Prevalence of Burnout Syndrome in Emergency Nurses: A Meta-Analysis

We conducted a meta-analytic study following the 2015 recommendations of Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocols.²⁵ 

The search was conducted in June 2016. To find the largest number of documents, the search did not include a limitation or result filter. The search terms were “emergency AND nurs* AND burnout” in English and “urgencias AND enfermería AND burnout” in Spanish. We consulted the following scientific databases: CINAHL, Cochrane, CUIDEN, IBECS, LILACS, PubMed, ProQuest Platform (ProQuest Health & Medical Complete), PsycINFO, SciELO, and Scopus. ProQuest Dissertation & Thesis and Google Scholar were used to find gray literature (ie, published and unpublished research material that is not available commercially; eg, conference papers/conference proceedings and theses).

We included research reports published in English, Spanish, and Portuguese, with original empirical data on burnout prevalence rates. We used the MBI for burnout assessment for emergency nurses.³  We excluded studies without sufficient statistical information for performing meta-analysis calculations and studies with data from other professions or multiple units but without independent information for emergency nurses.

We included only studies that used the MBI, because it is an instrument with good psychometric properties²⁶  and one of the most widely used by researchers studying burnout syndrome.²⁷  Including literature that used the MBI for assessing burnout facilitated data interpretation; although studies may have used different versions of the test, the results are based on similar cutoff values. The MBI use facilitated the integration and interpretation of the results of the studies because including results from other burnout questionnaires (with different cutoff values of burnout, or even different subscales) would not allow for an integrated reliability analysis.

Two members of the research team independently conducted the search and selection process; a third member was consulted in cases of disagreement. The selection process consisted of 4 phases. First the researchers made a selection after reading the title and abstract. They did a second screening after reading the full text of the selected studies. The researchers then assessed the studies using a checklist to ascertain the quality of the studies as well as the absence of methodological bias. Finally, the researchers conducted an inverse search (searching the included studies’ reference lists) and a forward search (looking for studies that cite the included studies). An inverse search was also carried out in systematic reviews located during the search and related to the topic.

To assess the methodological quality of the studies, we used Ciaponni’s²⁸  critical reading checklist, which included the items specifically related to the internal validity of the study. We used this checklist because it is specific for quantitative observational studies, as were all the included studies. No study was excluded because of methodological bias.

We included the following data from each study in the meta-analysis:

• Authors,

• Year of publication,

• Country where the study was performed,

• Type of MBI (MBI-Human Services Survey vs MBI-General Services),

• Study temporality (cross-sectional vs longitudinal),

• Type of sampling (intentional vs random),

• Mixed sample (yes vs no),

• Total sample of emergency nurses,

• Total sample of emergency nurses with high Emotional Exhaustion,

• Total sample of emergency nurses with high Depersonalization, and

• Total sample of emergency nurses with low Personal Accomplishment.

The sample number or percentage with low or high values in each burnout subscale was gleaned from the studies’ results. Low, medium, and high values were determined by the MBI cutoff points of each study.

Data coding was conducted independently by 2 members of the team, and then the mean degree of agreement was assessed. For categorical variables, Cohen κ coefficient was calculated, yielding a mean value of 0.91 (minimum = 0.86; maximum = 1). The intraclass correlation coefficient was used for continuous variables; the mean value was 0.93 (minimum = 0.88; maximum = 1).

We used StatsDirect software²⁹  and its meta-analysis package for statistical analysis.

Three independent meta-analyses (1 for each burnout subscale) were performed. A random-effects model meta-analysis—including the total sample and the sample with high Emotional Exhaustion, high Depersonalization, and low Personal Accomplishment—was chosen to calculate mean prevalence rates and 95% CI of each subscale because the number of included studies was higher than 10.^30,31 

We used the Cochran Q test and the I² index to assess the degree of homogeneity.

A sensitivity analysis, removing 1 study from the analysis at a time, was done to ensure that none of the studies included in the meta-analysis produced significant variations in the mean prevalence rates obtained.

We used Egger linear regression for assessment of publication bias.

We initially selected 1049 studies for evaluation. After reading the title and abstract of each article and excluding duplicates and studies not related to the study topic, without emergency nurses, not fitting publication language criteria, or not using the MBI, we included 80 documents for full-text reading. The documents were reduced to a final sample of 13 studies for Emotional Exhaustion and Depersonalization subscales and to 11 studies for Personal Accomplishment after applying inclusion and exclusion criteria. The selection process is detailed in Figure 1.

All selected studies were cross-sectional with convenience sampling. Eleven (84.6%) studies used the Human Services Survey version of the MBI and 2 (15.4%) used the General Services version. Nine (69.2%) studies were done in Europe (Spain or Brazil), 3 (23.1%) were done in the United Kingdom or the United States, and 1 (7.7%) in Australia. The characteristics of each study are detailed in the Table.

Sensitivity analyses showed no statistically significant changes in the prevalence values for any of the 3 subscales when any of the studies were omitted from the analysis. The value of Egger statistic to assess publication bias was 6.0 with P = .01 for Emotional Exhaustion, 4.6 with P = .4 for Depersonalization, and 4.5 with P = .3 for Personal Accomplishment, showing no publication bias.

The prevalence rate of high Emotional Exhaustion estimated by the meta-analysis was 31% (95% CI, 20–44). The estimation of high Emotional Exhaustion of each study, and the overall meta-analytic estimation, is shown in the forest plot in Figure 2. Heterogeneity analysis showed a 286.6 Cochran Q value with P < .001 and an I² index of 95.8% (95% CI, 94.6–96.6), indicating high heterogeneity.

