In 1993, the American Medical Women’s Association surveyed its resident members on how residency programs provided for maternity leave: 36% of residents took 4 to 8 weeks of leave, and 30% took less than 4 weeks.1 Since that study, the number of women in medical school, residency, and academic faculty has increased.2,3 Yet recent studies suggest that the length of maternity leave among female physicians has not changed, still averaging less than 8 weeks.4–7
Research has consistently shown that longer maternity leaves are associated with better maternal and infant health outcomes. For example, greater than 12 weeks of maternity leave is associated with fewer postpartum depressive symptoms and lower rates of self-reported poor maternal health.8,9 For infants, longer maternity leaves are associated with longer duration of breastfeeding and increased well-baby care visits and vaccination rates.10 Yet few graduate medical education (GME) programs provide explicit policies on the length of maternity leave allowed.11 In the absence of clear standards and practices to dictate what a resident contemplating parenthood can or should expect, many of these decisions are determined on an ad hoc basis.
To better understand the determinants and effects of maternity leave in GME, we characterized maternity leave across specialties at 6 institutions, examined factors associated with its length, and assessed correlations between the length of maternity leave and resident and infant well-being.
We invited all female residents (n = 1,537) enrolled across 78 GME programs at 6 institutions (University of California San Diego, Emory University, University of Michigan, University of South Florida, University of Utah, and University of Washington) to participate in a voluntary, anonymous, 5- to 10-minute survey using REDCap, a secure web-based application.12 With the exception of one institution (University of Utah), a $5 gift card, not conditional on response, was provided with the email invitation. We collected data over a 6-week period in 2017, with 3 email reminders sent at 48 hours, 3 weeks, and 5 weeks.
We developed a survey tool based on a focus group of physicians with personal experience in pregnancy during medical training, a literature review of pregnancy and parenthood in GME identifying potential factors that influence maternity leave length,10,13 and relevant existing survey items.6,7,14 Wellness outcome measures were duration and satisfaction with breastfeeding, burnout, postpartum depression, perceptions of emotional and logistical support, satisfaction with maternity leave length, and satisfaction with resident parenthood.15,16 Duration of breastfeeding was recorded in months as a free-text response. Three-point Likert-like scales (“less than I would like,” “about right,” and “more than I would like”) were used to assess satisfaction with length of breastfeeding and satisfaction with maternity leave length. Burnout was assessed using a single-item measure shown to be a reliable substitute for the Maslach Burnout Inventory domain of emotional exhaustion.17 Residents who had given birth within the past 12 months were also assessed using the Edinburgh Postnatal Depression Scale, the most common and extensively validated screen for postpartum depression.18 Based on prior literature showing that the support of coworkers toward a resident’s pregnancy could modify the resident’s perceptions of stress, we included perceptions of support from colleagues, attendings, and program leadership as an indicator of well-being.19 These perceptions were measured using a 5-point Likert scale (with response categories ranging from “not supported” to “well supported”).
Finally, to identify other determinants of maternity leave length, we asked respondents to select up to 3 options from a list of choices, including a free-text response, which was then categorized into themes.
The survey underwent content validation followed by cognitive interviewing of a sample of female resident volunteers, prior to its pilot administration to 214 female residents at the University of Washington over the 2015–2016 academic year. Based on these results, we further refined the survey and conducted a second round of content validation by 8 experts in survey design or the topic at hand from our participating institutions, as well as cognitive interviewing of 5 volunteers representative of our target population. This was followed by pilot testing among 21 resident mothers at nonparticipating institutions. Analysis of pilot data showed an appropriate range of variance. The final version was subsequently distributed to our study population. The complete survey is available as Supplemental Digital Appendix 1 at http://links.lww.com/ACADMED/A682.
