Introduction

Burnout syndrome has been characterized by emotional exhaustion, depersonalization, and a diminished sense of personal accomplishment and is typically observed to be the result of chronic workplace stress.¹ Numerous meta-analyses have estimated the prevalence of nurse burnout, which ranged from 11% to 56%^2,3 and was even higher during the COVID-19 pandemic given hospital overcrowding and understaffing.^4,5 Meta-analyses have also explored the correlates of nurse burnout, including younger age,⁶ male sex,⁷ single or divorced marital status,⁷ not having children,⁷ low staffing levels,⁸ and workplace violence⁹ as risk factors and resilience,¹⁰ job control,¹¹ social support,¹¹ and nurse empowerment¹² as protective factors. A few meta-analyses have examined the association of burnout with nurses’ well-being and career, including burnout and sleep problems,¹³ depression,¹⁴ and turnover intention.¹⁵ However, to our knowledge, no meta-analysis has estimated the association between nurse burnout and health care quality and safety and patient morbidity and mortality. This absence is interesting as there have been several meta-analyses investigating the association between physician burnout and patient safety, satisfaction, and quality of care.^16,17

Nurses are on the front line of patient care in hospitals and other settings, often present when physicians are not, and charged with making important clinical decisions.¹⁸ Because of the critical role of nurses in delivering and, in some cases, overseeing patient care, nurse burnout may be associated with many dimensions of patient outcomes.

One systematic review involving 20 studies¹⁸ concluded that there was an association between nurse burnout and health care quality and safety, but it did not include a quantitative synthesis of the evidence to estimate the magnitude of the association. Moreover, due to the nature of the analysis, we do not know whether the association was moderated by subcomponents of burnout or by nurse demographics, clinical specialty, geographic area, or when the survey was conducted. Answering these questions may help to clarify the nature and severity of the adverse effects of nurse burnout and may help to inform nurse burnout interventions.¹⁹

This study evaluates through a systematic review and meta-analysis the magnitude and moderators of the association between nurse burnout and patient safety, patient satisfaction, and quality of care. Patient safety is defined as the absence of preventable harm to a patient,²⁰ patient satisfaction as patients’ positive evaluations of distinct dimensions of the health care,²¹ and quality of care as the degree to which health services for individuals and populations increase the likelihood of desired health outcomes.²²

Methods

Inclusion and Exclusion Criteria

This systematic review and meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guideline.²³ The population, exposure, comparator, outcomes, and study design framework was used to specify eligibility criteria.²⁴ The protocol was pre-registered at PROSPERO (CRD42024512637 and CRD42024512684). The Stanford Institutional Review Board exempted the study from review and informed consent as there were no identifiable human participants.

The population was nurses. The inclusion criteria were (1) any type of nurse (eg, registered nurse, nurse practitioner, or nurse supervisor); (2) any specialty; (3) any geographic region; and (4) any age, sex, work experience, or educational level. Exclusion criteria were populations consisting of (1) midwives, (2) nursing assistants, (3) nursing students, and (4) mixed samples of nurses and physicians.

The exposure was burnout. The inclusion criteria were (1) any validated measure of burnout; (2) any validated measure of burnout subcomponents, including emotional exhaustion, depersonalization, and low sense of personal accomplishment; (3) any validated measure of widely recognized synonyms of burnout subcomponents, such as cynicism and low professional efficacy; or (4) any study-specific measure that directly referred to burnout. Exclusion criteria were exposures related to but distinct from burnout, including (1) occupational stress, (2) fatigue, and (3) depressive symptoms. The comparator was (1) a low level of burnout if a continuous measure of burnout was used or (2) nonburnout if a dichotomous characterization of burnout was reported (eg, prevalence).

For outcomes, the inclusion criteria were (1) any measure of patient safety, including safety climate or safety culture, safety grade, patient safety incidents (any medical error or adverse event similar to those selected in the Agency for Healthcare Research and Quality indicators²⁵); (2) any measure of patient satisfaction, including patient satisfaction survey rating, patient complaint, and patient abuse; and (3) any measure of quality of care, including nurse-assessed overall quality of care and commonly used objective quality indicators, such as length of hospital stay and 30-day standardized mortality rate. Exclusion criteria were structural antecedents of burnout (eg, nursing foundations for quality of care²⁶).

