While the average age of the Western populations is rising, recent trends of early retirement have been estimated to pose a threat to national economies.1 Work disability on the basis of mental health reasons in particular has increased in European countries during the past decade.2 For example, in 2006, mental disorders were the most common cause for disability pensions in Finland; of all disability pensions, 44% were granted on the basis of mental health, the most prominent single-diagnosis group being depressive disorders.3 At the same time, evidence is accumulating on psychosocial stress factors at work constituting a possible risk for mental health.4

The concept of occupational burnout was originated from studying human service professionals among whom contacts with other people constitute the major part of the task and can become a source of stress.5 Later, it has been discovered that burnout can result from prolonged work stress in a wide range of occupations.6^–8 Although burnout has shown high co-occurrence with mental disorders and physical illnesses9 10 and predicted work disability indicated by self-reported,11 company-registered,12 and medically certified sickness absences,13 recent studies have proposed that burnout would not be redundant to ill health but might mediate the association between work stress and health.12 14 15 The relationship between burnout and health can also be reversed. Employees with health problems may face difficulties in meeting their job demands, drift into jobs with low resources, or experience their work situation more negatively, and thus show higher levels of burnout.16

The core of burnout is exhaustion which, in some definitions of burnout, has been suggested to form the syndrome together with resulting doubts about the value of one’s work (cynicism) and competence (diminished professional efficacy).8 The dimensional definition of burnout, which forms the basis for the most used assessment measure, the Maslach Burnout Inventory (MBI), has been criticised for lacking theoretical arguments for grouping together an individual state (exhaustion), a coping strategy (cynicism), and an effect (diminished professional efficacy).7 So far, this controversy, on whether burnout should be defined as a state of exhaustion or as a process of resource loss, lacks consensus.17

Burnout is considered to result from long-term involvement in unfavourable psychosocial work conditions, including high quantitative and qualitative work load, role conflict and ambiguity, low predictability, and lack of participation and social support.6 18 Most models of burnout also share the assumption presented in stress theories, according to which the combination of adverse environmental conditions and dysfunctional ways of coping enhance negative health-related outcomes.6 19 The combination of high demands and low resources at work in particular has been shown to associate with burnout.20 21

There is evidence indicating that high work stress may predispose to work disability, as indicated by long-term sickness absence22 and disability pension.23 In a study conducted in the Netherlands, one-third of workers who were absent from work due to work stress problems dropped out of work life permanently after 4 weeks of absence.24 So far, burnout has been shown to predict temporary work disability, indicated by sickness absence,11^–13 but there are no studies addressing the predictive power of burnout on permanent work disability. If burnout can be shown to lead to permanent work disability, it would be reasonable to screen it as a part of workplace surveys and use it as a risk marker for adverse health-related work stress and an early alarm signal for the need for intervention.

The aim of this study was to investigate whether burnout predicts new disability pension during a follow-up period of approximately 4 years in a population-based sample of Finnish employees, after taking the health status of employees at baseline into account. We also explored the different dimensions of burnout (ie, exhaustion, cynicism, and diminished professional efficacy) separately as predictors of disability pension and the diagnosis for which the disability pension was granted in cases of burnout.

METHODS

Study population

A multidisciplinary epidemiological health survey, the Health 2000 Study, was carried out in Finland during 2000–2001. The two-stage stratified cluster sample was representative of the Finnish population and included 8028 participants aged 30 years or over.25 Stratification and sampling were conducted as follows. The strata were the five university hospital districts, each serving about one million inhabitants and differing in several features in relation to geography, economic structure, health services, and the sociodemographic characteristics of the population. First, the 15 largest towns in Finland were included with a probability of 1. Next, within each of the five districts, 65 other areas were sampled applying the probability proportional to population size (PPS) method. Finally, from each of these 80 areas, a random sample of individuals was drawn from the National Population Register so that the total number of persons drawn from each stratum was proportional to the population size of the area in question.

Procedure

The data collection phase started in August 2000 and was completed in June 2001. The participants were interviewed at home, where they were given a questionnaire to be returned approximately 4 weeks later at the clinical health examination encompassing a structured interview on mental health. During the interview the respondents received an information leaflet and their written informed consent was obtained. Approval of the Ethics Committee of Epidemiology and Public Health in the Hospital District of Helsinki and Uusimaa was obtained for the study. Information about disability pensions granted until the end of 2004 was extracted from the registers of the Social Insurance Institution of Finland and the Finnish Centre for Pensions. It was linked to the data by means of each participant’s personal identification number received by all Finns at birth and used for all contacts with healthcare institutions.

