aThe total responses to each stigma item including “neither agree nor disagree” responses.bResults for women vs men by 2 × 2 Fisher exact test.
aThe total responses to each stigma item including “neither agree nor disagree” responses.bResults for first- and second-year vs third- and fourth-year by 2 × 2 Fisher exact test.
Schwenk TL, Davis L, Wimsatt LA. Depression, Stigma, and Suicidal Ideation in Medical Students. JAMA. 2010;304(11):1181-1190. doi:10.1001/jama.2010.1300
Author Affiliations: Departments of Family Medicine (Drs Schwenk and Wimsatt), Medical Education (Dr Schwenk), the Depression Center (Dr Schwenk), and Medical School (Ms Davis), University of Michigan, Ann Arbor.
Context There is a concerning prevalence of depression and suicidal ideation among medical students, a group that may experience poor mental health care due to stigmatization.
Objective To characterize the perceptions of depressed and nondepressed medical students regarding stigma associated with depression.
Design, Setting, and Participants Cross-sectional Web-based survey conducted in September-November 2009 among all students enrolled at the University of Michigan Medical School (N = 769).
Main Outcome Measures Prevalence of self-reported moderate to severe depression and suicidal ideation and the association of stigma perceptions with clinical and demographic variables.
Results Survey response rate was 65.7% (505 of 769). Prevalence of moderate to severe depression was 14.3% (95% confidence interval [CI], 11.3%-17.3%). Women were more likely than men to have moderate to severe depression (18.0% vs 9.0%; 95% CI for difference, −14.8% to −3.1%; P = .001). Third- and fourth-year students were more likely than first- and second-year students to report suicidal ideation (7.9% vs 1.4%; 95% CI for difference, 2.7%-10.3%; P = .001). Students with moderate to severe depression, compared with no to minimal depression, more frequently agreed that “if I were depressed, fellow medical students would respect my opinions less” (56.0% vs 23.7%; 95% CI for difference, 17.3%-47.3%; P < .001), and that faculty members would view them as being unable to handle their responsibilities (83.1% vs 55.1%; 95% CI for difference, 16.1%-39.8%; P < .001). Men agreed more commonly than women that depressed students could endanger patients (36.3% vs 20.1%; 95% CI for difference, 6.1%-26.3%; P = .002). First- and second-year students more frequently agreed than third- and fourth-year students that seeking help for depression would make them feel less intelligent (34.1% vs 22.9%; 95% CI for difference, 2.3%-20.1%; P < .01).
Conclusions Depressed medical students more frequently endorsed several depression stigma attitudes than nondepressed students. Stigma perceptions also differed by sex and class year.
Medical students experience depression, burnout, and mental illness at a higher rate than the general population, with mental health deteriorating over the course of medical training.1- 6 Medical students have a higher risk of suicidal ideation7 and suicide,8 higher rates of burnout,6,9 and a lower quality of life than age-matched populations.5,10 Burnout and depressive symptoms have been associated with suicidal ideation.4,6,9,10 Medical students are less likely than the general population to receive appropriate treatment despite seemingly better access to care.11- 13 Students may engage in potentially harmful methods of coping, such as excessive alcohol consumption, and, despite their training, may fail to recognize that depression is a significant illness that requires treatment.11
Stigma associated with depression and the use of mental health care services may represent a barrier to seeking treatment.2,12- 14 One study identified stigma as an explicit barrier to the use of mental health services by 30% of first- and second-year medical students experiencing depression. In addition, 37% identified lack of confidentiality and 24% cited fear of documentation in their academic record as barriers to treatment.2 Students may worry that revealing their depression will make them less competitive for residency training positions or compromise their education,2,12,13 and physicians may be reluctant to disclose their diagnosis on licensure and medical staff applications.15,16 The fear of professional sanctions may lead to inappropriate and possibly dangerous approaches to seeking care such as self-prescription of antidepressants.17 No studies to our knowledge have addressed in more specific detail the perceptions of stigma by depressed medical students that may serve as barriers to receiving appropriate mental health care.
We conducted a study of medical students at the University of Michigan Medical School to assess the prevalence of self-reported depression and suicidal ideation and to assess the perceptions of depression stigma by both depressed and nondepressed students.
We conducted a cross-sectional survey from September-November 2009 of all 769 medical students enrolled at the University of Michigan Medical School, a large, research-intensive public medical school. The medical school curriculum emphasizes basic science teaching in the first year with early clinical exposure, clinical and basic science correlations in the second year, required clinical rotations in the third year, and required and elective clinical and basic science rotations in the fourth year. The study was approved by the University of Michigan institutional review board. As part of written informed consent, students responding to the survey were presented with a full description of the study, the potential harm (no more than minimal) of the survey, their rights as potential participants, their right to decline participation without prejudice, and the opportunity to continue or terminate study participation. Participants received no financial compensation, but the private survey research firm (see below) conducted a lottery among all participants, in which a single pair of airline tickets was awarded.
Focus group discussions with students informed survey content development, item selection, and survey implementation. Participants were presented with a standardized oral script that described the nature of the proposed study and their voluntary participation. The focus groups included a convenience sample of students not specifically identified as depressed, and addressed general experiences and opinions about depression, views on health care seeking, career-related concerns about stigma and depression, and guidance about achieving optimal survey participation. The survey was pilot-tested and further modified using a convenience group of recent medical school graduates to avoid contaminating the intended study sample.
An independent survey research firm was used to administer the Web-based survey and manage data collection. The identity of survey respondents was confidential to the survey research firm, which followed approved protocols for security and confidentiality related to locked offices and computers and password-protected files. Only de-identified data were delivered to the investigators to preserve participant anonymity. These procedures were emphasized to potential participants during face-to-face presentations, in e-mail and letter invitations, and in introductory comments on the survey.
