Medical Students

Numerous studies have found a high prevalence of burnout among students in medical schools. Estimates from a comprehensive narrative review of articles about medical student burnout (1990–2015) were that 35 to 45 percent of medical students had high emotional exhaustion, 26 to 38 percent had high depersonalization, and 45 to 56 percent had at least one symptom of burnout (Dyrbye and Shanafelt, 2016). It is difficult to ascertain whether the prevalence of burnout increased over this time period given the lack of large, multi-institutional or national studies using similar methodologies and the existence of only one publication on burnout among medical students prior to 2005 (Guthrie et al., 1998). A review of existing data from large cross-sectional studies conducted over the past decade using similar methodologies found that the prevalence of burnout has had an upward trajectory overall (Dyrbye and Shanafelt, 2016). This review included results from a large study of more than 4,000 U.S. allopathic medical students, which reported that 56 percent of medical students had at least one symptom of burnout. In addition, high emotional exhaustion, high depersonalization, and overall burnout were more prevalent among medical students than in a national sample of age-similar college graduates not studying medicine (Dyrbye et al., 2014). Similarly, a study of nearly 1,300 osteopathic medical students found that more than 40 percent had symptoms of burnout (Lapinski et al., 2016).

Little is known about the course of burnout over time. Limited cross-sectional data suggest that the prevalence of burnout increases as students move into the more advanced years of medical school training (Dyrbye and Shanafelt, 2016). In a 1-year longitudinal study of medical students, 36 percent did not have burnout at either baseline or follow-up, 16 percent did not have burnout at baseline but did at follow-up (new burnout), 34 percent had burnout at both time points (chronic burnout), and 12 percent had burnout at baseline but not at follow-up (recovered) (Dyrbye et al., 2008). In a longitudinal study of medical students in Sweden, emotional exhaustion at the end of medical school, as measured by the Oldenburg Burnout Inventory, independently predicted the students' level of emotional exhaustion during their internships (Dahlin et al., 2007).

Resident Physicians

During postgraduate specialty training, also called residency, the “resident physician” works and learns under the supervision of a senior physician licensed in that specialty. There is evidence showing a high prevalence of burnout among resident physicians; this is present at the beginning of residency training and continues into subsequent years of residency (Willcock et al., 2004).

In a national study of second-year residents (n = 3,588), 45 percent were reported to have burnout symptoms, with a large variation in prevalence by specialty that was similar to the variation found in studies of practicing physicians (post-training) (Dyrbye et al., 2018). A multi-variate analysis found that training in urology, neurology, emergency medicine, ophthalmology, and general surgery was independently associated with higher relative risks (RRs) of burnout symptoms relative to training in internal medicine (the RRs ranged from 1.23 to 1.48), after controlling for other factors.

Several prevalence studies assessing burnout in other years of training or within a single discipline also indicate that residents commonly experience burnout. The percentage of residents reporting burnout symptoms across these studies is high: 32 to 38 percent have high depersonalization, 44 to 50 percent have high emotional exhaustion, and more than 60 percent have at least one symptom of burnout (Attenello et al., 2018; Dyrbye and Shanafelt, 2016; Moradi et al., 2015; Prins et al., 2007a; Ripp et al., 2011; West et al., 2011). Similar to the case with medical students, findings from a national sample of 1,701 residents conducted in 2012 reported that the prevalence of high emotional exhaustion, high depersonalization, and overall burnout was higher among residents than among age-similar college graduates not studying medicine (Dyrbye et al., 2014).

Nursing, Dental, and Pharmacy Learners

Compared to the research on physician trainees, there are fewer studies measuring the symptoms of burnout in students and trainees in nursing, dentistry, and pharmacy. Nonetheless, there is evidence to show that training in these fields is stressful and causes distress (Del Prato et al., 2011; Dutta et al., 2005; Edwards et al., 2010; Labrague et al., 2018). For example, Zinurova and DeHart found higher levels of stress among pharmacy residents than among other individuals in the U.S. population (Zinurova and DeHart, 2018).

In general, existing studies of burnout among nursing, dentistry, and pharmacy learners are small, single-institutional, or conducted outside the United States, limiting the generalizability of the findings. Nonetheless, the data suggest that burnout among non-physician learners is a serious problem. For example, a 2010 study of 436 nursing students at a single institution found emotional exhaustion, depersonalization, and personal accomplishment scores (Maslach Burnout Inventory–Human Services Survey for Medical Personnel [MBI–HSS MP]) of 18.2–22.3, 5.2–6.7, and 32.8–36.2, respectively, with mean scores lowest in the first-year nursing students (Michalec et al., 2013). A longitudinal study of nursing students in Sweden reported that the exhaustion and disengagement (as measured by the Oldenburg Burnout Inventory) increased over 3 years of training (Rudman and Gustavsson, 2012). A small, single-institutional study of 151 dental and dental hygiene students found that 40 percent had burnout symptoms (Deeb et al., 2018). A single-institution study of 629 pharmacy students showed significant associations between older age, year in school (higher in second than first year), and being at the main campus (versus a branch campus) and higher emotional exhaustion scores (using MBI–Student Survey [MBI–SS]) (Ried et al., 2006).

Demographic Characteristics

Studies of learners have reported differences in the prevalence of burnout by individual demographic variables, but the findings are not consistent. Some studies suggest that women learners are at higher risk for burnout than their male colleagues (Dyrbye et al., 2009b, 2018; Elmore et al., 2016). For example, in a national study of 3,588 second-year residents, female residents had a 19 percent increased odds of burnout (versus males) after controlling for specialty and other characteristics (Dyrbye et al., 2018). Longitudinal research, however, has produced conflicting findings. One study in medical students and another in internal medicine residents found no relationship between gender and the risk of developing burnout (Dyrbye et al., 2010b; Ripp et al., 2011). In a 3-year longitudinal study of 86 internal medicine residents, a logistic regression model indicated that male residents had a three-fold increased risk of persistent burnout, defined as having burnout at each time point across 3 years, after controlling for other factors (Campbell et al., 2010). There may also be subtle sex differences in the way that burnout is manifested among learners, with burnout more likely to manifest as emotional exhaustion in women than in men and men more likely than women to experience depersonalization (Golub et al., 2007; Liébana-Presa et al., 2018; Prins et al., 2007c; Ried et al., 2006).

