Introduction

There are major concerns about the workforce that provides prevention, treatment, and recovery oriented services in the mental health field.1 Many of these concerns focus on access to relevant and effective education and training. The most pressing seem to center on difficulties recruiting and retaining staff. High rates of staff turnover have led to the perception that members of the workforce are dissatisfied or “burned out” with their jobs and careers in this field.

The retention of staff has been identified repeatedly as a major issue within the mental health workforce. For example, Blankertz and Robinson2 cited a 20% annual turnover rate for community mental health workers. In a study of 848 rehabilitation direct service workers, these same authors reported that 50% of those surveyed indicated that it was “somewhat” to “highly” likely that they would leave the psychosocial rehabilitation field within the next 2 years.2–3 In a meta-analysis, Mor Barak et al.4 found that reported rates of turnover in child welfare, social work, and community mental health ranged from 30% to 60% in any given year. Similarly, Ben-Dror5 concluded that the turnover rate of employees exceeded 60% each year in the mental health field. High turnover rates compromise continuity of care and create organizational instability and financial drain related to the costs of employee separation and recruiting and training replacements.4

Burnout is often cited as a major problem in its own right and as a potential cause of turnover.4,6–9 There is a widely held perception that jobs within this field have become more demanding and stressful, leading to burnout and, eventually, to turnover. However, there is no comprehensive review of the burnout construct and the impact of burnout in the mental health field. Given the pressing concerns about staff turnover in behavioral health, this article provides a review and analysis of the existing literature, encompassing the history and definition of the construct as well as methodological and measurement issues. Data on the prevalence and correlates of burnout are reviewed as are potential interventions to reduce burnout. The implications of these findings for the field and for future research are discussed.

Method

Search strategy and results

A review of the relevant mental health literature from 1990 to 2009 was conducted. Numerous databases, such as PsycINFO, MEDLINE, CINAHL, BIOSIS, AMED, Cochrane Database of Systematic Reviews, Web of Science, Infotrac, Global Health, PubMed, and EMBASE, were utilized, yielding 145 articles related to issues of burnout in behavioral health. It is of note that 87 of the articles focused exclusively on international findings (particularly the UK), while 16 combined US and international findings. Only 38 focused solely on the USA.

Findings

Conceptual, measurement, and methodological issues

Burnout was first introduced in the literature by Herbert J. Freudenberg as a concept related to frontline human service workers.10 Through his experiences working in a free clinic, he articulated a set of symptoms commonly associated with burnout, identified the types of individuals who were susceptible, and prescribed preventive measures to attenuate the effects of this phenomenon. As a result of the work begun by Freudenberg, there has been a growing interest within mental health and other human services fields in burnout and its relationship to variables, such as turnover, absenteeism, and the psychological and physical health of employees.

Due to the potential negative impact of burnout, a number of researchers developed measures that have further defined this construct. The most commonly cited measures include the Staff Burnout Scale for Health Professionals,11 the Burnout Measure,12 and the Maslach Burnout Inventory.13

The Staff Burnout Scale for Health Professionals contains 30 items on a 6-point Likert scale, in which 20 items measure burnout and ten assess for social desirability. The instrument examines four dimensions of burnout, including cognitive, affective, psycho-physiological, and behavioral, with a total burnout score than can range from a low of 20 to a high of 140. The measure obtained a Spearman–Brown split-half reliability coefficient of 0.93 for internal consistency. Alpha coefficients for the subscales were cognitive (0.73), behavioral (0.59), affective (0.50), and psycho-physiologic (0.44).11,14–15

According to Enzmann et al.,16 the Burnout Measure is the second most frequently used self-report measure of burnout. The instrument is made up of 21 items utilizing a 7-point Likert scale (1—never to 7—always) that assesses levels of physical, emotional, and mental exhaustion; a score of 4 is considered indicative of burnout. The measure has been demonstrated to be a valid and reliable instrument, with good internal consistency coefficients (Cronbach’s alpha ≥ 0.90) and 1–4 month test–retest coefficients ranging from 0.66 to 0.89.

The Maslach Burnout Inventory (MBI) has become the “almost universally accepted gold standard to assess burnout.”17 It was designed as a general measure to be used for all human service professions and has been shown to have high reliability and validity.18,19 Data from a preliminary version of the MBI were subjected to a factor analysis using principal factoring with iteration and an orthogonal rotation. From the ten factors that emerged, three factors had eigenvalues greater than unity.19 These now define the concept.

