Burnout and health: Current Knowledge and Future Research Directions

Transcription

1 Arie Shirom and Shmuel Melamed Burnout and health: Current Knowledge and Future Research Directions Burnout has received increased research attention in recent years. During the period of , annually about 150 articles that concerned burnout have appeared in journals covered by PsychInfo. As evident, burnout has been a major focus of researchers efforts. A recent review of the area of burnout (Schuafeli & Enzmann, 1998) found about 5,500 entries with burnout as a key word between 1975 and Notwithstanding this large number of studies, the relationships between burnout and physical health, including physiological risk factors and physical disease states, have hardly been explored. The literature on burnout and well being or mental health is substantial, but has not been reviewed with sufficient attention to the instruments used to gauge burnout and their respective construct validity. The objectives of this chapter are to review current knowledge on the above issues and to provide a perspective on future directions of research into burnout health linkages. We start out by discussing the conceptual meaning of burnout. Burnout is viewed as an affective reaction to ongoing stress. We contend that core content of this affective reaction is the gradual depletion over time of individuals intrinsic energetic resources, leading to feelings of emotional exhaustion, physical fatigue, and cognitive weariness (Shirom, 1989). Given the multidimensionality of the construct and the controversy over its operational definition (Maslach, Schaufeli, & Leiter, 2001), this conceptual analysis is essential for understanding the possible health consequences of burnout. We seek in this chapter to propose mechanisms that link burnout and health (cf. Schaufeli & Greenglass, 2001). The following sections cover the empirical literature on burnout s linkages with mental and physical health, respectively.

2 Typically, empirical studies on the health consequences of burnout are based on a cross-sectional study design and measure burnout, and also mental health, by asking respondents to complete a self-report questionnaire. In this review, the emphasis is on longitudinal studies on burnout s impact on health since they provide more credence to cause and effect statements. The voluminous empirical research on burnout has already been reviewed by several by meta-analytic studies (Collins, 1999; Lee & Ashforth, 1996; Schaufeli & Enzmann, 1998). Most of this research has measured burnout by the Maslach Burnout Inventory (MBI). In references made to this body of studies, we will focus on results reported for the emotional exhaustion scale. This review covers burnout of employees in work organizations, excluding research that deals exclusively with non-employment settings (e.g., athletes burnout: Dale & Weinberg, 1990). Also excluded is research that deals with burnout in life domains other than work, like crossover of burnout among marital partners (e.g., Pines, 1996; Westman & Etzion, 1995). The Conceptual Basis of Burnout During the 1980s and early 1990s, research on burnout, regardless of the conceptual approach employed, dealt almost exclusively with people-oriented professionals (e.g., teachers, nurses, doctors, social workers, and policepersons). People-oriented professionals often enter their mostly public sector profession with service-oriented idealistic goals. They typically work under norms that expect them to continuously invest emotional, cognitive and even physical energy in service recipients. In most of today s advanced market economies, the public sector has to adjust to consumers growing demands for quality service, downsizing, and budgetary retrenchments. Inevitably, such a context of overloading and conflicting demands is a fertile ground for creating a process of emotional exhaustion, mental

3 weariness and physical fatigue that Freudenberger, who pioneered in scientifically investigating this phenomenon, labeled as burnout. Freudenberger (1974, 1980) pioneering clinically oriented work on burnout inspired three different conceptual approaches to burnout, each with its distinct measure. We refer to the conceptual schemes and measures of Maslach and her colleagues (Maslach, 1982; Maslach & Leiter, 1997), of Pines and her colleagues (Pines & Aronson, 1988; Pines, Aronson & Kafry, 1981) and of Shirom and Melamed (Shirom, 1989; Hobfoll & Shirom, 1993, 2000; Melamed, Kushnir & Shirom, 1992). In this review, we will emphasize issues related to the validity of the first conceptual approach toward burnout, including the measurement instrument constructed by Maslach and her colleagues, the MBI (Maslach, Jackson, & Leiter, 1996). The reason for this focus is that the MBI was one of the very first scientifically validated burnout measurement instruments, and it has been the most widely used in scholarly research (Schaufeli & Enzmann, 1998). The first version of the MBI reflected the field s preoccupation with professionals in people-oriented occupations. Subsequently, the construction of newer versions of the popular MBI, applicable to other occupational groups (Maslach et al., 1996) extended the study of burnout to other categories of employees. The Maslach Burnout Model and Inventory According to this conceptualization (Maslach & Jackson, 1981; Maslach, 1982; Maslach & Leiter, 1997), burnout is viewed as a syndrome that consists of three dimensions: emotional exhaustion, depersonalization, and reduced personal accomplishment. Emotional exhaustion, which refers to feelings of being depleted of one s emotional resources, is regarded as the basic individual stress component of the syndrome (Maslach, Schaufeli, & Leiter, 2001). Depersonalization, referring to negative, cynical or excessively detached responses to other people at work, represents the interpersonal component of burnout.