For study design, a study had to report a quantifiable association between nurse burnout and any of the 3 outcomes to be included; both cross-sectional and longitudinal studies were eligible. Grey literature included dissertations. Qualitative studies and conference proceedings were excluded. Publications in languages other than English were excluded.

Search Strategy

Two reviewers (L.Z.L. and P.Y.) applied any search term from each of the nurse, burnout, and quality of health care categories using specific search strings listed in eMethods 1 in Supplement 1, which were developed in consultation with a research librarian. In February 2024, the reviewers independently searched major databases from January 1, 1994, to February 29, 2024, including MEDLINE (via PubMed), Web of Science, Scopus, Embase, CINAHL (via EBSCOhost), PsycINFO (via EBSCOhost), and ProQuest. Using prespecified eligibility criteria, they independently included studies, identified additional studies from the reference lists, and extracted data. All discrepancies were resolved through discussion. The data were recorded in an online spreadsheet, and citations were managed with EndNote (Clarivate).

Statistical Analysis

Due to expected heterogeneity, we conducted a random-effects meta-analysis to quantitatively synthesize the association between nurse burnout and quality of care. Statistics were converted to standardized mean differences (SMDs) (Cohen d) using conventional methods (eMethods 2 in Supplement 1).

Subgroup meta-analysis and meta-regression were performed using prespecified variables to examine heterogeneity. We used forest plots to compare outcome subtypes, burnout measures and subcomponents, clinical specialties, and regions. We used random-effects meta-regression to examine the stratification by age, sex, work experience, and educational level. Self-reported race and ethnicity were collected but not analyzed because of low comparability across countries and low power among US studies. We also grouped regions into World Bank–defined geographic areas for graphical display.²⁷

All analyses were performed using R, version 4.3.2 (R Foundation). The significance level was defined as a 2-tailed P < .05 by t test.

Bias Assessment and Sensitivity Analysis

Two reviewers (L.Z.L. and P.Y.) independently used the Risk Of Bias in Nonrandomized Studies of Exposures (ROBINS-E) to assess 7 risks of bias.²⁸ We used the serial exclusion method to assess small-study effect bias. As sensitivity analyses, we used forest plots and meta-regressions to examine alternative variable coding, sample selection, and confounding adjustment.

Results

Descriptive Statistics

Nurse Characteristics

The included studies involved 288 581 nurses from at least 5322 hospitals in 32 countries (mean [SD] age, 33.9 [2.1] years; 82.7% female and 17.3% male). The mean (SD) percentage of US nurses who self-identified as White was 70.9% (6.0%) and 29.1% (6.0%) for other races and ethnicities. The mean (SD) percentage of nurses holding college degrees (bachelor’s degrees or higher) was 42.0% (15.7%), and the mean (SD) length work experience was 10.4 (1.6) years. The mean (SD) prevalence rate of burnout was 30.7% (9.7%) based on study-specific cutoffs. The ascertainments of burnout are described in eTable 1 in Supplement 1, and the burnout measures are compared in subgroup analyses later in the Results.

Study Characteristics

We identified 6231 studies from the initial search. Using prespecified criteria, 85 studies (81 cross-sectional and 4 longitudinal) were included for meta-analysis (eTables 1-5 in Supplement 1).^{29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113} The studies were published between 1994 and 2024 and had a median sample size of 458 individuals (IQR, 234-1352 individuals). A flow diagram of study inclusion and exclusion is provided in eFigure 1 in Supplement 1.

Main Analyses

From the random-effects meta-analysis, we found negative associations between nurse burnout and various patient safety outcomes (Figure 1 and Figure 2). Specifically, nurse burnout was associated with a lower safety climate or safety culture (SMD, −0.68; 95% CI −0.83 to −0.54; I² = 97.3%), lower safety grade (SMD, −0.53; 95% CI, −0.72 to −0.34; I² = 97.1%), more frequent nosocomial infections (SMD, −0.20; 95% CI, −0.36 to −0.04; I² = 97.0%), more frequent patient falls (SMD, −0.12; 95% CI, −0.22 to −0.03; I² = 62.2%), more frequent medication errors (SMD, −0.30; 95% CI, −0.48 to −0.11; I² = 87.6%), more incidences of medical errors (SMD, −0.19; 95% CI, −0.32 to −0.05; I² = 0.0%), more frequent adverse events or patient safety incidents (SMD, −0.42; 95% CI, −0.76 to −0.07; I² = 97.3%), and more incidences of missed care or nursing care left undone (SMD, −0.58; 95% CI, −0.91 to −0.26; I² = 92.2%). No association was found for frequency of pressure ulcers (SMD, −0.12; 95% CI, −0.24 to 0.01; I² = 70.7%) or for the 1 study⁶¹ that investigated frequency of adverse infusion- and transfusion-related reactions (SMD, −0.01; 95% CI, −0.14 to 0.12).