With respect to race, several multi-institutional studies involving more than 5,700 medical students have found that self-identifying as white was an independent predictor of burnout, after controlling for age, gender, and parenting and marital status (Dyrbye et al., 2007, 2009b, 2010b). Likewise, in a longitudinal multi-institutional study of 1,701 medical students, being non-white was an independent predictor of higher likelihood of recovering from burnout within 1 year (Dyrbye et al., 2010b). Although two cross-sectional studies involving family medicine residents and hematology–oncology fellows also reported a relationship between race and burnout (Michels et al., 2003; Mougalian et al., 2013), a longitudinal study involving more than 3,500 second-year residents found no independent relationship between burnout and race, ethnicity, or country of birth (United States or not) (Dyrbye et al., 2018).

Less is known about the experience of learners who belong to other minority groups. In a study of 1,294 osteopathic medical students, a univariate analysis showed that those who self-identified as lesbian/gay/bisexual/asexual were more than twice as likely to report symptoms of burnout than heterosexual students. There were no differences in the mean scores for emotional exhaustion or depersonalization between the groups and no adjusted analysis (Lapinski et al., 2016). In a large national study of allopathic medical students, sexual minority students (lesbian, gay, transgender, bisexual, questioning [LGTBQ]) had a greater risk of depressive symptoms, anxiety symptoms, and low self-rated health, after adjusting for relevant covariates (Przedworski et al., 2015).

The prevalence of burnout among U.S. medical residents may vary according to where the individual trained prior to residency. A large study involving nearly three-fourths of all U.S. internal medicine residents reported that international medical graduates were less likely to have burnout, a finding that persisted on multi-variate analysis that controlled for debt and other factors (West et al., 2011). Similarly, a small study of 150 medical residents from 13 specialties training in two hospitals reported residents who had completed medical school outside of the United States had significantly lower emotional exhaustion and depersonalization on univariate analysis (Eckleberry-Hunt et al., 2009).

Care Quality and Patient Safety

The evidence shows that burnout affects the quality of patient care delivered by medical trainees. Several studies have found that residents with burnout are more likely to report delivering sub-optimal care or having committed a medical error (Shanafelt et al., 2002; Wallace and Lemaire, 2009; West et al., 2006, 2009). In the seminal 2002 study on burnout (Shanafelt et al., 2002), burnout among a sample of internal medicine residents (n = 115) significantly predicted self-reported sub-optimal patient care practices, such as dismissing patients from the inpatient service to make the service more manageable, not fully discussing treatment options with patients, and paying little or no attention to the social or personal impact of illness on patients. Longitudinal studies of internal medicine residents have similarly found burnout to be an independent predictor of residents perceiving that they committed a medical error over the subsequent 3 months (West et al., 2006, 2009).

Negative emotions can impede learning, recall, and the application of knowledge and skills (McConnell and Eva, 2012). Consistent with this, medical residents with burnout are also more likely to report greater struggles with concentrating at work (Fahrenkopf et al., 2008). In one qualitative study, residents described how fluctuations in their overall well-being affected their motivation at work (Ratanawongsa et al., 2007). Burnout's adverse effects on learning and performance are further supported by a study of more than 16,000 U.S. internal medicine residents that reported a step-wise reduction in Internal Medicine In-Training Examination (IM-ITE) scores as burnout symptoms worsened (West et al., 2011). In addition, in a study of 58 emergency medicine residents from six institutions, residents with burnout showed lower cumulative performance scores on high-fidelity simulation scenarios used for assessing performance (Lu et al., 2017).

Other findings about the well-being and the experiences of medical trainees include focus group results from 26 medical residents in internal medicine, psychiatry, surgery, emergency medicine, anesthesia, obstetrics and gynecology, and pediatrics. Affirming the claim that well-being influences decisions and social interactions, the study reported that participants perceive having better decision-making capacity and higher-quality patient discussions when their well-being is high and more sub-optimal exchanges with patients when their well-being is low (Ratanawongsa et al., 2007). However, a small, single-institution study of 202 internal medicine residents found that those with burnout had higher supervisor ratings of their communication skills with patients, families, and team members (Beckman et al., 2012).

Few studies have examined the relationship of learner burnout in nursing, pharmacy, and dentistry with patient outcomes. A longitudinal study of more than 1,700 nursing students in Sweden, using the Oldenburg Burnout Inventory found that baseline levels of burnout and the subsequent experience of burnout during nursing school predicted lower levels of self-reported, in-class learner engagement as well as a lower perception of occupational preparedness in the final year of nursing school. Earlier development of burnout during nursing school also predicted lower self-reported mastery of occupational tasks and less use of research in every day clinical practice 1 year after graduation (Rudman and Gustavsson, 2012).

The evidence strongly indicates that burnout can impede the acquisition of professional knowledge and skills and thus diminish the systems' capacity to provide the best patient care, placing patients at risk in both the short term and the long term for sub-optimal care and medical errors.

Professionalism

Chapter 2 contains a discussion that places the tenets of professionalism in the context of this report. Here, studies examining the attributes of professionalism and burnout among learners document the negative consequences for patient care. In a multi-institutional study of 2,682 medical students, burnout was independently associated with self-reported unprofessional conduct, with the potential to undermine both student competency (by cheating and plagiarism) and good patient care (by lying about aspects of the physical exam done and tests ordered, etc.) (Dyrbye et al., 2010a). The evidence suggests that medical students with burnout have lower altruism and integrity, are less likely to endorse correct attitudes about managing conflicts of interests with industry, are less likely to have correct attitudes about appropriate prescribing behaviors, and are less likely to believe that they have a personal responsibility to report impaired colleagues (Dyrbye et al., 2010a, 2015b).

