Burnout as a Mediator in the Relationship between Work-Life Balance and Empathy in Healthcare Professionals

Study design and target population

A cross-sectional, observational study was conducted from August to October 2018 using a convenience sampling method. The target population constituted healthcare professionals, including nurses and doctors in their residency training programs, who had been working in the Seoul National University Hospital for more than six months. We recruited participants through an online, in-hospital recruiting board and offline postings that included a SurveyMonkey^TM link on designated hospital boards. The participants were required to provide informed consent before participating, with a total of 213 health care professionals consenting, and responses being recorded anonymously. No financial compensation was given to participants. Among respondents, 108 provided incomplete responses, which resulted in an eligible sample of 105. The study was approved by the Institutional Review Board of Seoul National University Hospital (IRB No C-1804-173-944).

Work-life balance (WLB)

Participants were asked to report their subjective WLB satisfaction on a scale from 0 to 100 with a single question, “How would you rate your satisfaction with the balance between work and life (out of 100)?” Despite there being several previous scales on WLB, “a (subjectively) perceived balance between work and the rest of life” as a definition per se for WLB was chosen as a single-item health indicator related to empathy and burnout since previous scales are not considered highly relevant to subjective well-being [24]. Stress resilience and psychological attitude toward workload and private life are individually different, and subjective evaluation of WLB may be the most reliable measurement for subjective well-being [25]. The “subjective” balance measure is not merely rating how satisfied or content providers are regarding time and workload. Respondents may engage in the process of reconsidering how they would ideally use their energy and time in a given situation [26].

Empathy scale

The Jefferson Scale for Physician Empathy (JSPE) is widely used among health professionals and trainees [27,28]. The 20 items ask about the respondent’s current situation on a 7-level Likert scale from strongly disagree to strongly agree (total score range: 20–140).

Burnout scale

Burnout was measured using the Korean version of the 22- item Maslach Burnout Inventory (MBI), originally invented by Maslach and Jackson [29] and formally validated in Korean [30]. The inventory questions are answered based on the participant’s current situation on a 7-point scale for each question. It is divided into three subscales, Emotional Exhaustion (nine items to measure feelings of being emotionally exhausted), Depersonalization (five items to measure an impersonal response to patients), and Personal Accomplishment (eight items to measure feelings of competence and achievement).

Mental health indices

The brief Patient Health Questionnaire (PHQ-9), composed of nine items corresponding to the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV)’s major depressive disorder criteria, was used to measure depressive symptoms over the past two weeks with a score ranging from 0 to 3 for each question. While the threshold score has varied in each study, a score of 8 or higher is considered to warrant clinical concern, and a higher score reflects even greater severity [31].

The validated Korean version of the Beck Anxiety Inventory (BAI), a 21-item questionnaire listing anxiety symptoms in the past week, with ratings from 0 (not at all) to 3 (severe) for each item, was administered. The maximum score is 63, and a score of 8 or higher was considered to represent mild anxiety [32,33]. Measures of depressive symptoms and anxiety were included since they represent the leading causes of mental disability, costing the global economy US$ 1 trillion each year in lost productivity [34].

Sleep problems over the past two weeks were assessed by the Korean version of the Insomnia Severity Index (ISI), composed of seven items on a 5-point Likert scale and validated in Korean [35]. A score of 15 or higher was considered to indicate insomnia [36].

Possible current alcohol problems were also screened by the Korean version of the Alcohol Use Disorder Identification Test (AUDIT-K), with scores ranging from 0 to 40. A score of 12 or higher signified an alcohol problem [37,38].

Other variables

Demographic variables including age, sex, length of time working at the hospital, current department, education level, and marital status were gathered.

Statistical analyses

The mean values and standard deviations of subjective WLB, MBI, and JSPE were calculated. These values were calculated according to gender, age (<30, 30–39, 40–49, and ≥50 years), work duration at the hospital (0.5–2, 3–4, 5–9, ≥10 years), education level (up to college and above college), and marital status (single and married). We used independent t-tests and analysis of variance (ANOVA) to compare the differences in scores of WLB, MBI, and JSPE according to the demographic variables.

