Physician well-being is essential to professionalism in medicine. Fundamental aspects of the physician’s identity—such as integrity, empathy, altruism, and service to others—best flourish when physicians care adequately for themselves (1–3). The estimated 15% lifetime prevalence rate of impairment among physicians (4), with resultant suboptimal care to patients (4, 5) and related concerns about physician burnout, depression, and suicide (6–11), further highlight the importance of physician health. Additionally, physicians who practice healthy habits are more likely to provide appropriate preventive care and counseling to their patients (12–14). Although self-care might seem inherent to professionals who have devoted their careers to the care of others, physicians often do not make their own health a priority (15). Physicians are likely to avoid care by seeking curbside consultation or self-treating (16–19) and are more likely than other professionals to work through illness episodes (19, 20). This phenomenon is likely the result of personal, attitudinal, and cultural factors in medicine that promote self-reliance and denial (17, 21). Self-management of illness also appears to be at least partially a learned behavior, picked up by medical students early in training and increasing with each stage of training (22–25).

Early in their professional lives, residents form attitudes about personal health, illness, and acquisition of medical care. Much of this is learned, explicitly or implicitly, from the established medical culture into which they are assimilating (26, 27). Denial and minimization characterize residents’ attitudes about their own health (28). Not only can these prevailing attitudes detrimentally affect personal health, but they also can influence more junior trainees and possibly invite problematic attitudes and behaviors regarding self-care for the next generation of physicians. Even more overtly, distorted boundaries for health care may be modeled by attending faculty members. Half of residency training directors reported having provided care for their trainees for acute and/or chronic conditions (29).

Residents have the additional complication of having multiple roles within their training institution: student/trainee, health care provider, employee, and when they become ill, potential patient. These conflicts—inherent to occupying multiple roles—can create additional complexity for residents when seeking personal medical care. The basic conflict of interest for residents considering care at the home institution lies in the tension between two primary interests: the desire to take care of personal health needs and the need to protect one’s professional reputation and confidentiality. Long, unpredictable work hours, lack of time for appointments, and perceived lack of support from training programs have been shown to be obstacles to obtaining care (18). These concerns and conflicts affect residents’ choices about whether to pursue care at all, where to seek care, and whether care-seeking is delayed (30).

Other important factors related to residents’ health care behaviors include stigma associated with certain medical conditions and breaches of confidentiality (25, 31). Residents who seek treatment at their training institution report concerns about supervisors or colleagues knowing their personal health information and how that may influence others’ opinions of their medical competence (31). Seeking treatment from the same people who evaluate their medical skills raises the possibility that personal information will be shared for the sake of providing care but may ultimately influence perceptions of residents’ capabilities. Although it is known that some illnesses carry increased stigma among patient populations (32–34) and the medical community (35) and that stigma negatively impacts patients’ ability to seek care appropriately (33), it is not known what role stigma may play in residents’ concerns about confidentiality, professional reputation, training status, and decision making for personal health care.

Much of the research pertaining to the relationship between stigma and utilization of health care has either been theoretical (36, 37) or focused solely on medical residents’ utilization of services for mental health issues (38). The potential for dual roles to influence residents’ decisions about seeking treatment for physical health concerns is evident but has been minimally studied (25). We therefore looked at the impact of stigma associated with both physical and mental health issues on the likelihood of care-seeking among residents. We specifically examined the relationship between stigma and concerns about confidentiality and how stigma may influence treatment seeking. We hypothesized that residents would see some personal health problems as stigmatizing and potentially jeopardizing to their training status. Consequently, we expected residents to indicate that they would avoid seeking necessary health care at their training institution for stigmatizing health problems.

A self-report survey on special issues in personal health care of residents during postgraduate training was created and pilot tested at the University of New Mexico School of Medicine in 2001. The survey included 241 items with scaled or counted responses, 11 demographic questions, and two open-ended questions. Survey content domains included personal health and health care behaviors and preferences, experiences as trainees with confidentiality and overlapping roles, attitudes concerning compassion and empathy with ill patients, and six vignettes addressing stigma and fears of professional jeopardy after becoming ill or seeking treatment. The medical illnesses, mental illnesses, and pregnancy condition were used in the vignettes primarily because of their common occurrence. Here we report findings from the survey of vignettes of residents in dual roles as patients and trainees.

The survey was distributed to all 217 residents in postgraduate years (PGY) 2–4 at the University of New Mexico School of Medicine in 2001. The survey was voluntary and confidential; respondents received $50 in compensation for their time and effort. This study was funded by the Arnold P. Gold Foundation and approved by the University of New Mexico Health Sciences Center institutional review board.