Similarly, several studies have found that medical students with burnout have lower empathy, including a multi-institutional study of 545 medical students (Thomas et al., 2007), a single-institution study of 127 fourth-year medical students (Brazeau et al., 2010), and a national study of 1,350 Brazilian medical students (Paro et al., 2014). These findings are important because empathy during medical school predicts subsequent risk of burnout during residency (Dyrbye et al., 2018). In a longitudinal study of internal medicine residents, reduced empathy was an independent predictor of higher odds of self-perceived error in the following 3 months (West et al., 2006). A study of 254 hematology–oncology fellows suggests that lower levels of burnout may promote more compassionate patient care (Mougalian et al., 2013).

Societal Consequences

Attrition in health professions education has the potential to contribute to shortages in some professions and specialties. When learners do not finish their training, the adequacy of the workforce is threatened, and institutions and individual learners bear the cost (Jones and Korn, 1997). There can be many reasons students choose to discontinue training, and studies suggest that half of those who choose to discontinue medical school do so for nonacademic reasons (Garrison et al., 2007).

Burnout may influence learner attrition and turnover. A multi-institutional study of medical students (n = 858) found that burnout independently predicted students having serious thoughts of dropping out of medical school over the course of 1 year. The frequency of serious thoughts of dropping out returned to baseline level with recovery from burnout, suggesting a causal association (Dyrbye et al., 2010c). In addition, an earlier development of burnout during school was reported to be associated with higher turnover intentions during the first year post training in a longitudinal study of burnout in nursing students (n = 1,702) in Sweden that used the Oldenburg Burnout Inventory (Rudman and Gustavsson, 2012).

Burnout is associated with an increased likelihood of career choice regret among residents (Becker et al., 2006; Campbell et al., 2010; Dyrbye et al., 2018; Shanafelt et al., 2002). In a cross-sectional study of second-year residents (n = 3,588), career regret was reported by 14 percent and was significantly associated with burnout (Dyrbye et al., 2018). These findings parallel those in studies of practicing physicians, which report that burnout is associated with turnover intention (Sinsky et al., 2017).

In addition, some studies suggest that burnout may influence medical students' specialty choices, raising the possibility of societal consequences since an inadequate specialty distribution among physicians in the workforce can negatively affect patients' access to care. For example, one study reported that medical students with high emotional exhaustion were more likely to choose a specialty with a more controllable lifestyle (e.g., dermatology versus internal medicine) and that those with low personal accomplishment were more likely to choose a higher-income specialty (e.g., anesthesiology versus pediatrics) (Enoch et al., 2013). Other studies found that burnout among medical residents increased the likelihood of seriously considering changing specialty and abandoning medicine altogether (Blanchard et al., 2010) and that medical residents who developed burnout over the course of intern year were less likely to plan to pursue subspecialty training (Campbell et al., 2010).

Personal Health

Cross-sectional studies in medical students and residents show a relationship between burnout and suicidal thoughts (Dyrbye et al., 2008; van der Heijden et al., 2008). A multi-institutional longitudinal study of U.S. medical students (n = 1,321) found that burnout was an independent predictor of medical students developing thoughts of suicide over the course of 1 year (Dyrbye et al., 2008). Even medical students without depression were at substantially higher risk of developing thoughts of suicide if they had burnout. In a cross-sectional study of residents training in the Netherlands (van der Heijden et al., 2008) and the United States (Lebares et al., 2018), suicidal thoughts were also more prevalent among residents with burnout.

Similar to findings concerning alcohol use among U.S. physicians (Oreskovich et al., 2012), cross-sectional studies of learners report that burnout is associated with higher alcohol intake. For example, burnout was found to be an independent predictor of alcohol abuse and dependence in a multi-institutional study of 4,402 medical students (Jackson et al., 2016), and higher emotional exhaustion and depersonalization scores were associated with greater alcohol intake in a study of 168 family medicine residents (Lebensohn et al., 2013).

There are some data to suggest burnout in learners may increase the likelihood of them subsequently developing poor mental health, but, notably, poor mental health may not increase the likelihood of burnout. In a study of 218 nursing students, emotional exhaustion scores (measured by the MBI–HSS) at baseline predicted psychological well-being (measured by the General Health Questionnaire [GHQ]-12) 18 months later; however, baseline psychological well-being did not predict subsequent emotional exhaustion, suggesting that the relationship is not bidirectional (Ríos-Risquez et al., 2018). Concordant with those findings, a multi-institutional longitudinal study of 185 first-year internal medicine residents found no relation between past psychiatric history and the incidence rate of burnout over the course of 1 year (Ripp et al., 2011).

Burnout also appears to adversely affect learners' physical health. A longitudinal study of internal medicine residents (n = 340) provides evidence of a dose–response relationship between burnout and adverse risk to personal health; each 1-point decrease in a personal accomplishment score was associated with an 8 percent increase in the odds of a self-reported blood and body fluid exposure in the subsequent 3 months (West et al., 2012). Moreover, after controlling for fatigue, each 1-point increase in an emotional exhaustion score or a depersonalization score was associated with a 3 percent increase or 4 percent increase, respectively, in the odds of reporting any motor vehicle incident in the subsequent 3 months. This magnitude of effect was large enough to meaningfully affect resident safety and potentially public safety as well, if others were involved in the motor vehicle incidents.

Work Hours, Overnight Call, and Workload

Among practicing physicians, studies show that work hours and workload are major drivers of burnout; their role in the experience of burnout among learners is less clear, however. Among medical students, two studies suggest no relationship between work hours and workload and burnout. A study of first- and second-year (pre-clinical) medical students reported no significant association between hours spent in lectures and small groups, hours of clinical experiences, hours and number of exams, or weeks of vacation and any measure of student well-being, including burnout, quality of life, or depressive symptoms (Reed et al., 2011). In a study of third- and fourth-year medical students, workload characteristics (e.g., number of patients cared for in the past week, call schedule, inpatient/outpatient rotation) and the current specialty of the clinical rotation were not independent predictors of burnout (Dyrbye et al., 2009b).

