The training for and practice of medicine has long been acknowledged as arduous. Some of the harmful effects associated with the rigors of medical training and practice are reduced patient safety (e.g., increases in medical mistakes resulting from fatigue or work overload), trainees’ safety (e.g., impaired driving resulting from fatigue), and unnecessarily high levels of stress, depression, and burnout among medical professionals (1–3). The ACGME’s 2003 rule to limit residents’ duty hours is one consequence of this concern (4). A growing body of research also indicates increasing attention to this topic. This study advances knowledge about important issues concerning resident burnout by describing the results of a study of burnout in residents in the family medicine and psychiatry residency programs at one institution.
Burnout is a pathologic syndrome resulting from prolonged occupational stress. Maslach (13), in developing a psychological measure for burnout, has identified three components: emotional exhaustion, depersonalization (detachment), and loss of feelings of personal accomplishment. A recent comprehensive review of the literature on resident burnout discussed five issues: the level of clinically significant burnout among residents; factors associated with the development of resident burnout; health and performance consequences of resident burnout; resources to assist residents to cope with burnout; and recommendations concerning areas of additional research and reform (5). A major finding reported in this review was that burnout appears to be high among residents. There is some evidence that burnout has a “natural course” that peaks during the first postgraduate year, although Michels et al. (6) found that some burnout symptoms in a group of family medicine residents worsened over 3 years of residency. Strong conclusions about the causes of resident burnout were not possible from this review. Reviewed studies used different definitions and measures of burnout, covered varying time frames and addressed different foci (e.g., the natural history of burnout versus the efficacy of interventions to address burnout). In addition, most existing research had methodological weaknesses such as small samples, cross-sectional methodology, or limited populations (e.g., 7 of the 15 studies considered appropriate for the review involved only internal medicine residents). These conclusions provide clear impetus for continuing to explore factors related to resident burnout.
Many questions remain unanswered about resident burnout. The fact that women have higher rates of depression has led to speculation that women residents might experience higher levels of stress and burnout; but, the literature in this area is sparse. Thomas (5) reported that only 4 of the 15 studies reviewed analyzed data by gender and none reported higher burnout in women. Another demographic variable that has been reported but that has not received significant consideration is parental status. One hypothesis is that parenting produces an additional drain on a resident’s practical and emotional resources and would increase burnout. Several reports indicate that having children is either unrelated to burnout or associated with decreased burnout (7–10). Finally, the issue of burnout among international medical graduates has not been studied. In 2005, approximately 25% of all residency positions in U.S. allopathic medical schools were held by international medical graduates (11). It seems reasonable to hypothesize that international medical graduates would experience increased stress and burnout compared to U.S. medical graduates as a result of factors like adjusting to unfamiliar customs and language, practical issues like relocation, emotional factors such as separation from family and friends, and the general problems of having to accommodate to a new culture. A search of the literature on international medical graduates in U.S. training programs does not reveal any studies focusing on the effects of acculturation stress on these students. Recently, medical training curricula have begun to reflect a greater awareness of cultural diversity and acculturation stress in patient populations and have begun to include programs to prepare trainees to practice in a culturally sensitive manner (12). It is ironic that the literature concerning international medical graduates in U.S. residencies neglects to address questions about the effects of acculturation stress on the students themselves.
The current study addresses many of these unanswered questions. Two specialties not well represented in the existing literature on resident burnout—family medicine and psychiatry—provide the samples. We specifically focus on understudied issues such as gender, parental status, and international status to elaborate existing knowledge about resident burnout.
Following approval by the Institutional Review Board at East Tennessee State University, family medicine and psychiatry residents at the College of Medicine voluntarily completed a series of questionnaires twice a year over the course of 3 years. The questionnaires included the Maslach Burnout Inventory and the Work Environment Scale, form R (13, 14). We also collected demographic data (age, gender, years lived in the United States, etc.). All sets of questionnaires were administered at the beginning and the middle of each of three consecutive residency years (2002–2005). Residents in each of the programs were asked to complete the questionnaires during a noontime lecture. In order to satisfy Institutional Review Board concerns about confidentiality and coercion, there were no names or identifying codes on the questionnaires and it was made clear to residents that they had the option to not participate without fear of consequences from the training program. It was our hope that anonymous questionnaires would encourage residents to be frank in their responses and also improve completion rates.
