The role of psychosocial working conditions on burnout and its core component emotional exhaustion – a systematic review

Contemporary workplaces are characterized by increasing work pace, increasing expectations for self-actualization, increasing reliance on interpersonal coordination in the execution of work tasks, increasing pace of change, and increasing job insecurity. Against this background, the influence of psychosocial working conditions on mental health, but also on cardiovascular and musculoskeletal disorders is under discussion [1–5].

On a general level, mental disorders are characterized by multifactorial genesis in which genetic as well as environmental factors play an etiologic role [6]. In this context, it is increasingly recognized that working conditions are also important determinants of mental health as it can have both beneficial and harmful influences on the individual worker’s health status [7, 8].

In the mid-1970s, the investigation of the so-called burnout phenomenon was established at first with respect to service professionals [9]. Since then the burnout syndrome, particularly its relationship to prolonged interpersonal work-related stress has been controversially discussed from various perspectives (e.g. in the fields of psychiatry, psychosomatics, occupational medicine and health psychology) [10–12]. Maslach et al. divided the burnout syndrome into three dimensions: emotional exhaustion (feelings of being “empty”) as the core component, depersonalization (negative, cynical attitude toward the work or the recipients) and reduced personal accomplishment / professional efficacy (negative evaluation of the achievements at work) as additional components [10, 13–16]. Pines et al. [17] defined burnout not only as a state of emotional, but also physical and mental exhaustion in long-lasting emotionally demanding situations. It is becoming apparent that especially emotional exhaustion can be understood as the core component of burnout. In understanding burnout as a stress-related adaptive process, Nil et al. [11] described the development of burnout “… as going through several phases, from increased working efforts to cope with external demands, which can lead to mental and physical exhaustion and demotivational affective states, and on to psychosomatic complaints and finally depressive state” (p. 72). Burnout is not regarded as a medical or psychiatric diagnosis as defined by the ICD-10 or DSM-IV manuals. In the ICD-10 burnout has the status of a residual category [18]: “Z73.0: problems associated with difficulties in coping with life, that affects the health and leads to a demand of health services”. The differential diagnosis of burnout to clinical diseases is maldefined and includes ruling out other diseases with similar clinical components, e.g. depression (ICD-10 diagnosis code F32.), neurasthenia (F48.0, also fatigue syndrome), chronic fatigue syndrome / benign myalgic encephalomyelitis (G93.3), insomnia (F51.0) or posttraumatic stress disorder (F43.1) [19].

Although there are currently no consistent standardized diagnostic criteria for burnout, estimates on the prevalence of burnout were made in some studies, specifically related to specific occupational groups. Research findings about the prevalence of burnout exist primarily for service-related professions (e.g. teachers, nurses or physicians). Depending on the professional group, the prevalence estimates differ significantly from each other. Nil et al. [11] investigated the status quo of burnout prevalence rates, estimating that between 3.5% and 12% of Swiss primary care doctors [20], and up to 50% of emergency room workers were affected [21, 22]. Arora et al. [23] reported a burnout prevalence among orthopaedic surgeons between 50-60% and among surgeons in general between 30-40%. Of course the determinability of the prevalence for a phenomenon that is not consistently defined might be questioned. However, burnout affected persons clearly suffer from the symptoms of this syndrome [19]. Moreover, burnout potentially constitutes a developmental state of a depressive disorder [19]. Therefore, the high – although inconsistent – estimations on the prevalence of burnout point to a high public health impact.

In general the burnout “diagnosis” is based on self-reported screening instruments. The Maslach Burnout Inventory (MBI) including its specific versions and subscales (e.g. MBI-General Survey - GS) is the most commonly used burnout measure, which is applied in most of all studies concerning burnout [24, 25]. Other established burnout questionnaires are the Copenhagen Burnout Inventory (CBI [26]), the Oldenburg Burnout Inventory (OLBI [27]), the Shirom Melamed Burnout Measure (SMBM [28]), Gillespie-Numerof Burnout Inventory (GNBI [29]) or the Tedium Measure (TM [17, 30]). Given the different definitions of burnout in these listed questionnaires, the comparability of the results is somewhat limited.