In a study of first-year pharmacy residents, a relationship was found between higher work hours and greater levels of stress (Zinurova and DeHart, 2018). Among medical residents, work hours, excessive workload, and overnight call frequency have been shown to increase the risk of burnout (Sargent et al., 2004; van Vendeloo et al., 2018), but the studies are inconsistent (Dyrbye and Shanafelt, 2016). Several studies examined the Accreditation Council for Graduate Medical Education's (ACGME's) common duty-hour standards for resident training in the United States. In 2003, ACGME set resident duty-hour and institutional oversight standards, which it revised in 2011 and again in 2017 (Nasca, 2017).

Only three of seven studies using historical cohort controls found that the 2003 ACGME mandated work-hour reforms led to significant reduction in burnout among residents (Dyrbye and Shanafelt, 2016; Fletcher et al., 2011). A 2005 study of 684 residents in otolaryngology–head and neck surgery reported a strong positive linear relationship between emotional exhaustion and hours worked which persisted after adjusting for potential confounders (Golub et al., 2007). However, in a 2008 national study of nearly three-quarters of all U.S. internal medicine residents (>16,000 residents), the prevalence of burnout was similar to rates prior to 2003 ACGME work-hour reforms (West et al., 2011). Additionally, no relationship was found between self-reported workload, work hours, call frequency, and burnout in two longitudinal studies of residents conducted in 2003–2008 (Campbell et al., 2010; Ripp et al., 2011, 2015).

Studies also reported mixed findings related to resident well-being and the 2011 ACGME-mandated changes to shift length, night float rotations, and protected sleep time (Reed et al., 2010). For example, a study of first-year internal medicine residents found similar year-end prevalence of burnout in cohorts that trained before and after the 2011 ACGME changes (Ripp et al., 2015). Similarly, there were no differences in burnout among 47 residents in Canada who were randomly assigned in 2-month blocks to an intensive care unit rotation with overnight in-house calls in schedule blocks of 24, 16, or 12 hours (Parshuram et al., 2015). In a 2012 single-institution study (Krug et al., 2017), internal medicine residents reported a lower prevalence of burnout (61 percent) than had been reported in 2001 (76 percent), but there was no statistically significant difference in prevalence when compared with the 2004 (64 percent) historical cohort. Only 23 percent of the residents in 2012 thought that the 2011 duty-hour policy had a positive impact on their well-being, and most thought the policy had a neutral (42 percent) or negative (41 percent) impact on patient care (Krug et al., 2017).

In light of the difficulty in proving benefit across several dimensions and concern among program directors about the rigidity of the policy, ACGME reviewed the 2011 duty-hour standards and revised the policy in 2017 (ACGME, 2017). However, one recent study (Desai et al., 2018) showed no significant differences in mean emotional exhaustion or depersonalization scores between cohorts of residents in 63 internal medicine programs who trained under a flexible policy that resembles the 2017 ACGME requirements and those who trained under the 2011 ACGME standard duty-hour policies. The potential impact of a flexible policy for surgical trainees is less clear since studies to date have not included a validated measure of wellbeing (Bilimoria et al., 2016). However, a qualitative study to explore some of these findings in greater depth found that residents perceived a flexible duty-hours policy to have several advantages, including giving them the ability to schedule time off for personal needs (Kreutzer et al., 2017).

The literature offers several possible reasons for the lack of consistently documented benefit from mandated work-hour restrictions on resident well-being. There is the possibility that the potential benefits are obscured as a result of work-compression, that is, that educational requirements and patient care duties remain the same despite a reduction in work hours. Or, alternatively, learners may feel a sense of increased stress due to the perception of being less well prepared for clinical tasks (owing to less repetition and exposure to clinical encounters or impressions given by supervisors) (Ludmerer, 2010). Eckleberg-Hunt (2009) reported that not having enough time in the day to complete the workload and perceptions of having excessive paperwork were associated with burnout among 150 residents in two hospitals. Similarly, Domaney and colleagues (2018) found that self-reported hours spent on EHRs (outside of work, total time, and reviewing notes) correlated with emotional exhaustion in a cross-sectional study of 40 psychiatric residents.

Curriculum and Training Experiences

Little is known about the relationship between school or clinical training program characteristics (e.g., size, competitiveness, research intensiveness, curriculum design [subject-centered or problem-based centered], faculty-to-student ratio, longitudinal clerkships) and burnout among learners, although recent longitudinal studies have explored how these factors relate to depression among learners (Dyrbye et al., 2019b; Pereira-Lima et al., 2019).

Licensure and Hospital/Clinic Credentialing and Privileging

Multiple organizational, accreditation, and licensing boards have a role in influencing the types, frequency, and stakes of assessments, which results in substantial variability across health profession training programs. Licensure requires passing national standardized examinations, and exam stress is a widely acknowledged stressor for learners (Jenkins et al., 2018; Jevtic et al., 2012). How well learners perform on these exams may have a large impact on their future careers (National Resident Matching Program, 2018). For example, a medical student's USMLE Step 1 score has a strong impact on that student's chances of getting a residency within his or her desired specialty. The competition within select residency training specialty programs is steep, particularly in more desirable training locations (Mullan et al., 2015). Although well accepted as an enormous stressor for medical students, the way in which the USMLE Step 1 exam affects self-care behaviors and mental health has not yet been formally studied. For residents, licensure typically requires a specified period of training. Those residents who take a medical leave due to illness or maternity may experience escalations of their stress as rigid training requirements constrain the time available for self-care and personal life events (Magudia et al., 2018; Varda et al., 2018).

In addition, state medical license boards and hospital credentialing and privileging processes commonly inquire about previous emotional problems, including help-seeking behaviors (Dyrbye et al., 2017c). Although those types of questions are well-established barriers to help-seeking for practicing physicians experiencing emotional difficulties (Dyrbye et al., 2017c). Less is known about their role in learners' reluctance to seek help for emotional problems. However, factors such as a lack of time, a lack of confidentiality, the stigma associated with using mental health services, cost, a fear of documentation on academic record, and a fear of unwanted intervention all appear to be major barriers to help-seeking among medical students (Chew-Graham et al., 2003; Dyrbye et al., 2015a; Givens and Tjia, 2002; Schwenk et al., 2010).