Of the total sample of participants aged 30 years or over, 5380 persons were in active working age, that is, between 30 and 60 years, and comprised the base population of our study. Of this sample, 87% were interviewed, 84% returned the questionnaire, 83% participated in the health examination, and 80% produced a valid mental health interview. Altogether, 3307 of these participants reported that they were currently working and not on maternity or parental leave. Of these, 93 were excluded due to more than one missing value per burnout inventory dimension, 63 due to missing values in the health-related variables, and 26 for the reason of already been granted a pension. Therefore, the final study population was reduced to 3125 persons. Those with one missing value per burnout inventory dimension were included in the study, and the missing value was replaced by the mean of the respondent’s existing values in that dimension.

Measures

Burnout

Burnout was measured with the Maslach Burnout Inventory-General Survey (MBI-GS).8 26 The MBI-GS consists of the following three subscales: exhaustion (five items), cynicism (five items), and (diminished) professional efficacy (six items). Satisfactory reliability and validity of the MBI-GS have been confirmed.27 28 The items were scored on a seven-point frequency rating scale ranging from 0 (never) to 6 (daily). High scores for exhaustion and cynicism and low scores for professional efficacy are indicative of burnout. The items of professional efficacy were reversed (diminished professional efficacy).

We calculated a weighted burnout sum score of the dimensional sum scores so that exhaustion, cynicism, and lack of professional efficacy had different weights in the syndrome (0.4 × exhaustion + 0.3 × cynicism + 0.3 × lack of professional efficacy).26 29 For descriptive purposes we categorised burnout as follows: no burnout (sum score 0–1.49), mild burnout (sum score 1.50–3.49), and severe burnout (sum score 3.50–6). According to this categorisation, the symptoms were experienced approximately daily or weekly with severe burnout, monthly with mild burnout, and only a few times a year or never in cases of no burnout.26 29

Mental disorders

Mental disorders were assessed with the computerised version of a Composite International Diagnostic Interview (M-CIDI) of the World Health Organization (WHO) at the end of the clinical health examination.30 The M-CIDI was carried out by healthcare workers who were trained for the interview by psychiatrists and physicians who had themselves been trained by a WHO authorised trainer. The program uses operationalised criteria for diagnoses in Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) and allows the estimation of DSM-IV diagnoses for major mental disorders. The participant was identified as having a mental disorder if he/she fulfilled the criteria for a depressive, anxiety, or alcohol use disorder.

Physical illnesses

At the beginning of the clinical health examination, a symptom interview was carried out. After several measurements the research physician took a history and performed a standard 30-minute clinical examination. The diagnostic criteria of the physical illnesses were based on current clinical practice. In the present study, the participant was identified as having a physical illness if he/she fulfilled the diagnostic criteria for musculoskeletal disorder, cardiovascular disease, respiratory disease, or other physical illness.

Work disability pensions

There are two complementary pension systems in Finland: earnings-related pension linked to past employment and national pension linked to residence in Finland. Of the four main types of pensions (ie, old-age, disability, unemployment, and part-time pension) only the work disability pensions are health related. A full-time or part-time work disability pension can be granted due to a chronic illness, handicap, or injury which has resulted in decreased work ability.31 The pension is granted as temporary if rehabilitation or treatment can be reasonably assumed to improve work ability and return to work life seems plausible. A special form of disability pension, “individual early retirement pension”, has been available for older employees born before 1944, who have had a long work career and whose work capacity has substantially decreased, but who have not fulfilled the criteria for ordinary disability pension.31

All pensions granted to the participants until December 2004 were extracted from records of the Social Insurance Institution of Finland and the Finnish Centre for Pensions. A variable for new disability pension, including full-time and part-time work disability pensions and individual early retirement pensions, was formed for pensions granted during the follow-up, that is, after the date of the clinical health examination in the Health 2000 Study until December 2004. The main diagnosis for which the disability pension was granted was categorised according to the International Statistical Classification of Diseases and Related Health Problems, the ICD-10, as: neoplasm and diseases of the blood (C00-D89), mental and behavioural disorders (F00-F99), diseases of the circulatory system (I00-I99), diseases of the musculoskeletal system (M00-M99), injuries (S00-T98), and other diseases (all other ICD-10 codes), reflecting the main causes for work disability in Finland.3

Demographic factors

Information on the following sociodemographic factors was collected in a home interview: sex, age, marital status, occupation, and type of business and sector. Age was classified as 30–39, 40–49, and 50–60 years for descriptive purposes and used as a continuous variable in the analyses. Marital status was divided into the following two groups: married or cohabiting (married) and other (unmarried). Occupational grade was formed on the basis of occupation and type of business or sector: upper-grade non-manual, lower-grade non-manual, manual, and self-employed. Sector was dichotomised as human service sector and non-human service sector.

