Job stress and burnout affecting the mental health of Korean medical faculty members: constructing causality among latent variables

1. Background

A recent nationwide survey on burnout reported that many medical faculty members in Korea suffer from high levels of burnout. The 34.0% of medical faculty members in the survey have experienced emotional exhaustion, 66.3% depersonalization, and 92.4% reduced personal accomplishment of burnout. More seriously, 31.5% responded to a high level of burnout in two subdimensions, and 30.5% responded to a high level of burnout in all three sub-dimensions in the survey [1].

The 11th Revision of the International Classification of Disease indicates workplace stress being a key factor for burnout by defining burnout as “a syndrome conceptualized as resulting from chronic workplace stress that has not been successfully managed” [2]. Most medical schools and hospitals represent stressful workplaces where their members cannot successfully manage stress, thus experience burnout. Many studies at home and abroad demonstrate job stress and its meaningful relationship with burnout syndrome among medical school personnel [3-8].

Internal working conditions such as extra workloads and rigid working systems affect job-related stress and burnout levels of medical school personnel. For example, having night-duty calls was the most vital risk factor for stress and its following burnout among medical oncologists in Korea [4]. External factors such as health care policy and governmental regulation influence the medical school environment and its members [3,4]. Moreover, an outbreak of pandemic like Middle East respiratory syndrome and coronavirus disease 2019 drastically changes the medical school system and thus influences job stress and burnout levels of their members [9-12]. Therefore, job stress and its consequential burnout are prevalent among Korean medical school personnel [1,4,9]. However, rare studies have scrutinized job stress, burnout, and mental health of medical school personnel in Korea with a national data until Seo et al. [1] conducted the first nationwide survey on burnout of medical faculty members in 2020.

Job stress and burnout have adverse effects on the mental health of medical school personnel, such as depression [11,13], work-related dissatisfaction [8,14], retirement [1], and even suicide [10]. Most studies in Korea have examined the relationships between only two variables, i.e., stress and depression [13], burnout and suicide [1], and so forth, but neglected structural relationships among three or more variables.

2. Causal relationships among job stress, burnout, and adverse mental health

First, many studies highlight that workplace or job stress is significantly associated with burnout in medical school personnel. Job stress was commonly associated with burnout among Iranian medical faculty members [5] and gastroenterology clinicians from the Association of Southeast Asian Nations member states [7]. Duration of shift and relationships with other personnel significantly affected burnout of Indonesian medical residents [6]. Excessive time on daily electronic health records was associated with increased burnout and decreased work-life satisfaction among American clinical faculty [8]. Work intensity and job stress significantly affect burnout of Korean neurosurgeons [3].

Next, job stress and burnout are likely to affect mental health among medical school members negatively. Job stress significantly impacted the depressive mood among interns and residents in a tertiary hospital in Korea [13]. Burnout also negatively impacted the emotional and psychological health, such as anxiety and depression, of American emergency room physicians [11]. Burnout was significantly associated with decreased job satisfaction [4] and a higher perception of job risk among healthcare workers [12]. Further, personal accomplishment burnout significantly decreased work-related satisfaction among Hungarian female physicians [14].

Nurikhwan et al. [6] theoretically constructed causal relationships with intrinsic and extrinsic factors for burnout yielding physical and psychological problems and underperformance among medical residents. Based on the above discussions, it is hypothesized that job stress will increase the level of burnout, which will increase the level of adverse mental health, and burnout will mediate the relationships between job stress and mental health of medical faculty members. It is essential to identify causal relationships among three variables, i.e., job stress as an independent variable, mental health as the dependent variable, and burnout as a mediating variable in this study, as it is possible to intervene in burnout as well as job stress to prevent adverse mental health of medical school members.

1. Data

This study utilized the survey data of the research project on “2020 Burnout of Faculty Members of Medical Schools in Korea” with a grant from the Korean Association of Medical Colleges in 2020. A total of 996 faculty members in 40 medical schools nationwide voluntarily participated in the survey on an anonymous online questionnaire from October to December 2020. The questionnaire includes questions about job-related stress, burnout, and mental health in addition to demographic characteristics and working conditions. This study selected 855 faculty members who filled out all burnout questions as the study sample [1].

2. Measurements

Job stress was measured with 11 items: bureaucratic tasks, working hours, respect from colleagues/staff, compensation (money), lack of autonomy, government or university regulation, daily life, research, education, medical treatment, and chart recording. Each item was scored on a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree) points [1]. Factor analyses reduced 11 items into three sub-dimensions and named managerial burden, organizational pressure, and daily task for three stressors. Reliability analyses yielded moderate internal consistencies, 0.775 for managerial burden, 0.691 for organizational pressure, and 0.664 for daily tasks (Cronbach’s α).