To test the differences in mental health, including anxiety (BAI ≥16), depressive symptoms (PHQ-9 ≥8), insomnia (ISI ≥10), and alcohol use problems (AUDIT-K ≥12) according to WLB, participants were classified into good WLB group (scored ≥50th percentile) and bad WLB group (score <50th percentile) to increase the contrast between the two groups. A chi-square test was used to examine the differences between mental health indices, according to the WLB group. Next, multivariate linear regression models were constructed to investigate the association between WLB and empathy. We used quartiles of subjective WLB score class (categorical variable) as the independent variable and JSPE score (20–140, continuous) as the dependent variable. In Model 1, we adjusted for demographic variables, including work duration, education, and marital status. In Model 2, we additionally adjusted for mental health indices, such as anxiety, depressive symptoms, insomnia, and alcohol use problems. Finally, in Model 3, three dimensions of MBI (Burnout, Depersonalization, and Personal Achievement) were further adjusted.

Finally, a mediation model was hypothesized and explored for the role of the sub-dimensions of MBI between WLB and empathy. Bootstrapped mediation analyses were performed to examine the direct and indirect effects of WLB on empathy through the MBI subdomains [39]. Mediation analyses involve the estimation of three equations [40]:

In this analysis, b₁₁, b₂₁ and b₃₁ stand for constants, while c₁₂, c₂₂, c₃₂ and c₃₃ refer to the mean coefficients. The coefficient in Equation 1(c₁₂) was considered as the direct effect of the causal variable X´ (WLB score) on the outcome variable Y (JSPE score), without the effect of the mediator M (subdomains of Maslach Burnout Inventory). It presents 95% confidence intervals (two-tailed p<0.05) of indirect effects in the model. Mediation analyses were performed with adjustments for age, work duration, education, marital status, and mental health indices (i.e., anxiety, depressive symptoms, insomnia, and alcohol use problems). Statistical analyses were performed using SAS (Version 9.4; SAS Institute Inc., Cary, NC, USA).

Demographic and clinical characteristics

Table 1 shows the characteristics of the study participants. Although most incomplete respondents discontinued the survey at an early stage, no difference in demographic characteristics was found between complete and incomplete respondents. A total of 105 health professionals participated in our study, and 96 (91.4%) were female. Most participants were below 30 (51.4%), had worked for two years or less (50.5%), were college-educated or below (78.1%), and single (71.4%). The mean subjective WLB score was 50.9 (SD=24.3), and the mean JSPE score was 88.3 (SD=16.4). The subjective WLB score was positively and significantly associated with older age, longer work duration, higher education, and married status. Higher MBI scores were related to a younger age, shorter work duration, lower education, and unmarried status. Higher JSPE scores were also significantly associated with older age, longer work duration, and higher education level, but marital status was not significantly associated.

Table 1.

The work-life balance, empathy and burnout scores by the characteristics of study participants

Table 2 shows mental health indices by subjective WLB score groups. Compared to the low WLB group, the high WLB group consisted of fewer nurses with anxiety (44.8% vs. 21.3%, p<0.011), depressive symptoms (65.5% vs. 29.8%, p<0.001), and insomnia (70.7% vs. 51.1%, p=0.023). WLB level groups did not significantly differ in terms of alcohol use problems.

Table 2.