Existing systematic reviews on the psychologic effects of psychosocial working conditions mainly focused on depression [31, 32], mental disorders in all [33], and stress-related disorders [34]. There are also some systematic reviews that focus on the relationship between (inter alia) work-related factors and burnout [35–41]. However, the mentioned systematic reviews on burnout address very specific occupational groups: medical residents [36, 37], pediatric oncology staff [38], correctional officers [35, 40], elderly care staff [39], or nurses [41]. Thus, the generalizability of the main results to other occupational groups might be restricted. In addition, a quality assessment of included studies, which is an essential work step preparing a systematic review, was unfortunately conducted in only three of the fore cited previous systematic reviews [37, 38, 40]. In contrast to the topic of psychosocial working conditions and cardiovascular diseases – where several longitudinal studies can be found – most studies on burnout/emotional exhaustion apply a cross-sectional approach. All mentioned reviews therefore included cross-sectional studies which might have introduced substantial cause-and-effect bias. The few reviews that also included some longitudinal studies found limited or unclear evidence for the relationship between psychosocial working conditions and burnout: In the review published by Thomas [36], only one of 12 included observational studies on the association between psychosocial working conditions and burnout applied a longitudinal design; Prins et al. [37] identified three longitudinal studies with inconsistent findings concerning the impact of long working hours.

The aim of this systematic review is to provide a comprehensive overview about the effects of psychosocial working conditions on the development of burnout and its core symptom, emotional exhaustion. In contrast to the mentioned previous reviews the provided systematic review on psychosocial working conditions and burnout excludes cross-sectional designs; moreover, this systematic review allies a comprehensive quality assessment with standardized checklists.

Of 5,599 initially identified search hits and 16 articles found by hand search, 749 publications were excluded by publication type (comments, chapters in books, dissertations) or as duplicates at this stage. The titles and abstracts of 4,866 citations were screened by two independent reviewers, and 4,320 articles were excluded after review. Of the resulting 546 articles, 387 were excluded after full-text review because they did not meet the eligibility criteria, so that a total of 159 remained (Figure 1), 65 for burnout / emotional exhaustion and 95 for mental disorders (one study reported about both burnout / emotional exhaustion and depression). After quality assessment, a further 109 publications were discarded due to an inadequate quality score (-). The main reasons for an inadequate quality score were as follows:

– To avoid selection bias, the study base has to be clearly defined. Ad-hoc samples or the recruitment of health conscious subjects do considerably affect the external validity of the study. Furthermore, the studies should have a sufficiently high initial response (of about 50% or more). Moreover, to avoid substantial selection bias, the loss to follow-up should be low. Studies with a loss to follow-up of 50% or more are prone to substantial selection bias, so they were classified as methodologically inadequate, even if other sources of bias were lacking.

– To avoid reverse causation, the exposure had to be measured at baseline.

– The study population should not suffer from the outcome of interest at baseline to allow the distinction between before-and-after effects. At least baseline mental health had to be taken into account in the analyses (by adjustment).

– Confounding should be adequately considered. As a minimum, age and gender had to be taken into account (e.g. by adjustment or stratification).

Additional file 3 gives a list of the studies that were excluded based on inadequate quality assessment; the concrete reasons for inadequate quality scores are given by Thinschmidt et al. [55]. Of the remaining 50 studies, six methodologically adequate longitudinal studies dealt with emotional exhaustion or burnout and were therefore included in this systematic review. The characteristics of these six cohort studies are summarized in Table 2 and described narratively below. All of these six studies fulfilled some of the above mentioned quality criteria (quality score +), there were no studies that fulfilled all or most of the criteria (quality score ++).

The follow-up study of Ahola and Hakanen [45] recruited 3,255 of 4,588 dentists employed in clinical work in Finland. 2,555 subjects (74% women) participated at a three-year follow-up in 2006. The study is characterized by a high initial response and a relatively low loss to follow-up. Burnout was assessed with the Maslach Burnout Inventory (MBI). This study is the only included publication that did not consider the different burnout subscales separately, but considered burnout as total score. Furthermore, this publication examined the potentially mediational relationships between burnout and depression under the condition of job strain (measured by the Job Content Questionnaire, JCQ). Using logistic regression models (all models adjusted at least for sex, age, marital status, baseline burnout and depressive symptoms) job strain at baseline was associated with burnout at follow-up among those participants who were free of depressive symptoms at baseline (OR adjusted for – inter alia – baseline burnout = 1.8, 95% CI 1.0-3.1 for each one-point increase in job strain), and also for those with existing depressive symptoms at baseline (OR adjusted for baseline burnout = 2.2, 95% CI 1.4-3.4). Odds ratios were remarkably higher among those free of burnout at baseline and differed between men and women: job strain predicted burnout stronger in men (adj. OR for each one-point increase in job strain = 22.3, 95% CI 5.1-98.1, adjusted for baseline depression) than in women (adj. OR 4.0, 95% CI 2.0-8.0, adjusted for baseline depression).