The random-effects meta-analysis showed an association between nurse burnout and lower patient satisfaction survey ratings (SMD, −0.51; 95% CI, −0.86 to −0.17; I² = 98.1%) (Figure 3). However, the association was not significant for 2 studies^61,104 on the frequencies of patient complaints (SMD, −0.15; 95% CI, −0.43 to 0.12; I² = 89.4%) and patient abuse (SMD, −0.16; 95% CI, −0.42 to 0.11; I² = 88.8%).

The random-effects meta-analysis also showed negative associations between nurse burnout and quality of care outcomes (Figure 4). Specifically, nurse burnout was associated with lower nurse-assessed quality of care (SMD, −0.44; 95% CI, −0.57 to −0.30; I² = 97.2%), a higher rate of tube feeding in nursing home patients (SMD, −0.20; 95% CI, −0.27 to −0.12),⁴⁹ and a higher rate of urinary catheter use in nursing home patients (SMD, −0.22; 95% CI, −0.30 to −0.14).⁴⁹ No association was found for length of stay in 1 study⁹¹ (SMD, −0.01; 95% CI, −0.04 to 0.02), failure to rescue rate in 1 study⁹¹ (SMD, −0.03; 95% CI, −0.05 to 0.00), and 30-day standardized mortality rate in 2 studies^79,91 (SMD, −0.27; 95% CI, −0.75 to 0.20; I² = 99.2%).

Subgroup Analyses

We conducted subgroup analyses of the planned variables for the patient safety quality-of-care outcomes. These analyses were stratified by measurement scale of burnout; subcomponents of burnout; nurse age, sex, work experience, and education; clinical specialty; geographic region; and survey time.

Stratification by Measurement Scale of Burnout

An association between nurse burnout and patient safety and quality of care was found for most burnout measurement scales (eFigures 2 and 3 in Supplement 1), including the Maslach Burnout Inventory (safety: SMD, −0.51 [95% CI, −0.61 to −0.41]; quality: SMD, −0.33 [95% CI, −0.45 to −0.21]), Oldenburg Burnout Inventory (safety: SMD, −0.85; 95% CI, −1.04 to −0.66), Shirom-Melamed Burnout Measure (safety: SMD, −0.80; 95% CI, −1.06 to −0.53), and Bergen Burnout Inventory (safety: SMD, −2.32; 95% CI, −3.08 to −1.56). The Professional Quality of Life Questionnaire did not find an association of nurse burnout with patient safety (SMD, −0.81; 95% CI, −1.63 to 0.02), possibly due to low power, but found a significant association with quality of care (SMD, −0.92; 95% CI, −1.19 to −0.66). The Copenhagen Burnout Inventory found a significant association of nurse burnout with patient safety (SMD, −0.15; 95% CI, −0.27 to −0.02) but did not find an association with quality of care (SMD, −0.84; 95% CI, −1.72 to 0.05).

Stratification by Subcomponent of Burnout

Forest plots (eFigures 4 and 5 in Supplement 1) showed that among the studies that used the Maslach Burnout Inventory as the measurement tool, the low personal accomplishment subcomponent of burnout had a smaller association with patient safety (SMD, −0.28; 95% CI, −0.38 to −0.19) than the emotional exhaustion subcomponent (SMD, −0.52; 95% CI, −0.63 to −0.41; z = −3.15; P = .002) and the depersonalization subcomponent (SMD, −0.45; 95% CI, −0.57 to −0.33; z = −2.10; P = .04). However, the forest plots did not show any difference in the association between nurse burnout and quality of care across the burnout subcomponents.