Three pairs of vignettes presented three dual-role contexts with two health issue examples for each context (Table 1). There were three questions of perceived stigma, concern about training status jeopardy, and likelihood of avoiding care at the training institution for each vignette. Separate analyses were performed for each question within each pair of vignettes. For each analysis, nine-point scaled rating responses were subjected to repeated measures: vignette (within subjects repeated measures) × training level (PGY-2 versus PGY-3 and PGY-4 combined, between subjects) × gender (between subjects) × residency area (primary care or family practice, internal medicine, and pediatrics versus specialty or all other areas, between subjects) multivariate analyses of variance (MANOVA). No significant effects of training level were found. Fisher’s least significant difference values were calculated to allow post hoc comparison of means where appropriate. We examined correlations among the three questions within each vignette and among responses to vignette items, and we examined a composite measure of concern with confidentiality derived from additional survey items.

Participant Characteristics (Table 2)

The 155 residents who returned surveys (response rate 71%) were 45% women and 55% men. About half (53%) were PGY-2 training level, 45% PGY-3, and 3% PGY-4. About half (46%) were in the primary care areas of family practice, internal medicine, or pediatrics, and about half (54%) were in specialty areas, including psychiatry (12%). Gender groups did not differ by training level or area of residency. About two-thirds of participants (69%) were white, 16% Hispanic, and 15% of other or unreported ethnicity. Gender groups differed by ethnicity (p<0.05), with a greater proportion of Hispanics among women than men (24% versus 9%) and a lower proportion of whites (61% versus 75%). More than half of respondents (59%) were married or living with a partner. Mean age was 31.8 (SD=4.5) years (range=26 to 54 years).

Resident’s Personal Physician as Ward Attending Physician (Table 3)

For the pair of vignettes presenting a resident whose personal physician was also a ward team attending physician, respondents rated the stigma they would feel with a health issue of either hypertension or panic attacks (1=no stigma, 5=some stigma, 9=great stigma). Residents rated the stigma associated with hypertension as low (mean=2.45, SD=1.71) and panic attacks as moderate (mean=5.52, SD=2.22) and greater than the stigma of hypertension (vignette main effect F=284.10, df=1, 147, p<0.0001, Cohen’s d=1.55). For the same pair of vignettes, respondents also rated their concern regarding jeopardy to their training status if the health issue were known to their residency training director or clinical supervisor (1=no concern, 5=some concern, 9=great concern). Residents rated their concern about jeopardy from hypertension as low (mean=1.79, SD=1.28) and panic attacks as moderate (mean=4.49, SD=2.38) and greater than that from hypertension (vignette main effect F=226.22, df=1, 147, p<0.0001, d=1.41). Residents in specialty areas indicated greater concern about jeopardy to training status than did residents in primary care areas across both health issues (means=3.44 and 2.85, SDs=1.57 and 1.53, respectively; residency area main effect F=26.36, df=1, 147, p<0.02, d=0.39).

Respondents also rated their likelihood of avoiding care at their training institution (1=no chance, 5=somewhat likely, 9=certain) for each health issue. Residents indicated that they were unlikely to avoid care at their institution for hypertension (mean=2.25, SD=1.75), somewhat likely to avoid care at their institution for panic attacks (mean=5.71, SD=2.39), and more likely to avoid care at their institution for panic attacks than for hypertension (vignette main effect F=359.73, df=1, 147, p<0.0001, d=1.65). Women were more likely to avoid care at their institution for panic attacks than men (means=6.25 and 5.18, SDs=2.36 and 2.42, respectively; d=0.51), but women and men did not differ in avoidance of care at their institution for hypertension (means=2.27 and 2.23, SDs=1.72 and 1.77, respectively; d=0.02; vignette × gender interaction F=7.94, df=1, 147, p<0.01).

In support of our hypothesis, residents indicated greater concern about training status and greater likelihood of avoiding care at their training institution for the situation perceived as more stigmatizing in this pair of vignettes.

Respondents showed greater diversity of opinion for the moderate stigma than for the low stigma situation for all three questions (SDs=2.22 to 2.39 for panic attacks versus 1.28 to 1.75 for hypertension, p<0.01 in all cases). For each vignette, greater perceived stigma was associated with greater concern about training status and greater likelihood of avoiding care at the training institution (r=0.57 to 0.71, mean r=0.62 for the hypertension vignette; r=0.54 to 0.69, mean r=0.61 for the panic attack vignette; p<0.0001 in all cases), again supporting our hypothesis.