4 Reduced personal accomplishment, referring to feelings of decline in one s competence and productivity and to one s lowered sense of self-efficacy, represents the self-evaluation component of burnout (Maslach, 1998, p. 69). The three dimensions were not deducted theoretically but resulted from labeling exploratory factor-analyzed items initially collected to reflect the range of experiences associated with the phenomenon of burnout (Maslach, 1998, p. 68; Schaufeli & Enzmann, 1998, p. 51). Subsequently, Maslach and her colleagues modified the original definition of the latter two dimensions (cf. Maslach et al., 2001, p. 399). Depersonalization was renamed cynicism, though it still referred to the same cluster of symptoms involving other people at work rather than service recipients. However, the new label for this dimension of the syndrome poses new problems. As an emerging new concept in psychology and organizational behavior, the term cynicism is used to refer to negative attitudes involving frustration from, disillusionment with, and distrust of organizations, persons, groups or objects (Andersson & Bateman, 1997; Dean, Brandes, & Dharwadkar, 1998). Abraham (2000) has suggested that work cynicism, one of the forms of cynicism that she had identified in her research, tends to be closely related to burnout. Garden (1987) has argued that this dimension of the syndrome of burnout gauges several distinct attitudes, including distancing, hostility, rejection, and unconcern. It follows that the discriminant validity of this component of burnout, relative to the current conceptualizations of employee or work cynicism or relative to the other distinct attitudinal concepts noted by Garden (1987) has yet to be established. The third dimension was re-labeled as reduced efficacy or ineffectiveness, depicted to include the self-assessments of low self-efficacy, lack of accomplishment, lack of productivity, and incompetence (Leiter & Maslach, 2001). Each of these concepts, namely self-efficacy, accomplishment or achievement, personal productivity or performance, and

5 personal competence, represents a distinct field of research in the behavioral sciences and the authors of the MBI have yet to clarify on what theoretical grounds they can be grouped together to represent a single conceptual entity. Does reduced efficacy refer to one s confidence in one s capability to successfully execute courses of action required to deal with prospective tasks, as self-efficacy is customarily defined (e.g., Lee & Bobko, 1994; Stajkovic & Luthans, 1998)? Does the third dimension of burnout reflect one s belief in one s knowledge and skills, as competence is often conceptualized (Foschi, 2000; Sandberg, 2000)? As an additional alternative, does it relate to self-assessed job performance or performance expectations (e.g., Stajkovic & Luthans, 1998)? It appears that each of the second and third dimensions of the MBI, as currently defined, probably represents each several multifaceted constructs, each having different theoretical implications with regard to the emotional exhaustion component of burnout suggested by the authors of the MBI (cf. Moore, 2000, p.341). Clearly, the conceptualization of burnout as tapped by the MBI relates to it as a multidimensional construct. A construct is multidimensional when it refers to several distinct but related dimensions that are viewed as a single theoretical construct (Law, Wong, & Mobley, 1998). The proponents of this multidimensional view of burnout (e.g., Maslach, 1998) argue that it provides a holistic representation of a complex phenomenon, broadly conceived as referring to the process of wear and tear or continuous encroachment upon employees resources. However, they have yet to provide convincing theoretical arguments as to why the three different clusters of symptoms that comprise their conceptualization of burnout should hang together (cf. Maslach et al., 2001). They further argue that their conceptualization allows researchers to use broadly conceived types of stress in both the work and the family domains as potential antecedents of burnout, thus increasing its explained variance. However, there is a paucity of evidence that there are specific antecedent

6 variables or mechanisms leading to all of the three clusters of symptoms included in the MBI (Collins, 1999; Lee & Ashforth, 1996; Schaufeli & Enzmann, 1998). A case in point is the phase model of burnout, developed by Golembiewski and his colleagues and tested in a series of studies (see, for example, Golembiewski & Boss, 1992; Golembiewski, Munzenrider, & Stevenson, 1986; Golembiewski & Munzenrider, 1988). One of the theoretical assumptions upon which this model was based was that individuals experiencing burnout on the dimension of emotional exhaustion do not necessarily experience either of the other two clusters of symptoms. Indeed, Golembiewski and his colleagues (Golembiewski et al., 1986,1988, 1992) have provided the notion that each phase or dimension of burnout may develop independent of each other. Maslach (1998, p. 70) has argued that adding the dimensions of cynicism and reduced personal efficacy to the core dimension of emotional exhaustion was justified in that they add the interpersonal aspect of burnout to the conceptualization of the phenomenon. However, the emotional exhaustion scale of the MBI already includes items that tap interpersonal aspects of work, like working with people all day is really a strain for me, and Working with people directly puts too much stress on me (Maslach & Jackson, 1981). Conceptually, therefore, the view of burnout as a syndrome consisting of three clusters of symptoms lacks theoretical underpinnings, has not been supported by evidence demonstrating a common etiology for the three dimensions, and includes two clusters of symptoms, cynicism an reduced personal effectiveness, that appear to be too heterogeneous for advancing our knowledge on burnout. In sum, the MBI, the measurement scale whose process of construction has led inductively to the above conceptualization, has been the most popular instrument for measuring burnout in empirical research (for reviews of studies using it, see Collins, 1999; Lee & Ashforth, 1996; Schaufeli & Enzmann, 1998). It contained items purportedly assessing

