Burnout and Emotional Well-Being Among Nephrology Fellows: a National Online Survey

SPECIAL ARTICLE www.jasn.org

Burnout and Emotional Well-Being among Nephrology Fellows: A National Online Survey

Varun Agrawal,1 Laura Plantinga ,2 Khaled Abdel-Kader,3 Kurtis Pivert ,4 Anthony Provenzano,5 Sandeep Soman,6 Michael J. Choi,7 and Bernard G. Jaar 8,9,10

Due to the number of contributing authors, the afﬁliations are listed at the end of this article.

ABSTRACT Background Physician burnout and emotional distress are associated with work response to stressors at work.2 Burnout dissatisfaction and provision of suboptimal patient care. Little is known about burn- among attending physicians in all spe- out among nephrology fellows. cialties combined is estimated at 44%.1 Methods Validated items on burnout, depressive symptoms, and well being were Burnout is also common among medical included in the American Society of Nephrology annual survey emailed to US trainees (medical students, residents, and nephrology fellows in May to June 2018. Burnout was defined as an affirmative fellows) and is concerning because of its response to two single-item questions of experiencing emotional exhaustion or association with self-reported medical depersonalization. errors, suboptimal patient care, and sui- cidal ideation.4 Results Responses from 347 of 808 eligible first- and second-year adult nephrology A 2017 survey of 15,543 United States fellows were examined (response rate542.9%). Most fellows were aged 30–34 years physicians reported a burnout preva- (56.8%), male (62.0%), married or partnered (72.6%), international medical gradu- lence of 40% among nephrologists ates (62.5%), and pursuing a clinical nephrology fellowship (87.0%). Emotional (range 23% [plastic surgeons] to 48% exhaustion and depersonalization were reported by 28.0% and 14.4% of the fellows, [intensivists]).5 However, information respectively, with an overall burnout prevalence of 30.0%. Most fellows indicated about burnout in nephrology fellows re- having strong program leadership (75.2%), positive work-life balance (69.2%), pres- mains a critical knowledge gap. This is ence of social support (89.3%), and career satisfaction (73.2%); 44.7% reported a especially relevant because the high pro- disruptive work environment and 35.4% reported depressive symptoms. Multivari- portion of unfilled positions in nephrol- able logistic regression revealed a statistically significant association between fe- ogyfellowshipprogramsislikelyto male gender (odds ratio [OR], 1.90; 95% confidence interval [95% CI], 1.09 to 3.32), compromise the future workforce.6 poor work-life balance (OR, 3.97; 95% CI, 2.22 to 7.07), or a disruptive work envi- Medical students, residents, and non- ronment (OR, 2.63; 95% CI, 1.48 to 4.66) and burnout. nephrology fellows perceive nephrology Conclusions About one third of US nephrology fellows surveyed reported experi- practice to be based on complex patho- encing burnout and depressive symptoms. Further exploration of burnout— physiology concepts, comprising man- especially that reported by female physicians, as well as burnout associated with agement of medically complex patients poor work-life balance or a disruptive work environment—is warranted to develop with chronic illnesses and a heavy work- targeted efforts that may enhance the educational experience and emotional well load that may cause distress at work.7 being of nephrology fellows. Nephrology fellows constituted the

JASN 31: 675–685, 2020. doi: https://doi.org/10.1681/ASN.2019070715 greatest proportion (22.3%) of the 121 fellows in 20 internal medicine subspecialties who left before completing their

Burnout is a state of mental distress char- care) levels.2 Physician burnout is consid- acterized by the presence of emotional ex- ered a public health crisis and a threat to Published online ahead of print. Publication date haustion, depersonalization, and sense of future medical practice.3 Causes of burn- available at www.jasn.org. low accomplishment at work.1 Burnout out include a host of factors at work Correspondence: Dr. Varun Agrawal, Division of among physicians is associated with ad- (workload, regulatory pressures, poor Nephrology and Hypertension, Larner College of Medicine, University of Vermont, 1 South Prospect verse effects at both personal (relationship work-life balance, lack of social support, Street, UHC 2311, Burlington, VT 05401. Email: difﬁculties, substance abuse, depression) loss of control) and at the individual level [email protected] and professional (decreased productivity, (compulsivity, poor personal coping Copyright © 2020 by the American Society of work dissatisfaction, suboptimal patient mechanisms) that may exacerbate the Nephrology

JASN 31: 675–685, 2020 ISSN : 1046-6673/3104-675 675 SPECIAL ARTICLE www.jasn.org fellowship training in 2016–2017.8 due to a lack of a centralized accredita- externally and we edited the survey to Hence, we performed this exploratory tion organization.8 Pediatric nephrology enhance readability and ensure capture study to investigate the prevalence of fellows were excluded because their prac- of key variables of interest. Content val- burnout and other emotional-distress tice (although renal specific) is not a di- idity was established by seven nephrolo- measures using standardized, validated rect analogue of adult nephrology, with a gists in the National Kidney Foundation questions in a cross-sectional survey. smaller fellow pool (101 pediatric versus (NKF) Education Committee. Approval We sought to identify personal and 808 adult nephrology fellows), different for the study was obtained from the work-related risk factors associated training protocols (three accredited years Committee on Human Research in the withburnoutinnephrologyfellows,a of training with a dedicated research Medical Sciences, Institutional Review high-risk group. year), differing etiologies and complexity Board, University of Vermont, Burling- of renal disorders, and higher attrition in ton, Vermont (CHRMS 17-0516) under pediatric nephrology fellows.8,11 Emails the exempt category. METHODS with a unique link to the online survey were sent to each potential respondent by Study Variables Study Design and Population ASN in collaboration with the George Burnout Thetargetpopulationinthiscross- Washington Health Workforce Institute Questions from previously validated sectional study was all first- and second- (George Washington University Institu- survey instruments were selected to mea- year fellows in adult nephrology training tional Review Board #051430; principal sure different aspects of emotional programs in the United States. Our 11- investigator Edward Salsberg). The con- health. Two single-item measures of item study questionnaire on burnout and sent page made no mention of “burnout” burnout adapted from the 22-item Mas- well being was approved by the American or “distress.” Participation was voluntary, lach burnout inventory (MBI), the refer- Society of Nephrology (ASN) Workforce responses were anonymous, and partici- ence standard for measuring burnout, and Training Committee for inclusion in pants were eligible for incentives (ASN were used.13 The two-item burnout the annual ASN fellow survey that sought educational programs), with winners questions have similar efficacy as the to capture information on demograph- chosen randomly from participants at full-length MBI and have been effectively ics, educational perceptions, job market the survey’s conclusion. used in physician surveys due to its brev- experiences, and future employment After informed consent was provided, ity.4,14 These items test how frequently characteristics for all trainees.9 The sur- the respondent could start answering the the respondent perceived emotional ex- vey had a total of 83 questions (inclusive questions, and responses were collected haustion (“I feel burned out from my of our questions) and was distributed by Research Electronic Data Capture work”) and depersonalization (“I’ve be- electronically to all nephrology fellows (REDCap).12 Confidentiality was en- come more callous toward people since I in training who received complimentary sured by separating identifying informa- started this job”) on a seven-point Likert ASN membership (n51329). Because tion (emails, internet protocol addresses) scale ranging from “never” to “every day.” ASN extends fellow membership to all from the rest of the data in a secure server Choosing “once a week” or more fre- current trainees, the survey audience at the completion of each survey. Fre- quently to either item was considered a comprised all or nearly all adult nephrol- quent email reminders (approximately positive response and indicated burn- ogy, pediatric nephrology, research, weekly) to the respondents and two re- out.14 A total of 650 online licenses for transplant, nephrology critical care, in- minders to the fellowship program direc- use of these proprietary questions were terventional nephrology, and postdoc- tors were sent. Fellows could complete purchased from Mindgarden.com with toral fellows. All respondents outside the survey over multiple sessions but research funds provided by the Division the target population were excluded as once it was finalized and submitted they of Nephrology and Hypertension, Uni- follows. Responses from third-year fel- could not make further changes. The sur- versity of Vermont. lows (whose third year of fellowship is vey opened on May 1, 2018 and closed on almost always dedicated to research) June 11, 2018. In addition to third-year Depressive Symptoms and Other Well were censored because their clinical adult and pediatric nephrology fellows, Being Measures load is minimal (comprising coverage participants who did not respond to any Questions on depressive symptoms, for other fellows or a few night or week- question on the survey and whose burn- positive influence at work, quality of end calls) and not representative of ser- out status (primary outcome of interest) program leadership, work-life balance, vice demands experienced during the could not be clearly determined were ex- quality of life, and career satisfac- 2 years of accredited nephrology fellow- cluded. A pilot study for system valida- tion were included in the survey ship training.10 Furthermore, there are tion was performed on two nephrology (Supplemental Table 1).4,15–18 All the sparse data on fellows pursuing addi- fellows in May 2017 at the University of questions were multiple-choice ques- tional training beyond the 2 years ac- Vermont and these data were excluded tions and one best response had to credited by Accreditation Council for from subsequent analysis. The question- be chosen. Responses were on a Likert Graduate Medical Education (ACGME) naire was reviewed both internally and scale and addressed level of agreement

