The prevalence of burnout, a syndrome of emotional exhaustion, depersonalization, and decreased sense of personal accomplishment, is high in residents and in practicing physicians 1 in the United States. It is associated with depression, impairment, and perceived medical errors, and is more prevalent among female clinicians. 2 While the introduction of additional duty hour restrictions may attenuate the prevalence of burnout, 3 it has not been eliminated. Recent iCOMPARE (Individualized Comparative Effectiveness of Models Optimizing Patient Safety and Resident Education) trials—sponsored by the Accreditation Council for Graduate Medical Education—that assessed outcomes with standard vs flexible duty hours found that, using moderate to high burnout scores on a standardized measure, over 70% of first-year internal medicine (IM) residents were burned out. 4 Many factors in the workplace and in residents’ personal lives contributed to this high burnout rate.
While a growing body of literature describes the effects of discrete burnout-reduction interventions on trainees and practicing clinicians, little is written about longitudinal processes for measuring resident wellness and burnout, assessing work conditions, introducing interventions, and remeasuring to track progress. In this report, we discuss an innovative strategy for measuring and responding to burnout in residents at Hennepin Healthcare.
Our strategy is based upon 2 conceptual frameworks. The first is the demand–control–support model of job stress. 5 In this model, work demands are balanced by control and social support: When control and support are lacking, a person’s stress level rises. Thus, burnout, a long-term stress reaction, may be mitigated by improved support and work control. The second framework identifies 6 core domains of work stress: control, community, workload, balance in effort and reward, fairness, and shared values. 6 The monitoring and intervention approaches we discuss were guided by these 2 overlapping frameworks.
Setting and subjects
Hennepin Healthcare is an urban safety-net hospital in Minneapolis, Minnesota, with 470 beds and a population of predominantly vulnerable patients. In this report, we focus on Hennepin’s internal medicine (IM) residency program, which has approximately 60 residents in all 3 years of training with 10 additional residents in a combined emergency medicine–internal medicine (EM–IM) program.
Since 2015, all IM and EM–IM first-year residents have rotated through a 1-month ambulatory immersion experience. Subsequent subspecialty consult blocks are interspersed with ward months. Ward and intensive care unit (ICU) teams admit patients every third or fourth day until 7 or 8 pm. Residents’ patients are located near each other geographically on the ward to facilitate daily multidisciplinary rounding with nurses, care coordinators, and social workers. Night teams arrive at 8 pm to achieve separation of day and night duties. The average work week is 60 hours for postgraduate year 1 (PGY-1) residents, 55 hours for PGY-2s, and 50 hours for PGY-3s.
Survey and sampling
Starting in 2014, we surveyed residents on whether they were experiencing feelings of fulfillment and burnout, and the possible predictors of each. Because few surveys at the time covered multiple domains of burnout and fulfillment, and even fewer had been used to longitudinally track burnout, we administered a survey developed by Mickey Trockel and colleagues at Stanford University. 7 The 62-item instrument was designed to efficiently assess well-being using both hypothesized and previously validated determinants of burnout and fulfillment. The survey is available as Supplemental Digital Appendix 1 at https://links.lww.com/ACADMED/B67. Burnout was also measured using a single-item self-assessment question from the Physician Worklife Study 8—a U.S. national survey designed to assess the job satisfaction of primary care and specialty physicians. A waiver of consent document for this work was provided by the institutional review board of the Hennepin Healthcare Research Institute.
Residents were surveyed several times during their training, usually annually; because the survey was designed to protect respondents’ anonymity, there was no way to determine how many times each resident responded. The 6-page questionnaire contained queries related to 5 domains associated with burnout, including: (1) peer support, (2) program recognition, (3) professional fulfillment, (4) negative effects on relationships, and (5) sleep-related impairment. Items that comprised each domain were assessed individually and with summary indices. Questions were typically scored using the following responses: not at all, somewhat, moderately, very, and completely. For examples of items within each domain, see Table 1.
Our approach engaged residents in making programmatic changes (which are outlined below) to improve their well-being. We viewed this relatively low-cost process as a conversation between program administrators and frontline clinicians and residents. This way, trainees were not just recipients of “burnout interventions” but rather initiators of change within their own community.
Program leaders hoped to reduce rates of burnout and improve the residency experience not only by reducing workload but also by improving the learning environment in important ways: eliminating duplication of work, streamlining tasks, promoting peer support and community, and creating schedule flexibility to help meet individual needs. We have made a number of changes over the years to achieve our goals, which we describe below.
In 2014–2015, we instituted full-time jeopardy resident coverage (where a house officer is called in to replace a resident at short notice) for essential life events, such as caring for a sick child or taking a mental health day. We accomplished this by moving residents from quality improvement and medicine consult rotations as needed. We also upgraded the resident exercise room, disseminated an electronic newsletter from the program director to honor accomplishments and life events of residents and their families, created a clinic immersion experience for first-year residents to improve their relationships with clinic staff and familiarity with patient panels, added questions related to essential life events to the annual wellness survey distributed to all physicians, and included out-of-hospital “golden weekends” for senior residents during non-ICU months. We also reduced time pressure on residents by allowing more time for patient visits during clinic and reorganized ward teams geographically (with patients clustered in one area) for better communication about patient care and patient safety between nurses, care coordinators, and residents.