According to Maslach and Jackson,19 burnout is a syndrome comprised of a set of symptoms related to these three empirically derived factors, which the authors have labeled emotional exhaustion (EE), depersonalization (DP), and reduced personal accomplishment (PA). These symptoms are hypothesized to produce a generalized negative outlook toward self and others. Specifically, a high level of depersonalization (cynicism) refers to “an unfeeling and impersonal response toward recipients of one’s service, care, treatment, or instruction”; a low level of personal accomplishment (inefficacy) suggests unsatisfactory “feelings of competence and successful achievement in one’s work with people”; and a high level of emotional exhaustion indicates “feelings of being emotionally overextended and exhausted by one’s work.”13

The original measure designed for service providers in the human services field [Maslach Burnout Inventory-Human Services Survey (MBI-HSS)] is made up of 22 items, categorized by those three factors: emotional exhaustion (nine items), depersonalization (five items), and personal accomplishment (eight items). The items are rated using a 7-point response scale ranging from never experienced such a feeling (0) to experienced such feelings every day (6).20 The scores for each factor are computed separately and can be coded as low, average, or high using cutoff scores that separate the distribution into thirds.21 The range of burnout experienced is broken down as follows: EE (low, ≤16; average, 17–26; high ≥27); DP (low, ≤6; average, 7–12; high, ≥13); and PA (low, ≥39; average, 38–32; high, ≤31). Burnout is associated with higher scores on the EE and DP subscales and a lower score on the PA subscale. The overall sample utilized to norm the measure (n = 11,067) included individuals from the following subgroups: elementary and secondary education, postsecondary education, social services, medicine, and mental health. Initial data on the reliability of this measure included Cronbach’s alphas of 0.90 for emotional exhaustion, 0.79 for depersonalization, and 0.71 for personal accomplishment. In addition, test–retest reliability coefficients were 0.82 for EE, 0.60 for DP, and 0.71 for PA.22 Maslach, Jackson, and Leiter13 have summarized evidence of the instrument’s convergent and discriminant validity. For example, in a questionnaire survey of 142 police officers, their MBI-HHS scores were highly predictive of intention to quit [R (6, 135) = 0.68, p < 0.001].

While burnout is addressed frequently in the professional literature, the first finding of this review is that there is somewhat limited empirical research on the topic in behavioral health. Much of the existing literature is based on casual observation, anecdotal reports, case studies, or informal surveys and questionnaires. Among the studies that formally measured this construct, many used samples of convenience that were often small and poorly defined. Even though the MBI appears to be the most widely administered tool for the assessment of burnout, there is inconsistency in the manner in which researchers have scored and analyzed data obtained with this measure. Furthermore, in many studies examining the relationship of burnout to other variables of interest, the methods for defining and measuring those other variables have been imprecise. While all of these limitations suggest that caution be exercised in drawing conclusions from this literature, there are findings that warrant consideration. These are presented and discussed below. The absence of key information about study samples or statistical results is generally attributable to its absence or to ambiguity in the original published reports.

Prevalence of burnout

The literature is replete with claims that burnout is a significant problem within the mental health workforce. A number of studies have provided evidence regarding its prevalence, and these are reviewed briefly below. The studies are remarkable for their diversity in terms of country, work setting, and type of personnel. Selected studies have compared burnout levels between different professional groups, while one examines burnout within a profession, comparing those working in behavioral health and non-behavioral health roles. It is noteworthy that all of these studies used the MBI as their measure of burnout.