7 each of the three clusters of symptoms included in the syndrome view of burnout, namely emotional exhaustion, cynicism or depersonalization, and reduced effectiveness or lowered professional efficacy. It asks respondents to indicate the frequency over the work year with which they have experienced each feeling on a 7-point scale ranging from 0 (never) to 6 (every day). Three indices are usually constructed, relating to each of the above dimensions (for a recent psychometric critique, see Barnett, Brennan, & Careis, 1999). The factorial validity of the MBI has been extensively studied (Byrne, 1994; Handy, 1988; Lee & Ashforth, 1996; Schaufeli & Dierendonck, 1993; Schaufeli & Buunk, 1996). Most of the researchers examining this aspect of MBI validity have reported that a three-factor solution better fits their data than does a two-dimensional or a one-dimensional structure (for recent examples, see Boles, Dean, Ricks, Short & Wang, 2000; Schutte, Toppinen, Kalimo, & Schaufeli, 2000). Researchers using the MBI have most often constructed three different indices corresponding to the three dimensions of emotional exhaustion, cynicism, and reduced personal effectiveness. Several studies have argued, on both theoretical and psychometric grounds, that the use of a total score to represent total burnout should be avoided (e.g., Moore, 2000; Kalliath, O Driscoll, Gillepsie & Bluedorn, 2000; Koeske & Koeske, 1989). The emotional exhaustion dimension has been consistently viewed as the core component of the MBI (e.g., Moore, 2000; Cordes, Dougherty & Blum, 1997; Burke & Greenglass, 1995). Most studies have shown it to be the most internally consistent and stable of the three components (Schaufeli & Enzmann, 1998). In meta-analytic reviews, it has been shown to be the most responsive to the nature and intensity of work-related stress (Lee & Ashforth, 1993; Schaufeli & Enzmann, 1998 Pines Burnout Model and Measure Pines and her colleagues define burnout as the state of physical, emotional and mental exhaustion caused by long term involvement in emotionally demanding situations (Pines &

8 Aronson, 1988, p. 9). This view does not restrict the application of the term burnout to the helping professions, as was initially the case with the first version of the MBI (Winnubst, 1993). Indeed, Pines and her colleagues have applied it not only to employment relationships (Pines, Aronson & Kafry, 1981) and organizational careers (Pines & Aronson, 1988), but also to marital relationships (Pines, 1988, 1996) and to the aftermath of political conflicts (Pines, 1993). Much like the case of the MBI, the Pines et al. conceptualization and measure of burnout, the BM (Burnout Measure) emerged from clinical experience and case studies. In the process of actually constructing the BM measure, Pines and her colleagues have moved away from their original conceptualization to an empirical definition that regards burnout as a syndrome of co-occurring symptoms that include helplessness, hopelessness, entrapment, decreased enthusiasm, irritability, and a sense of lowered self-esteem (cf. Pines, 1993). None of these symptoms is anchored in the context of work or employment relationships. The BM is considered a one-dimensional measure yielding a single composite burnout score. As we have noted, the overlap between the conceptual definition and the operational definition of the BM is minimal (cf. Schaufeli & Enzmann, p. 48). In addition, the discriminant validity of the BM, relative to depression, anxiety, and self-esteem, is in doubt (cf. Shirom & Ezrachi, 2003). This has led researchers to describe the BM as a general index of psychological distress that encompasses physical fatigue, emotional exhaustion, depression, anxiety, and reduced self-esteem (e.g., Schaufeli & Dierendonck, 1993, p. 645; Shirom & Ezrachi, 2003). Therefore, it appears irrelevant to assess the linkage between burnout and indicators of mental health like depression or anxiety using questionnaire measures, since the overlap between the items used to gauge burnout by the BM and depression or anxiety is considerable (Shirom & Ezrachi, 2003). As we were unable to find any evidence linking the BM to

9 disease end states or to physiological risk factors for physical disease, we have not included studies that have used it in this review. Shirom-Melamed Burnout Model and Measure (SMBM) The conceptualization of burnout that underlies the Shirom-Melamed Burnout Measure (SMBM) was inspired by the work of Maslach and her colleagues and Pines and her colleagues, and views burnout as an affective state characterized by one s feelings of being depleted of one s physical, emotional, and cognitive energies. Burnout follows prolonged exposure to chronic stress. Relative to chronic stress, event-based conceptualizations of stress, like those that relate to critical life events or acute stress and to episodic stress or hassles, derive from different theoretical approached (Derogatis & Coons, 1993) and has been found to be differently related to physiological risk factors in coronary heart disease (Kahn & Byosiere, 1992). Theoretically, the SMBM was based on Hobfoll s (1989, 1998) Conservation of Resources [COR] theory. COR theory s fundamental tenets are that people have a basic motivation to obtain, retain, and protect that which they value, including material, social, and energetic resources. According to COR theory (Hobfoll, 1989, 1998) stress at work occurs when individuals are either threatened with resource loss, lose resources, or fail to regain resources following resource investment. One of the corollaries of COR theory is that stress does not occur as a single event, but rather represents an unfolding process, wherein those who lack a strong resource pool are more likely to experience cycles of resource loss. The affective state of burnout is likely to exist when individuals experience a cycle of resource loss over a period of time at work (Hobfoll & Freedy, 1993). For example, a reference librarian who comes to work every morning to face yet another line of students impatiently awaiting her help, lacking opportunities to replenish her resources, is likely to cycle into a forceful spiral of resource loss and as a result feel burned out at work.