We generated descriptive statistics to characterize study participants. A linear regression analysis could not be performed because of the distribution and variability across reported maternity leave lengths resulting in a noncontinuous variable. Instead, we dichotomized maternity leave length as 6 weeks or less and greater than 6 weeks. This length was chosen based on the distribution of reported maternity leaves among respondents. To assess whether hypothesized factors contribute to maternity leave length, we compared categorical variables using the Pearson χ2 test, and continuous variables using independent t tests. We performed a logistic regression analysis to determine associations using factors with a P value of < 0.5. Household income was normalized by the number of members supported by that income and adjusted to the site’s cost of living. Residency program specialties were categorized according to medical, surgical, and hospital-based specialties as designated by the Accreditation Council for Graduate Medical Education (ACGME).20 We also categorized specialties by controllable versus uncontrollable lifestyle as designated by Dorsey and colleagues.21 Program size was categorized as large for ≥ 40 residents, and small if < 40, based on the distribution of participating program sizes. We normalized paid leave across institutions by calculating the percentage of base salary a resident could receive over a 12-week leave, not including sick leave or vacation, and regardless of the actual length of leave taken. For example, paid leave for an institution that provided 6 weeks of full salary support while on maternity leave was categorized as 50% paid leave.
We stratified wellness outcomes by length of maternity leave. For breastfeeding duration, a target of 6 months or more was chosen per American Academy of Pediatrics recommendations.22 We dichotomized satisfaction scales for length of breastfeeding and length of maternity leave for analysis as satisfied (“about right,” “more than I would like”) versus unsatisfied (“less than I would like”). Scales measuring perceived support were also dichotomized for analysis as positive support (“well supported,” “somewhat supported”) versus no support (“neutral,” “somewhat not supported,” “not supported”). Stata SE 13.1 statistical software for Windows (StataCorp, College Station, Texas) was used for the analysis, with P < .05 denoting statistical significance.
At the end of the survey, we gave participants the option of a free-text response for additional thoughts on pregnancy and parenthood in GME. These answers provided context for our quantitative results.
Each site’s institutional review board deemed this study exempt (University of California San Diego, University of Michigan, University of South Florida, University of Utah, and University of Washington) or approved (Emory University).
The survey was administered to female residents in 78 residency programs representing 25 unique specialties at 6 academic medical centers in the United States. All 6 centers are urban, tertiary care teaching hospitals associated with large medical schools. Among 1,537 residents invited to complete the survey, 57% (876) were medical, 23% (351) surgical, and 20% (310) hospital-based trainees. The distribution of 804 respondents was similarly 57% (458) medical, 23.5% (189) surgical, and 19.5% (157) hospital-based trainees. The overall response rate was 52% (804 of 1,537) and ranged from 48% to 57% among institutions, 51% to 53% among program categories, and 33% to 89% among individual programs.
Among participating institutions, 4 provided institution-funded leave, 1 provided state-funded leave, and 1 provided no paid leave. The percentages of paid leave for 12 weeks across institutions were 0%, 17%, 30%, 50%, and 100%.
Characteristics of the residents who completed the survey are listed in Table 1. Among the respondents, 16% (126) were mothers. Of residents who were mothers, 14% (18) were interns (postgraduate year 1) and 50% (63) reported having their first child during residency. Among the 43 residents who were pregnant at the time of the survey, 53% (23) were not currently mothers, and 16% (7) had previously given birth during residency. Of resident mothers, nearly 33% (41 of 126) had 2 or more children, and across all resident mothers, most children were toddlers (60%).
Ninety-eight percent (123) of mothers were married compared with 40% (271) of nonmothers. Mothers were also more likely to have a PhD (20 of 126; 16%) compared with nonmothers (19 of 678; 2.8%) and to have a lower household income per household member when adjusted for site cost of living.
Seventy-seven maternity leaves were reported (Figure 1) among 77 residents who gave birth during residency. Three of these residents gave birth twice during residency but only reported details for their most recent maternity leave. Length of leave ranged from 2 to 40 weeks with a median of 7 weeks (interquartile range, 6–10 weeks). More residents took 6 weeks of leave (25; 32%), then 8 weeks (10; 13%) and 10 weeks (8; 10%).
The associations between maternity leave length and trainee, program, and institutional characteristics are reported in Table 2. Those who used neither sick leave nor vacation were significantly more likely to take ≤ 6 weeks of leave (P = .04). Institution was also a significant determinant on bivariable analysis (P = .02). The percentage of salary a resident was eligible to receive over a 12-week maternity leave was also found to be significant, with less paid leave associated with > 6 weeks of maternity leave. However, this finding was largely influenced by the 45% of residents in the cohort taking > 6 weeks of leave from one institution, where at the time, maternity leave was unpaid.