In 1993, the Sainsbury Centre for Mental Health in the UK conducted two large-scale national surveys that examined the mental health workforce. Several measures were utilized in these studies, including the MBI, to obtain data from 445 respondents. In total, 44% of the study participants fell into the “high” burnout category for emotional exhaustion. The rates on this subscale did not vary greatly among the subgroups, with 45% of community mental health nurses, 54% of social workers, and 63% of psychiatric consultants (psychiatrists) falling into the high category. While rates of emotional exhaustion were high, scores for the sample revealed low levels of depersonalization and high levels of personal accomplishment, both of which are desirable findings.23–25

Rupert and Morgan conducted a study in which they randomly selected 1,200 psychologists from the membership rosters of the American Psychological Association.22 These individuals were asked to complete the Psychological Practice Experiences and Attitudes Survey, which contained six sections and combined several instruments, including the MBI-HSS. They also provided diverse information and ratings that included basic demographics, years of experience, total hours worked, time spent in various professional activities, theoretical orientation, attitudes toward workload, income satisfaction, and sources of both job satisfaction and stress. With regards to experienced burnout, and utilizing the MBI categorization data for mental health workers, emotional exhaustion (mean score of 19.99; SD = 9.83) and depersonalization (mean score of 5.21; SD = 4.26) were in the average range, while PA was in the low (desirable) range (mean score of 41.64; SD = 4.78).

Prosser et al.26 conducted a longitudinal study examining job satisfaction and burnout in a sample of mental health staff working in three adult mental health facilities in inner city London. Over a 3-year period (1994–1996), the authors sent to participants annual questionnaire packets, which included the 12-item General Health Questionnaire, the MBI, and a general job satisfaction measure. The response rate ranged from 60% to 76% each year. It is worth noting that turnover was high during each of these 3 years, averaging over 30% annually, so only 25 individuals completed the questionnaire in all 3 years. The authors found a relatively high level of emotional exhaustion and depersonalization in year 1 but an absence of problems with respect to personal accomplishment. These findings remained largely unchanged in years 2 and 3.

Similar results were found in a cross-sectional questionnaire survey conducted by Wykes, Stevens, and Everitt.9 The respondents were a non-random sample of 61 members of six multidisciplinary mental health teams in Great Britain, made up of psychiatrists, nurses, social workers, occupational therapists, clinical psychologists, health/support workers, and administrative and secretarial personnel. Multiple measures were used, including the MBI, General Health Questionnaire 28 (GHQ-28), Beck Anxiety Inventory, Daily Hassles Scale, and general demographic and work-related information. The authors found that among the sample, 57% and 42% experienced a high degree of emotional exhaustion and depersonalization, respectively; however, 59% experienced a positive sense of personal accomplishment.

Sweeney and Nichols27 summarized the findings of several studies examining the issue of stress and burnout in occupational therapists (OT) working in the mental health field. The reviewed studies utilized a variety of instruments, including the MBI. The findings suggested that OT experienced a moderate degree of burnout but at lesser levels than that experienced by other mental health professions. Within the OT profession, there was no evidence that those working in the mental health field experienced a higher degree of burnout than those working in physical rehabilitation.

Evans and colleagues distributed postal surveys containing the General Health Questionnaire, MBI, Karasek Job Content Questionnaire, and a job satisfaction measure to 610 mental health social workers in England and Wales. In total, 237 questionnaires met criteria for inclusion.28 Using the MBI categorization data for mental health workers, the authors found a high level of emotional exhaustion (EE mean score of 26.3; SD = 10.1) and depersonalization (DP mean score of 7.3; SD = 5.2), with a low (desirable) range on personal accomplishment (PA mean score of 33.9; SD = 6.8).

Similarly, Blankertz and Robinson29 surveyed 2,000 psychosocial rehabilitation workers and obtained 848 usable responses. The surveys assessed job and personal characteristics, background in psychosocial rehabilitation, opinions about their field and career future, job satisfaction as measured by the Rehabilitation Job Satisfaction Inventory, and burnout as measured by the MBI. Emotional exhaustion and depersonalization were in the average range, while personal accomplishment was in the high (undesirable) range of experienced burnout.

Potential correlates of burnout

Researchers in the USA and Europe have identified various professional, environmental, and personal factors believed to be associated with burnout. Since the studies are generally methodologically weak and multiple studies examining the relationship of burnout to a single, common variable do not exist, the available studies can be considered suggestive of “potential” correlates of burnout.

As cited earlier, Rupert and Morgan22 examined the relationship between burnout and work setting by comparing psychologists employed in agencies to those in independent practice.