10 The conceptualization of burnout formulated by Shirom (1989) based on COR theory (Hobfoll & Shirom, 1993, 2000) relates to energetic resources only, and covers physical, emotional, and cognitive energies. Burnout thus represents a combination of physical fatigue, emotional exhaustion, and cognitive weariness, three closely interrelated factors (Hobfoll & Shirom, 2000) that can be represented by a single score of burnout. Physical fatigue refers to feelings of tiredness and low levels of energy in carrying out daily tasks at work, like getting up in the morning to go to work. Emotional exhaustion refers to feeling too weak to display empathy to clients or coworkers and lacking the energy needed to invest in relationships with other people at work. Cognitive weariness refers to feelings of slow thinking and reduced mental agility. Each component of burnout covers the draining and depletion of energetic resources in a particular domain. There are three reasons for the focus on the combination of physical fatigue, emotional exhaustion and cognitive weariness in the conceptualization of burnout that has led to the construction of the SMBM. First, these forms of energy are individually possessed, and theoretically are expected to be closely interrelated. COR theory postulates that personal resources affect each other and exist as a resource pool, and that lacking one is often associated with lacking the other (Hobfoll & Shirom, 2000). Empirical research conducted with the SMBM has supported the linkage among physical fatigue, emotional exhaustion and cognitive weariness (e.g., Melamed, Kushnir & Shirom, 1992; Shirom, Westman, Shamai, & Carel, 1997). Second, the three forms of individually possessed energy included in the SMBM represent a coherent set that does not overlap any other established behavioral science concept, like depression and anxiety or like aspects of the self-concept such as selfesteem and self-efficacy. Third, the conceptualization of the SMBM clearly differentiates burnout from stress appraisals anteceding burnout, from coping behaviors that individuals may engage in to ameliorate the negative aspects of burnout like distancing themselves from

11 client recipients, and from probable consequences of burnout like performance decrements. This stands in contrast to the two other conceptualizations of burnout outlined above. A series of studies that confirmed expected relationships between the SMBM and physiological variables have lent support to its construct validity. In these studies, respondents total score on the SMBM was used to predict risk factors for cardiovascular disease (Melamed, Kushnir & Shirom, 1992; Shirom, Westman, Shamai, & Carel, 1997), quasi-inflammatory factors in the blood (Lerman, Melamed, Shargin, Kushnir, et al., 1999), salivary cortisol levels (Melamed, Ugarten, Shirom, Kahana, et al., 1999) and upper respiratory infections (Kushnir & Melamed, 1992). These studies are covered in more detail below. However, the convergent validity of the SMBM relative to the MBI and the BM has yet to be established, as has its discriminant validity relative to other types of possible emotional reactions to chronic stress at work, like anger, hostility, anxiety, and depressive symptomatology. The factorial validity of the SMBM needs to be investigated in additional occupational categories. Also, there is a paucity of evidence with regard to the possibility that different types of stress may have varying effects on physical fatigue, emotional exhaustion, and cognitive weariness, thus casting doubt on the use of a single composite score of the SMBM to represent burnout. There is some indirect evidence suggesting that each of the three components of the SMBM may be related to a different coping style (Vingerhoets, 1985). Models of Burnout and Health Past reviews of the burnout literature (i.e., Burke & Richardson, 2000; Cordes & Dougherty, 1993; Moore, 2000; Schaufeli & Enzmann, 1998; Hobfoll & Shirom, 2000; Shirom, 1989) view it as a consequence of one s exposure to chronic job stress. The chronic stresses that may lead to burnout include qualitative and quantitative overload, role conflict and ambiguity, lack of participation, and lack of social support. Burnout has been shown to

12 be more job-related and situation-specific relative to emotional distress such as depression (Maslach et al., 2001). Among the major theoretical approaches to burnout reviewed in Cooper (1998), none focuses on the burnout- health relationship. Our theoretical view of stress and burnout is based on Hobfoll s COR theory (Hobfoll & Shirom, 1993, 2000). Thereafter, we will link this theoretical view to burnout s relations with mental and physical health. According to COR theory (Hobfoll, 1989, 1998), when individuals experience loss of resources, they respond by attempting to limit the loss and maximizing the gain of resources. To achieve this, they usually employ other resources. When circumstances at work or otherwise threaten people's obtaining or maintaining resources, stress ensues. As indicated, COR theory postulates that stress occurs under one of three conditions: (1) when resources are threatened, (2) when resources are lost, and (3) when individuals invest resources and do not reap the anticipated return. Insofar as COR theory (Hobfoll, 1988, 1998) further postulates that because individuals strive to protect themselves from resource loss, loss is more salient than gain, in a work situation employees are more sensitive to workplace stresses that threaten their resources. Thus for teachers, for example, having to discipline students and face negative feedback from their supervisors will be more salient than any rewards that they might receive. The stress of interpersonal conflict has been shown to be particularly salient in the burnout phenomenon (Leiter & Maslach, 1988). A meta-analysis by Lee and Ashforth (1996) examined how demand and resource correlates and behavioral and attitudinal correlates were related to each of the three scales that comprise the MBI. In agreement with the COR theory-based view of stress and burnout outlined above, these authors found that both the demand and the resource correlates were more strongly related to emotional exhaustion than to either depersonalization or personal accomplishment. These investigators also found that consistent with COR theory of stress, emotional exhaustion was more strongly related to the demand correlates than to the resource