Finally, while not statistically significant, ≤ 6 weeks of leave was more frequently observed in residents planning on fellowship, and in programs with ≤ 40 residents. The mean partner parental leave for each group was nearly equivalent to the overall mean partner parental leave of 2 weeks. Thus, no comparison could be made. In a multivariable logistic regression model, no factors were significantly correlated with length of leave.
For maternity leave logistics, among the 77 reported leaves, 81% (62) included use of vacation time, 64% (49) included use of sick leave, 9.1% (7) included use of neither, and 60% (46) indicated that they received another form of paid leave. Forty-seven percent (36) of respondents reported that their maternity leave required that they pay back call; 13% reported being “unsure” whether call must be paid back. Upon initial return to work, 9.1% (7) had less inpatient time and 9.1% (7) returned to a research year, while 86% (66) had no adjustment to their schedule. Fifty-three percent (41 of 77) reported that their residency training was not extended, while 9% (7) were unsure whether training would be extended. Residents who took > 6 weeks of leave had to extend their training a median of 4 weeks, while those who took ≤ 6 weeks had a median training extension of 0 weeks.
Self-reported determinants of maternity leave length, which were not mutually exclusive, are shown in Figure 2. The most frequently reported factors reported by the 77 residents who took maternity leave were, in order, desire to avoid extension of residency training (59; 27%), finances (27; 12%), newborn bonding (26; 12%), and concern for repercussions to their colleagues (24; 11%). The least reported factors were infant health complications (0; 0%), partner paternal leave (6; 3%), and maternal health complications (6; 3%). Free-text responses were added to existing categories where appropriate. Uncategorized responses include concern for decline of clinical skills (1), being asked to return early because another resident was taking maternity leave (1), stress of breastfeeding as an intern and caring for other children (1), concern for jeopardizing the spouse’s career by using partner leave (1), loss of a partner (1), and visa requirements (1).
Resident-reported measures of well-being are shown in Table 3 stratified by ≤ 6 weeks versus > 6 weeks of maternity leave. Both groups achieved high rates of reaching 6 months of breastfeeding and had similar rates of negative postpartum depression and burnout screens and positive emotional support. Greater than 6 weeks of maternity leave was associated with statistically significant longer duration of breastfeeding (P = .01) and greater perceived logistical support from their program administration (P = .04). Greater satisfaction with maternity leave length and recommending childbearing during training were favored by those who took > 6 weeks of leave compared with ≤ 6 weeks of leave.
This multicenter survey of female residents training in a variety of different medical specialties in the United States provides the most comprehensive recent evaluation of experiences with pregnancy and motherhood during graduate medical training to our knowledge. Given that women now constitute more than half of incoming medical students, the medical profession must consider how women can integrate professional expectations of GME with the biological constraints of female childbearing. Our findings that 1 in 5 responding women residents were mothers, and that half of these experienced their first childbirth during residency, suggest that this is an issue of considerable importance to the community of graduate medical educators. The observation that maternity leave length varied considerably across the 6 institutions studied, along with correlations between maternity leave duration and key wellness outcomes, including duration of breastfeeding and satisfaction with childbearing, is particularly important.
Our findings of the prevalence of pregnancy during residency training add to findings from prior studies limited to single institutions or specialties. In a recent survey of Mayo Clinic residents, 35.3% of female respondents were mothers, and over 75% of reported pregnancies occurred during residency training.7 Among trainees in surgical specialties, 32% to 46% bear children during residency, often during research time.23–25 Our study provides a broader view of pregnancy in GME that could help institutions in planning for maternity leave and parental benefits.
Paid leave policies varied among the institutions studied, with 4 of the institutions providing funded leave, 1 institution relying on state funding, and another with no provisions for paid leave. Excluding use of vacation and sick leave, trainees at only 1 of the 6 studied institutions could receive a full 12 weeks of paid leave; all other institutions provided 6 weeks or less. A recent survey of faculty maternity leave policies at 12 academic U.S. medical schools found a more generous, yet similar variability among institutions.26 In that study, the length of faculty salary support for maternity leave ranged from 6 to 16 weeks, with a mean of 8.5 weeks. Paid faculty maternity leave at our 6 participating institutions was higher, ranging from 9 to 12 weeks, with a mean of 11.5 weeks. The variability in paid leave for both residents and faculty could be a result of the lack of a national mandate on paid family leave. Notably, this absence of a standard policy persists despite demonstrated benefits of paid leave for maternal, infant, and early childhood health as well as women’s economic outcomes and academic retention and productivity.27–30 As such, a standardized paid maternity leave policy for residents that allows for sufficient time for childbirth recovery and infant bonding has the potential to support women’s growth as future academic faculty. This is a key element of advancing gender equity in academic medicine.