Using longitudinal data from the Maastricht Cohort Study on “Fatigue at Work” (n = 5,256 male and female employees, numbers of females and males not given), Janssen and Nijhuis [46] analyzed the effects of positive changes in perceived work characteristics on fatigue, but also on emotional exhaustion over the period of one year (1998–1999) in a heterogeneous working population including all sectors. The initial response was low, but loss to follow-up was only moderate. As the selection of the final sample is clearly described, the study was classified as methodologically acceptable despite of the low initial response. The study examined the influence of a positive change in job demands, decision latitude, and social support (measured by the JCQ) on a reduction in emotional exhaustion (measured by the MBI-GS, subscale EE; one-year-span). Hierarchical regression models were performed. Emotional exhaustion decreased with decreasing job demands, increasing decision latitude and increasing social support compared with a situation without changing work characteristics. The authors did not conduct gender-specific analyses.

In their two-wave panel study Langballe et al. [47] observed a random sample of 291 female and 232 male Norwegian doctors (from each of 500 subjects) over a follow-up-period of two years (2003–2005). The authors examined the associations between work characteristics (workload, autonomy, measured by a three- or a four-item scale based on the Job Stress Questionnaire, JSQ) and physicians’ burnout (measured by the OLBI-subscales exhaustion) considering individual factors and work-home-interaction. The initial response and loss to follow-up was moderate for both genders. In hierarchical multiple regression models the effects were examined separately for males and females. Control variables (age, marital status, number of children younger than six years, working hours), individual variables (job performance-based self-esteem, goal orientation, value congruency), and work-home-interaction (work-home/home-work conflict, work-home/home-work facilitation) were included in the analysis, controlled for baseline exhaustion. Workload was significantly positively associated with exhaustion at follow up in both sexes, and autonomy had a buffering effect on exhaustion only in male physicians.

Lorente Prieto et al. [48] investigated the relationship of job demands and burnout in 274 teachers (57% females) drawn from secondary schools in Spain. Initial response and loss to follow-up were acceptable. The period between the two surveys was only eight months. Exposure was operationalized by job demands (i.e., quantitative overload, mental demands, emotional demands, role ambiguity, and role conflicts). Burnout was measured by the subscale exhaustion of the Spanish version of the MBI-GS. After the stepwise introduction of the different independent variables into the hierarchical regression model and control for baseline exhaustion in the final model (adjusted for age and gender, job demands, job resources, personal resources), only quantitative overload at baseline was significantly related to exhaustion at follow-up. Women felt more exhausted than men, even when controlling for the baseline levels of burnout symptoms. However, the authors did not conduct gender-specific analyses of the relationship between psychosocial working conditions and exhaustion.

The two-wave panel study of Taris et al. [50] examined the effects of job characteristics (job demands, job control) on emotional exhaustion (MBI-GS) in Dutch police officers (n = 5,277; 17% females) during one year. After exclusion of subjects who refused participating at the follow-up investigation (48%) and of subjects with serious burnout symptoms at baseline, 1,000 subjects were randomly selected to be included in the follow-up, of whom 828 participated. As the sampling procedure is well-described and as subjects with serious burnout symptoms were excluded from baseline, the overall study quality was assessed as just acceptable despite of the relatively high loss to follow-up. Using structural equation modeling (SEM) high demands at baseline were related to higher emotional exhaustion at follow-up. In the summary model no significant association was found between job control at baseline and emotional exhaustion. The authors do not report gender-specific analyses.