In 2016–2017, we removed residents from after-hours consult pager call by slightly shifting some of their responsibilities to faculty members and we standardized weekend consult service coverage by scheduling fellows to work on Saturdays and residents to work on Sundays. During recruiting season, we gave higher priority to applicants whose personal statements in their applications showed a clear interest in caring for poor and vulnerable patients. We presented an abstract on our burnout-reduction program at the International Conference on Physician Health in 2016, gaining insights from the international community; expanded the number of venues where we shared wellness data with all residents to include annual sessions, noon conferences, chats with program chiefs, and meetings of the Program Evaluation Committee; began an annual survey asking PGY-2s which of their peers would make good chief residents; and instituted simple interventions to increase the percentage of residents who are able to publish abstracts or manuscripts during training.
We also improved residents’ research skills in 2016–2017 by: adding protected time in their schedules for them to complete human subjects training, matching residents with research mentors, and making it easier for them to access online research resources. We helped residents manage fatigue by distributing care packages to those who worked night shifts (packages contained melatonin, sleep masks, sunglasses for driving home, and educational materials about sleep impairment). We also removed end-of-day clinic appointments (4:30 pm) from residents’ schedules to improve their work–home balance.
In 2018–2020, we replaced less essential personnel on the medical emergency team with an emergency medicine faculty member to improve supervision of residents and airway support for patients; we put special lamps (created to help people suffering from seasonal affective disorder) in rooms used by residents on night teams to mimic the light of the sun, helped build residents’ camaraderie with their peers by sponsoring group events (e.g., surprise brunches for on-call teams), added resiliency training to didactics, developed a new process for residents to rapidly and confidentially obtain mental health support, and chose one staff physician to provide primary care to all residents with protected time set aside for wellness visits. We also improved resident satisfaction with feedback by holding faculty training sessions and encouraging residents to seek more feedback directly from faculty members.
Also in 2018–2020, program staff overhauled the residency website to emphasize diversity among residents and faculty and our safety-net mission. We continued to prioritize wellness by changing from every third- to every fourth-day call for several ward teams; worked to reduce workplace racism; provided residents with training on trauma-informed care; and created Resident Wellness Week, which includes instructions on how to meditate, tips on how to better make use of technological resources, presentations on how to maintain mental health, a “bring your family to work” session, a support session to help residents learn how to manage their in-basket in the electronic health record, and a group art project.
Most interventions and changes that we instituted did not require us to shift responsibilities away from residents; this helped preserve their ownership over patient care and our faculty’s time for teaching. We chose these interventions because we believe that the success of the residency program is tied to engaged residents; a supportive, community-focused culture; preservation of resident autonomy; and faculty engagement in teaching and mentoring.
Survey data were downloaded from the SurveyMonkey (SVMK, Inc., San Mateo, California) website. Responses were recorded from text to numeric values of 1 to 5. Yearly averages were calculated for each question. Significance levels were calculated by comparing data from 2015 (the second year of measurement, and the one with some of the least favorable scores) with the most recent data from 2019. All recoding and analyses were performed in SAS version 9.4 (SAS Institute, Inc., Cary, North Carolina).
Response rates ranged from 40/66 (60.6%) in 2014 to 62/73 (84.9%) in 2019 (average over 6 years = 300/416 [72.1%]). High professional fulfillment scores (classified as greater than 3.5 on a 5-point scale) were documented annually from 2014 to 2019. Self-reported rates of burnout fluctuated between 25% and 35% on the single-item measure, and we felt that annual changes in rates of burnout were the result of interventions that we had put in place as well as external events in residents’ lives, which changed from year to year.
One area of improvement in burnout was in perception of empathy, where a lower score is better: The score in 2015 was 2.43 (standard deviation [SD] 1.02, 95% confidence interval [CI] 2.09–2.77); the score in 2019 was 1.84 (SD 0.86, 95% CI 1.60–2.08, P < .005). Seven of 8 sleep-impairment items and 3 of 4 peer-support items showed statistically significant favorable changes from 2015 to 2019 (P < .05). For example, one sleep item—“Did you have problems during the day due to poor sleep?”—improved from a score of 2.49 in 2015 (SD 1.35, CI 2.04–2.94) to 1.71 in 2019 (SD 0.90, CI 1.45–1.96, P < .002). When residents were asked, “Did the program make it harder to nurture personal relationships?” scores improved (dropped) substantially from 3.32 in 2015 (SD 1.25, CI 2.91–3.74) to 2.39 in 2019 (SD 1.17, CI 2.06–2.72, P < .001). Findings were remarkably consistent between PGY-1s, PGY-2s, and PGY-3s. Other outcomes included continued strong recruitment to the program and low rates of adverse events (i.e., daily absences or withdrawal from the program due to impairment).