The impact of leave on timing of residency completion is also an important consideration. Many of the member boards of the American Board of Medical Specialties (ABMS) have specific restrictions regarding leave from residency training.31 Our observation that residents who took leave—and particularly those who took longer leaves—extended their training likely reflects these stipulations. Whether this in turn affects board eligibility timing, job options (including the ability to enter fellowship training, given typically rigid start dates and schedules), and salary would be worthy subjects of future research. Regardless, that a desire to avoid extension of residency training was the most commonly reported determinant of maternity leave duration suggests that ABMS member policies strongly influence women’s choices and options. This supports the need for revision of these policies to allow flexibility for the integration of adequate professional training and motherhood, without compromising the competency and milestone expectations of each specialty. As such, the rigidity of the residency training timeline should be reconsidered. One potential solution is to permit earlier program completion contingent on successful demonstration of competency and milestone attainment without a minimum elapsed training time as currently stipulated by ACGME specialty review committees. Alternatively, policies that allow for delayed fellowship (or other employment) start dates could also help mitigate this concern.
This study also illuminates several other targets for further policy clarification and programmatic support. The vast majority of resident mothers returned to a full-time, unadjusted work schedule after a median of only 7 weeks of maternity leave. This suggests that programs do not employ creative solutions, such as parenting electives, research rotations, or part-time schedules, to ease resident mothers back into the workplace. Instead, many mothers returned to more demanding schedules as the majority of respondents had to either “pay back” call or were unsure whether this would be required of them. This lack of clarity on whether call duties should be “paid back” after maternity leave leaves resident mothers particularly vulnerable and must be rectified with explicit, written policies. The observed high expectation of repayment of call for maternity leave should be evaluated carefully by GME programs. This stipulation may be inconsistent with employment law or contractual obligations for some trainees. Program directors should maintain vigilance to ensure that expectations do not differ for trainees on maternity leaves as opposed to other medical leaves or vacation, depending on how the maternity leave was structured. Clarifying this point may also recalibrate expectations of peers and faculty, as perceived lack of support from these groups has been shown to be a major source of stress in pregnant residents.32 Moreover, a noteworthy minority of residents did not perceive logistical support from the program director or administration, suggesting that logistical support may be a strategic target to help residents harmonize training and childbearing.
Remarkably, the vast majority of respondents continued breastfeeding for 6 months or longer. This contrasts with prior surveys where less than half of residents who initiated breastfeeding after childbirth continued beyond 6 months.33,34 The association between longer duration of breastfeeding and longer maternity leaves suggests that identifying mechanisms to facilitate longer leaves for women experiencing childbirth during residency may lead to improved health for this subset of residents in both the short and long term. Although our findings are heartening, they should not be taken to suggest that ongoing attention to ensuring adequate scheduling flexibility and facilities for lactation support is unnecessary.
Similarly, observations that residents who took maternity leaves overwhelmingly perceived emotional support from their co-residents, attending physicians, and program directors are encouraging. Yet this finding also suggests the possibility that the very decision to embark on pregnancy during residency training may be more likely among residents training in already-supportive environments. This is consistent with prior research demonstrating that residents who delay childbearing do so because of perceived career threats13 and negative stigma attached to pregnancy in residency.35
About half of the respondents who had taken maternity leave indicated emotional burnout compared with 44.4% of residents nationally,36 and a nontrivial minority had a positive postpartum depression screen. Residency programs must anticipate these threats to resident wellness and develop resources to support resident mothers, who may be especially vulnerable to these challenges. Of course, burnout and depression are not unique to mothers, and interventions need not be specific to this subgroup given that this is an area yet to be studied. Many initiatives that address these issues more generally among residents may be useful to those who face these issues in the context of motherhood. Still, recognition of the unique challenges that resident mothers face, by creating supportive environments and informal support networks, may be useful in reducing the prevalence of symptoms in this group.