The authors found that greater emotional exhaustion was positively related to variables such as long working hours (r = 0.29, p < 0.01), time spent in administrative and paperwork tasks (r = 0.26, p < 0.01), lack of control over work activities (r = −0.29, p < 0.01), higher rates of managed care (r = 0.15, p < 0.01) versus self-pay clients (r = −0.23, p < 0.01), higher rates of “negative” client behaviors (r = 0.30, p < 0.01), and over involvement with clients (r = 0.30, p < 0.01). Depersonalization was associated with negative client behaviors (r = 0.29, p < 0.01) and over involvement with clients (r = 0.15, p < 0.01) while inversely related to the percentage of self-pay clients (r = −0.14, p < 0.01) and perceived control within the work setting (r = −0.21, p < 0.01). Finally, a sense of personal accomplishment was positively associated to the number of therapy hours conducted (r = 0.35, p < 0.01), the percentage of self-pay clients (r = 0.11, p < 0.01), and perceived control within the work setting (r = 0.33, p < 0.01). The sense of personal accomplishment was inversely related to the time spent on testing (r = −0.12, p < 0.01) and to administrative and paperwork tasks (r = −0.15, p < 0.01).22

Evans and colleagues, in a study described above focusing on social workers, used multivariate analyses to study the correlates of burnout. They reported that high emotional exhaustion was associated with higher job demands (β = 0.38, p < 0.001), while a sense of personal accomplishment was associated with greater latitude or independence in decision-making (β = 0.29, p = 0.001). Depersonalization was associated with dissatisfaction regarding the state of the social work field in mental health (β = −0.22, p = 0.010), not feeling valued by one’s employer (β = −0.19, p = 0.024), and male gender (β = −0.18, p = 0.016).28

In the study reported above on psychosocial rehabilitation workers, Blankertz and Robinson29 found a negative relationship between age and both emotional exhaustion (r = −0.20, p < 0.001) and depersonalization (r = −0.17, p < 0.001), suggesting that older workers were less likely to experience these dimensions of burnout. Non-white respondents reported not only a lower level of emotional exhaustion (t = −4.94, df = 8,804, p = 0.0001) but also a lesser sense of personal accomplishment (t = −2.72, df = 789, p < 0.01). Individuals with a master’s degree had higher mean scores on emotional exhaustion than those individuals with a high school diploma or some college education (F = 7.03, p < 0.0001). Salary was positively correlated with personal accomplishment (Spearman’s R = 0.16, p < 0.001) and emotional exhaustion (R = 0.12, p < 0.001), suggesting that those individuals with higher salaries not only felt that they were more effective on the job but also felt more emotionally exhausted. Tenure in a job was also related to an increased level of emotional exhaustion (R = 0.09, p < 0.01). On the other hand, number of years in the field was correlated with lower levels of depersonalization (R = −0.07, p < 0.05).29

Geurts, Schaufeli, and De Jonge30 examined burnout and intention to leave their employment among 208 mental health care professionals using several survey measures, including the Dutch version of the MBI. Utilizing covariance structure modeling, the authors found that negative communication between employees about their employer was associated with perceptions of inequity in their employment situation (path 1, β = 0.25, p < 0.05). The experienced perception of inequity was, in turn, positively related to intention to leave employment (path 2, β = 0.54, p < 0.001) and to emotional exhaustion (path 3, β = 0.64, p < 0.001). This is the only study that explicitly examined variables related to turnover.

In a study cited above on community mental health workers, Wykes, Stevens, and Everitt found a significant correlation between the GHQ-28 total score and the three subscales of the MBI (GHQ total with EE, r = 0.651, p < 0.001; with DP, r = 0.532, p < 0.001; and with PA, r = −0.368, p < 0.01). This indicates a relationship between the presence of burnout and psychological symptoms. A substantial percentage of individuals working in inner city teams displayed a high degree of emotional exhaustion (57%) and depersonalization (42%) while exhibiting a sense of personal accomplishment (59%).9

DeStefano, Clark, and Potter31 compared the burnout levels of 827 employees to a normative sample of mental health workers utilizing the MBI. Of those surveyed, 73% were employed with their agency for 3 years or less, while only 10% had been employed for 8 years or more. Compared to the normative sample of mental health workers represented in the MBI norm group, the rural mental health workers in this study demonstrated significant differences on both emotional exhaustion [t (827) = 8.958, p < 0.001] and personal accomplishment [t (827) = 36.335, p < 0.001)]. In essence, rural mental health workers reported higher levels of emotional exhaustion (M = 20.73) and personal accomplishment (M = 39.45) than the norm group, with no significant differences in terms of depersonalization. Analysis of demographic and work-related variables with the MBI scales suggested that significantly higher levels of emotional exhaustion were experienced by case managers, therapists, and psychiatrists as compared to paraprofessionals and behavioral health technicians.