Stratification by Nurse Age, Sex, Work Experience, and Education

Meta-regression results indicated that the association between nurse burnout and patient safety and quality of care was consistent across various demographic characteristics. Specifically, the association was not moderated by nurses’ age (safety: meta-regression coefficient, −0.01 [95% CI, −0.03 to 0.02; P = .63]; quality: meta-regression coefficient, 0.01 [95% CI, −0.04 to 0.02; P = .48]), sex (safety: meta-regression coefficient, −0.43 [95% CI, −1.39 to 0.54; P = .39]; quality: meta-regression coefficient, 0.34 [95% CI, −1.63 to 2.30; P = .74]), or work experience (safety: meta-regression coefficient, −0.02 [95% CI, −0.05 to 0.01; P = .20]; quality: meta-regression coefficient, 0.01 [95% CI, −0.02 to 0.05; P = .46]).

The association between nurse burnout and quality of care was not moderated by nurses’ educational level (meta-regression coefficient, −0.01; 95% CI, −0.49 to 0.47; P = .97) (eFigure 6 in Supplement 1). However, the association between nurse burnout and lower patient safety was smaller for the nurses with a higher percentage of college degrees (meta-regression coefficient, 0.52; 95% CI, 0.15-0.89; P = .006) (Figure 5). Stratification by graduate degrees (master’s degree or higher) is presented in eMethods 3 in Supplement 1, with the caveat of lower power.

Stratification by Clinical Specialty

The meta-regression showed negative associations between nurse burnout and patient safety and quality of care in most nursing specialties. Associations between nurse burnout and patient safety and quality of care by clinical specialty (eFigures 7 and 8 in Supplement 1) were as follows: nurses without a reported specialty (safety: SMD, −0.50 [95% CI, −0.62 to −0.38]; quality: SMD, −0.34 [95% CI, −0.49 to −0.19]), primary care (quality: SMD, −1.05; 95% CI, −1.29 to −0.80), acute care (safety: SMD, −0.55 [95% CI, −0.83 to −0.26]; quality: SMD, −0.34 [95% CI, −0.74 to 0.06]), emergency department or urgent care (safety: SMD, −0.16; 95% CI, −0.48 to 0.16), intensive care unit or critical care (safety: SMD, −0.62 [95% CI, −1.10 to −0.14]; quality: SMD, −0.66 [95% CI, −0.82 to −0.51]), pediatrics (safety: SMD, −0.29; 95% CI, −0.56 to −0.01), geriatrics (safety: SMD, −0.43 [95% CI, −0.93 to 0.06]; quality: SMD, −0.60 [95% CI, −1.26 to 0.07]), psychiatry (quality: SMD, −0.05; 95% CI, −0.22 to 0.12), maternity (safety: SMD, −0.76; 95% CI, −0.88 to −0.64), and oncology (safety: SMD, −0.93 [95% CI, −1.02 to −0.84]; quality: SMD, −0.55 [95% CI, −1.07 to −0.02]). For patient safety, the negative association was significant for all but geriatric and emergency nurses, possibly due to low power. For quality of care, the negative association was significant for all specialties except geriatrics, psychiatry, and acute care.

Stratification by Geography and Survey Time

The meta-regression showed negative associations between nurse burnout and health care quality and safety in most countries. A significant association between nurse burnout and patient safety was found in Australia, Belgium, Brazil, Canada, China, Ecuador, Germany, Iran, Ireland, Japan, Korea, Norway, Oman, the Slovak Republic, South Africa, Spain, Thailand, Turkey, the United Kingdom, and the US but not in Greece, Italy, Jordan, and Saudi Arabia (eFigure 9 in Supplement 1). A significant association between nurse burnout and quality of care was observed in Belgium, Brazil, Canada, China, Egypt, Iceland, Iran, Italy, South Africa, Thailand, and the US but not in Germany, Japan, New Zealand, and the United Kingdom (eFigure 10 in Supplement 1). eFigure 36 in Supplement 1 presents a graph of the mean association by geographic area.

The association between nurse burnout and patient safety was not moderated by when the survey was conducted (meta-regression coefficient, −0.00; 95% CI, −0.02 to 0.02; P = .98 (eFigure 11 in Supplement 1) over a 33-year survey completion period from 1991 to 2023. However, the association between nurse burnout and quality of care was more negative over time (meta-regression coefficient, −0.03; 95% CI, −0.05 to −0.009; P = .006) (eFigure 12 in Supplement 1), even after excluding the studies published after 2019 to account for the COVID-19 pandemic (eFigures 13 and 14 in Supplement 1).