676 JASN JASN 31: 675–685, 2020 www.jasn.org SPECIAL ARTICLE

Table 1. Characteristics of responding Table 1. Continued for the In-Training Exam—taken by all first- and second-year nephrology adult nephrology fellows (.98%offellows n5 fellows in the United States ( 347) Study Variable n (%) in 2014) and administered by ASN—we Study Variable n (%) Number of patients seen in a believe ASN’s 2018 fellow survey was Age weekday directed to almost every first- and , #29 yr 23 (6.6) 10 17 (4.9) second-year adult nephrology fellow.21 – 30–34 yr 197 (56.8) 11 15 65 (18.7) – Demographic data for the nonre- 35–39 yr 78 (22.5) 16 20 110 (31.7) – sponders were derived by subtracting $40 yr 45 (13.0) 21 25 83 (23.9) – the categorical data of the study sample Missing 4 (1.2) 26 30 35 (10.1) . from the ACGME data set and com- Gender 30 35 (10.1) pared with the responders by chi- Male 215 (62.0) Missing 2 (0.6) Female 131 (37.8) Census region squared test of independence (ethnicity Missing 1 (0.3) Midwest 68 (19.6) was not compared due to numerous Current fellowship type Northeast 94 (27.1) missing responses in the ACGME Data Clinical nephrology 302 (87.0) South 107 (30.8) Resource Book). The mean age of first- Research nephrology 38 (11.0) West 44 (12.7) year fellows who did not respond to the Transplant nephrology 1 (0.3) Missing 34 (9.8) survey was imputed from similar avail- a Interventional nephrology 1 (0.3) Number of responses for race does not add to able data from the ACGME data set and Other 3 (0.9) 347 because multiple options could be chosen. bFive fellows attended medical school in study sample and then compared with Missing 2 (0.6) Canada. the responding fellows by one-sample Fellowship year t test. Chi-squared and unpaired t tests First-year fellow 158 (45.5) (“definitely yes” to “definitely no,” or were used to compare categorical data Second-year fellow 174 (50.1) “strongly agree” to “strongly disagree”) Missing 15 (4.3) and continuous variables, respectively. Marital status or frequency (every day to never). Per- Missing responses were reported as Married 232 (66.9) missions were obtained from the authors such. Logistic regression analysis with Partnered 20 (5.8) of the original questions for adoption in adjustment for all available covariates Single 90 (25.9) this study. Questions on demographics (mutually adjusted multivariable Divorced 4 (1.2) and information on the respondent’s fel- model) was performed to determine de- Other 1 (0.3) lowship program were also included in mographic and emotional health factors Missing 0 the analysis. After an iterative process of associated with burnout. Cases that had a Race internal review, one question each on missing variables in the model were American Indian or 3 disruptive behavior and social support dropped by the statistical program Alaska Native were considered pertinent to the survey when performing the analysis, leaving Asian or Pacific Islander 142 and included in the final draft before sur- Black 24 only complete cases and no imputation White 125 vey dissemination (Supplemental was performed. Apriori, the following 19,20 Other 55 Table 1). 10variableswerechosenforinclusion Missing 0 in the multivariable model: age, gender, Ethnicity Statistical Analyses fellowship year, international medical Hispanic or Latino 34 (9.8) Data are presented as descriptive statis- graduate (IMG) versus United States Not Hispanic or Latino 313 (90.2) tics. Proportions were calculated with medical graduate (USMG), workload, Missing 0 the denominator being the total number relationship status, strong program Location of medical school of respondents included in the analysis. leadership, social support, disruptive United States 126 (36.3) Response rate was calculated as the pro- environment, and work-life balance, be- Other countries and 217 (62.5) b portion of eligible adult nephrology cause we predicted 100 burnout events Canada fi Missing 4 (1.2) rst- and second-year fellows in the (based on 25% response rate and 50% 22,23 Educational debt United States who provided responses burnout prevalence). Due to nota- fi None 176 (50.7) in this survey to conclusively determine ble ndings on depressive symptoms, $1–$200,000 100 (28.8) burnout status. ASN’s membership da- additional regression analysis using the .$200,000 67 (19.3) tabase does not capture fellowship year, same 10 variables was performed to Missing 4 (1.2) and the number of eligible United States identify predictors of depressive symp- Nephrology as first choice adult first- and second-year nephrology toms. We did not include depressive Yes 126 (36.3) fellows were obtained from the ACGME symptoms in the regression model for No (other specialties or 219 (63.1) Data Resource Book for the academic burnout (and vice versa)becausea general practice) year 2017–2018 (n5808).8 Because strong correlation between these two Missing 2 (0.6) ASN membership is required to register constructs is known and overlapping

JASN 31: 675–685, 2020 Burnout in Nephrology Fellows 677 SPECIAL ARTICLE www.jasn.org