Statistical analysis

The associations between sociodemographic and clinical factors, burnout, and disability pension were described by cross-tabulations and χ²-tests. Logistic regression analyses were used to calculate the odds ratios and their 95% confidence intervals for new disability pension during the follow-up separately in relation to the burnout syndrome and individual burnout dimensions (ie, exhaustion, cynicism, and diminished professional efficacy) used as continuous variables. The analyses were adjusted stepwise for sociodemographic factors (ie, gender, age, marital status, occupational grade, and sector), mental disorders, and physical illnesses at baseline. Interaction terms were applied to test whether the association between burnout and disability pension is dependent on gender, because some earlier studies have detected gender differences in work disability.32 33

Weighting adjustment and sampling parameters were used in the analyses to account for complexities in the survey design, including clustering in a stratified sample, and the loss of participants.25 34 The data were analysed using the Sudaan 9.0.1 statistical program package, which is specifically designed for analysing cluster-correlated data in complex sample surveys.

RESULTS

The study population comprised 1564 men and 1561 women. Altogether 2% of the participants had severe burnout and 25% mild burnout. The level of burnout was related to age, marital status, occupational grade, and the presence of a mental disorder or physical illness. Those with severe burnout were more often older, unmarried, engaged in manual work and had a mental or physical disorder more often compared with the others (table 1).

Of the participants, 113 persons were granted a new disability pension during the follow-up. A new disability pension was more common among those with higher levels of burnout at baseline (df = 2, χ² = 25.6, p<0.001). Of those with severe burnout, 22% were granted a new disability pension during the follow-up, whereas the corresponding percentages among those with mild and no burnout were 6% and 2%, respectively. Table 2 shows that each one-point increase in the overall burnout sum score was related to 70% increase in the odds for a future disability pension when sociodemographic factors were adjusted for. After additional adjustments for baseline mental disorders and physical illnesses, the odds for a new disability pension attenuated but still remained statistically significant (table 2). The interaction effect between gender and overall burnout on disability pension did not reach statistical significance (p = 0.187).

The crude associations between the dimensions of burnout and new disability pension were statistically significant but after adjustments for sociodemographic factors, only exhaustion and cynicism predicted new disability pension (table 2). A statistically significant interaction effect was found between gender and exhaustion predicting disability pension (p = 0.044), and therefore the association between exhaustion and disability pension was analysed by gender. After adjustment for mental disorders, the prospective association between exhaustion and disability pension remained statistically significant among both genders, but when physical illnesses were also accounted for, the exhaustion dimension of burnout predicted disability pension only among men. The association between cynicism and new disability pension remained significant after all adjustments in the combined group of men and women.

The diagnosis on which a new disability pension was based also related to the level of burnout (table 3). Compared with employees without burnout, a new disability pension was very often granted on the basis of mental and behavioural disorders among those with severe burnout and on the basis of musculoskeletal disorders among those with mild burnout.

DISCUSSION

This study on a large population-based cohort sample of Finnish employees aged 30 to 60 years with a follow-up of approximately 4 years showed that burnout predicts permanent work disability. To the best of our knowledge, this is the first study to examine this relationship. Our study benefited from a high rate of participation, standardised assessment of common mental disorders and physician-confirmed diagnoses of physical illnesses at baseline, and nationwide register data on disability pensions.

Burnout has previously been shown to predict sickness absence among specific occupational branches.11^–13 In our previous report on the same population-based sample as used in the present study, we found that burnout is associated with long-term medically certified absences irrespective of the co-occurring common mental disorders and physical illnesses.33 Now we showed that burnout also predicts new disability pension after mental and physical health at baseline is accounted for. Because burnout as such does not qualify as a cause for disability pension in Finland, it is possible that burnout has contributed to the outbreak of a disorder, worsening of an already existing illness, or to decreased resources to cope with an illness at work. Even though previous studies have shown prospective association between burnout and risk factors of illnesses,35 symptoms of disorders,36 or medically certified sickness absences,13 we cannot altogether rule out the possibility that subclinical phase of an illness has contributed to both burnout and work disability. Therefore, more longitudinal evidence is needed on whether burnout can actually predict new disorders or illnesses in addition to predicting decrease in work ability.

We found that mental and behavioural disorders as well as musculoskeletal disorders were the most common diagnosis groups behind burnout-related disability pensions, as they are also the commonest causes behind work disability in general in Finland.3 This is in line with earlier results of burnout associated with these groups of common disorders.9 10 However, more research is called for to clarify whether burnout-related work disability is cause specific or more heterogeneous in nature.