The Maslach Burnout Inventory was developed to measure burnout in human services workers and was validated in a variety of occupations (teachers, social service workers, medical workers, mental health workers, ministers, attorneys, police etc.). The Maslach Burnout Inventory has become the most commonly used instrument to measure burnout in physicians. In the review article on burnout by Thomas, 12 of the 15 articles selected for review used the Maslach Burnout Inventory. The Maslach Burnout Inventory measures three aspects of burnout: emotional exhaustion, depersonalization, and personal accomplishment. High scores on emotional exhaustion indicate emotional overextension and work exhaustion. High scores on depersonalization indicate unfeeling and impersonal responses toward patients. Personal accomplishment assesses competence and successful achievement in work with people. Note that low scores on personal accomplishment indicate burnout.
The Work Environment Scale, Real Form, assesses work setting social environments along three main dimensions: relationships, personal growth, and system maintenance and system change. The relationship dimension assesses concern about jobs, collegiality, and management support, and is comprised of the subscales involvement, peer cohesion, and supervisor support. Personal growth assesses encouragement toward self-sufficiency, independent decision-making, efficiency; work pressure and time urgency and is measured by the subscales autonomy, task orientation, and work pressure. The system maintenance and system change dimension consists of the clarity, control, innovation, and physical comfort scales, and measures knowledge of daily work routines, rule and policy clarity and control, and physical surroundings.
We introduced a variable we termed “acculturation status” based on the number of years the resident had lived in the United States. The choice of “number of years lived in the U.S.” is an attempt to identify those residents who are adapting to the U.S. culture while in residency. The term international medical graduates includes residents who have only attended medical school in another country and would have little difficulty adapting to U.S. culture on their return. Similarly, country of birth or citizenship may not reflect the foreign nature of the U.S. culture for the resident. Examining the data revealed that all but three of our residents had either resided in the U.S. for more than 20 years (N=78) or 10 years or less (N=24). We selected 10 years as a dividing point based on this observation. We have not seen this measure used elsewhere and have no information concerning its validity or reliability.
Because no personal identifiers were collected, residents who completed more than one questionnaire were detected by matching the demographic information (age, sex, marital status, number of children and years lived in the United States). Data were analyzed using the SPSS (SPSS, v.13.0). Individual items on the Maslach Burnout Inventory were analyzed along with all subscales on both the Maslach Burnout Inventory and Work Environment Scale. Contrasts with various demographic variables were made using parametric and nonparametric statistics.
A total of 155 psychiatry and family medicine residents were surveyed over a 3 year period (July 2002 – February 2005). A total of 250 sets of questionnaires were completed and scored. The response rate was 56% (N=227)(61% for psychiatry residents and 53% for family medicine residents). These response rates were calculated by dividing the number of questionnaires returned by the total number of residents enrolled in the residency at that point in time. Residents may have failed to return a questionnaire for a variety of reasons. Residents may have had obligations elsewhere that prevented them from attending the noontime lecture, may have been on leave, or declined to complete the questionnaire for personal reasons.
The group’s average age was 35 years (SD=7.5), 43% (N=150) of the group was female, 50% (N=105) were married, and 36% (N=107) had children. Family medicine residents comprised 57% of the sample and psychiatry residents 43% (N=67).
Data from residents who completed more than one set of questionnaires were averaged so that these residents were not over represented by appearing multiple times in the data set.
Gender and Parental Status
Women had lower scores than men on the depersonalization scale of the Maslach Burnout Inventory (t=3.37, p=0.001) (Figure 1
). Residents (men and women combined) with children had lower scores on two Maslach Burnout Inventory scales when compared to residents without children: emotional exhaustion (t=−2.59, p=0.011) and depersonalization (t=−3.98, p<0.001) (Figure 2
). There were no significant differences on the Work Environment Scale based on gender or parental status.