Figure 1. Burnout symptoms reported by nephrology fellows were more frequent in the female as compared to the male gender. “Once a week” or more frequently to either item indicates burnout. features (such as exhaustion and feeling 54, who were censored. The final study applied to other specialties or general down) make it very difficult to differen- sample comprised 347 of the 808 eligible practice before choosing to pursue ne- tiate between cause and effect.24,25 A P adult nephrology first- and second-year phrology (63.1%) and three quarters of value ,0.05 was considered statistically fellows in the United States, yielding a the fellows saw .15 patients on average significant. Statistical analysis was per- response rate of 42.9%. in a weekday (75.8%). No statistically formed with SPSS, Inc., version 25.0 More than half of the fellows were significant difference was noted between (Chicago, IL). 30–34 years old (56.8%) and male our study sample and the nonresponders (62.0%) (Table 1). Most of the fellows in distribution of fellowship year or gen- were in clinical nephrology fellowship der, although the proportion of IMGs RESULTS (87.0%) and the postgraduate years versus USMGs and age of first-year fel- were about equally distributed (45.5% lows was higher in the nonresponders The survey was distributed to 1329 ne- first-year and 50.1% second-year fel- when compared with the responding fel- phrology fellows in the United States lows). More than half of the respondents lows (Supplemental Table 2). with ASN fellow membership, with 494 identified themselves as IMGs (62.5%). fellows consenting to participate. Partic- Respondents were commonly training in Burnout and Emotional Well Being ipants with no responses (n512), pedi- the southern (30.8%) and northeastern On the two-item burnout question atric nephrology fellows (n541), and (27.1%) United States. Two thirds of the items, emotional exhaustion more than those who identified themselves in their fellows were in a relationship (married once weekly and depersonalization more third year of fellowship (n551) were ex- or partnered, 72.6%), and nearly half of than once weekly were reported by cluded. Among the remaining fellows, the respondents had no educational debt 28.0% and 14.4% of the fellows, respec- burnout could not be determined in (50.7%). Two thirds of the fellows had tively (Figure 1, Table 2). This yielded a

678 JASN JASN 31: 675–685, 2020 www.jasn.org SPECIAL ARTICLE

Table 2. Burnout and emotional well being among surveyed first- and second-year burnout prevalence of 30.0% among the nephrology fellows in the United States (n5347) respondents. On evaluating for depres- Study Variable n (%) sive symptoms, depressed mood and an- fi Burnout—emotional exhaustion hedonia were identi ed in 31.1% and Every day 18 (5.2) 23.9% of the fellows, respectively; A few times a week 47 (13.5) 35.4% of the respondents were positive Once a week 32 (9.2) for depressive symptoms. A few times a month 90 (25.9) About half of the fellows agreed that Once a month 46 (13.3) their work had a positive effect on their A few times a year or less 97 (28.0) patients (54.8%). Three quarters of fel- Never 17 (4.9) lows identified strong leadership in their Missing 0 fellowship program (75.2%). Satisfac- Burnout—depersonalization tory work-life balance was noted by Every day 11 (3.2) A few times a week 23 (6.6) 69.2%, adequate social support by Once a week 16 (4.6) 89.3%, and good quality of life by 86.2% A few times a month 49 (14.1) of the respondents. Career satisfaction Once a month 35 (10.1) and disruptive environment at work was A few times a year or less 115 (33.1) reported by 73.2% and 44.7% of the fel- Never 96 (27.7) lows, respectively. Missing 2(0.6) — Burnout present/absent Characteristics Associated with Burnout present 104 (30.0) Burnout and Depressive Symptoms Burnout absent 243 (70.0) Demographic and emotional well being Missing 0 factors stratified by presence of burnout Depressed mood (feeling down) Yes 108 (31.1) are presented in Supplemental Table 3. No 238 (68.6) Fellows with burnout were more likely Missing 1(0.3) than those without burnout to be fe- Anhedonia (little interest) male, see .15 patients per weekday, Yes 83 (23.9) have depressive symptoms, and report No 263 (75.8) a disruptive environment at work. Fel- Missing 1(0.3) lows negative for burnout were statisti- Positive for depressive symptoms 123 (35.4) cally significantly more likely than those Negative for depressive symptoms 223 (64.3) positive for burnout to identify a strong Missing 1(0.3) program leadership, appropriate work- Positive influence life balance, adequate social support, Every day 61 (17.6) A few times a week 90 (25.9) good quality of life, and presence of ca- Once a week 39 (11.2) reer satisfaction. Multivariable logistic A few times a month 78 (22.5) regression analysis revealed poor work- Once a month 33 (9.5) life balance (odds ratio [OR], 3.97; 95% A few times a year or less 38 (11.0) confidence interval [CI], 2.22 to 7.07; Never 8(2.3) P,0.001), presence of a disruptive envi- Missing 0 ronment (OR, 2.63; 95% CI, 1.48 to Program leadership 4.66; P50.001), and female gender Strongly agree 125 (36.0) (OR, 1.90; 95% CI, 1.09 to 3.32; Agree 136 (39.2) P50.024) to be statistically significantly Neither agree nor disagree 52 (15.0) associated with burnout (Figure 2, Disagree 22 (6.3) Strongly disagree 12 (3.5) Supplemental Table 4). Because there Missing 0 was a sizable number of fellows with de- Work-life balance pressive symptoms, a multivariable logis- Very satisfied 55 (15.9) tic regression analysis was also performed Somewhat satisfied 103 (29.7) to identify factors associated with depres- Neutral 82 (23.6) sive symptoms (Supplemental Table 5). Somewhat dissatisfied 71 (20.5) Presence of social support (OR, 0.08; Very dissatisfied 35 (10.1) 95% CI, 0.03 to 0.24; P,0.001), adequate Missing 1(0.3) work-life balance (OR, 0.17; 95% CI, 0.10

JASN 31: 675–685, 2020 Burnout in Nephrology Fellows 679 SPECIAL ARTICLE www.jasn.org