In a one-year follow-up study, Van Vegchel et al. [52] included 2,255 Swedish human services employees (59% females) of Social Insurance Organizations. Initial response was good (76%) and loss to follow-up was moderate (29%). The primary aim of the study was to examine the moderating effect of both social support and job control on the relationship between quantitative or emotional job demands and burnout, assessed by the subscale emotional exhaustion (MBI). In the hierarchical regression analysis, quantitative demands and emotional demands were separately examined with respect to burnout and its dimensions. The authors found positive associations between quantitative as well as emotional demands with burnout at follow-up. Burnout was negatively associated with control and social support at baseline. Emotional demands had a slightly stronger effect on burnout than quantitative demands. Job control was found to constitute an effect modifier of the relationship between emotional demands and emotional exhaustion. The authors do not report gender-specific analyses of the association between psychosocial working conditions and burnout or emotional exhaustion.

The studies included in our systematic review varied greatly with respect to their population, study design, the number of occupational groups included, the instruments applied to measure exposure and outcomes and in the method of data analysis. Because a sufficient number of studies with comparable exposure measurements (e.g. job content questionnaire JCQ and single items), outcome measures and comparable analytic approaches was lacking, we qualitatively summarized the evidence for an association between the single psychosocial work exposures and burnout.

All six included studies examined the relationship between work task characteristics and burnout or its core component emotional exhaustion. In five of these studies [46–48, 50, 52], the association between high workload / quantitative demands / overload and the development of the burnout core-component emotional exhaustion was examined. All of these studies found a statistically significant increase in emotional exhaustion among persons with high workload or high quantitative demands/ overload. Moreover, statistically significant associations between high emotional demands and emotional exhaustion could be found by van Vegchel et al. [52]. Concerning the influence of low job control on burnout dimensions, we found inconsistent results: While Van Vegchel et al. [52] as well as Janssen and Nijhuis [46] found a significant relationship between low job control or the analogous decreased decision latitude and emotional exhaustion, Langballe et al. [47] revealed a statistically significant relationship between high autonomy and increased emotional exhaustion among male physicians.

Three studies examined the influence of social conditions on single burnout dimensions. In two studies low social support predicted increased emotional exhaustion [46, 52]. Role conflicts were not found to be associated with burnout subscales [48].

Work organizational factors were assessed only in one study [47]. In this study no significant influence of the number of working hours on emotional exhaustion was found for both genders.

Altogether, most studies applying an adequate methodology found a relationship between “poor” psychosocial working conditions and burnout/emotional exhaustion. According to the Oxford Centre for Evidence Based Medicine (CEBM) Levels of Evidence Working Group [56], the level of evidence formally is classified as level 3 because exclusively non-randomized controlled cohort/follow-up studies could be identified.

In our comprehensive systematic review, only six longitudinal studies of adequate methodological quality could be identified that investigated the relationship between psychosocial working conditions (categories: work organization, work task and social conditions) and the development of burnout/emotional exhaustion. We found a relatively consistent association between “unfavorable” psychosocial working conditions (high workload, high quantitative, mental or emotional demands, low social support) and burnout and emotional exhaustion as the core burnout component. The influence of low job control on burnout dimensions was inconsistent. Altogether, with the exception of one study [47] which found that autonomy was a risk factor for later emotional exhaustion among male physicians, we found no statistically significant associations between “favorable” psychosocial working conditions and burnout or emotional exhaustion.

This systematic review of psychosocial working conditions and burnout and its core component emotional exhaustion found very few methodologically adequate studies. We therefore have to state a strong discrepancy between the high presence of burnout risks in the present public discussion and the low number of corresponding methodologically sound studies. Selection effects, a possible consequence of high loss-to-follow-up, were one of the most common limitations present in the original studies.

Nevertheless, the resulting evidence of our systematic review together with the evidence of a simultaneously conducted review on mental disorders [55] indicates that mental health is related to psychosocial working conditions. Therefore the creation of healthy working conditions appears to be essential for preventing psychological disease. Both objectifiable as well as “subjective” stress might impair mental health. Of course, redesigning working conditions can naturally only influence the “objective” stressors. Identifying effective preventive approaches requires the realization of methodologically impeccable intervention studies. This in mind, additional intervention studies are greatly needed. No one isolated workplace factor could be identified that could serve as the focus of such intervention studies. More probably - analogous to other diseases caused by a multitude of factors - comprehensive multidimensional approaches to designing working conditions may prove to be most effective for preventing psychological diseases. Psychosocial working conditions can not only influence emotional exhaustion/burnout, but also a wide variety of other health outcomes as cardiovascular diseases and musculoskeletal disorders. Therefore the improvement of psychosocial working conditions might not only serve to prevent burnout, but is also a relevant approach to the promotion of public health.