As the separate dimensions of burnout syndrome were examined, diminished professional efficacy was not related to work disability after adjustments. This is in accordance with our earlier finding, based on the same sample, that diminished professional efficacy was not related to long-term sickness absence.33 Even though all three dimensions of burnout have earlier been found to associate with health indicators separately,9 10 other previous results have supported the central status of exhaustion and cynicism in the burnout syndrome.14 The obtained results suggest that the exhaustion and cynicism dimensions constitute the core of burnout, at least when work disability is considered. However, the burnout syndrome, expressed by the overall burnout sum score including the scores on all three dimensions of burnout, showed in our study a strong association with new disability pension among both genders even after the associations were adjusted for baseline health status. This finding indicates the usefulness of the “total burnout” concept, compared with mere exhaustion dimension, in predicting work disability. Similar results have also been found earlier concerning overall burnout and medically certified sickness absences.13

A gender interaction emerged in the present study regarding the associations of the exhaustion dimension of burnout. After adjustment for physical illnesses, exhaustion predicted new disability pension only among men. This finding is in line with earlier results which have indicated that the relationships between burnout and health tend to be stronger for men than for women. Among men, even mild burnout was associated with depressive disorders9 and severe burnout was related to delayed return to work.33 This phenomenon might result from the stigma and social consequences faced by men at work when showing symptoms of exhaustion, as elaborated by Hensing et al.37 Because, according to cultural role expectations, men are not supposed to show weakness, they might not obtain social support from other people and seek help too late, leading to the consequences of work-related problems becoming more invalidating to them than to women. On the other hand, it has been established that the threshold for being granted disability pension is higher for women compared with men: the applications are more often rejected in the case of women.31

Both individual and work-related factors have previously been found to affect work disability.38 39 Age has been found to prospectively increase the risk in random population-based samples, as also female gender to some extent.32 39 Among employees with a depressive disorder, also low vocational education, the presence of physical illnesses, low self-esteem, and hopelessness contributed to disability pensioning.40 41 According to present results, burnout could be used as a proxy for the risk of work disability. Burnout can be screened at a group level as a part of work place surveys by human resource management and occupational health professionals. High prevalence of burnout would then signal the need for preventive intervention, before it develops into depression.36

Yet, high quality of work (ie, work experienced as meaningful and including good possibilities to influence one’s work and develop in it), satisfaction with working hours, and interaction with others at work served as buffers against early exit from work in a 16-year follow-up study among Finnish municipal workers.42 Social support at work was also found to decrease the risk for disability pension in a population-based Finnish sample43 and in a random Danish sample.39 As psychosocial work conditions also relate to the development of burnout,6 11 continuous assessment and improvement of work conditions are called for to extend individuals’ participation in the labour force. Studies have shown that change in working conditions can generate positive effects on decreasing levels of burnout,44 45 experienced fatigue,46 and even sickness absence.22

Individual-focused intervention programmes targeted at risk groups have also had success in decreasing burnout47 and sickness absences48 in randomised controlled trials. However, the results of individual-focused interventions do not always last long47 or go beyond the experienced well-being.49 Therefore, it seems that comprehensive programmes combining interventions at individual and organisational levels might help to obtain effective and far-reaching outcomes.50

There is no consensus on the definition of burnout, and therefore several conceptualisations and operationalisations exist.7 17 In the present study, burnout was measured with a general version (MBI-GS) of the most widely used burnout instrument, the MBI,8 but also exhaustion and other dimensions were used as outcomes in the analyses as well. The nationally established procedure to form a weighted sum score for the burnout syndrome has been published, but the clinical validity of the three-level categorisation has not been confirmed.26 29

Because of the relatively short follow-up period of approximately 4 years, the number of cases with new disability pension was relatively low and did not allow for stratified analyses by background variables or by separate diagnosis groups. It would be important in the future to explore further the gender differences in pensioning and the associations between burnout and work disability due to various health problems and in different occupational groups with a longer follow-up period. The diagnoses due to which the disability pensions are granted for may cause slight imprecision in the results, because the accepted cause for a pension may not unambiguously account for all problems the employee is suffering from. Therefore the cause-specific results between burnout and disability in the present study are suggestive, especially because we only took the main cause for disability into consideration. The comorbidity between mental disorder and musculoskeletal disorder diagnoses was low in this sample (three cases) and did not occur at all in the severe burnout subgroup. In addition, all possible diseases the participants might have suffered from at baseline were not assessed; only the most prominent disease groups related to work disability were covered. Therefore, the result concerning the independent effect of burnout on work disability in the present study is an approximation, and needs to be verified in future research.

CONCLUSION

Burnout was an independent predictor of work disability. Therefore it could be used as a marker of chronic health-related work stress. To prevent early exit from workforce, work conditions and employee burnout could be regularly assessed with the help of occupational health services. Working conditions should be continuously developed and support offered to risk groups in order to facilitate health-enhancing work places.