13 correlates, suggesting that workers might have been sensitive to the possibility of resource loss. These meta-analytic results were subsequently reconfirmed by additional studies, like Demerouti et al. (Demerouti, Bakker, Nachreiner, & Schaufeli, 2000) who used a burnout scale that focused on energy depletion. Applying these notions to the relations of burnout relations with mental health, we argue that individuals feel burned out when they perceive a continuous net loss, which cannot be replenished, of the physical, emotional, or cognitive energy that they possess. This feeling of ongoing net loss of any combination of physical vigorousness, emotional robustness, and cognitive agility represents an emotional response to the experienced stresses. Moreover, expanding other resources, borrowing, or gaining additional resources by investing in existing resources cannot compensate for the net loss. Indeed, burned-out individuals risk entering an escalating spiral of losses (Hobfoll & Shirom, 2000), culminating in an advanced stage of burnout, in which depression may become the predominant emotion. They may even reach advanced stages of burnout that manifested by symptoms of psychological withdrawal such as acting with cynicism toward and dehumanizing their customers or clients. As noted by Schaufeli and Enzmann (1998), longitudinal studies to date have not supported the notion of a time lag between the stress experience and the feelings of burnout. It could be that stress and burnout affect each other simultaneously, which could explain the failure of the eight longitudinal studies examined by Schaufeli and Enzmann (1998) to reproduce the effects of stress on burnout found in most cross-sectional studies. This theoretical perspective has direct implications with regard to the linkages among burnout, anxiety, and depression. COR theory implies that during its early stages burnout will be characterized by a process of depletion of energy resources directed at coping with the threatening demands, that is with work-related stresses. During this stage of coping, burnout may occur concomitantly with a high level of anxiety, due to the direct and active coping

14 behaviors that usually entail a high level of arousal. When and if these coping behaviors prove ineffective, the individual may give up, and resort to emotional detachment and defensive behaviors that may lead to depressive symptoms (cf. Shirom & Ezrahi, 2003). Cherniss (1980a, b) has found that in the later stages of burnout individuals behave defensively and hence display cynicism toward clients, withdrawal, and emotional detachment (for empirical support, see Burke & Greenglass, 1989,1995). These attempts at coping have limited effectiveness and often cycle to heighten burnout and problems for both the individuals and the organizations in which they work. The unique core of burnout, as posited above, is distinctive in content and nomological network from either depression or anxiety (Corrigan, Holmes, Luchins et al., 1994; Leiter & Durum, 1994). Measures of depression, such as the Beck Depression Inventory (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) include items that gauge passivity and relative incapacity for purposeful action. In addition, as proposed above, later phases of burnout may be accompanied by depressive symptomatology. These two considerations may explain the often-reported high positive correlation between burnout and measures of depression (e.g., Meier, 1984; Schaufeli & Enzmann, 1998). Based on these theoretical arguments, we expect burnout to be conceptually distinct from depression. Depressive symptomatology is affectively complex, and includes lack of pleasurable experience, anger, guilt, apprehension, and physiological symptoms of distress. Moreover, cognitive views of depression regard it as related primarily to pessimism about the self, capabilities, and the future (Fisher, 1984). This theoretical position may be exemplified by burnout among people-oriented professionals- such as teachers, social workers and nurses. When faced with overload and interpersonal stress on the job on an ongoing basis, the key issue for these individuals is the amount of emotional energy they need to meet the job demands. When they feel emotionally exhausted, direct or problem-focused coping, which invariably requires that they invest

15 emotional energy, is no longer a viable option. Presumably, they employ emotion-focused coping in an effort to ameliorate their feelings of emotional exhaustion, and attempt to distance themselves from their service recipients, psychologically withdraw from their job tasks, or limit their exposure to their clients. This may explain the often found linkage between emotional exhaustion and cynicism (Lee & Ashforth, 1996). In a recent study of the process of burnout among general practitioners, a study that used a five-year longitudinal design, Bakker and his colleagues (Bakker, Schaufeli, Sixma, et al., 2000) found that repeated confrontation with demanding patients over a long period of time depleted the GPs emotional resources, with perceptions of inequity or lack of reciprocity mediating the process. This study (Bakke et al., 2000) also reported that emotional exhaustion evoked a cynical attitude towards patients. However, the linkage of emotional exhaustion and cynicism does not mean that emotional exhaustion is necessarily followed by cynical attitudes or indirect coping styles like distancing. Nor does it necessarily follow from this linkage that burnout s core meaning and ways of coping with advanced stages of it belong to the same conceptual space (cf. Maslach et al., 2001, p. 403). The Health Consequences of Burnout: Mental Health Burnout has been linked to several negative organizational outcomes, including increased turnover and absenteeism (e.g., Jackson, Schwab, & Schuler, 1986; Parker & Kulik, 1995), lower organizational commitment (Maslach & Leiter, 1997), increase in suicidal feelings (Samuelson, Gustavsson, Petterson, et al., 1997) and the self-reported use of violence by policepersons against civilians (Kop, Euwema, & Schaufeli, 1999). Indicators of mental health whose linkages with burnout were investigated include depression symptoms, anxiety, and somatic complaints. Somatic complaints refer to subjectively reported healthrelated problems reported by individuals, including circulatory and heart problems, muscloskeletal pains, excessive sweating and gastrointestinal problems. Depression