Finally, in a longitudinal study, Corrigan and colleagues6 used a cross-lagged panel design to study the relationships of the three factors of burnout outlined by Maslach and Jackson, with state anxiety and collegial support. Thirty-five staff members (nursing, clinical, and administrative), working in a state psychiatric hospital, were assessed using the MBI, Modified Social Support Questionnaire, and State Trait Anxiety Inventory (state version) at baseline and 8 months. Mean, standard deviations, and Cronbach alphas for the MBI factors at times 1 and 2 were as follows: EE (1, M = 15.94, SD = 12.44, alpha = 0.90; 2, M = 16.00, SD = 11.42, alpha = 0.91), DP (1, M = 5.82, SD = 5.30, alpha = 0.67; 2, M = 5.91, SD = 5.83, alpha = 0.80), and PA (1, M = 34.88, SD = 7.90, alpha = 0.70; 2, M = 34.88, SD = 7.78, alpha = 0.71). Significant positive synchronous correlations were found between emotional exhaustion and state anxiety at both time points (T1, z = 0.56, p < 0.05; T2, z = 0.49, p < 0.05), suggesting that emotional exhaustion may lead inpatient staff to experience generalized anxiety. Furthermore, the authors also found a significant association between depersonalization and collegial support at both time points (T1, z = −0.51, p < 0.05; T2, z = −0.66, p < 0.05), suggesting that those individuals who perceived greater support from their colleagues were less likely to exhibit depersonalization.

Interventions to decrease burnout

While there are many published works that discuss interventions to decrease burnout, there are only two that have made even a basic attempt to evaluate the impact of interventions. The methodology in these studies, which are described below, was relatively weak.

Group formats are frequently recommended as useful in countering professional isolation and burnout.32 One group format that has been tested and demonstrated some potential to reduce burnout is based on equity theory and cognitive behavioral principles. A team of researchers evaluated a 5-week group-based burnout intervention program with direct care professionals who were treating mentally disabled clients.33 The group was aimed at improving the sense of fairness in interpersonal relationships with the recipients of their care and the organization in which they were employed. The researchers utilized a quasi-experimental design that included pre-, post-, and follow-up measures (6 and 12 months) of one experimental group (36 individuals) and two control groups (39 and 74 individuals). Measures included the MBI and assessments of equity, perceived social support, turnover intention, and absenteeism. The intervention program consisted of weekly meetings with six to eight participants, which were led by a psychologist. The first two sessions explored the reasons group members chose their career as well as a discussion of the burnout concept and the relationship of burnout to their work. The third session focused on their present work environment, work-related goals and expectations, and training in relaxation exercises. The fourth session included a discussion of self-image, while the final session focused on developing an “action plan” for the subsequent year. The plan either outlined ways in which the present job could be changed or steps that could be taken to find another job more in line with one’s goals and expectations. After 6 months, a follow-up meeting took place to re-examine the plan.

At the 1-year follow-up, several promising trends were noted. First, levels of emotional exhaustion (F = 2.15, p < 0.10) decreased, and perceived levels of organizational equity (F = 2.41, p < 0.10) increased, as compared to the control groups. There was a similar trend related to decreases in the duration of absences from work (F = 2.00, p > 0.10). On the other hand, there was no change in levels of depersonalization, perceptions of interpersonal equity, and the frequency of absences. In addition, feelings of personal accomplishment initially increased but returned to baseline levels after the 1 year follow-up.

In the second empirical study, Scarnera, Bosco, Soleti, and Lancioni34 examined the effectiveness of a group intervention utilizing a cognitive-behavioral framework. Twenty-five individuals providing psychiatric and therapeutic/rehabilitative services in Italy participated. The Italian versions of the MBI and Occupational Stress Inventory were administered at pre-, post- (6 month), and follow-up (18 months). The intervention consisted of a series of six monthly workshops aimed at developing assertiveness techniques to manage interpersonal relationships with other staff, clients, and family members. In terms of outcome, depersonalization demonstrated significant main effects of time [F (2, 48) = 5.04; p < 0.05; mean T0 = 4.52; T1 = 3.76; T2 = 2.80]. Specifically, it was found that depersonalization within the group decreased from a medium to high to medium to low level of burnout subsequent to the training, and this effect persisted at the 18-month follow-up. The authors concluded that the intervention positively impacted depersonalization.

