patient safety: medication errors and ulcers
Path analysis 1) Very good final model fit
2) No association between pressure ulcers and emotional exhaustion
3) Positive association between medication errors and emotional exhaustion
1) χ2 = 217.6, p < 0.001, SRMR = 0.054, CFI = 0.947, RMSEA = 0.047, PCLOSE = 0.67
2) NS
3) β = 0.14, p < 0.05
12 (16)
Teng et al., 2010 [14] Interactions between time pressure and burnout on patient safety yes 458 nurses, 90 units, 2 medical centers, Taiwan Cross sectional self- report questionnaire Well-being: MBIa
Patient safety: frequency of adverse events scale
Multiple linear regression 1) Burnout negatively predicts patient safety
2) The interaction of burnout and time pressure negatively predict adverse events
1) β = −0.25, p = 0.00
2) β = −0.08, p = 0.03
R2 = 0.06
[f2 = 0.06]b, c (7 predictors altogether)
13 (16)
Welp et al., 2015 [117] Relationships between burnout, demographic and unit characteristics, and patient safety yes 1425 nurses and physicians, 54 intensive care units, Switzerland Cross-sectional self-report questionnaire,
hospital records
Well-being: MBIa
Patient safety: standardized mortality ratios, length of stay, clinician-rated patient safety
Hierarchical (multilevel) linear regression 1a) Emotional exhaustion and
1b) depersonalization are negatively associated with clinician-rated patient safety;
c) personal accomplishment is positively associated with clinician-rated patient safety
2a) Emotional exhaustion, but not
2b) depersonalization or
2c) personal accomplishment is positively associated with standardized mortality ratios
3a) Emotional exhaustion,
3b) depersonalization, and
3c) personal accomplishment are not associated with length of stay
1a) B = −0.13, p < .001
1b) B = −0.07, p < .05
1c) B = 0.16, p < .01
2a) β = 0.39, p < .05
2b) NS
2c) NS
3a) NS
2b) NS
2c) NS
15 (16)
West et al., 2006 [113] Relationships between distress, quality of life and medical errors yes 184 internal medicine residents, teaching hospital, USA Longitudinal cohort study,
self-report questionnaire
Well-being: MBIa, fatigue and sleepiness: 2 items
Patient safety: medical errors
Generalized estimation equations (GEE) 1) Higher levels of
a) emotional exhaustion
b) depersonalization are associated with major medical errors in the
c) previous
d) following 3 months
2) Lower levels of personal accomplishment are associated with higher levels of major medical error in the
a) previous
b) following 3 months
1 ac) PE = 4.58, p = 0.002
1bc) PE = 2.45, p = 0.002
1ad) OR = 1.07, p = < 0.001, 95 % CI 1.03 – 1.12
1bd) OR = 1.10, p = 0.001, 95 % CI 1.04 – 1.16
2a) PE = −2.59, p = 0.002
2b) OR = 0.93, p = 0.02, 95 % CI 0.88 – 0.99
12 (16)
West et al., 2009 [23] Relationships between fatigue, distress, and medical errors yes 380 internal medicine residents, teaching hospital, USA Longitudinal cohort study, self-report questionnaire Well-being: MBIa, fatigue and sleepiness: 2 items
Patient safety: medical errors
Generalized estimation equations (GEE) Higher levels of
1) sleepiness
2) fatigue
3) emotional exhaustion
4) depersonalization and
5) lower levels of personal accomplishment
are associated with subsequent medical errors
1) OR = 1.10, p = 0.002, 95 % CI 1.03 – 1.16
2) OR = 1.14, p < 0.001, 95 % CI 1.08 – 1.21
3) OR = 1.06, p < 0.001, 95 % CI 1.04 – 1.08
4) OR = 1.09, p < 0.001, 95 % CI 1.05 – 1.12
5) OR = 0.94, p < 0.001, 95 % CI 0.92 – 0.97
13 (16)
Wetzel et al., 2010 [115] Relationships between stress and surgical performance yes 30 surgeons, 1 hospital, UK Cross-sectional self-report questionnaire,
observation of simulated operations
Well-being: State-Trait Anxiety Inventory (STAI)a, heart rate, cortisol, oberver rating
Patient safety: OTASa, End Product Assessment Rating Scale (EPA)
Linear regression Non-crisis simulation:
No relationship between
1) STAI
2) heart rate
3) cortisol
4) observer stress rating
and
a) OTAS
b) EPA
Crisis simulation:
5) no results reported on relationships between the above variables
6) Interaction between low experience and “stress” (not clear how variable was calculated) predicts lower
a) EPA and
b) OTAS
1a) NS
1b) NS
2a) NS
2b) NS
3a) NS
3b) NS
4a) NS
4b) NS
5) N/A
6a) β = .54, p < .01
6b) β = .65, p < .001
10 (15)
  1. We report not only significant but also non-significant relationships between predictor and outcome variables of interest in this review as hypothesized in the reviewed studies; even if not explicitly stated in the original publication
  2. avalidated instrument
  3. beffect sizes calculated by authors, calculation not possible if brackets empty
  4. cCohen’sƒ2 based on R2 instead of ΔR2
  5. din brackets: maximal possible score