16 symptomatology, as distinct from the clinical state of depression, includes feelings of sadness, emptiness, hopelessness, helplessness dysphoric feelings and low energy. It is the latter component of depressive symptomatology that gave rise to the conjecture that burnout may overlap with depression (Schaufeli & Buunk, 2003). Theoretically, the two constructs are different each from the other. Depression signifies a generalized distress encompassing all life domains, whereas burnout is context specific in that it refers to the depletion of individuals energetic resources at work (the SMBM) or to a set of work-related attitudes (the BMI). Empirically, factor analytic studies of all items measuring burnout and depression (for references, see Schaufeli & Enzmann, 1998, and also Nadaoka, Kashiwakura, Oiji et al., 1997) have generally found each construct to load on different factors, indicating that these measures probably tap different conceptual domains. Two theoretical considerations support the hypothesis that burnout and depression share a significant amount of their respective variance. First, these two conceptual entities share some antecedent variables: chronic stress influences both burnout and depression. Second, depressive symptomatology is regarded as being a component of one of the Big Five personality factors, neuroticism (McCrae & John, 1992), and this personality trait has been shown to be closely related to burnout (Zellars, Perrewe, & Hochwarter, 2000 ). Third, Leiter and Durup (1994) argued that the MBI s emotional exhaustion overlaps the lowered energy and chronic fatigue symptoms, regarded as symptoms of depression (dysthymic disorder). Especially noteworthy is Glass and McKnight s (1996) review of 18 studies that measured depression and burnout. Only one out of the 18 studies used the BM to measure burnout, and therefore the conclusions are relevant to burnout as gauged by the BMI. Glass and McKnight s (1996) review suggested that depressive affect and burnout may share a common etiology, and that their shared variance may be due to their concurrent development. They (Glass & McKnight, 1996) concluded that burnout and depressive symptomatology are

17 not mutually redundant and that their shared variance does not indicate complete isomorphism. Schaufeli and Enzmann (1998) quantitatively reviewed 12 studies (including the above six) and concluded that burnout and depression, while sharing appreciable variance, do not represent mutually redundant concepts denoting the same underlying dysphoric state. The meta-analytic study by Schaufeli and Enzmann (1998) reported that the emotional exhaustion component of the MBI and depression shared on the average 28% of their variance, while depersonalization and reduced personal accomplishment shared on the average 13% and 9%, respectively, of their variance with deperssion. Several studies have compared the construct validity of burnout and depression. A study of hospital nurses (Glass, McKnight, & Valdimarsdotter, 1993) found that under certain conditions burnout may develop into depression. However, McKnight and Glass (1995) found that burnout and depression were reciprocally related rather then one causally related to the other. In a recent study (Bakker, Schaufeli, Demerouti, Janssen et al., 2000) in which burnout and depression among school teachers was investigated, lack of reciprocity in relations with students was found to predict burnout but not depression whereas lack of reciprocity in relations with one s partner was found to predict depression but not burnout. Therefore, while burnout and depression share some dysphoric symptoms, including low energy, fatigue, and inability to concentrate, and therefore have been found to be empirically related each to the other, the evidence summarized above supports the conclusion that they are two distinct and separable constructs. It may also be possible that negative emotions, including anxiety, depression and burnout, reinforce each other, particularly in people who are susceptible to emotional stimuli (cf. Bakker & Schaufeli, 2000). The discriminant validity of burnout measures in relation to anxiety has been investigated in several studies (Bakker & Schaufeli, 2000; Brenninkmeyer, Van Yperen & Buunk, 2001; Leiter & Durup, 1994). Burnout may overlap with anxiety since

18 high levels of emotional exhaustion may raise individuals level of anxiety in stressful situations and weaken their ability to cope with anxiety (Winnubst, 1993). Richardson, Burke, and Leiter s (1992) study assessed the extent to which trait anxiety predicted each of the MBI components, and concluded that it may function as a relatively stable individual difference in the burnout process. Likewise, Turnispeed (1998), using both trait and state anxiety scales to predict each of the MBI components, suggested that both are significant contributors to burnout, especially the emotional exhaustion component of the MBI. Several studies have found burnout to be associated with a variety of somatic symptoms, including sleep disturbances, recurrent headaches, and gastro-intestinal problems (e.g., Kahill, 1988; Gorter, Eijkman & Hoogstraten, 2000). Such somatic symptoms may merely reflect a personal disposition of negative affectivity (Watson & Pennebaker, 1989) rather than poor physical health. In sum, as could be expected based on the above theoretical arguments, burnout, depression, anxiety, and somatic complaints were found to be significantly associated (Schaufeli & Buunk, 2003; Schaufeli & Enzmann, 1998). However, there is hardly any support for the contention that the construct of burnout is mutually redundant with any of the other indicators of poor mental health, including depression, anxiety, and somatic complaints. The Health Consequences of Burnout: Physical Health Researchers that have used the BMI have made the claim (e.g., Schaufeli & Enzmann, 1998, p. 87) that burnout does not reflect the type of psychophysiological arousal that can lead to health problems and that it is not likely to be a precursor of disease states such as high blood pressure or diabetes. As we will show in this section, the bulk of available evidence does not support this view. The wear and tear of energetic resources, considered to be the core dimension of burnout, is implicated in several health problems via other etiological pathways, including the immune system and inflammatory processes. The research literature