Mental health indices according to the subjective worklife balance

Multivariate regression analyses: the association between WLB and empathy

Table 3 presents the association between WLB and empathy, which were analyzed using multivariate linear regression models. Model 1 used age, work duration, education level, and marital status as covariates. The results of regression indicated that subjective WLB group and the covariates explained 29.0% of the variance [R²=0.29, F (8, 104)=4.98, p< 0.001]. In this model, compared to the best subjective WLB group (fourth quartile), the good group (third quartile), poor group (second quartile), and the worst group (first quartile) had significantly lower JSPE scores at -13.32 (p=0.005), -12.35 (p=0.005), and -21.50 (p<0.001), respectively. Higher education level was significantly associated with a higher empathy level. Model 2 used anxiety, depressive symptoms, insomnia, and alcohol use problem as additional covariates to Model 1. The results of regression indicated that Model 2 explained 34.6% of the variance, R²=0.35, F (12, 104)=5.89, p<0.001. In Model 2, the good group (β=-11.21, p=0.017) and the worst group (β=- 16.67, p=0.002) also had significantly lower JSPE scores than the best group. In Model 3, three dimensions of the burnout scale (Emotional Exhaustion, Depersonalization, and Personal Achievement) were additionally used as covariates to Model 2, and Model 3 explained 54.0% of the variance, R²=0.54, F (15, 104)=8.42, p<0.001. In this model, the association between subjective WLB and JSPE score disappeared. Only personal achievement among the burnout sub-dimensions was associated with a better JSPE score (β= 0.56, p<0.001).

Table 3.

Results of multivariate linear regression models for the association between health professionals’ work-life balance and empathy

Mediation analyses: WLB, empathy, and burnout

Figure 1 shows the results of mediation analyses performed to investigate the indirect effects of subjective WLB on empathy through the three sub-dimensions of burnout. We found significant mediating effects of burnout between WLB and Empathy. The direct effect of WLB on Empathy was not significant (β=0.06, 95% CI: -0.06–0.20, p=0.450); however, the total indirect effect was significant (β=0.17, 95% CI: 0.06–0.29, p=0.003). More specifically, Personal Achievement (p=0.011) and Depersonalization (p=0.047) among the subscales of burnout were found to be mediating variables.

Figure 1.

Results of mediation analyses.

Association between burnout versus WLB & between burnout versus empathy

The relationship between subjective WLB and burnout in our study is compatible with other studies. A study targeting engineers and nurses reported that burnout is a direct consequence of work-home conflict for both groups [41]. Another study with 750 Italian nurses found that work-family conflict is significantly linked to burnout [42]. Although most studies were cross-sectional, which cannot test for causality, a longitudinal study with 242 Spanish soldiers has provided some insight on the issue, showing the reciprocal relationship between work-family conflict and emotional exhaustion as a facet of burnout [43]. Although the causal linkage between subjective WLB and qualitative/quantitative measurements were not confirmed in many comprehensive studies, life balance measures individuals’ overall life pattern and represents one’s life system holistically [26]. Instead of controversial measures of WLB, we chose the simplest form of subjective evaluation on WLB. Neurotic people may complain about disturbed WLB despite having objectively healthy lifestyles. Conversely, optimists may handle heavy schedules and workloads with serene perspectives if they believe their lives are fulfilling and meaningful [44]. Regarding the limitless confounding factors in judging WLB, a subjective assessment may be the most reliable variable for predicting burnout.

The negative association between burnout and empathy among primary physicians and nurses was established in other studies. A recent systematic review revealed that burnout mostly has a negative association with empathy, with 8 out of 10 studies included showing a negative correlation [45].

Feelings of personal achievement & higher educational attainment: a possible booster of empathy?

Among burnout scales, Personal Achievement items were found to be significantly related to empathy, and higher education level was significantly associated with higher empathy levels. Several studies showed similar results where high levels of empathy were reported to correlate with the personal accomplishment factors in the MBI [46-48]. The personal accomplishment factor is related to intrinsic personal factors, such as being positive, happy, and sociable. A feeling of strong personal accomplishment may lower the risk of compassion fatigue and burnout, which are often labeled as emotional labor [49].

In our study, those with higher education and older people who have a better chance of personal accomplishment reported being more empathetic and having less burnout, which is compatible with the above-mentioned findings. To have better hospital management, various strategies to enhance empowerment, such as active participation in decision-making, research on important issues, promotion, and prizes, should be developed.