Although this study has numerous strengths, including its methodological development and validation of survey measures, substantial response rate, and multicenter design, it also has several limitations of note. First, as in any observational study, correlations may not necessarily indicate causation. For example, the observation that residents who took longer maternity leave also breastfed for longer might partly reflect the influence of a confounding underlying desire of certain individuals both to spend more time with their infants and to breastfeed. Still, we believe that at least part of this association is likely to reflect the challenges of maintaining lactation after reentry into the residency training workplace. Second, our estimates may reflect selection bias due to survey nonresponse—for example, if mothers were more likely to respond than nonmothers, then our estimates of frequency of motherhood would be inflated. We intentionally did not convey in the invitation or survey title the purpose of our study, to avoid selection of those most interested in issues relating to pregnancy and childbearing; the survey was simply titled “Women in Residency Survey.” In addition, our substantial rate of response helps to improve the likelihood that our respondents actually were reflective of the underlying population targeted. Third, as a survey study, all responses are self-reported and could be influenced by measurement or recall biases. Our use of previously validated measures of key wellness outcomes and careful validation of new measures helps to limit this concern.
Finally, although we included many individual programs across a diversity of specialties, the residents in our study trained at only 6 institutions; we observed a significant association between institution itself and duration of leave, but given the number of institutions participating, we were unable to definitively identify factors promoting leave. We caution readers not to overinterpret the curious finding of a bivariable association between a higher amount of paid leave and a shorter duration of leave taken. This finding appeared to be driven almost entirely by the propensity of residents at a single institution without paid leave to take longer leaves; it was not significant on multivariable analysis that included institution, suggesting that something else in the cultural or policy environment was an important driver of the observed association. Further investigation—perhaps using qualitative methods—is therefore necessary to evaluate cultural and policy factors in the institutional environment that encourage longer maternity leaves.
In summary, the current study provides insights into the community of medical educators. Pregnancy during residency training is not uncommon, and resident mothers’ experiences vary widely. The GME community has widely embraced the mandate to attend more closely to the physical and psychological well-being of trainees. Our findings underscore the need for clear, consistent, transparent, and easily accessible policies that support the many women already attempting to integrate GME and motherhood, as well as the many others who may be considering it.
The authors gratefully acknowledge the project development support of Christy McKinney, PhD, MPH, associate professor of pediatrics and associate director of the Institute of Translational Sciences KL2 Career Development program at the University of Washington School of Medicine.
1. Bongiovi ME, Freedman J. Maternity leave experiences of resident physicians. J Am Med Womens Assoc (1972). 1993;48:185–188, 193.
2. Bickel J, Galbraith A, Quinnie R. Women in Academic Medicine: Statistics. 1993.Washington, DC: Association of American Medical Colleges.
3. Association of American Medical Colleges. AAMC Data Book: Statistical Information Related to Medical Schools and Teaching Hospitals. 2017.Washington, DC: Association of American Medical Colleges.
4. Sandler BJ, Tackett JJ, Longo WE, Yoo PS. Pregnancy and parenthood among surgery residents: Results of the first nationwide survey of general surgery residency program directors. J Am Coll Surg. 2016;222:1090–1096.
5. Hutchinson AM, Anderson NS 3rd, Gochnour GL, Stewart C. Pregnancy and childbirth during family medicine residency training. Fam Med. 2011;43:160–165.
6. Holliday EB, Ahmed AA, Jagsi R, et al. Pregnancy and parenthood in radiation oncology, views and experiences survey (PROVES): Results of a blinded prospective trainee parenting and career development assessment. Int J Radiat Oncol Biol Phys. 2015;92:516–524.
7. Blair JE, Mayer AP, Caubet SL, Norby SM, O’Connor MI, Hayes SN. Pregnancy and parental leave during graduate medical education. Acad Med. 2016;91:972–978.
8. Dagher RK, McGovern PM, Dowd BE. Maternity leave duration and postpartum mental and physical health: Implications for leave policies. J Health Polit Policy Law. 2014;39:369–416.
9. Gault B, Hartmann H, Hegewisch A, Milli J, Reichlin L. Paid Parental Leave in the United States: What the Data Tell Us About Access, Usage, and Economic and Health Benefits. 2014Washington, DC: Institute for Women’s Policy Research.