Sensitivity Analysis

Risk of bias in nonrandomized studies of exposure scores for each included study are provided in eTable 6 in Supplement 1 and a bubble plot in eFigure 15 in Supplement 1. The Begg rank test (z = −1.16; P = .25) (eFigures 16-19 in Supplement 1) did not indicate small-study effects, and a meta-analysis of nonaffirmative results suggested that the results were highly robust to hypothetical worst-case publication bias and p-hacking (eFigure 20 in Supplement 1).¹¹⁴ Serial exclusion of the studies found that each study contributed to a mean (SD) 0.00% (0.97%) of the Cohen d estimate (eTable 7 in Supplement 1). We show strength of evidence under effect heterogeneity¹¹⁵ in eFigures 21 and 22 in Supplement 1. We used robust variance estimation to account for multiple outcomes from the same study (eTable 8 in Supplement 1). Sensitivity analyses (eFigures 23 and 24 in Supplement 1) showed that burnout levels or prevalence rates did not moderate the association, suggesting that the results were not driven by divergent ascertainments of burnout.¹¹⁶ We found similar results using alternative variable coding (odds ratio conversion) (eFigures 25-30 in Supplement 1), sample selection (mixed professions) (eFigures 31-33 in Supplement 1), and confounding adjustment (eFigures 34 and 35 in Supplement 1).

Discussion

This meta-analysis shows a negative association between nurse burnout and patient safety, patient satisfaction, and quality of care. The association between nurse burnout and these outcomes was consistent independent of nurses’ age, sex, work experience, and geography. The associations and effect size between nurse burnout and quality of care were also consistent with previous meta-analyses of burnout and quality of care in physicians and pooled studies of health care professionals.^16,17

The emotional exhaustion and depersonalization dimensions of burnout, more so than personal accomplishment, were observed to be the most closely associated with patient safety. This finding is consistent with previous research suggesting that the personal accomplishment dimension may be a somewhat distinct construct.¹¹⁷

The association of nurse burnout with patient safety was persistent over time, and the association with quality of care was increasingly negative over 3 decades, even after accounting for the COVID-19 pandemic. This finding is concerning considering decades of national and organizational efforts for quality improvement.

The moderation of having a bachelor’s degree on patient safety and having a graduate degree on quality of care were consistent with a previous finding that higher levels of nurse education were associated with lower patient mortality.¹¹⁷ Accordingly, investments in nursing education and training may be considered as a component of burnout mitigation efforts.

To date, most antiburnout efforts have focused on individual interventions, such as mindfulness or personal resilience training designed to help people more effectively cope with stress, instead of interventions to reduce stress and burnout in the workplace.¹¹⁸ Many effective interventions are at the work unit level, where health care workers experience teamwork, feelings of community, professional development, and recognition. Numerous health care organizations have begun to take action, including appointing senior leaders to develop an organizational strategy to address the root cause issues in the clinical practice environment, such as low staffing levels and long work hours or overtime.¹¹⁹ Hospital accrediting bodies have also begun evaluating such organizational efforts, which may encourage more widespread adoption.¹²⁰

The US Surgeon General has prioritized the mitigation of burnout for the nation’s health care delivery system.¹²¹ The National Academy of Medicine has launched a holistic action collaborative involving stakeholder groups, including health care payers, technology companies, government agencies, professional societies, and health care organizations to promote well-being for health care workers.¹²² Congress has passed legislation allocating more than $100 million of funding to promote mental health, foster resilience, and reduce stigma among health care professionals.¹²³ Allocation of even more substantive funding, commensurate with the magnitude and adverse effects of health worker burnout, seems necessary to support research and implementation of evidence-based approaches to reduce clinician burnout.

Limitations

This study is subject to several limitations. One key limitation of the meta-analysis is the heterogeneity across the included studies, which is common among studies of psychosocial factors. Another limitation is that most included studies used cross-sectional research designs, which prevented us from determining the causal direction of the association between nurse burnout and the outcomes. We also combined 2 separate preregistrations because we determined that patient safety is an inseparable dimension of quality of care.¹²⁴ Finally, despite many countries included, the comparisons by geography did not represent all countries in an area and should be interpreted with that qualification.

Conclusions

In this systematic review and meta-analysis of 85 studies, nurse burnout was associated with lower health care quality and safety and lower patient satisfaction. These associations were consistent across nurses’ age, sex, work experience, and geography and have been persistent over time.

References