Burnout was measured using a modified and revalidated version of the Maslach Burnout Inventory-Human Service Survey [15]: seven items for each sub-dimension, emotional exhaustion, depersonalization, and sense of reduced personal accomplishment. Each item was scored on a 7-point Likert scale from 0 (not experiencing at all) to 6 points (experiencing certain emotions, feelings, and thoughts at least 6 times a week). Reliability analyses yielded high internal consistencies, 0.938 for emotional exhaustion, 0.911 for depersonalization, and 0.873 for personal accomplishment (Cronbach’s α).

Mental health was measured with five items of depressive feelings, retirement intention, retirement attempt, suicidal ideation, and suicidal attempt [1]. Each item was scored on a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree) points. Reliability analysis yielded moderate internal consistency, 0.780 for mental health (Cronbach’s α).

Demographic characteristics of medical faculty members included gender, age, marital status, and regional locations, and working conditions included institutional affiliation (hospital/school), department (basic versus clinical), academic rank (assistant/associated/professor), employment period, and weekly working hours.

3. Statistical analyses

This study conducted descriptive statistics to examine demographic characteristics and working conditions of medical faculty members. This study used T-test and analysis of variance to compare perceived levels of job stress, burnout, and mental health of medical faculty members according to demographic characteristics and working conditions with ad hoc test (Bonferroni method). This study also used correlation analyses to examine the relationships among variables using Pearson’s correlation coefficients with statistical significance at a p-value of <0.05.

This study employed structural equation modeling (SEM) to investigate the direct and indirect effects of job stress and burnout on the mental health of medical faculty members by constructing a causality among three variables. The SEM is a valid and reliable statistical tool to construct causal relationships of the data collected on a cross-sectional design. The normality test found that all the variables are within the recommended criteria of skewness (<3.00) and kurtosis (<10.00). Goodness-of-fit indexes of the path model used absolute fit measures such as chi-square (χ²) and root mean squared error of approximation (RMSEA), and incremental fit measures such as normal fit index (NFI), Tucker-Lewis index (TLI), and comparative fit index (CFI) to show a good fit between the path model and the data. RMSEA below 0.05 indicates a good fit, but below 0.08 shows a reasonable fit. NFI, TLI, and CFI greater than 0.90 indicate a good fit [16]. IBM SPSS Statistics for Windows ver. 25.0 (IBM Corp., Armonk, USA) was used to conduct statistical analyses.

4. Ethics

The Institutional Bioethics Committee of Gyeongsang National University approved our study (GIRB-A20- Y-0014). This study conformed to the ethical guidelines of the World Medical Association Declaration of Helsinki. All the participants submitted the informed consent when they replied.

1. Sample description

The 855 medical faculty members comprised men 66.7% (n=570) and women 33.3% (n=285). Ages ranged from 32 to 68 years old with an average of 46.1 years old. The ages of 41–50 years old was the largest group at 48.3% (n=414), next the 31–40 years old at 25.4% (n=218), the 51–60 years old at 22.8% (n=195), and 61 or more years old at 3.3% (n=28). Faculty members who are married comprised 75.7% (n=647) and unmarried ones 6.7% (n=57). The locations of the institutions were the largest in Seoul metropolitan area including Gyeonggi and Incheon at 42.8% (n=366), next Gyeongsang area including Gyeongbuk, Daegu, Gyeongnam, and Busan at 29.6% (n=255), Jeolla area including Jeonbuk, Jeonnam, Gwangju, and Jeju at 12.0% (n=103), Chungcheong area including Chungbuk, Chungnam, and Daejeon at 9.7% (n=83), and Gangwon area at 4.7% (n=40), respectively. A majority, 83.9% (n=717), responded that they belonged to both hospitals and schools, schools only at 8.1% (n=69), and hospitals only at 3.9% (n=33). Approximately three-quarters, 72.4% (n=619), were affiliated with clinical departments, and basic departments 9.9% (n=85). Professors accounted for 34.2% (n=292), associate professors 19.3% (n=165), assistant professors 22.2% (n=190), and clinical professors 6.5% (n=56) of the academic ranks. However, 152 faculty members (17.8%) did not disclose their academic rank. Those having worked more than 15 years were the largest group, 26.9% (n=230), next 6 to 10 years at 19.8% (n=169), 11 to 15 years at 19.6% (n=168). Approximately 38.9% (n=333) replied that their weekly working hours were 52–80 hours, over 80 hours per week at 30.4% (n=260), and less than 52 hours per week at 13.0% (n=111) (Table 1).

Table 1.