Parental status was examined for women and men as subgroups. Female residents with children (N=20) had decreased levels of burnout with lower scores on the emotional exhaustion scale (t=2.22, p=0.029) as well as on the depersonalization scale (t=−0.81, p=<0.001) when compared to the rest of the sample (Figure 3
). Six female residents who reported being single and having children had burnout scores that were not significantly different from the remainder of the group. Maslach Burnout Inventory scores of both men with children (N=36) and single men with children (N=2) did not differ from the rest of the sample.
Residents from the United States culture experienced significantly more burnout than those from other cultures with higher scores on both emotional exhaustion (t= −3.85, p<0.001) and depersonalization (t=−4.73, p<0.001) (Figure 4
). They also reported less task orientation (t=2.89, p=0.005), less control (t=2.24, p=0.027), and higher work pressure (t=−2.79, p=0.006) on the Work Environment Scale than residents not reared in the United States culture.
Psychiatry residents reported significantly less burnout when compared with family medicine residents on two scales of the Maslach Burnout Inventory; emotional exhaustion (t=2.05, p=0.042) and depersonalization (t=2.49, p=0.014) (Figure 5
). When family medicine and psychiatry residents’ scores on the Work Environment Scale were compared, psychiatry residents reported higher physical comfort scores (t=−2.60, p=0.011); and family medicine residents reported higher peer cohesion (t=3.41, p=0.001), supervisor support (t=2.38, p=0.019), and autonomy (t=2.27, p=0.025).
Pearson correlations were used to examine the relationship between the residents’ age and the three Maslach Burnout Inventory scales. Age and the depersonalization scale were inversely correlated, indicating that younger residents had higher depersonalization scores (r=−0.21, p=0.020). No differences between married and single residents were found.
We found gender differences in burnout among residents that indicate women residents experience less burnout than their male counterparts. Previous speculation had led to the hypothesis that women residents would have higher stress and be at greater risk of burnout as a result of conflicts between traditional gender roles and professional practice and because of the climate of medical education (competitive, oriented toward masculine traditions, etc.). Our data do not support this hypothesis. It is possible that the results reflect a response bias such that women residents are less comfortable disclosing that they depersonalize patients. However, none of the studies reviewed by Thomas (5) that analyzed resident burnout data by gender determined that women had a higher rate of burnout. In a study of factors affecting work-home interference in resident physicians, Geurts et al. (15) looked at the effect of gender and found emotional exhaustion higher in females but found no effect of gender on depersonalization and no overall effect of gender on measures of work-home interference. Two studies of practicing physicians (not residents) provide conflicting results. One failed to note any gender difference in burnout (16) while another suggested that women had increased levels of burnout; however, a five point self-report scale for burnout was used rather than the Maslach Burnout Inventory (17).
Our findings regarding the impact of parenting roles on burnout risk are quite provocative and are in agreement with those reported in previous studies by Collier et al., Martini et al., and Lemkau et al. (9, 18, 19) Although Lemkau et al. (19) and Martini et al. (18) found that having children had no effect on burnout in a group of family medicine residents, Lemkau et al. (8) did find that having children was associated with lower depersonalization scores (similar to our findings) in a later study of physicians in early practice.
Collier et al. (9) found residents with children scored lower on depression and cynicism and higher on humanistic feelings. Although it may seem counterintuitive, parenting during a residency appears to be a protective factor for female residents. Our data suggest that male residents do not accrue a significant protective effect from parenting. The “tend-and-befriend” stress response, a tendency to affiliate and nurture others as opposed to “fight or flight,” proposed by Taylor et al. (20) may partially explain this result. This hypothesis proposes that the stress response for women is characterized by a secretion of oxytocin to a greater extent than catecholamines. A study using animal models suggests that the effect of parenting may be even more fundamental. In this study, female rats with offspring showed lowered neurochemical and behavioral responses to stress than controls (21). One also may speculate that the responsibilities of parenting may create a different set of internal priorities for female residents, helping them to avoid being consumed by the daily stresses and conflicts within residency programs. In addition, parenting may have a humanizing effect, helping residents to avoid depersonalization.