19 strongly suggests that chronic exposure to work and life stress may negatively affect physical health. A wide range of physical morbidity manifests this: cardiovascular disease, infectious illness, cancer, diabetes, rheumatoid arthritis, and musculoskeletal disorders (Dougall & Baum, 2001; Shirom, 2003). Several reviews have suggested possible mechanisms that underlie stress and disease associations (including Bailey Merz, Dwyer, Nordstrom, et al., 2002; Baum & Posluszny, 1999; Steptoe, 1991; and Kelly, Hertzman & Daniels, 1997). As noted above, there is evidence suggesting that burnout results from ineffective coping with work and life stresses and may well be chronic, lasting over long periods of time. Therefore, it seems logical to assume that burnout will also be associated with degraded physical health and psychological well-being, as suggested by recent reviewers of the burnout literature (Burke & Richardson, 2000; Cordes & Dougherty, 1993). Burnout has been linked to self-reported ill health or disease states. One study found small but significant associations between the MBI and self-reported episodes of cold or flu, but failed to find a significant correlation with cholesterol ratio (Hendrix, Steel, Leap, & Summers, 1991). Appels and his colleagues (e.g., Apples & Mulder, 1989) pioneered in the first systematic research in this area using objective indicators of physical morbidity. The predictor variable in their series of studies was vital exhaustion (VE), representing a construct that to some extent overlaps with burnout. VE, as assessed by the Maastrict Questionnaire, was defined to include unusual fatigue, increased irritability, and feelings of demoralization (Appels & Mulder, 1988), and was shown to be distinct from depression (van Diest & Appels, 1991). In a series of studies, Appels and his colleagues found that VE was associated with sleep disturbances ( van Diest, 1990; van Diest & Appels, 1994), and cardiac symptoms (angina pectoris and unstable angina) (Apples & Mulder, 1989). Moreover, it was predictive of future myocardial infarction (MI) in men and women, independent of the classic risk factors (Appels & Mulder, 1988; 1989; Appels, Falger & Schouton, 1993). To illustrate, in

20 4.2 years of follow-up of healthy men, VE was predictive of future MI, even after controlling for blood pressure, smoking, cholesterol levels, age and the use of anti-hypertensive drugs (Appels & Mulder, 1988). In a case-control study of women with first MI it was found that the relative risk for MI associated with VE was 2.75, when adjusting for several potent confound variables (Appels et al., 1993). In another study, VE was also found to be a precursor of sudden cardiac death (Appels & Otten, 1992). Using data from the prospective study of healthy men mentioned above, Appels and Schouten (1991a) found that a single question measuring burnout, Have you ever been burned out? was found to be predictive of MI risk [RR (relative risk) = 2.13, p<.01]. To the best of our knowledge, the relationships between the VE measure and either of the burnout measures have never been explored. Nonetheless, this last finding suggests that not only VE, but burnout too, may be a risk factor for coronary heart disease (CHD). Our program of research started in the early 1990s to explore the association between burnout and physical health (Shirom, 2002). In this programmatic research effort, performed among white and blue-collar workers, burnout was assessed by the SMBM. The findings provided the first evidence that burnout, as assessed by one of the measures specifically constructed to gauge it, may have a negative influence on physical health. In a study of 104 disease-free male employees of a high-tech company, Melamed, Kushnir and Shirom (1992) found burnout to be associated with elevated risk factors for cardiovascular disease. Specifically, this study reported that the combination of high burnout and tension was significantly associated with increased total cholesterol, low-density lipoprotein (LDL), triglycerides, and uric acid, and marginally with ECG abnormality. No association was found for systolic and diastolic blood pressure. In addition, in that study, consistent with findings of other studies (e.g., Gorter, et al., 2000), a high level of tense-burnout was also associated with poor health habits, including smoking and lack of participation in leisure physical activities.

21 In another prospective study of healthy male and female employees (Shirom, Westman, Shamai & Carel, 1997), emotional exhaustion (as measured by the SMBM) in men was found to be predictive of cholesterol changes, evidenced 2 to 3 years later. Among female employees emotional exhaustion was positively correlated with cholesterol and triglycerides levels whereas the correlation with but physical fatigue was negative (Shirom et al., 1997). Type 2 diabetes mellitus (DM) is a complex disorder characterized by impaired secretion of insulin and increased resistance to insulin, and associated with increased risk of coronary heart disease, peripheral vascular disease, renal failure, and blindness (Bailes, 2002; Becman, Creager & Libby, 2002). The past two decades have witnessed an explosive increase in the number of people diagnosed with diabetes worldwide (Seidell, 2000; Zimmet, Alberti & Show, 2001), primarily type 2 DM (Zimmet et al., 2001). In the US nearly ten million persons are affected with type 2 DM and the prevalence is increasing (Nelson, Everhart, Knower & Bennett, 1988). It is argued that diabetes is now becoming one of the main threats to human health in the 21 st Century (Zimmet et al., 2001). The most important risk factor in the onset of type 2 diabetes is obesity, in particular abdominal obesity, and obesity is on the rise worldwide (Visscher & Seidell, 2001). The correlation of increased diabetes with increased obesity has led to the adoption of the term diabesity (Zimmet et al., 2001). Other established risk factors for type 2 DM include age and family history of diabetes. Additional factors found to be associated with this condition are alcohol intake, smoking, reduced physical activity, diets with a high glycaemic load and a low cereal fiber content (Rimm, Chan, Stampfer et al., 1995; Helmrich, Ragland, Leung et al., 1991; Pan, Li, Hu et al., 1997; Nakanishi, Nakamura, Matsuo et al., 2000; Salmeron, Manson, Stampfer et al., 1997). Furthermore, certain risk factors for coronary heart disease, such as hypertension and dyslipidemia, are also known to be associated with risk of type 2 DM (Jacobsen, Bonan & Njolstad, 2002; Beckman et al., 2002).