Data show that medical students are more reluctant to seek help for a serious emotional problem than the general population (and age-matched individuals) (Dyrbye et al., 2015a). A 2015 study found that perceived stigma likely explains why medical students with burnout do not actively seek help and, in addition, reported that faculty and peer behaviors may also influence a learner's help-seeking (Dyrbye et al., 2015a). Although burnout is not a mental health problem, it can lead to one (Hakanen and Schaufeli, 2012; Ríos-Risquez et al., 2018), which underscores the urgency to eliminate barriers to help-seeking for all students, trainees, and practicing clinicians.

Rapidly Changing Health Care Delivery

For students, trainees, and practicing clinicians, the unprecedented rate of change in health care is one source of feelings of uncertainty about their professional future, and feelings of uncertainty appear to be more common among residents meeting burnout criteria (Shanafelt et al., 2002). Health care systems changes have led to additional required competencies for the 21st-century physician (Crosson et al., 2011; Lucey, 2013), resulting in an increase in the amount of content one must master during medical school. This expanding content adds to workload. A study conducted at a single institution a decade ago found that medical students would need to devote 5 to 7 days of 6 hours each week to complete the assigned basic science reading only once (Klatt and Klatt, 2011). Adding content to an already dense educational program is occurring in an era when some medical schools are reducing their 4-year curriculum to 3 years, at least for some learners.

Patient Factors

Interactions with patients are a critical component of competency development for health care professional students and trainees. Surprisingly little is known about patient-related factors, interpersonal relationships between patients and their families, and learners' experiences. The interplay between the level of complexity of a patient's medical problem, the learner's level of preparedness to handle the situation, available support and resources, and the learner's professional development and well-being is largely unexplored.

A cross-sectional study of 150 residents from 13 specialties in two U.S. hospitals found an association between self-reported difficult and complicated patients and a higher risk of burnout (Eckleberry-Hunt et al., 2009). On the other hand, patients can also be a source of renewed energy for learners (Mata et al., 2016).

Organizational Culture

Many factors arising from an organization's culture are positive and supportive, but there are many other factors that can contribute to the risk of burnout, particularly among certain groups of learners. For example, organizational culture influences how many same-gender, same-race, same-ethnicity, or same–sexual orientation advocates and role models are available to provide personal and professional support to learners. Learners may struggle even more without same-gender role models. In one qualitative study, for example, 35 female and 63 male surgical residents generally agreed that women residents had to try harder and received less respect from hospital staff and patients (Dahlke et al., 2018).

Another cultural factors that may affect learner well-being is the racial and ethnic inclusiveness of the organization. Although non-white medical students are overall at lower risk for burnout, a multi-institutional study found that this changes when non-white medical students perceive that their race has adversely affected their medical school experience; that is, they were more likely than non-white students who reported no such experience to have burnout as well as a lower quality of life and depressive symptoms (Dyrbye et al., 2007). Similarly, a longitudinal study of medical students found that greater exposure to a negative medical school diversity climate was associated with an increase in depressive symptoms among students of all races and ethnicities groups (Hardeman et al., 2016).

Flexibility and Empowerment

Studies show that the degree of control that learners have over their education and daily lives also affects stress level. Medical students who perceive having little control over their daily schedule or life are more likely to have burnout (Mazurkiewicz et al., 2012; Santen et al., 2010), and among a sample of residents from 13 specialties in two hospitals, perceptions of a lack of control over office processes and schedule increased the likelihood of burnout (Eckleberry-Hunt et al., 2009). A qualitative study of 26 residents from seven specialty training programs (Ratanawongsa et al., 2007) found that schedule flexibility was more important to some residents' sense of well-being than an overall limitation of work hours. Overly rigid schedules amplify the challenges of completing personal tasks, sending the message that personal needs are inconsequential, which has been found to be a major stressor more often for residents with burnout than for those without (Shanafelt et al., 2002).

Educator Behaviors

Learners in the health professions have a variety of teachers and clinical supervisors (educators), including other learners who are more advanced in their training and faculty within and outside their intended professions. Faculty members are at various stages of their own careers and have disparate abilities and interests in educating learners as well as different levels of investment and connection to the training program.

Learning interactions occur in a variety of settings, including classrooms, laboratories, simulation centers, and clinical settings—each with its own unique attributes, challenges, and constraints. For example, in most clinical settings the faculty are expected to teach and assess multiple learners at various stages of professional development while also having patient care responsibilities. A faculty member's ability to shape clinical interactions and provide an optimal levels of professional challenge and support to a learner's professional growth depends on many factors within and outside the scope of faculty's influence, such as the learner's previous experiences and formal curriculum, the types of patients available, and the faculty member's skills and own job demands (Sanford, 1962).

Many learning environment elements that can adversely affect learners can also have negative effects on educators. A national study of U.S. second-year residents found that the clinical specialty areas with the highest prevalence of resident burnout were similar to the clinical specialty areas with the highest rates of burnout among practicing physicians (Dyrbye et al., 2018). The authors say further study is needed to determine whether this finding is due to unique work environment factors intrinsic to certain specialties or to the behaviors of burned-out supervising physicians that model burnout behavior to residents or otherwise negatively affect the learning environment. On that latter point, a national study found that clerkship directors with burnout were more likely to report an attitude of not caring what happens to some of their students, suggesting that faculty well-being may influence teaching behaviors and the experience of learners (Dyrbye et al., 2009a). In a qualitative study, medical students identified negative role models as a major stressor in the learning environment and reported that positive role models helped them reframe stressors, making the challenges inherent to professional development seem more worthwhile (Jenkins et al., 2018).