Demographic Characteristics and Working Conditions of Medical Faculty Members (n=855)

2. Perceived levels of stress, burnout, and mental health by demographic characteristics and working conditions

Perceived levels of job stress, burnout, and mental health of medical faculty members significantly differed according to demographic characteristics and working conditions. First, women responded the higher stress levels in daily tasks than men. Faculty members aged 40s responded the highest stress levels of managerial burden and organizational pressure contrary to those aged 60s and more responding the lowest stress levels for the two stressors. Faculty members whose institutions locate in the Gyeongsang area responded the higher stress levels for all three stressors than any other area. In contrast, those in the Gangwon area responded the lowest stress levels for organizational pressure and daily task. Faculty members working at both hospitals and schools responded the highest stress levels for all three stressors contrary to those working at schools only with the lowest stress levels for all three stressors. Associate professors and assistant professors responded the higher stress levels for managerial burden and daily task than clinical professors. Faculty members having worked 10 to 14 years and those working 52 to 79 hours per week responded the highest stress levels with daily tasks (Table 2).

Table 2.

Job Stress Levels of Medical Faculty Members by Demographic Characteristics and Working Conditions (N=855)

Women report higher burnout levels in emotional exhaustion and depersonalization than men. The ages of the faculty members show linear relationships with burnout levels. Those aged 30s and 40s experienced higher burnout levels for emotional exhaustion and depersonalization than other age groups, contrary to those aged 60s and more with the lowest burnout levels in two sub-dimensions. Those not married responded higher levels of emotional exhaustion than those married. Faculty members whose institutions locate in the Chungcheong area experienced the highest burnout levels in emotional exhaustion and depersonalization, and next came the Gyeongsang area. In contrast, those in the Gangwon area responded the lowest burnout levels in two subdimensions. Faculty members who work at both hospitals and schools and at hospitals only experienced higher burnout levels for emotional exhaustion and depersonalization than those working at schools only. Associate professors and assistant professors experienced higher burnout levels of emotional exhaustion contrary to clinical professors with the lowest levels. Department and employment periods did not show any significant group differences for burnout levels, but those working 52–79 hours per week showed the lowest personal accomplishment burnout (Table 3).

Table 3.

Burnout Levels of Medical Faculty Members by Demographic Characteristics and Working Conditions (N=855)

Women significantly perceived more depression and retirement than men. Faculty members aged 30s perceived the highest levels of depression and retirement, contrary to those aged 60s and more with the lowest levels. Faculty members whose institutions locate in the Geyongsang area and the Chungcheong area perceived higher levels of depression than the Gangwon area. Those working at both hospitals and schools perceived higher levels of depression and retirement than those working at schools only. Marital status, department, academic rank, employment period, and weekly working hours did not show any significant group differences in the mental health of medical faculty members (Table 4)

Table 4.

Mental Health Levels of Medical Faculty Members by Demographic Characteristics and Working Conditions (N=855)

3. Correlation analyses between variables

Correlation analyses resulted that managerial burden was positively associated with emotional exhaustion (r=0.621, p<0.01) and depersonalization (r=0.469, p<0.01) but negatively associated with personal accomplishment (r=-0.072, p<0.05). Further, managerial burden was positively associated with depression (r=0.382, p<0.01), retirement (r=0.383, p<0.01), and suicide (r=0.128, p< 0.01). Organizational pressure and daily task showed similar correlations with burnout and adverse mental health to managerial burden. On the other, emotional exhaustion was positively associated with depression (r=0.555, p<0.01), retirement (r=0.428, p<0.01), and suicide (r=0.211, p<0.01). Depersonalization also showed all positive correlations with mental health. Meanwhile, personal accomplishment showed all negative correlations with mental health (Table 5).

Table 5.

Pearson’s Correlation Coefficients between Variables

4. Causality among latent variables

This study constructed a hypothetical causality with job stress as the independent variable, mental health as the dependent variable, and burnout as the mediating variable to investigate the direct and indirect effects of job stress and burnout on mental health using the SEM. Goodness-of-fit indexes, which are within the recommended criteria, showed a good fit between the path model and the data (χ²=128.329, degrees of freedom=21, NFI=0.958, TLI=0.939, CFI=0.965, RMSEA=0.077).

All the coefficients (standardized regression weights) in the path were statistically significant at the 0.01 or 0.001 levels. First, job stress in the path model directly affected mental health (β=0.215). Next, job stress affected burnout (β=0.809) which in turn affected mental health (β=0.609). Thus, job stress indirectly affected mental health via burnout which partially mediated this relationship. The indirect effect (β=0.493) is from 0.809 ×0.609 (Table 6). Therefore, the indirect effect of job stress is relatively bigger than the direct effect on mental health. Fig. 1 illustrates the path model with three variables.

Table 6.

Regression Weights of the Latent Variables in the Path Model

Fig. 1.

The Path Model with Three Main Variables of Job Stress (MB, OP, and DT), Burnout (EE, DP, and PA), and Mental Health (DPR, RPM, and SCD)

MB: Managerial burden, OP: Organizational pressure, DT: Daily task, EE: Emotional exhaustion, DP: Depersonalization, PA: Personal accomplishment, DPR: Depression, RPM: Retirement, SCD: Suicide. **p<0.01. ***p<0.001.