Table 2. Continued design precludes any conclusion on how burnout among nephrology fellows is Study Variable n (%) different when compared with other in- Social support ternal medicine subspecialties, our find- All of the time 138 (39.8) ings do warrant further studies into this Most of the time 119 (34.3) concerning burden affecting physicians Some of the time 53 (15.3) in training. A little of the time 29 (8.4) We sought to speculate on why the None of the time 7 (2.0) Missing 1(0.3) burnout prevalence rate in our study Overall quality of life was lower than expected by looking at fi As good as it can be 92 (26.5) the study methodology and ndings. Somewhat good 141 (40.6) The survey was distributed toward the Neutral 66 (19.0) end of the academic year when clinical Somewhat bad 37 (10.7) workload may be more efficiently han- As bad as it can be 9 (2.6) dled, as compared with the start of the Missing 2(0.6) year, due to familiarity with the health Career satisfaction system processes and enhanced medical fi De nitely yes 114 (32.9) knowledge. Improved seasonal affective Probably yes 140 (40.3) symptoms or a secured employment af- Not sure 44 (12.7) Probably no 31 (8.9) ter graduation may have also affected Definitely no 16 (4.6) our study results. Nephrology fellowship Missing 2(0.6) programs are unique in having a higher Disruptive behavior proportion of IMGs.8 We did not find Daily 5(1.4) IMGs to have greater burnout than Weekly 13 (3.7) USMGs, possibly due to low educational 1–2 times per month 36 (10.4) debts from their home countries and re- 1–5 times per year 101 (29.1) silience from navigating the competitive Never 191 (55.0) application process for training and Missing 1(0.3) work-related acculturation, although cultural differences and visa restrictions to securing a job may add to emotional to 0.31; P,0.001), and strong program among nephrology fellows to be similar distress.4 Many fellows were in a rela- leadership (OR, 0.46; 95% CI, 0.24 to or even worse than that among medicine tionship and indicated having good 0.88; P50.018) were statistically signifi- residents (45.2%–60.3% prevalence) social support, a known deterrent to cantly associated with depressive when measured in a questionnaire sim- burnout.20,22 About two thirds of the re- symptoms. ilar to ours.4,27 In this first study of sponding fellows had applied to another burnout in a national sample of ne- specialty besides nephrology, a finding in phrology fellows in the United States, line with a 2015 report where 59% of the DISCUSSION the prevalence of burnout (30.0%) nephrology fellows applied simulta- was, importantly, much lower in com- neously to another specialty.29 In our The need to recognize and address phy- parison with that in medicine residents. study, we did not find a difference in sician burnout to improve the joy of ne- Asimilarsurveyofhematology- burnout prevalence between fellows phrology practice has been recently oncology fellows in 2013 revealed a who applied only to nephrology or a emphasized.26 Clinical workload, ad- burnout prevalence of 34.1%, which is second specialty, suggesting a complex ministrative burden in dialysis care, closer but still higher than our study re- interplay between work-related and in- and electronic medical records (EMRs) sults.28 The multi-institutional random dividual factors in causing burnout. In- at multiple sites were speculated to be sample of nephrology fellows with terestingly, we also found social support some of the drivers of burnout in ne- almost equal distribution between first- to be negatively associated with depres- phrology attendings.26 While going and second-year fellows and representa- sive symptoms. Promoting a culture of through the rigors of medical training, tiveness of the United States nephrology support from faculty, cofellows, staff, nephrology fellows provide care to pa- fellow pool improves the generalizabil- and family during physician training is tients with medically complex issues and ity of our findings.8 We designed the sur- likely to help decrease burnout and en- commonly work long hours in a consul- vey from validated instruments in the hance resilience.22 Strong program lead- tant role to the primary medical team.6 research literature, thus ensuring valid- ership reported by many fellows in our We expected the burnout prevalence ity to a large extent. Although our study survey may have lowered the burnout

680 JASN JASN 31: 675–685, 2020 www.jasn.org SPECIAL ARTICLE

Figure 2. Forest plot of personal and work-related factors showing poor work-life balance, female gender and presence of disruptive environment to be significantly associated with burnout in United States adult nephrology fellows. prevalence because an effective leader counterparts.31–34 Low satisfaction with 71% of physicians in a 2011 survey com- who can engage and inspire physicians work-life balance was associated with monly in the surgical, emergency room, has a negative effect on burnout.30 burnout in our survey and other studies and intensive-care-unit settings.41,42 Be- Nephrology program directors form a in residents and attendings.4,34,35 This is cause nephrologists commonly collabo- cohesive group, striving to improve the especially relevant to the field of ne- rate with physicians in multiple medical educational experiences of the fellows, phrology because having an adequate and surgical settings in a consultant role, it streamline the fellowship application lifestyle and work-life balance is impor- is possible that nephrology fellows are ex- and matching process, and enhancing tant in influencing career decisions.36 posed to disruptive behaviors. Our survey interest in nephrology in close collabo- Similar to our findings, the 2017 ASN could not delineate if the disruptive be- ration with the ASN Workforce and Nephrology Fellow Survey reported haviors originated from patients or other Training Committee.6 Furthermore, a 17%ofsecond-yearfellowshadpoor health care providers, or whether it was national survey of United States physi- or very poor work-life balance.9 We physical or psychologically traumatic. cians reported that burnout prevalence could not identify the reason behind The high proportion of IMGs in nephrol- in 2017 may actually be decreasing when the perception of poor work-life balance. ogy fellowship programs may possibly ex- compared with 2011, although work sat- Large case workload, as expected, was pose them to racist behaviors by patients, isfaction continues to be low.1 associated with burnout in bivariate thus contributing to workplace disrup- We identified factors associated with analysis, possibly due to the burden of tion.43 The psychologic stress of experi- burnout in nephrology fellows, although patient interaction, long work hours, encing or witnessing the disruptive causality could not be established due to and increased documentation require- behavior and the fear of a future similar the cross-sectional design of our study. ments.37 The effect of work-related fac- event may lead to burnout.44 We found female nephrology fellows tors such as work hours, workload, pace, Concerning findings in our survey were more likely to experience burnout working conditions, and EMR docu- were that 35.4% of fellows reported de- than their male colleagues, an observa- mentation—especially at home (“pa- pressive symptoms and about a quarter tion also reported among residents and jama time”)—on burnout need to be of the nephrology fellows reported ca- attendings. Our finding may possibly be studied.6,38–40 Presence of disruptive be- reer dissatisfaction. We used the two- explained by under-reporting by males, havior at work was strongly associated item Primary Care Evaluation of Mental perceived lack of control over work with burnout in our survey. Disruptive Disorders questionnaire that inquires schedule, patients having different ex- behavior—actions that violate the per- about the presence or absence of de- pectations from female versus male phy- petrator’s standard of respectful behav- pressedmoodandanhedoniainthe sicians, or more work-home conflicts by ior and results in a perceived threat to the past month. A positive (“yes”)answer female as compared with their male victim (or witness)—was witnessed by to either of these questions constitutes

JASN 31: 675–685, 2020 Burnout in Nephrology Fellows 681 SPECIAL ARTICLE www.jasn.org