Health Clinic Physician as Resident’s Clinical Supervisor (Table 4)

For the pair of vignettes presenting a resident seeing a health clinic physician who will also be his or her clinical supervisor, respondents rated the stigma they would feel with a health issue of either severe nausea or pregnancy as moderate (means=5.48 and 5.69, SDs=2.56 and 1.97, respectively; d=0.09). Residents in specialty and primary care areas perceived similar stigma from severe nausea (means=5.58 and, 5.79, SDs=1.99 and 1.94, respectively; d=0.09), but residents in specialty areas perceived greater stigma from pregnancy than did residents in primary care areas (means=5.83 and 5.14, SDs=2.58 and 2.52, respectively; d=0.30; vignette × residency area interaction F=5.65, df=1, 147, p<0.02).

For the same pair of vignettes, respondents also rated their concern with jeopardy to their training status if the health issue were known as moderate for both health issues (means=3.95 and 4.15, SDs=2.23 and 2.59, respectively; d=0.08). Residents in specialty areas indicated greater concern about jeopardy to training status than did residents in primary care areas (means=4.52 and 3.58, SDs=2.09 and 2.04, respectively; residency area main effect F=8.03, df=1, 147, p<0.01, d=0.46), particularly for pregnancy (means=4.88 and 3.42, SDs=2.61 and 2.55, respectively; d=0.61; vignette × residency area interaction F=6.67, df=1, 147, p<0.02). Respondents also indicated they would be somewhat likely to avoid care at their training institution for pregnancy and nausea (means=4.94 and 4.98, SDs=2.22 and 2.79, respectively; d=0.02). Women indicated a greater likelihood of avoiding care at their training institution than men (means=5.42 and 4.50, SDs=2.07 and 2.12, respectively; d=0.44; gender effect F=7.39, df=1, 147, p<0.01), particularly for pregnancy (means=5.55 and 4.34, SDs=2.75 and 2.82, respectively; d=0.48, p<0.02 by Fisher’s least significant difference). Residents in specialty areas were more likely to avoid care at their institutions for pregnancy than residents in primary care areas (means=5.50 and 4.38, SDs=2.82 and 2.75, respectively; d=0.44), but less likely than primary care residents to avoid care at their institutions for severe nausea (means=4.63 and 5.33, SDs=2.24 and 2.19, respectively; d=0.28; vignette × residency area interaction F=16.44, df=1, 147, p<0.0001). Consistent with our hypothesis, residents perceived moderate stigma for both situations and also indicated moderate concern about training status and moderate likelihood of avoiding care at their institution.

Respondents showed substantial diversity of opinion for all three questions in both moderately stigmatizing situations, more so for the pregnancy vignette than for the nausea vignette (SDs=2.56 to 2.79 versus 1.97 to 2.23, p<0.04 in all cases). For each vignette, greater perceived stigma was associated with greater concern about training status and greater likelihood of avoiding care at the training institution (r=0.58 to 0.66, mean r=0.61 for the nausea vignette; r=0.65 to 0.71, mean r=0.68 for the pregnancy vignette; all p<0.0001), again supporting our hypothesis.

Erratic Performance by a Resident Under the Care of an Attending Physician (Table 5)

For the pair of vignettes presenting a resident who has been followed closely by an attending physician and who is performing erratically, respondents rated the stigma they would feel with a health issue of severe diabetes as moderate (mean=4.81, SD=2.33) and rated severe alcohol abuse as high (mean=8.09, SD=1.28) and greater than diabetes (vignette main effect F=356.37, df=1, 147, p<0.0001, d=1.75). For the same pair of vignettes, respondents also rated their concern with jeopardy to their training status if the health issue were known to their residency training director or clinical supervisor as moderate for diabetes (mean=4.15, SD=2.42) and high for alcohol abuse (mean=7.78, SD=1.62) and greater for alcohol abuse than for diabetes (vignette main effect F=423.93, df=1, 147, p<0.0001, d=1.76). Women indicated greater concern with training status than men (means=6.33 and 5.60, SDs=1.71 and 1.76, respectively; gender main effect F=6.91, df=1, 147, p<0.01, d=0.42).

Respondents also rated their likelihood of avoiding care at their training institution as somewhat likely for diabetes (mean=4.01, SD=2.60), very likely for alcohol abuse (mean=7.51, SD=1.83), and more likely for alcohol abuse than for diabetes (vignette main effect F=287.43, df=1, 147, p<0.0001, d=1.56). Women indicated a greater likelihood of avoiding care at their training institution than men (means=6.09 and 5.42, SDs=1.81 and 1.86, respectively; d=0.36; gender effect F=5.12, df=1, 147, p<0.03), particularly for alcohol abuse (means=8.06 and 6.96, SDs=1.80 and 1.85, respectively; d=0.49, vignette × gender interaction F=4.37, df=1, 147, p<0.04). In support of our hypothesis, residents indicated greater concern about their training status and greater likelihood of avoiding care at their training institution for the situation perceived as more stigmatizing in this pair of vignettes.