Substantial evidence suggests that various aspects of the learner–educator relationship, such as faculty support, relate to a learner's risk of experiencing burnout symptoms. In a study of more than 1,100 medical students attending five different medical schools, Dyrbye and colleagues (2009b) found that dissatisfaction with the level of faculty support was an independent predictor of burnout among first- and second-year students, while perceptions of residents being cynical and the students' dissatisfaction with the level of resident supervision were independent predictors of burnout among third- and fourth-year students (Dyrbye et al., 2009b). In contrast, students reporting that education was a priority for faculty members was an independent predictor of “never” having burnout (i.e., not having burnout at baseline or 1 year later) and of “recovering” from burnout (i.e., having burnout at baseline but not 1 year later) in a multi-institutional study of medical students (Dyrbye et al., 2010b). In a study of residents training in the Netherlands, dissatisfaction with the emotional support received from supervisors was a strong predictor of emotional exhaustion and depersonalization, and this finding persisted upon multi-variate analysis controlling for other factors (Prins et al., 2007c). Among otolaryngology–head and neck surgery residents (n = 684), greater perceptions of demands from attending physicians and lack of independence correlated with higher emotional exhaustion scores (Golub et al., 2007). Similarly, several studies have demonstrated a relationship between a perception of insufficient autonomy and burnout, including a study of 193 emergency medicine residents in eight residency programs (Kimo Takayesu et al., 2014) and a smaller sample of residents in the Netherlands (Ringrose et al., 2009). Residents with burnout are also more likely to describe their relationships with supervisors as stressful (Prins et al., 2007b; Sargent et al., 2004).

Reports of harassment and belittlement are common among all learners, particularly among minorities. Not surprisingly, perceptions of recurrent mistreatment by faculty or residents are associated with an increased risk of burnout among medical students (Cook et al., 2014). Within dentistry, Rowland and colleagues (2010) found that intimidation by instructors was significantly associated with perceived stress, a finding that persisted after controlling for school, age, and gender. Similarly, in a study of 395 neurosurgical residents, perceptions of working with hostile faculty was an independent predictor of burnout that persisted on multivariable analysis (Attenello et al., 2018).

Other studies suggest a relationship between faculty teaching behaviors and learner burnout. For example, more favorable ratings of teachers, reported explicit teaching about certain topics, and direct observation of goals-of-care discussions was associated with lower emotional exhaustion scores and better overall teaching quality, while more frequent observation of the residents' skills was associated with lower depersonalization scores in a sample of hematology–oncology residents (Mougalian et al., 2013). Although the direction of the relationship cannot be determined from these cross-sectional studies, one longitudinal study involving 186 final-year medical students in Sweden found that positive perceptions of the first year of residency training environment, driven by supervisors who incorporated residents' needs for education, feedback, and support, was negatively associated with the development of emotional exhaustion (measured using Oldenburg Burnout Inventory) during residency (Dahlin et al., 2010). Similarly, when residents perceived their relationships with supervisors to be one of mutual support and benefit—that is, the resident benefited from a supervisor's teaching and support, and the supervisor benefited from the work done by the resident—residents had lower emotional exhaustion and depersonalization scores (Prins et al., 2008).

Peer Behaviors

Peers can be a tremendous source of support for learners, but in some cases poor peer interactions (e.g., competitiveness) can affect a learner's well-being. For example, perceptions of poor peer collaboration and poor relationships with colleagues have been associated with an increased risk of burnout among orthopedic residents training in the Netherlands (van Vendeloo et al., 2014) and among U.S. residents training in 13 specialties in 2 hospitals (Eckleberry-Hunt et al., 2009). A qualitative study involving 26 residents from multiple specialties also found that residents attributed conflict with colleagues to lower states of well-being (Ratanawongsa et al., 2007).

Less is known about the frequency of harassment or belittlement among peer learners and how these relate to the risk of burnout, but a study of first-year dental students found that perceptions of bullying by fellow students was significantly associated with higher perceived stress (Rowland et al., 2010).

Work–Home Conflict

Work–home conflicts are a well-established contributor to burnout among practicing physicians (Dyrbye et al., 2011a,b, 2012). Few studies have explored learners' satisfaction with their work–life integration or experiences of work–home conflict. Small qualitative studies have examined the dual-role responsibilities experienced by residents and how work hours contribute to work–life imbalance (Dahlke et al., 2018; Rich et al., 2016). A study of 252 residents training in Belgium reported that satisfaction with work–life balance was an independent predictor of burnout (van Vendeloo et al., 2018). Similarly, in a study of 2,115 residents training in the Netherlands, conflicts between work and home were found to be important contributors to burnout (Verweij et al., 2017). Conflict between personal and professional obligations was also associated with burnout in a study of pediatric residents (Sagalowsky et al., 2018). Other smaller studies have also reported an association between work–home conflict and higher emotional exhaustion scores, higher depersonalization scores, and overall burnout among residents (Ringrose et al., 2009; Sargent et al., 2004).

By contrast, a qualitative study of 26 residents from multiple specialties indicated that they “viewed residency as a time for temporary imbalance, during which they invested in professional development at the expense of other domains” (Ratanawongsa et al., 2007, p. 273). Although some re-balancing of personal priorities for the sake of professional development may be warranted, the authors noted that excessive self-sacrifice can be detrimental. This point is underscored by findings by Shanafelt and colleagues (2002), which found that among internal medicine residents, those who reported adopting a survival attitude as a significant or essential approach for managing stress were more likely to have burnout.

Personality and Personal Disposition

In a prospective study of learners in the United Kingdom, a weak but statistically significant association was found between higher levels of neuroticism (measured by an abbreviated questionnaire assessing the “big five” personality dimensions), measured both at the time of the study and 5 or more years prior, and emotional exhaustion and depersonalization (measured using a modified MBI) as well as lower career satisfaction once in practice. Individuals who were extroverts also reported lower levels of emotional exhaustion subsequently (McManus et al., 2004). A more recent study of 185 internal medicine residents, which included a 10-item personality instrument assessing extroversion, agreeableness, conscientiousness, emotional stability, and openness to experience, reported that a disorganized personality style was associated with increased odds of developing burnout over the course of the first year of residency (Ripp et al., 2011).