Table 3. Potential strategies to address burnout and depressive symptoms in nephrology training programs Potential Strategy Support from program and organizational leadership: Program directors need to be cognizant of the national and local trends and risk factors for burnout and depressive symptoms among physicians in training. Programs should emphasize the importance of personal and professional well being at orientation and offer support throughout training. Prioritizing work-life balance may be achieved by review of work hours, patient load, documentation needs, EMR responsibilities, work environment, call rooms, inpatient and outpatient rotations, and collaborating with fellows in creating work schedules. Other strategies to enhance work-life integration such as mentoring programs, leadership training, promoting physical health and diet, and offering personal self-care tools to manage work and stress (such as time management, limiting EMR work at home) may be needed in fellowship programs. Commitment from national organizations such as ASN, NKF, and the American College of Physicians to address physician burnout and emotional distress is highly encouraging. Assessment: Periodic assessment of burnout through the MBI or other validated surveys such as the Mini-Z in collaboration with the institution’s graduate medical education office can help identify and track longitudinal trends in well being, while identifying solutions at the personal level, workplace, or training environment based on available resources. Academic curriculum: Fellowship curriculum should incorporate sessions on emotional well being, financial planning, clinical expectations and practice, and responsibilities as medical director to allow a smooth transition into the attending role where burnout may be higher. Training in palliative care and end-of-life discussions in nephrology may reduce burnout and fellows by empowering them with essential communication skills. Addressing clinical workload: Interventions—such as having nurse practitioners, nephrology hospitalists, change in overnight call—need to be studied as each strategy may produce different results in a program. Access to mental health support: Prompt access to confidential resources such as counselor or psychologist support while fostering a culture of social support and destigmatization is essential to prevent, identify, and treat depressive symptoms early, when present. Addressing disruptive work environment: Disruptive work environment needs to be addressed through program directors who serve as a liaison between fellows and hospital administration. Hospital should adopt zero tolerance for any kind of violence in the workplace and promote confidential reporting in addressing workplace issues, while providing physicians with tools for coping and enhancing resilience. a positive screen for depression (i.e.,pos- exposure to severe burnout stressors at during fellowship.38 Physicians dissatisfied itive for depressive symptoms) and has a work may lead to depression.48 Career with their career or specialization have in- sensitivity of 97% and specificity of 57%, dissatisfaction is also of concern because dicated the desire to leave medical prac- respectively, for diagnosing major de- it is associated with physician burnout.34 tice, which highlights the need to look pression.16 Our finding is in line with In a survey of .6500 physicians in the into career dissatisfaction to maintain a 2015 meta-analysis data that showed United States, nephrology ranked 39 nephrology workforce for the future.34 prevalence of depressive symptoms to (out of 42) in career satisfaction, which Our study has important implica- be 28.8% (95% CI, 25.3% to 32.5%) correlated with long work hours.49 A tions for nephrology fellowship program among resident physicians and sur- 2011 survey by Shah et al.38 found that directors and teaching faculty members geons.45 A positive screening result for approximately 16% of the 204 respond- who make concerted efforts to ensure depression is of major concern because ing nephrology fellows were slightly or fellow well being.28,50–53 Fellowship pro- this requires evaluation with the Patient notatallsatisfied with their career grams need to implement steps to min- Health Questionnaire-9 or direct psychi- choice, especially IMGs or those who in- imize burnout, identify trainees at high atric interview to diagnose major dicated nephrology not to be their first risk of burnout, and provide fellow sup- depression.46 Prompt psychiatric in- career choice. Fellows who identified pos- port to achieve work satisfaction and volvement is needed in an attempt to itively for burnout in our study were more high-quality learning experiences prevent self-harm, especially as medical likely to have career dissatisfaction on bi- throughout training (Table 3). Our trainees do not necessarily seek help in variate analysis. Career dissatisfaction study has limitations because we are un- depression due to guilt or shame in pur- among nephrology fellows was reported sure how many fellows received the sur- suing mental health services.47 Some ex- to be due to poor income potential after vey due to email addresses being inactive perts propose that burnout is a depressive fellowship, poor job opportunities, long or behind institutional firewalls. Survey symptom and that prolonged period of work hours, and overall poor experiences fatigue may have affected our survey

682 JASN JASN 31: 675–685, 2020 www.jasn.org SPECIAL ARTICLE results as suggested by the missing val- reasonable workload, constructive and for purchase of the questionnaire instrument. ues.54 However, the response rate in our supportive program leadership, and ad- Travel support to Dr. Agrawal was provided by study was similar to the 2017 ASN sur- equate social support) to enhance the the Teaching Academy at the Larner College of Medicine, University of Vermont. vey.9 Although we have shown that our educational experience and emotional study sample is reasonably well repre- well being of nephrology fellows. sentative of the nephrology fellow pool, SUPPLEMENTAL MATERIAL we do not have the complete data on the nonresponders and we could not fully ACKNOWLEDGMENTS This article contains the following sup- account for this bias. Although we can- plemental material online at http://jasn. not exclude the possibility that nonre- The authors thank the surveyed physicians for asnjournals.org/lookup/suppl/doi:10.1681/ sponders may have more burnout, the their participation. They also thank Alan Ho- ASN.2019070715/-/DCSupplemental. participants were not aware that the ward (University of Vermont) for biostatistical Supplemental Table 1. Questions on de- ASN survey included questions on burn- support,Dr.RobertW.Rope(OregonHealth pressive symptoms and emotional well-being out and well being because this was not & Science University), the ASN Workforce and in the survey. mentioned in the cover letter or consent Training Committee (chaired by Dr. Scott Supplemental Table 2. Comparison of form. Our study was not designed to J. Gilbert, Tufts Medical Center), Edward Sals- demographic characteristics of responders identify the drivers of burnout and we berg, Master of Public Administration, the NKF and non-responders. did not evaluate whether fellows’ sched- Education Committee (chaired at the time of Supplemental Table 3. Parameters strati- ules, night calls, or EMR use were asso- the study by B.G.J., Johns Hopkins University), fied by burnout status. ciated with burnout. Weused the validated Dr.ShadenT.Eldakar-Hein(UniversityofVer- Supplemental Table 4. Association of two-item burnout from the MBI to iden- mont), and Dr. Jeffrey S. Berns (University of characteristics with burnout among re- tify burnout. Due to the large variability in Pennsylvania) for critical review of the ques- sponding US nephrology fellows. prevalence of physician burnout reported tionnaire, as well as Dr. Richard J. Solomon Supplemental Table 5. Multivariable lo- (from 0% to 80.5%) with various ques- fi (University of Vermont) for nancial support. gistic regression model for predictors of de- tionnaire instruments and between differ- An abstract of this work was presented in pressive symptoms. ent specialties, there exists an opinion part at the NKF’s 2019 Spring Clinical Meeting. whether the two items (emotional ex- All authors reviewed and approved the haustion and depersonalization) can final version of the manuscript. Dr. Agrawal REFERENCES reliably assess burnout among physi- was responsible for conception, study design, 55,56 cians. We evaluated the fellows for data analysis, interpretation, drafting, analy- 1. Shanafelt TD, West CP, Sinsky C, Trockel M, depressive symptoms but were unable sis, and was accountable for the work and Tutty M, Satele DV, et al.: Changes in to offer support or resources to those af- burnout and satisfaction with work-life accepts all responsibilities. Dr. Plantinga and integration in physicians and the general fected due to anonymous data collection. Dr. Abdel-Kader were responsible for data US working population between 2011 About a third of nephrology fellows analysis, interpretation, drafting, and critical and 2017. Mayo Clin Proc 94: 1681–1694, in the United States experienced burn- revision of the manuscript for important in- 2019 out and depressive symptoms. Future tellectual content. Mr. Pivert was responsible 2. Shanafelt TD, Noseworthy JH: Executive research studies need to evaluate ne- leadership and physician well-being: Nine for acquisition of study data, editing, and organizational strategies to promote en- phrology fellows for burnout using figure design. Dr. Provenzano was responsible gagement and reduce burnout. Mayo Clin questionnaires such as the full-length for conception of the study and study design. Proc 92: 129–146, 2017 MBI, Mini-Z, or Copenhagen Burnout Dr. Soman, Dr. Choi, and Dr. Jaar were re- 3. Shanafelt TD, Dyrbye LN, West CP, Sinsky Inventory longitudinally or using quali- sponsible for conception of the study, study CA: Potential impact of burnout on the US tative research methods to study the physician workforce. Mayo Clin Proc 91: design, interpretation, and editing. 1667–1668, 2016 prevalence and drivers of burnout in 4. West CP, Shanafelt TD, Kolars JC: Quality of 57 training. Further research needs to life, burnout, educational debt, and medical measure the burden of depressive symp- DISCLOSURES knowledge among internal medicine resi- toms in nephrology fellows and identify dents. JAMA 306: 952–960, 2011 factors (including burnout) that could 5. Peckham C: Medscape nephrologist lifestyle Mr. Pivert is an employee of the ASN Alliance for report. Medscape, New York, NY, 2017. Avail- result in serious mental health problems Kidney Health. All remaining authors have nothing able at: https://www.medscape.com/features/ and major depression. Burnout in ne- to disclose. slideshow/lifestyle/2017/nephrology#page52. phrology fellows—especially that per- Accessed April 30, 2019 ceived by female physicians and due to 6. Parker MG, Ibrahim T, Shaffer R, Rosner MH, Molitoris BA: The future nephrology work- disruption in work environment or poor FUNDING — force: Will there be one? Clin J Am Soc work-life balance need to be further Nephrol 6: 1501–1506, 2011 explored to evaluate causality and de- This study was supported by the Division of Ne- 7. Jhaveri KD, Sparks MA, Shah HH, Khan S, velop targeted efforts (e.g.,byensuring phrology and Hypertension, University of Vermont Chawla A, Desai T, et al.: Why not nephrology?