Residents raised in a foreign culture reported less burnout despite the common belief that they experience increased levels of stress. Residents from other cultures had higher scores on task orientation, clarity, and physical comfort as well as lower scores on the experience of work pressure from the Work Environment Scale. One potential explanation is that relocating to the U.S. and adapting to the new culture might provide these residents with coping skills that serve them well in residency. They also may have a different perspective on their training situation that increases their adaptation (e.g., appraising the training as more valuable as a result of their cultural background and therefore willing to tolerate more hardship). A third potential explanation is that residents from other cultures may have stronger social networks through affiliation with fellow expatriates and families to form strong, supportive bonds. Another possible explanation is that residents from other cultures may not be as burdened by medical school debt as their U.S. graduate colleagues. Finally, administrative and other constraints may also serve to protect residents from other cultures. Since international medical graduates cannot be licensed until they complete residency, they cannot moonlight and the risk of overextension of time and energy may be reduced. Schaufeli et al. (22) compared Maslach Burnout Inventory scores of normative samples from the U.S. and the Netherlands, and found higher scores on the emotional exhaustion and depersonalization scales of the U.S. sample. He speculated that U.S. jobs may be more stressful, but also noted that different societal values such as a higher achievement orientation within the U.S. might be a factor. He cautions about comparing Maslach Burnout Inventory scores between nations and cultures. Medical residents who reside in the U.S. for but a short period might also be expected to reflect the values of their native country.
Our sample of psychiatry residents had less burnout than the family medicine residents on both emotional exhaustion and depersonalization scales. This finding differs from Martini et al. (18), who found no Maslach Burnout Inventory score differences among eight specialties (including psychiatry and family medicine). However, Martini et al. noted that a low response rate (35%) could have compromised their conclusions. In our study, family medicine residents reported significantly higher scores than psychiatry residents on the peer cohesion, supervisor support, and autonomy scales of the Work Environment Scale. Family medicine residents also reported lower physical comfort scores than psychiatry residents, a finding that could be associated with more intense on-call demands for family medicine residents and the pressure to see more primary care patients in shorter office visits.
In our sample, older residents had less burnout with the depersonalization score on the Maslach Burnout Inventory decreasing with increasing age. This finding can be contrasted with McCranie (16), who found no association between age and burnout. That study, however, measured burnout using a “Tedium Scale” rather than the Maslach Burnout Inventory and surveyed practicing physicians rather than residents. Lemkau et al. (8) used the Maslach Burnout Inventory with family medicine residents but found no correlation of burnout and age. Lemkau’s group was smaller (N=71) and the residents were both younger and had less variability in age [average age 29 (SD=3.8)] than our group [35 years (SD=7.5)]. It seems reasonable that our sample would be more sensitive to the effects of age. If average ages among residents are increasing, then our study may be more informative for current policies and plans.
We found no difference between married and single residents in the Maslach Burnout Inventory scales. Given the well established role of social support in buffering the effects of adverse life events, having the support of a spouse might be expected to protect against burnout in residents. Results of previous research have been mixed on this point, however. Although Martini et al. (18) found less burnout in married residents, others found no effect based on marital status (16, 19). Our results can be added to those which challenge the idea that marriage is a buffering factor against resident burnout. While we cannot rule out that the residents in our sample had relationships other than marriage (e.g., long term committed relationships) that provided a function that marriage is frequently hypothesized to provide, these results clearly question the role of marriage per se, in protecting against burnout.
There are weaknesses in our study design. The anonymous nature of the questionnaires is a potential source of selection bias that is difficult to assess. For example, residents with higher levels of burnout may have been less inclined to take on the additional task of completing the questionnaires, or less inclined to attend noontime conferences where the questionnaires were distributed. Also, despite the promise of anonymity, residents may have been anxious about providing answers critical of the training programs.
Important issues were identified for further exploration: the relationship of depression to burnout, how parenting may protect residents from burnout, and the possible protective effect of being raised in a foreign culture. We attempted to assess changes in burnout over time in our sample, but it proved methodologically impossible. Changes in the composition of resident samples from one survey administration to another produced data that were not meaningful. This remains an issue for further research.