While the number of empirical studies is extremely limited, the literature contains a host of practical strategies that have been recommended for decreasing burnout, though without support from research. A sampling of these strategies, which have been described at various levels of detail, includes the following: (1) competitive salaries, (2) financial and non-financial incentives to enhance staff motivation and morale, (3) opportunities for promotion and career advancement, (4) funding for increased staffing levels, (5) training staff on self-care strategies, (6) additional clinical supervision and mentoring, (7) clear job descriptions/expectations, (8) routine assessment of burnout, (9) flexible work schedules, (10) social events and informal support, (11) in service trainings, and (12) open-door policies with management.7,9,32,35–39

All of the above cited actions may have a potential impact in decreasing the negative consequences associated with burnout and its effect on turnover, but as previously stated, the empirical support for these interventions is lacking.

Implications for Behavioral Health

The concept of burnout has a long history in the health and human service field, dating back over 30 years. It is a construct that is frequently referenced in the general health and behavioral health literature due, perhaps, to its face validity. It is a seemingly intuitive and compelling concept to those who manage or study the workforce within health care systems.

There are a number of conclusions that can be drawn from this review. First, there have been substantial efforts to define, measure, and validate the construct using established psychological assessment techniques. Instruments such as the MBI have served to sharpen the construct and provide tools to systematically assess it. Looking forward, a primary challenge for the mental health field is to routinely use such valuable measures in order to build a more robust knowledge base about the prevalence, causes, and effects of burnout in this field, as well as the impact of interventions designed to decrease it.

Second, the literature on burnout in mental health, as a whole, is quite weak. Publications tend to rely principally on casual observation, anecdote, or informal survey to describe the phenomenon and the many interventions that have been employed to ameliorate it. Methodological rigor is most often absent.

While the data on burnout are limited, the studies identified in this review raise considerable alarm about the prevalence of burnout among the mental health workforce. Thus, the third conclusion is that there are consistently high levels on at least one factor of burnout among most of the workforce groups in behavioral health that have been studied. These statistics demonstrate a compelling need to redouble efforts within this sector of healthcare to better understand and address the high levels of distress among those providing mental health services.

Fourth, it is noteworthy that scores on the MBI tend not to be uniformly in the high range on all three burnout factors. This demonstrates the importance of examining each factor independently. Many studies found that staff members scored in the high (undesirable range) on the emotional exhaustion factor, feeling overextended and depleted. However, the variability of findings on the depersonalization and personal accomplishment factors suggests that individuals in the mental health workforce may often feel connected to others and experience a sense of competence and achievement in their work despite being emotionally drained. Again, looking forward, understanding the nature of the differences between these factors is essential to the development of more tailored and effective interventions.

As a fifth conclusion, the review of the literature revealed that empirical studies are lacking to establish a causal link between burnout and staff turnover. There are multiple variables that are potential correlates of burnout, including selected personal characteristics (i.e., age, gender, ethnicity, etc), and the nature of the work environment, yet definitive links have not been established.

Finally, there are only a couple of studies that evaluate the impact of interventions on burnout—too few to draw even preliminary conclusions. Burnout interventions have tended to focus on individual-centered approaches even though, in other fields, research has demonstrated that organizational and situational factors are more significant contributors to burnout than individual factors.40,41 Fixsen and colleagues,42 in their work on implementation of best practices, have similarly argued that organizational and systems change must complement interventions focused on the individual. In addressing burnout, the mental health field must turn increasing attention to the organization and systems levels.

In summary, there is ample evidence of a substantive burnout problem within the mental health workforce. Analysis of the factors that comprise burnout clearly indicates that it manifests itself in a variable, complex, and inconsistent manner, rather than as a unitary phenomenon. The field has, at its disposal, a number of validated tools to assess this workforce problem and the impact of efforts to ameliorate it. Concerted attention, evaluation, and intervention, using sound methods, are essential.