JASN 31: 675–685, 2020 Burnout in Nephrology Fellows 683 SPECIAL ARTICLE www.jasn.org

A survey of US internal medicine sub- https://www.asn-online.org/education/training/ specialty choice by US medical students. specialty fellows. AmJKidneyDis61: fellows/ite.aspx. Accessed November 7, JAMA 290: 1173–1178, 2003 540–546, 2013 2019 37. Elliott DJ, Young RS, Brice J, Aguiar R, Kolm 8. Accreditation council for graduate medical 22. Dyrbye L, Shanafelt T: A narrative review on P: Effect of hospitalist workload on the education data resource book 2017-2018. burnout experienced by medical students quality and efficiency of care. JAMA Intern Available at: https://www.acgme.org/Portals/ and residents. Med Educ 50: 132–149, Med 174: 786–793, 2014 0/PFAssets/PublicationsBooks/2017-2018_ 2016 38. Shah HH, Jhaveri KD, Sparks MA, Mattana J: ACGME_DATABOOK_DOCUMENT.pdf. 23. West CP, Dyrbye LN, Shanafelt TD: Physi- Career choice selection and satisfaction Accessed April 18, 2019 cian burnout: contributors, consequences among US adult nephrology fellows. Clin 9. Quigley L, Salsberg E, Mehfoud N, Collins A: and solutions. J Intern Med 283: 516–529, JAmSocNephrol7: 1513–1520, 2012 Report on the 2017 Survey of Nephrology 2018 39. Elmariah H, Thomas S, Boggan JC, Zaas A, Fellows. Washington, DC, American Society 24. Bianchi R, Schonfeld IS, Laurent E: Burnout- Bae J: The burden of burnout. Am J Med of Nephrology, 2017. Available at: https:// depression overlap: A review. Clin Psychol Qual 32: 156–162, 2017 www.asn-online.org/education/training/ Rev 36: 28–41, 2015 40. Sinsky C, Colligan L, Li L, Prgomet M, workforce/Nephrology_Fellow_Survey_ 25. Ahola K, Hakanen J, Perhoniemi R, Mutanen Reynolds S, Goeders L, et al.: Allocation of Report_2017.pdf. Accessed June 28, 2019 P: Relationship between burnout and de- physician time in ambulatory practice: A time 10. Brown RS: Is nephrology fellowship training pressive symptoms: A study using the and motion study in 4 specialties. Ann Intern on the right track? Am J Kidney Dis 60: person-centred approach. Burn Res 1: Med 165: 753–760, 2016 343–346, 2012 29–37, 2014 41. Villafranca A, Fast I, Jacobsohn E: Disruptive 11. Ferris M, Iglesia E, Ko Z, Amamoo A, Mahan J, 26. Williams AW: Addressing physician burnout: behavior in the operating room: Prevalence, Desai T, et al.: Wanted: Pediatric nephrolo- Nephrologists, how safe are we? Clin J Am consequences, prevention, and manage- gists! - why trainees are not choosing pediatric Soc Nephrol 13: 325–327, 2018 ment. Curr Opin Anaesthesiol 31: 366–374, nephrology. Ren Fail 36: 1340–1344, 2014 27. Dyrbye LN, Burke SE, Hardeman RR, Herrin 2018 12. Harris PA, Taylor R, Thielke R, Payne J, J, Wittlin NM, Yeazel M, et al.: Association 42. Sanchez LT: Disruptive behaviors among Gonzalez N, Conde JG: Research electronic of clinical specialty with symptoms of physicians. JAMA 312: 2209–2210, 2014 data capture (REDCap)--a metadata-driven burnout and career choice regret among 43. Chen PG, Auerbach DI, Muench U, Curry methodology and workflow process for pro- US resident physicians. JAMA 320: LA, Bradley EH: Policy solutions to ad- viding translational research informatics 1114–1130, 2018 dress the foreign-educated and foreign- support. J Biomed Inform 42: 377–381, 2009 28. Shanafelt TD, Raymond M, Horn L, born health care workforce in the 13. Maslach C, Jackson SE, Leiter MP: Maslach Moynihan T, Collichio F, Chew H, et al.: United States. Health Aff (Millwood) 32: burnout inventory manual,3rdEd.,Palo Oncology fellows’ career plans, expecta- 1906–1913, 2013 Alto, CA, Consulting Psychologists Press, tions, and well-being: Do fellows know what 44. Portoghese I, Galletta M, Leiter MP, Cocco P, 1996 they are getting into? JClinOncol32: D’Aloja E, Campagna M: Fear of future vio- 14. West CP, Dyrbye LN, Satele DV, Sloan JA, 2991–2997, 2014 lence at work and job burnout: A diary study Shanafelt TD: Concurrent validity of single- 29. Ross MJ, Braden G; ASN Match Committee: on the role of psychological violence and job item measures of emotional exhaustion and Perspectives on the nephrology match for control. Burn Res 7: 36–46, 2017 depersonalization in burnout assessment. fellowship applicants. Clin J Am Soc Nephrol 45. Mata DA, Ramos MA, Bansal N, Khan R, J Gen Intern Med 27: 1445–1452, 2012 12: 1715–1717, 2017 Guille C, Di Angelantonio E, et al.: Prevalence 15. Aiken LH, Patrician PA: Measuring organi- 30. Shanafelt TD, Gorringe G, Menaker R, of depression and depressive symptoms zational traits of hospitals: The Revised Storz KA, Reeves D, Buskirk SJ, et al.: among resident physicians: A systematic review Nursing Work Index. Nurs Res 49: 146–153, Impact of organizational leadership on phy- and meta-analysis. JAMA 314: 2373–2383, 2000 sician burnout and satisfaction. Mayo Clin Proc 2015 16. Whooley MA, Avins AL, Miranda J, Browner 90: 432–440, 2015 46. American Psychiatry Association: Practice WS: Case-finding instruments for depres- 31. Purvanova RK, Muros JP: Gender differences guideline for the treatment of patients with sion. Two questions are as good as many. in burnout: A meta-analysis. J Vocat Behav major depressive disorder, Third Edition. J Gen Intern Med 12: 439–445, 1997 77: 168–185, 2010 2010. Available at: https://psychiatryonline.org/ 17. Maslach C, Jackson SE: The measurement 32. McMurray JE, Linzer M, Konrad TR, Douglas pb/assets/raw/sitewide/practice_guidelines/ of experienced burnout. J Organ Behav 2: J, Shugerman R, Nelson K: The work lives of guidelines/mdd.pdf. Accessed May 3, 99–113, 1981 women physicians results from the physician 2019 18. Shanafelt TD, Balch CM, Bechamps GJ, work life study The SGIM Career Satisfaction 47. Guille C, Speller H, Laff R, Epperson CN, Sen Russell T, Dyrbye L, Satele D, et al.: Burnout Study Group. J Gen Intern Med 15: 372–380, S: Utilization and barriers to mental health and career satisfaction among American 2000 services among depressed medical interns: surgeons. Ann Surg 250: 463–471, 2009 33. Linzer M, Harwood E: Gendered expecta- A prospective multisite study. JGradMed 19. Rosenstein AH: Original research: Nurse- tions: Do they contribute to high burnout Educ 2: 210–214, 2010 physician relationships: Impact on nurse among female physicians? J Gen Intern Med 48. Bianchi R, Schonfeld IS, Laurent E: Physician satisfaction and retention. Am J Nurs 102: 33: 963–965, 2018 burnout is better conceptualised as de- 26–34, 2002 34. Dyrbye LN, Varkey P, Boone SL, Satele DV, pression. Lancet 389: 1397–1398, 2017 20. Hardeman RR, Przedworski JM, Burke SE, Sloan JA, Shanafelt TD: Physician satisfaction 49. Leigh JP, Tancredi DJ, Kravitz RL: Physician Burgess DJ, Phelan SM, Dovidio JF, et al.: and burnout at different career stages. Mayo career satisfaction within specialties. BMC Mental well-being in first year medical stu- Clin Proc 88: 1358–1367, 2013 Health Serv Res 9: 166, 2009 dents: A comparison by race and gender: A 35. Glasheen JJ, Misky GJ, Reid MB, Harrison 50. Linzer M, Poplau S, Babbott S, Collins T, report from the medical student change RA, Sharpe B, Auerbach A: Career satisfac- Guzman-CorralesL,MenkJ,etal.:Work- study. JRacialEthnHealthDisparities2: tion and burnout in academic hospital med- life and wellness in academic general in- 403–413, 2015 icine. Arch Intern Med 171: 782–785, 2011 ternal medicine: Results from a national 21. Nesbitt H: American Society of Nephrology | 36. Dorsey ER, Jarjoura D, Rutecki GW: Influence survey. JGenInternMed31: 1004–1010, fellows - in-training exam. Available at: of controllable lifestyle on recent trends in 2016