10. Ogbuanu C, Glover S, Probst J, Liu J, Hussey J. The effect of maternity leave length and time of return to work on breastfeeding. Pediatrics. 2011;127:e1414–e1427.
11. Magudia K, Bick A, Cohen J, et al. Childbearing and family leave policies for resident physicians at top training institutions. JAMA. 2018;320:2372–2374.
12. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)—A metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42:377–381.
13. Willett LL, Wellons MF, Hartig JR, et al. Do women residents delay childbearing due to perceived career threats? Acad Med. 2010;85:640–646.
14. Sarma AA, Nkonde-Price C, Gulati M, Duvernoy CS, Lewis SJ, Wood MJ; American College of Cardiology Women in Cardiology Leadership Council. Cardiovascular medicine and society: The pregnant cardiologist. J Am Coll Cardiol. 2017;69:92–101.
15. Sattari M, Serwint JR, Neal D, Chen S, Levine DM. Work-place predictors of duration of breastfeeding among female physicians. J Pediatr. 2013;163:1612–1617.
16. Merchant S, Hameed M, Melck A. Pregnancy among residents enrolled in general surgery (PREGS): A survey of residents in a single Canadian training program. Can J Surg. 2011;54:375–380.
17. Dolan ED, Mohr D, Lempa M, et al. Using a single item to measure burnout in primary care staff: A psychometric evaluation. J Gen Intern Med. 2015;30:582–587.
18. Sit DK, Wisner KL. Identification of postpartum depression. Clin Obstet Gynecol. 2009;52:456–468.
19. Phelan ST. Sources of stress and support for the pregnant resident. Acad Med. 1992;67:408–410.
21. Dorsey ER, Jarjoura D, Rutecki GW. Influence of controllable lifestyle on recent trends in specialty choice by US medical students. JAMA. 2003;290:1173–1178.
22. Johnston M, Landers S, Noble L, Szucs K, Viehmann L. Breastfeeding and the use of human milk. Pediatrics. 2012;129:e827–e841.
23. Lerner LB, Stolzmann KL, Gulla VD. Birth trends and pregnancy complications among women urologists. J Am Coll Surg. 2009;208:293–297.
24. Hamilton AR, Tyson MD, Braga JA, Lerner LB. Childbearing and pregnancy characteristics of female orthopaedic surgeons. J Bone Joint Surg Am. 2012;94:e77.
25. Eskenazi L, Weston J. The pregnant plastic surgical resident: Results of a survey of women plastic surgeons and plastic surgery residency directors. Plast Reconstr Surg. 1995;95:330–335.
26. Riano NS, Linos E, Accurso EC, et al. Paid family and childbearing leave policies at top US medical schools. JAMA. 2018;319:611–614.
27. Aitken Z, Garrett CC, Hewitt B, Keogh L, Hocking JS, Kavanagh AM. The maternal health outcomes of paid maternity leave: A systematic review. Soc Sci Med. 2015;130:32–41.
28. Lichtman-Sadot S, Bell NP. Child health in elementary school following California’s paid family leave program. J Policy Anal Manag. 2017;36:790–827.
29. Jou J, Kozhimannil KB, Abraham JM, Blewett LA, McGovern PM. Paid maternity leave in the United States: Associations with maternal and infant health. Matern Child Health J. 2018;22:216–225.
31. Jagsi R, Tarbell NJ, Weinstein DF. Becoming a doctor, starting a family—Leaves of absence from graduate medical education. N Engl J Med. 2007;357:1889–1891.
32. van Dis J. STUDENTJAMA. Residency training and pregnancy. JAMA. 2004;291:636.
33. Miller NH, Miller DJ, Chism M. Breastfeeding practices among resident physicians. Pediatrics. 1996;98(3 pt 1):434–437.
34. Dixit A, Feldman-Winter L, Szucs KA. “Frustrated,” “depressed,” and “devastated” pediatric trainees. J Hum Lact. 2015;31:240–248.
35. Turner PL, Lumpkins K, Gabre J, Lin MJ, Liu X, Terrin M. Pregnancy among women surgeons: Trends over time. Arch Surg. 2012;147:474–479.
36. 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.