Need for educational approach & working environment changes for healthcare professionals

Our finding showed a medium level of burnout compared to other studies targeting health professionals [50,51]. A few studies reported tentative findings on interventions to increase empathy and prevent burnout. Leisure and exercise programs to improve quality of life were developed to prevent burnout among physicians [52,53]. Preventative intervention focuses on developing workers’ engagement and putting more emphasis on organizations to take some responsibility in the detection and prevention of burnout [54]. West and colleagues noted that individual (e.g., mindfulness, discussion, and stress management) and organizational (e.g., work environment) interventions similarly produced substantial improvements in burnout [55]. A recent meta-analysis confirmed that burnout in health care professionals is a problem of the whole organization rather than of individuals [56]. The Mayo clinic presented nine organizational strategies to reduce physicians’ burnout by emphasizing more flexibility and facilitating leadership to ensure members’ resiliency and self-care [57]. The suggested strategies include 1)Acknowledging and assessing the problem, 2) Harnessing the power of leadership, 3) Developing and implementing targeted interventions, 4) Cultivating the community at work, 5) Using rewards and incentives wisely, 6) Aligning values and strengthening the work culture, 7) Promoting flexibility and work-life integration, 8) Providing resources to promote resilience and self-care, and 9) Facilitating and funding organizational science.

Fundamental reform of hospital settings has been suggested, decreasing the burden of documentation and delivery system. Physicians need to be trained as coaching/mentoringleaders who keep their team informed and enable them to actively participate in improving their work environment [58]. The National Academy of Medicine (NAM) [59] launched a national Action Collaborative on Clinician Well-Being and Resilience. NAM offers various workgroups for sharing information, optimal team-based care and documentation, and tool kits for a productive relationship [60].

Compassion fatigue or professionals who become mentally ill because of excessive workload often cause inefficient and inhumane health care, which brings about a less safe hospital environment because of anger issues. Dual responsibilities, both at home and work, may lead to burnout symptoms, which may explain the significant global shortage of health care workers. Nurses who are relatively less empathetic, not being aware of their burnout or mental illness may provoke various conflicts and eventually become involved with medical accidents, poor service, malpractice suits, and violent incidents [61]. Over the next 20 years, many health professionals may quit their jobs if proper preventive measures against burnout and emotional or physical overwork are not undertaken, which could threaten the wellness of society despite advancements in technological development [62].

Limitations

This study has several limitations. Although it was hypothesized that WLB would affect burnout, and burnout would affect empathy, the possibility of reverse causality should be considered. Burned-out workers may not try to pursue WLB. Workers who have less empathy for other psychological reasons or personality structures may feel more burnout related to their sacrificial job requirements. Sound evidence is needed in prospective cohort studies concerning WLB, burnout, and empathy. Second, although subjective satisfaction in WLB is critical when considering both intrinsic resilience and the external environments, quantification of WLB with a well-validated and objective scale is necessary for clarifying the relationship. Additionally, the item assessing WLB did not specify the specific period of time being measured, which may result in variability between respondents. Third, an integrated analysis of the determinants of empathy could not be performed in our study. As one of our hypotheses was to test the mediating roles of burnout in the relationship between WLB and empathy, we did not assess several empathy-related variables, such as personal traits or emotional intelligence. However, we performed mediation analyses adjusting for demographic variables and mental health indices. In the future, investigating the integrated analysis for empathy using structural equation modeling while considering various potential determinants of empathy is needed. Finally, most of the respondents were female; hence, gender differences could not be explored. Women caring for children and managing the household in a traditional society may be either more vulnerable or more resilient, which was not confirmed by this study. Fourth, the targeted workers, working in a big general hospital in Seoul, where the workload is typically very high and competitive, may have different profiles from workers in other environments.

Conclusions

We found that WLB correlated positively with empathy and personal achievement, with one sub-dimensions of burnout appearing to be a significant mediating factor. WLB was negatively associated with burnout, but the causal nature of the relationship between the two factors was not confirmed. Theoretically, subjective feelings of work-life imbalance may cause burnout symptoms, and burnout symptoms may bring about an imbalanced work-life balance in health care professionals. Further longitudinal studies are needed to explore the causal relationship between these variables.