22 It is believed that stress plays a significant role in the etiology of type 2 DM, but only a few studies have systematically tested this hypothesis (Surwit & Schneider, 1993). Some studies have shown that the risk of developing type 2 DM is higher in certain occupations, for instance among air traffic controllers and transport workers (Cobb & Rose, 1973; Morikawa, Nakagawa, Ishizaki, et al., 1997). Other studies, focusing on working hours, have yielded conflicting results (Kawakami, Araki, Takatsuka, et al., 1999; Nakanishi, Nishina, Yoshida, et al., 2001). There is paucity of prospective studies directly linking work-related stress and clinically diagnosed type 2 DM. Several studies, however, have examined the association with glycosylated hemoglobin A1c (Hb A1c) which is not used in the diagnosis of diabetes but provides a measure of chronic glycemia and is also used to assess the effectiveness of therapies provided to those inflicted with DM (Barr, Nathan, Meigs & Singer, 2002). Hb A1c appears to be a marker of the increased risk of developing atherosclerosis, MI, strokes, cataracts, and loss of the elasticity of arteries, joints and lungs (Kelly et al., 1997). The results have shown that greater job strain and lower social support at the work place may be associated with increased concentrations of Hb A1c (Kawakami, Akachi, Shimizu, et al., Thus, it is still unclear whether psychosocial stress is causally implicated in the onset of diabetes. A recent study on burnout and type 2 DM risk (Melamed, Shirom & Froom, 2003) provides initial support for such a possibility. This study was conducted among primarily white-collar Israeli workers. After excluding those who had a history of diabetes mellitus or other chronic diseases, 633 workers were followed up for a period of 3-5 years. During this period there were 17 new cases of treated type 2 DM. Burnout, as measured by the SMBM, was found to be associated with increased risk of type 2DM (OR=1.83, 95% CI ), even after controlling for age, sex, body mass index, smoking, period of followup and job category. Thus, this finding suggests that burnout might be a risk factor for type 2 DM in Israeli workers.

23 Findings from a study of blue-collar workers (Melamed, Ugarten, Shirom, Kahana, Lerman, & Froom, 1999) provided evidence supporting the notion that burnout is associated with increased somatic and physiological hyper-arousal. Among those reporting higher levels of burnout, there was a higher prevalence of unpleasant sensations of tension and restlessness at work, post-work irritability, sleep disturbances, complaints of waking up exhausted, and higher cortisol levels during the workday (Melamed et al., 1999). These findings suggest that burned-out persons may have an inability to unwind after working hours. Furthermore, these persons may suffer from insomnia and non-refreshing sleep. This may explain in part, the chronic fatigue experienced by burned- out persons. Insomnia in general (Carney, Freedland & Jaffe, 1990; van Diest, 1990), and waking up exhausted in particular (Appels & Schouten, 1991b), were found to be risk indicators of future MI. Therefore, the findings of this study suggest an additional pathway by which burnout may be associated with increased risk of cardiovascular disease (CVD). Taken together, the findings of the above studies suggest that burnout may be associated with CVD risk through multiple pathways: increased biochemical risk factors, development of diabetes (conceived to be an independent risk factor of CVD; Visscher & Seidell, 2001), and sleep disturbances. In recent years, following the findings that the classical risk factors (hypertension, poor lipids profile, smoking, lack of physical exercise and overweight) explain only in part the incidence (new cases) of MI, research efforts have been directed toward identifying new risk factors for CVD. One such risk factor is inflammation, which accumulating evidence indicates is linked to atherosclerosis and acute coronary syndromes (Libby, Ridker & Maseri, 2002; Rose, 1999; Koenig, 2000). There is also evidence that chronic inflammation may play an important role in linking diabetes and CVD as well as evidence supporting the relationships between markers of inflammation, abnormalities of glucose metabolism, and

24 CVD endpoints (Resnick & Howard, 2002). Findings from the study by Lerman et al.(1999), suggesting that burnout may be associated with inflammation condition, revealed a close association between burnout and leukocyte adhesiveness/aggregation (LAA). LAA is a sensitive marker that detects inflammation and assesses its intensity (Rotstein, Mardi, Justo, et al., 2002). The adoption of LAA as a marker of inflammation was based on the notion that white blood cells get activated and sticky during the inflammatory response (Frenetto & Wagner, 1996 a,b). LAA probably represents both enhanced expression of cell adhesion molecules during cell activation, as well as the appearance of plasmatic adhesive proteins during the acute phase response (Rotstein, et al., 2002). Additional evidence indicates that there is a high correlation between LAA and erythrocyte aggregation (Shapira, Rotstein, Fusman, et al., 2001), and there is also evidence regarding red blood cell aggregability in patients with hyperlipidemia, diabetes mellitus, hypertension and acute MI (Berliner, Zeltser, Rotstein, et al., 2001; Shapira, et al., 2001). Thus, by implication burnout may be associated with CVD risk through the presence of smoldering inflammation though this hypothesis awaits further research and confirmation. Research on the health implications of the burnout syndrome suggests that the adverse effect on health may go beyond diabetes and the cardiovascular system. Accumulating evidence points to the association between stress, immunity and susceptibility to infectious disease (Marsland, Buchen, Cohen & Manuck, 2001). Following this evidence, new studies have been initiated to explore the effect of burnout on the immune system and its possible association with infectious disease. In a study of office workers, one of the dimensions that form burnout as assessed by the MBI, namely depersonalization, was found to be associated with reduced cellular immunity (lower natural killer (NK) cell activity and lower proportionality of CD57+CD16 to total lymphocytes. This was independent of health behaviors (e.g., smoking, alcohol use,