684 JASN JASN 31: 675–685, 2020 www.jasn.org SPECIAL ARTICLE

51. Salles A, Liebert CA, Greco RS: Promoting workplace violence (WPV) overlap: Is this an 56. Rotenstein LS, Torre M, Ramos MA, Rosales balance in the lives of resident physicians: A opportunity to sustain prevention of both? RC, Guille C, Sen S, et al.: Prevalence of call to action. JAMA Surg 150: 607–608, 2015 Health 08: 531–537, 2016 burnout among physicians: A systematic re- 52. Melamed ML, Campbell KN, Nickolas TL: 54. Sullivan GM, Artino AR Jr: How to create a view. JAMA 320: 1131–1150, 2018 Resizing nephrology training programs: A bad survey instrument. J Grad Med Educ 9: 57. Dyrbye LN, Clinic M, Meyers D, Ripp J, Dalal call to action. Clin J Am Soc Nephrol 12: 411–415, 2017 N, Bird SB, et al.: A pragmatic approach for 1718–1720, 2017 55. Schwenk TL, Gold KJ: Physician burnout-A organizations to measure health care pro- 53. Privitera MR: Organizational contributions serious symptom, but of what? JAMA 320: fessional well-being. NAM Perspect 2018 to healthcare worker (HCW) burnout and 1109–1110, 2018 doi:10.31478/201810b

AFFILIATIONS

1Division of Nephrology and Hypertension, University of Vermont, Burlington, Vermont; 2Department of Medicine, Emory University, Atlanta, Georgia; 3Division of Nephrology, Vanderbilt University Medical Center, Nashville, Tennessee; 4Workforce, Training, and Career Advancement Department, American Society of Nephrology, Washington, DC; 5Division of Nephrology, Duke University, Durham, North Carolina; 6Division of Nephrology and Hypertension, Henry Ford Hospital, Detroit, Michigan; 7Division of Nephrology and Hypertension, Medstar Georgetown University Hospital, Washington, DC; 8Department of Medicine, Division of Nephrology, Johns Hopkins University, Baltimore, Maryland; 9Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health Welch Center for Prevention, Epidemiology, and Clinical Research, Baltimore, Maryland; and 10Nephrology Center of Maryland, Baltimore, Maryland

JASN 31: 675–685, 2020 Burnout in Nephrology Fellows 685 Supplemental file

Table of contents:

Supplementary table 1. Questions on depressive symptoms and emotional well-being in the survey.

Supplementary table 2. Comparison of demographic characteristics of responders and non-responders.

Supplementary table 3. Parameters stratified by burnout status.

Supplementary table 4. Association of characteristics with burnout among responding

US nephrology fellows.

Supplementary table 5. Multivariable logistic regression model for predictors of depressive symptoms.

Burnout in Nephrology Fellows—Agrawal et al.