In a longitudinal study involving more than 3,588 individuals followed from the first year of medical school through residency, higher reported levels of anxiety during the fourth year of medical school were associated with higher odds of burnout as a second-year resident (Dyrbye et al., 2018). In a longitudinal study that followed medical students as they transitioned from the Karolinska Institute Medical School into residency, a high degree of worry about the future during the final year of medical school predicted emotional exhaustion (Oldenburg Burnout Inventory) 6 to 10 months after graduation. The authors suggest that students who are anxious about workload, long hours, the volume of information to learn, and their ability to meet future responsibilities may be more vulnerable to burnout as they start residency (Dahlin et al., 2010). In a study of 342 Swedish medical students, a small to moderate relationship was found between performance-based self-esteem and exhaustion and disengagement (as measured by the Oldenburg Burnout Inventory) (Dahlin et al., 2007). Similarly, a correlation between confidence in having the knowledge and skills needed to become an intern was found to be related to burnout (as measured by MBI–General Survey [MBI–GS]) in a small study of 89 third-year medical students (Mazurkiewicz et al., 2012). Other studies suggest that medical students who feel like an “imposter” (Villwock et al., 2016) and residents who report a intolerance of uncertainty may be more likely to experience higher degrees of burnout (Kimo Takayesu et al., 2014; Simpkin et al., 2018).

Few studies have explored the relationship between the underlying reason for entering the medical profession and the risk of developing burnout. One study of 277 medical students in Brazil found that those who had applied for medical school motivated by illness/death experiences were at a great risk for burnout (MBI–SS) according to a multiple regression analysis that controlled for life events, age, and gender; no relationship was found between altruism, economic motivator, or professional esteem as primary career choice motivator and burnout (Pagnin et al., 2013).

Coping and Self-Care Strategies

Learners use a variety of coping and self-care strategies (e.g., exercise, hobbies) to manage stress (Labrague et al., 2018; Ratanawongsa et al., 2007). Greater use of approach-oriented coping strategies (as opposed to avoidant-oriented strategies) was shown to significantly decrease the risk of burnout in a cohort of 161 medical students (Thompson et al., 2016) and among 280 nursing students (Gibbons, 2010). In a study of 244 clinical dental students in Chile, the prevalence and severity of burnout was directly related to the use of “social withdrawal” as a coping strategy (Pérez et al., 2016). Residents who report lacking skills to cope with stress are also more likely to report burnout (Eckleberry-Hunt et al., 2009). Interventions that teach coping skills to reduce the intensity or number of stressors have shown promising results in samples of nursing students (Galbraith and Brown, 2011). How best to teach approach-oriented coping strategies and if coping skills learned early on are transferrable to later practice both warrant additional study.

Similarly, self-care strategies are important. For example, a study of 4,402 medical students found that being compliant with exercise guidelines from the Centers for Disease Control and Prevention was an independent predictor of a lower risk of burnout (Dyrbye et al., 2017a). These students also reported a higher quality of life. Smaller studies of residents have found associations between greater physical activity and both an improved quality of life and a lower risk of burnout (Lebensohn et al., 2013; Weight et al., 2013).

Social Support and Isolation

Social support is a well-established buffer against emotional distress. Relationships with others are an important source of strength for medical students (Jenkins et al., 2018) and residents (Ratanawongsa et al., 2007). Needing to move for training disrupts personal lives and often separates learners from their family and friends, leading to lack of social support outside of training (Rich et al., 2016). Studies have found associations between perceptions of lower social support and the risk of burnout (Santen et al., 2010; Thompson et al., 2016). A large survey involving 2,115 residents in the Netherlands found that social support from family, partner, or colleagues seemed protective against burnout (Verweij et al., 2017). Among U.S. medical residents from 13 specialties in two hospitals, perceived social support was also significantly associated with levels of emotional exhaustion, depersonalization, and personal accomplishment (Eckleberry-Hunt et al., 2009). Sources of social support, however, likely vary for individuals. One small study of 86 third-year medical students found no relationship between having family in the local area and burnout (MBI–GS) (Mazurkiewicz et al., 2012), and a study of 1,294 osteopathic medical students reported that being involved in extramural club activities was associated with a lower risk of burnout (Lapinski et al., 2016).

Personal Life Stressors

Life stressors unrelated to educational or training processes also affect learners' well-being. For example, the number of negative personal life events (e.g., serious illness in a family member) in the previous 12 months was found to be correlated with the risk of burnout in a multi-institutional study of 545 medical students. A multi-variate analysis indicated that personal life events had a stronger relationship with burnout than did year of training (Dyrbye et al., 2006). Similarly, a larger study of osteopathic medical students reported that personal and family stressors were strongly linked to overall burnout (Lapinski et al., 2016). On the flip side, in a subsequent cross-sectional study of 897 first- and second-year medical students, not having a positive life events (e.g., getting married) within the previous 12 months was an independent predictor of burnout (Dyrbye et al., 2009b). Finally, social stressors outside of work were an independent predictor of burnout in a national study of 395 neurosurgical residents (Attenello et al., 2018).

Learners often have a high educational debt with the accompanying financial stress. Although, in a large study of medical students, educational debt has not found to be an independent predictor of burnout (Dyrbye et al., 2010b), a multi-institutional longitudinal study of 1,701 medical students found that those students who worked for income were significantly less likely to recover from burnout over the course of 1 year than students who were not employed (Dyrbye et al., 2010b). In contrast to medical students, educational debt may feel “more real” to residents, as they are in a situation of having to pay back their educational loans. In a study of more than 16,000 internal medicine residents, educational debt was an independent predictor of burnout (West et al., 2011). Although smaller studies support a relationship among education debt, financial stress, and burnout (Sargent et al., 2004), education debt was not an independent predictor of burnout in a large multi-specialty sample of second-year residents (Dyrbye et al., 2018) or in a 1-year longitudinal study of internal medicine residents (Ripp et al., 2011).