Respondents showed greater diversity of opinion for the moderate stigma than for the high stigma situation for all three questions (SDs=2.33 to 2.60 for diabetes and 1.27 to 1.8 for alcohol abuse, p<0.0001 in all cases). For each vignette, greater perceived stigma was associated with greater concern about training status and greater likelihood of avoiding care at the training institution (r=0.58 to 0.76, mean r=0.66 for the diabetes vignette; r=0.44 to 0.68, mean r=0.53 for the alcohol abuse vignette; p<0.0001 in all cases), again supporting our hypothesis.

Concern With Confidentiality and Diversity of Opinion

For the four vignettes with moderate stigma ratings, responses to all three questions ranged across the entire scale, and standard deviations were consistently large, demonstrating substantial diversity of opinion among respondents. We examined responses to related items in other parts of the full survey, including general health, needing and seeking health care, importance of the availability of care outside the training institution, experiences in providing and receiving care, current stress level, and concern with confidentiality for relationships to responses to the vignette questions. A composite measure of five items related to concern with confidentiality (concern influencing the ability to get health care in the past year, having health concerns but not seeking professional attention, seeking care informally, avoiding care because a supervisor might find out, and concern about personal medical records being seen by others at the training institution) was positively correlated with rated stigma (r=0.24 to 0.45, mean r=0.34, p<0.01 in all cases), concern about training status (r=0.27 to 0.45, mean r=0.35, p<0.01 in all cases), and avoidance of care at the training institution (r=0.27 to 0.48, mean r=0.37, p<0.01 in all cases) for the four vignettes with intermediate stigma. For the lowest and highest stigma vignettes, confidentiality concern was also positively correlated with rated stigma (both r=0.28, p<0.01), concern about training status (r=0.25 and 0.29, p<0.01 in both cases), and avoidance of local care (r=0.27 and 0.32, p<0.01 in both cases).

The factors affecting residents’ concerns about their ability to seek care for health issues are multidimensional, spanning the areas of stigmatization of specific clinical issues, conflicting interests inherent in the resident’s role as patient and trainee, and resulting concerns about jeopardizing confidentiality and training status. We confirmed our hypotheses that residents would see some personal health problems as stigmatizing and potentially jeopardizing to their training status and that they would be more likely to avoid seeking care at their training institution for stigmatizing health problems. Although many supervising attendings would likely choose not to treat one of their residents due to the ethical issues involved, factors such as an attending not being a supervisor at the time resident treatment is needed and resident health care systems often not offering other treatment options may lead to attendings treating residents.

Impact of Stigma on Avoidance of Care

Residents from multiple specialties ranked vignettes according to perceived stigma, concern about how each scenario could affect status as a trainee, and the likelihood that they would avoid care at the training institution. One vignette was clearly viewed as having a high level of stigma (alcohol abuse), one as having low stigma (hypertension), and four as moderately stigmatizing (stress-related gastrointestinal symptoms, panic attacks, pregnancy, and diabetes with erratic behavior). The stigma associated with each played a powerful role in determining both concern about trainee status and avoidance of care (Figure 1). A clear pattern was observed correlating stigma with concern about training status, with the highest level of variance seen among the moderate-level stigma cases. A similar relationship was found between stigma and avoidance of care. Although we were unable to parse out any potential effect differences between the dual-role aspect of residency and stigma of the clinical presentation, one interpretation is that stigma drives the other concerns about dual roles. Another potentially influential factor was the dual role of the treating physician as either ward attending, consulting attending, or internal medicine attending in the department. Stigma overrode these subtle differences in the supervisory role of the treating physician. Because in all cases the treating physician was perceived as having a potential evaluative role over the resident, it is possible that concern regarding confidentiality and potential jeopardy to training status was pervasive.

Gender Differences in Perception of Stigma and Care Seeking

Historically, women have encountered discrimination and biases that make it more difficult for them to make career advancements, particularly to high-status positions (39, 40). Research suggests that women are often regarded as less capable than men in their jobs, a stereotype that is amplified when women become pregnant (39). Especially in higher level positions, the commitment of pregnant women to their jobs is often called into question, out of employers’ fears that women will leave their jobs or be less able to fulfill their duties (41). Women medical residents likely face similar concerns about job status and advancement. Consistent with this hypothesis, Roberts and colleagues (42) determined that female medical students worried more than their male counterparts about confidentiality in seeking health care.