The estimated prevalence of high Depersonalization was 36% (95% CI, 23–51). The forest plot in Figure 3 shows the meta-analytic estimation of high Depersonalization and the estimation of high Depersonalization of each study. The value of Cochran Q of the heterogeneity analysis is 363.4 with P < .001; the I² index was 96.7% (95% CI, 95.9–97.3), representing high heterogeneity.

The estimated prevalence for the third burnout sub-scale, low Personal Accomplishment, was 29% (95% CI, 15–44) (Figure 4). The Cochran Q value was 295.2 with P < .001, and a high heterogeneity was found with an I² index value of 96.6% (95% CI, 95.7–97.3).

Burnout syndrome is common among emergency nurses, but studies report that nurses in a wide variety of specialties are affected by this syndrome.^15,16  To the best of our knowledge, a meta-analysis that focuses on burnout prevalence in emergency nurses had never been published. In our meta-analysis of 13 studies involving emergency nurses, the prevalence of each of the 3 subscales of burnout according to the MBI ranged as follows: high Emotional Exhaustion, between 20% and 44%; high Depersonalization, between 23% and 51%; and low Personal Accomplishment, between 15% and 44%. These results are in line with studies of intensive care nurses conducted in other countries such as the United States.^45,46  Although the workloads in EDs and intensive care units differ (intensive care requires a greater use of technologies and devices, with a low nurse-to-patient ratio, whereas emergency care entails a higher workload in a less controlled environment), both units are likely to predispose their professionals to develop burnout. However, some US studies of intensive care nurses found higher levels of burnout than we found in emergency nurses: Emotional Exhaustion symptoms were reported in 61% to 67% of intensive care nurses, Depersonalization symptoms in 44% to 49%, and low Personal Accomplishment in 50% to 62% of nurses in intensive care units.^45,46 

The findings of our meta-analysis indicate that nurses working in EDs experience anxiety and stress, which in turn produce high Emotional Exhaustion scores. Unsatisfactory work conditions with insufficient time to develop caring activities⁴⁷  and an excessive workload may contribute to the high Emotional Exhaustion scores.⁴⁸  High Emotional Exhaustion can be further aggravated by the lack of experience in the case of younger nurses,⁴⁹  the hostility of patients who may physically and verbally assault the staff,^22,50  and a lack of assertiveness by the staff.⁵¹  Similarly, at the personal and family level, a lack of social support⁴⁸  and having an unemployed spouse⁴⁷  seem to increase the prevalence of high Emotional Exhaustion in emergency nurses.

Working conditions also contribute to the development of high Depersonalization scores among emergency nurses. Hostility toward workers,^22,50  excessive working hours,⁴⁸  contract conditions, or lack of assertiveness⁵¹  are just a few examples. On a personal level, the Depersonalization subscale adversely affects individuals in the same way as Emotional Exhaustion.⁴⁷  Low Personal Accomplishment can be aggravated by lack of assertiveness⁵¹  and again by personal factors,⁴⁷  probably as a result of Emotional Exhaustion and Depersonalization.

Burnout has been linked to increased absenteeism, the abandonment of jobs, a reduction in health care quality, an increase in errors, and a reduced level of patient safety^52–54 ; thus, identifying and treating professionals with burnout are important. In addition, the presence of high burnout prevalence rates has been associated with poor quality of nursing care, increased number of patient falls, medication errors, and increased incidence of infections⁵⁵ —factors that adversely affect patient care. Thus, future research should evaluate the results of preventive programs on the occurrence of burnout syndrome in emergency nurses⁵⁶  or treatment results if burnout syndrome develops.

Carrying out meta-analyses would be valuable to ascertain the most important risk factors associated with burnout in emergency nurses, as has been done for nursing staff in general.⁵⁷  Creating a suitable work environment is essential²¹  to minimize the impact of multiple risk factors that increase the development of burnout.^3,58  To achieve this goal, an improvement in labor agreements, a decrease in harassment, and an increase in incentives are needed.⁵⁹  Similar meta-analyses should be developed for other medical specialties to determine in what units the prevalence of burnout is high.

This study has some limitations. First, the number of studies with sufficient statistical information to perform the meta-analysis was low, because of the necessary homogenization of the tool used to assess burnout or the need to include only studies that had sufficient quantitative data to perform the analyses. Secondly, all the included studies were cross-sectional. Although adequate for prevalence assessment, this design provides less information about, for example, the number of nurses who leave the profession because of burnout. Professional attrition can be assessed more precisely with longitudinal studies, which will be necessary for future research about burnout in emergency nurses, because longitudinal studies could also identify how burnout can influence the development of other mental pathologies and, even, their economic impact on the institution.^60,61  Finally, although the psychometric cutoff points established for the different versions of the MBI or its adaptations^62,63  are similar in theory, we recommend caution in interpreting the results.

Burnout prevalence in emergency nurses is high and affects about one-third of the sample in each of the 3 burnout subscales —Depersonalization is the most affected subscale, followed by Emotional Exhaustion and Personal Accomplishment. These findings indicate that greater efforts may be needed to improve the mental health of emergency nurses and to prevent further complications. In addition to corrective interventions in the work environment, which has an important influence on burnout, interventions for professionals whose sociodemographic and psychological characteristics make them more vulnerable to burnout will also be necessary. Better workplace conditions and environment, together with the formation of professional groups in which nurses can express their emotions and feelings, may decrease the occurrence of burnout in these professionals.