In this 6-year program of continuous improvement to reduce burnout among IM residents, we found that a multifaceted program, informed by resident input, is feasible to implement at relatively low costs and may improve both the learning and work environments. These findings are in line with a 2016 editorial in which former Academic Medicine Editor-in-Chief David Sklar wrote about fostering trainee wellness to produce healthy doctors and a healthy population. 9 In the article, he asked institutions to improve their support networks for trainees and optimize learning environments. His specific recommendations included providing time off to promote health, enhancing trust between trainees and faculty, providing trainees with time to reflect after adverse events, and emphasizing education over service. He advocated for measuring one’s progress and committing resources to improvement. Our current program and next steps are modeled after many of these suggestions.
Guided by our conceptual frameworks, we developed a focused plan for minimizing burnout among our residents (see Table 2) using evidence-based domains relating to work stress. The table includes proposed methods of addressing each domain. One new item we will focus on is preventing moral injury, 10 which can result when physicians’ moral dilemmas and/or excessive workloads are not addressed. We believe moral injury can be attenuated by addressing moral distress through programs designed to help residents deal with stressful training issues. We intend to create a new set of metrics for this important approach to burnout prevention.
We have also identified a general internist to provide health care to residents, with same-day appointments available to attend to their urgent needs, and a psychiatrist to treat residents who are in distress. We have increased protected time for didactics; initiated resiliency, nutrition, and exercise training; and implemented strategies to acknowledge and address the negative effects of societal and institutional racism on patients and trainees. We seek to build a culture of well-being among residents and faculty, improve residents’ physical and mental health, and foster an accountable community capable of helping all of its members succeed. We hope that our graduates are resilient doctors who can care for complex patients, help their peers, and find workable solutions in challenging times.
The strengths of our evaluation include high response rates, the use of a standardized instrument, and the collection of several years of data for us to evaluate; weaknesses include the fact that the study only surveyed residents from one program and the fact that we were unable to distinguish the most effective interventions for minimizing burnout.
A key part of our strategy is a public health approach, where we consider rates of burnout and resident well-being to be indicators of residency program quality. If 75% of residents answered the survey and 60% of them indicated that they were burned out, we would know something is wrong; if 15% were burned out in later surveys when 30% were burned out before, we would know we are making positive strides in reducing resident burnout.
While it is impossible to remove all stressors from medicine, the ways in which residency program leaders anticipate and respond to them can mean the difference between a healthy environment and one that is perceived as toxic. Our commitment to a healthy workplace—and thus healthy residents and faculty—will drive our next steps.
The authors would like to thank Renee Marrero for her expert manuscript preparation, the internal medicine residents at Hennepin Healthcare for their participation in the survey, and Dr. Mickey Trockel and his colleagues at Stanford University, whose invaluable contributions included developing the survey measures used here. The authors would also like to acknowledge the anonymous reviewers for their superb assistance in improving an earlier version of this report.
1. Shanafelt TD, Hasan O, Dyrbye LN, et al. Changes in burnout and satisfaction with work-life balance in physicians and the general US working population between 2011 and 2014. Mayo Clin Proc. 2015;90:1600–1613.
2. Dyrbye LN, West CP, Satele D, et al. Burnout among U.S. medical students, residents, and early career physicians relative to the general U.S. population. Acad Med. 2014;89:443–451.
3. Busireddy KR, Miller JA, Ellison K, Ren V, Qayyum R, Panda M. Efficacy of interventions to reduce resident physician burnout: A systematic review. J Grad Med Educ. 2017;9:294–301.
4. Desai SV, Asch DA, Bellini LM, et al.; iCOMPARE Research GroupEducation outcomes in a duty-hour flexibility trial in internal medicine. N Engl J Med. 2018;378:1494–1508.
5. Johnson JV, Hall EM. Job strain, work place social support, and cardiovascular disease: A cross-sectional study of a random sample of the Swedish working population. Am J Public Health. 1988;78:1336–1342.
6. Jennings ML, Slavin SJ. Resident wellness matters: Optimizing resident education and wellness through the learning environment. Acad Med. 2015;90:1246–1250.
7. Trockel M, Bohman B, Lesure E, et al. A brief instrument to assess both burnout and professional fulfillment in physicians: Reliability and validity, including correlation with self-reported medical errors, in a sample of resident and practicing physicians. Acad Psychiatry. 2018;42:11–24.
8. Linzer M, Visser MR, Oort FJ, Smets EM, McMurray JE, de Haes HC; Society of General Internal Medicine (SGIM) Career Satisfaction Study Group (CSSG)Predicting and preventing physician burnout: Results from the United States and the Netherlands. Am J Med. 2001;111:170–175.
9. Sklar DP. Fostering student, resident, and faculty wellness to produce healthy doctors and a healthy population. Acad Med. 2016;91:1185–1188.
10. Poorman E. Depression and suicide: Occupational hazards of practicing medicine. J Patient Saf Risk Manag. 2019;24:181–183.