Structural Changes to the Learning Environment

Several recent publications have highlighted various approaches to improving the learning environment (Josiah Macy Jr. Foundation, 2018; Wasson et al., 2016) aimed at building a culture of respect (Morrissette and Doty-Sweetnam, 2010) and promoting well-being (Dyrbye and Shanafelt, 2016; Shiralkar et al., 2013). However, a 2016 systematic review found limited evidence showing that specific changes to the learning environment were associated with improved emotional well-being among medical students (Wasson et al., 2016). To improve the learning environment, the authors concluded that medical schools should consider a multi-faceted approach that includes pre-clinical pass–fail curricula and formal faculty advisor–mentor programs. Published studies suggest that pass–fail grading in the pre-clinical years is associated with enhanced student well-being without adverse impact on academic performance (Bloodgood et al., 2009; Reed et al., 2011; Rohe et al., 2006; Spring et al., 2011). Nonetheless, research assessing the relationship between how grades are assigned (e.g., norm-based grading versus criterion-based grading) and learner burnout or the relationship between grading schema and burnout among other health care professional learners does not seem to exist. Similarly there is no definitive evidence concerning whether new curricular models, such as longitudinal clerkships, or accelerated medical school training (e.g., 3-year tracks) lead to a lower risk of burnout.

There is an increasing body of evidence showing that faculty behaviors relate to the well-being of learners. Although some studies suggest that well-prepared preceptors may be able to facilitate new nurses transitioning into the work environment (Thomas et al., 2012), little remains known about how best to design and implement faculty development programming and organizational infrastructure in order to better equip and enable faculty to support and teach learners. The benefits of continuity of supervision (i.e., longitudinal relationships between faculty and learners) have been delineated (Hauer et al., 2012a,b; Hirsh et al., 2007; O'Brien et al., 2012), but the relationship between length of supervision by a faculty member and learner well-being has not been studied.

Lacking are studies focused on changes in the clinical work environment and learner well-being. Because learners and clinicians share the same environments, for at least a part of their experiences, there is a need to learn about the impact of clinician-focused system changes on learner burnout. In the most optimistic view, improvements in the clinician environment should benefit learners as well, but one could imagine that some changes aimed at improving clinician life could have an adverse effect on learners, or vice versa. Studies are also needed to determine how best to support learners involved in medical errors or emotionally laden events, such as unexpected deaths and illnesses in colleagues (Ripp et al., 2017). Some studies suggest some structural changes may effectively mitigate learner burnout. For example, providing medical residents with support from advanced practice providers and scribes (Holmes et al., 2017) and enhancing interprofessional collaboration between advanced practice nurses and junior doctors during overnight shifts (Johnson et al., 2017) have demonstrated some effectiveness.

As part of their overall well-being strategy, some medical schools and residency programs assess learner well-being (Hill and Smith, 2009; Oladeji et al., 2018) via internally or externally administered surveys or by reviewing data from Web-based self-assessment instruments completed by learners or collected nationally by external organizations, such as the Association of American Medical Colleges and ACGME. Doing so can be helpful as data suggest that program directors of residency programs underestimate the prevalence of burnout among their residents (Holmes et al., 2017) and accurately self-assessing one's level of distress is challenging (Shanafelt et al., 2014). Aggregated data from self-assessment instruments can provide organizations and programs with just-in-time information about the well-being of their learners and can allow for the identification of target groups or areas requiring focused attention and resources (Shanafelt and Noseworthy, 2017); self-assessment can also help learners more accurately self-calibrate their own well-being, which may promote health behavior change and help-seeking behavior before distress is severe. Effective strategies for measuring learner well-being use validated measurement tools (NAM, 2018) and ensure the protection of confidentiality, the consent of learners, transparency and honesty in reporting, and the regular evaluation and improvement of learner well-being as part of broader learning environment assessments.

Well-Being Curriculum

Medical schools and residency programs have introduced a breath of curricula to raise awareness, promote self-care, and teach positive coping skills and mindfulness-based stress reduction in an effort to help learners promote their well-being. In a national study of 27 U.S. medical schools, more than half had a well-being curriculum, and most offered a variety of emotional/spiritual, physical, financial, and social well-being activities intended to promote self-care, reduce stress, and build social support for medical students (Dyrbye et al., 2019a). Evaluation strategies consisted mostly of participation rates and student satisfaction. Resources and infrastructure varied substantially across the schools. In a 2012 survey of 212 family medicine residency directors, nearly all reported that they offered stress management lectures or workshops in addition to residency retreats and residency support groups or Balint groups (Gardiner et al., 2015). Similarly, in a study of 107 otolaryngology program directors, nearly all had had a wellness lecture within the past year, provided no-cost mental health resources with extended hours, and financially supported social events for their residents. Less than one-third held seminars in mindfulness or meditation, supported athletic or mental wellness activities, or provided healthy food options (Oladeji et al., 2018). Others have published models for curricula (Williamson et al., 2018) or described how to use a health behavior framework to help learners translate new self-care knowledge into action (Kushner et al., 2011).

Studies on the efficacy of curricula have primarily focused on mindfulness-based stress reduction and have used volunteer learners, with the majority reporting a reduction in burnout and stress and improvements in mood and empathy (de Vibe et al., 2013; Finkelstein et al., 2007; Hassed et al., 2009; Jain et al., 2007; Shapiro et al., 2000; Shiralkar et al., 2013; Warnecke et al., 2011). However, most studies did not include an appropriate control group and were vulnerable to volunteer bias. Several other studies have not found measurable improvements in learners' stress and emotional health as a result of wellness and stress management courses (Dyrbye et al., 2017b; Slavin and Chibnall, 2016) or facilitated small group discussion (Ripp et al., 2016).

A systematic review of the stress reduction and stress management literature targeting student nurses concluded that many work-site programs facilitated problem solving, self-management skills including relaxation and interpersonal skills, affective well-being, and work performance, although a number of design and evaluation inadequacies were identified (Edwards and Burnard, 2003). Additional strategies to reduce anxiety among nurses in the clinical environment have also been suggested. Studies testing the impact of curricular interventions on the rates of burnout among other health professional students are lacking (Moscaritolo, 2009).

Stronger evidence is needed to support the efficacy of wellness curricula aimed at improving learner well-being (Thomas et al., 2016). Learning organizations wishing to integrate wellness curricula should carefully consider the use of existing educational sessions so as to not further overburden learners. In most situations, a menu of offerings, rather than required experiences, is likely to be of greater benefit (Ratanawongsa et al., 2007; Ripp et al., 2017).