For the vignettes in which stigma was rated as moderate or high, female residents were consistently more concerned than male residents in rating stigma, concern about jeopardy to training status, and avoidance of care within one’s training institution (Figure 2). However, women’s concern was only significant for the avoidance of care at one’s training institution.

Although one might expect women to have greater empathy to the stigma associated with pregnancy, in this study both men and women rated the stigma as moderate (with no significance in difference). However, women were more inclined to rate the likelihood of avoiding care at their institution higher. Our findings are consistent with prior research on women medical student trainees having greater concern regarding academic jeopardy if the dean’s office found out they had a health problem (43). These gender differences are open to multiple interpretations. It is possible that women are more sensitive to privacy and confidentiality issues than men and that women feel more professionally vulnerable. Additionally, women’s reproductive health care needs may lead to greater preferences for outside care.

Primary Care and Specialist Residents’ Views of Stigma and Care Seeking

Similar levels of concern about stigma were expressed by primary care and specialty residents, but those in specialty programs expressed more concern about jeopardizing training status. Stigma may operate more powerfully among residents in non-primary care specialties, particularly in those viewed as the most highly competitive. Ironically, further support for the idea that stigma more strongly influences specialty residents comes from psychiatry. Fogel and colleagues (44) examined the effect of stigma related to personal psychiatric treatment among psychiatric residents in New York in 2002. Eight-seven percent of 130 residents thought that personal psychotherapy had little or no stigma, while only 34% had the same view of medication. Over 60% of respondents believed that trainee use of psychotropic medication carried moderate to significant stigma. The authors suggest that stigma among psychiatric trainees may be even more powerful than their results show because most residents considered psychotherapy an educational benefit and the use of medication an indication of a real psychiatric illness. If such negative attitudes toward the diagnosis and treatment of mental illness are found among psychiatric residents, it is likely that our finding that specialist residents view help-seeking for mental health problems as more stigmatized than obtaining care for physical illness is well founded. The paucity of data on the question of specialist versus primary care trainees’ views of stigma invites further research and underscores the importance of such work for the well-being of house officers in both specialty and primary care fields.

Limitations

This study was conducted at only one institution, and thus possible regional differences in attitudes toward stigma, dual-role concerns, and self-care were unexplored. Also, the design of the paired vignettes may have inadvertently accentuated stigma differences, because confidentiality and dual-role differences were not as dramatic between the pairs. If the questionnaire had asked about the likelihood of seeking care outside the institution, dual-role concerns might have been more readily differentiated because stigma would have been stable for a given clinical vignette; the difference would have been dual role (patient and trainee at home institution) versus single role (patient only at an outside clinic). Finally, the nature of the questions focusing on stigma, concerns about training status, and avoiding health care may have excluded other potentially influential factors in residents’ decision making regarding seeking health care from the home institution.

Medicine as a culture needs to evolve toward appropriate acknowledgment of the humanity of physicians in order to cultivate integrity, self-reflection, the capacity to admit weakness and mistakes, striving for continuous improvement and learning, altruism, and dedication to service—attributes central to the profession.

Several practical recommendations flow from our findings: that training directors, faculty, and residents actively combat stigma; that program directors clarify ambiguities and provide appropriate reassurances about confidentiality; and that residents be provided with the choice of seeking care outside the training institution and insurance policies that allow for such care. Confidential care programs modeled after the Employee Assistance Program have been well received and may address many of the concerns identified in this study (45). A house officer mental health program developed at the University of Michigan offered a system of care separate from the university for residents. Twice as many residents utilized this service compared with the previous 10 years’ utilization at the university program.

One model for policy change that attempts to address concerns about stigma and the dual roles of medical students as patients is the 2005 Liaison Committee for Medical Education mandate for students not to receive mental health care by faculty who could ever be in an evaluative role for them. Similar guidelines for residents do not exist but warrant serious consideration.

* means differ at p<0.05

Drs. Moutier, Cornette, Lehrmann, Geppert, and Green Hammond and Ms. DeBoard declared no competing interests. Dr. Laura Roberts’ work is funded through the Research for a Healthier Tomorrow-Program Development Fund, a component of the Advancing a Healthier Wisconsin endowment at the Medical College of Wisconsin. Disclosures of editors are published in each January issue. Manuscripts authored by an editor ofor by a member of its editorial board undergo the same editorial review process, including blinded peer review, applied to all manuscripts. Additionally, the Editor is recused from any editorial decision-making.