Supplementary table 1. Questions on depressive symptoms and emotional well- being in the survey. Emotional Well- Survey Question Answer Options Interpretation of Reference Being Measure Responses Depressive a. Depressed mood: ‘Yes’ or ‘no’ ‘Yes’ to either 16 symptoms ‘During the past question was month, have you considered positive often been bothered for depressive by feeling down, symptoms. depressed, or hopeless?’

b. Anhedonia: ’During the past month, have you often been bothered by little interest or pleasure in doing things?’ Positive Influence at ‘How often do you ‘Every day’, ‘a few ‘Once a week’ or 17 Work feel that you are times a week’, ‘once more frequently was positively influencing a week’, ‘a few times affirmative of other people’s lives a month’, ‘once a positive influence at through your work?’ month’ or ‘a few work. times a year or less’ Quality of Program ‘My program ‘Strongly agree’, ‘Strongly agree’ or 4,15 Leadership leadership inspires ‘agree’, ’neither ‘agree’ indicates me to do my best.’ agree nor disagree’, strong quality of To what extent do ‘disagree’ or program leadership. you agree or ‘strongly disagree’ disagree with this statement? Work-Life Balance ‘How satisfied are ‘Very satisfied’, ‘Very satisfied,’ 18 you with the balance ‘somewhat satisfied’, ‘somewhat satisfied,’ between your ‘neutral’, ‘somewhat or ‘neutral’ indicates personal and dissatisfied’ or ‘very adequate work-life professional life?’ dissatisfied’ balance. Quality of Life ‘Which of the ‘As good as it can ‘As good as it can 4,15 following best be’, ‘somewhat be,’ ‘somewhat describes your good’, ‘neutral’, good,’ or ‘neutral’ overall quality of ‘somewhat bad’ or indicates good life?’ ‘as bad as it can be’ quality of life. Career Satisfaction ‘If given the ‘Definitely yes’, ‘Definitely yes’ or 18 opportunity to revisit ‘probably yes’, ‘not ‘probably yes’ your career choice, sure’, ‘probably no’ indicates career would you choose to or ‘definitely no’ satisfaction. be a nephrologist again?’ Disruptive ‘How frequently ‘Daily’, ‘weekly’, ‘1 to ‘1 to 5 times per 19 Environment have you 2 times per month’, year’ or more experienced or ‘1 to 5 times per frequently indicates witnessed disruptive year’ or ‘never’ disruptive behavior of a environment. physician in your workplace?’

Burnout in Nephrology Fellows—Agrawal et al.

Social Support ‘Do you have ‘All of the time’, ‘All of the time,’ 20 someone you can ‘most of the time’, ‘most of the time,’ or count on to listen to ‘some of the time’, ‘a ’some of the time’ you when you need little of the time’ or indicates presence to talk?’ ‘none of the time’ of social support.

Burnout in Nephrology Fellows—Agrawal et al.

Supplementary table 2: Comparison of demographic characteristics of responders and non-responders.

Entire Responders Non- Test for nephrology responders differenceb fellow poola

US medical graduate [n (%)] 263 (32.5) 126 (36.7) 137 (29.5) X2 (1) = 4.76, International medical graduate 545 (67.5) 217 (63.3) 328 (70.5) p=0.03 [n (%)]

Female gender [n (%)] 278 (34.4) 131 (37.9) 147 (31.9) X2 (1) = 3.12, Male gender [n (%)] 529 (65.6) 215 (62.1) 314 (68.1) p=0.08

First year fellow [n (%)] 411 (50.9) 158 (47.6) 253 (53.2) X2 (1) = 2.42, Second year fellow [n (%)] 397 (49.1) 174 (52.4) 223 (46.8) p=0.12

Mean age of first year fellows 34.3 33.8 ± 4.3 34.6 t (155) = (in years) ± sd c 2.32, p=0.02 a: Data obtained from ACGME Data report 2017-2018.8 b: By chi-square or t test, as appropriate. C: Standard deviation (sd) of age was obtained for the responding fellows from the study sample data and was not available for the entire fellow pool in the ACGME Data report 2017-2018.

Burnout in Nephrology Fellows—Agrawal et al.

Supplementary table 3. Parameters stratified by burnout status.*

Variable Burnout (n=104) No Burnout (n=243)

Mean Age ± SD (years) 33.9 ± 4.3 34.4 ± 4.6

Female gender1 48 (46.2%) 83 (34.2%)

Second year of fellowship 54 (52.9%) 120 (52.2%)

IMG 58 (56.3%) 159 (66.3%) Nephrology as first fellowship choice 38 (36.9%) 88 (36.4%)

Educational debt present 53 (51.5%) 114 (47.5%) Workload (>15 patients/day)2 89 (86.4%) 174 (71.9%) Clinical nephrology fellowship 91 (87.5%) 211 (86.8%)

Being in a relationship 69 (66.3%) 183 (75.3%)

Positive Influence 56 (53.8%) 134 (55.1%) Positive Screen for Depression3 69 (67.0%) 54 (22.2%)

Strong Program Leadership3 60 (57.7%) 201 (82.7%) Adequate Work-Life Balance3 43 (41.3%) 197 (81.4%)

Social Support Present3 82 (79.6%) 228 (93.8%)

Good Quality of Life3 73 (70.9%) 226 (93.4%)

Career Satisfaction Present3 57 (55.3%) 197 (81.4%) Disruptive Environment Present3 66 (63.5%) 89 (36.8%) *IMG = international medical graduate.

P-values for significant differences were noted as:1 p=0.029, 2 p=0.01 and 3 p<0.001

Burnout in Nephrology Fellows—Agrawal et al.

Supplementary table 4. Association of characteristics with burnout among responding US nephrology fellows.*

Variable Odds Ratio 95% CI p value

Age (in years) 0.99 0.93 - 1.06 0.85

Female vs male gender 1.90 1.09 - 3.32 0.02

1st year vs 2nd year of fellowship 1.16 0.67 - 2.01 0.60

IMG vs USMG 0.57 0.31 - 1.03 0.06

Work load (>15 patients/day) vs (≤15 1.52 0.74 - 3.10 0.25 patients/day)

Relationship status (yes vs no) 0.67 0.36 - 1.25 0.21

Strong program leadership (yes vs no) 0.63 0.33 - 1.18 0.15

Social support (present vs absent) 0.47 0.19 - 1.13 0.09

Disruptive environment (yes vs no) 2.63 1.48 - 4.66 0.001

Work-life balance (poor vs adequate) 3.97 2.22 - 7.07 <0.001 *The model included 96 fellows with burnout and 226 fellows without burnout.

Burnout in Nephrology Fellows—Agrawal et al.

Supplementary table 5. Multivariable logistic regression model for predictors of depressive symptoms.*

Variable Odds Ratio 95% CI p value

Age (in years) 0.98 0.92 - 1.05 0.60

Female vs male gender 1.43 0.82 - 2.51 0.21

1st year vs 2nd year of fellowship 0.92 0.53 - 1.60 0.76

IMG vs USMG 0.56 0.31 - 1.01 0.06

Work load (>15 patients/day) vs (≤15 0.91 0.47 - 1.77 0.78 patients/day)

Relationship status (yes vs no) 0.77 0.41 - 1.44 0.42

Strong program leadership (yes vs no) 0.46 0.24 - 0.88 0.02

Social support (present vs absent) 0.08 0.03 - 0.24 <0.001

Disruptive environment (yes vs no) 1.16 0.65 - 2.06 0.62

Work-life balance (adequate vs poor) 0.17 0.10 - 0.31 <0.001 *The model included 115 fellows who were positive for depressive symptoms and 208 fellows who were negative for depressive symptoms.