Student participation was solicited primarily via an e-mail invitation to all students, plus printed invitations to first- and second-year students (the only students who have school mail boxes). Students were sent up to 4 e-mail reminders that ceased with completion of the survey or their declination of participation. Announcements encouraging participation were also made in several classes of first-, second-, and third-year students, and printed posters were widely distributed in medical center conference rooms and inpatient team rooms where third- and fourth-year students frequently meet.
Because the survey responses were entirely anonymous to the investigators, there was no mechanism for providing specific follow-up or referral to mental health resources for students who reported high levels of depression or recent suicidal ideation because the private survey research firm would not be properly trained to do this. However, existing medical school resources (accessible by telephone or Internet) were clearly made available to all participants in the invitations to participate, the informed consent process, and when respondents logged out of the survey. Medical students have access to mental health services through at least 3 resources on campus: a confidential medical student counseling service provided by the Department of Psychiatry, counseling and psychological services provided for all University of Michigan students, and the psychiatric emergency services. All medical students are required to have health insurance including coverage for mental health care, which may provide at least some reimbursement for counseling sessions. Each medical school class has an assigned counselor who provides initial counseling and triage services to direct students to appropriate resources. There is an anonymous blog for students to share their stresses and support with fellow students.
The survey instrument (eAppendix) included the Patient Health Questionnaire (PHQ-9, a validated self-report depression questionnaire18) plus questions about past and current diagnosis and treatment of depression, stress and coping in medical school, attitudes about mental illness and mental health care seeking, and demographic features. Race was classified by self-report using standard federally approved categories and was included because studies have shown racial or ethnic differences in prevalence of depression in medical students.4,12 Students were asked whether they had seriously considered committing suicide since enrolling in medical school and whether they considered dropping out of medical school within the past month.
The construction and scoring of the PHQ-9 has been described elsewhere.18 We used validated cut-off scores of 0 to 4 for no or minimal depression, 5 to 9 for mild depression, 10 to 14 for moderate depression, 15 to 19 for moderately severe depression, and 20 to 27 for severe depression.18 The PHQ-9 has been validated in a wide range of patient populations and correlates well with clinician assessments of depression using standard Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) (DSM-IV) criteria.19- 22
The survey included 27 statements regarding stigma and stigmatizing experiences related to depression with Likert-style responses. Some items, particularly those identified in prior studies,2,12,13 were included with both positive and negative framing as an internal consistency check. These statements were otherwise drawn from the existing literature on stigma in the general population, including several validated instruments,2,12,13,23- 28 and selected for their potential applicability to medical students. Stigma items considered for selection were drawn from generally accepted categories of stigmatization including self-perceived stigma, public stigma, and stigma related to being treated for depression. A total of 35 potential items were initially reviewed and reduced to 27 items based on focus group feedback to target items most pertinent to medical student experiences. Stigma items were adapted to the specific roles and professional responsibilities of medical students. For example, the statement “I believe that my friends would think less of me if they knew I was depressed” was changed to “I believe my fellow medical students would think less of me if they knew I was depressed.”
Frequencies and summary statistics were calculated on all variables of interest. The PHQ-9 scores were collapsed into 3 categories: 0 to 4 (no to minimal depression), 5 to 9 (mild depression), and 10 to 27 (moderate to severe depression). This categorization allowed us to compare responses of students who would most likely meet standardized criteria for major depressive disorder (likelihood ratio ≥7.1) compared with those of the remaining students.18- 22 The analysis of stigma responses involved dropping the neutral category of responses and recoding the stigma items as dichotomous variables (strongly disagree/disagree vs strongly agree/agree) based on standard analytical schemes for Likert scales. Given the explicit and sometimes provocative or controversial nature of the stigma items, we chose to focus on clear opinions of agreement or disagreement rather than diluting the analysis by, for example, using a simple mean of all responses that would give undue influence on the results to respondents who had no opinion.29
Class year was coded as a dichotomous variable in order to compare results for students in preclinical vs clinical education. Using χ2 analyses and Fisher exact tests as appropriate, we compared depression severity, perceptions of stigma, and suicidal ideation by several demographic variables and by depression diagnosis and treatment history. The χ2 tests investigated associations between stigma item responses and depression severity. Multiple comparisons with Bonferroni post hoc tests were used to limit Type I errors. With the experimental error rate set at 0.05, the individual error rate was reduced to 0.002 (0.05 divided by 27 stigma items). Quantitative analyses were performed using SPSS 17.0 statistical software package (SPSS Inc, Chicago, Illinois).
A formal prospective power analysis could not be made because there were no prior studies or data to indicate the likely pattern of responses and potential differences between groups. Based on general experience in the behavioral sciences, we estimated that a proportion difference of 15% would be of moderate size and worthy of further research and exploration.30 In order to find a proportion difference of 15% or more to be statistically significant with 80% power, with the smaller proportion ranging between 5% and 50%, we needed between 70 and 167 participants per group, corresponding to a response rate of approximately 14% to 44%. For 2-group comparisons across years in medical school, assuming an equal distribution across years, a 15% effect size would be detected with 80% power with the number of participants ranging between 36 and 114, corresponding to a response rate of approximately 19% to 59%.
The response rate for the PHQ-9 items was 65.7% (505 of 769 usable responses). The response rate for first- and second-year students was higher than that for third- and fourth-year students (78.6% and 82.9% vs 50.2% and 56.9%, respectively). The response rate for all stigma items was essentially the same as that for the PHQ-9 with minor exceptions (502-505 respondents, 65.3%-65.7%). Women comprise 49.3% of University of Michigan medical students and were overrepresented as participants, with the total survey response rate of women (299 of 379, 78.9%) higher than that of men (213 of 390, 54.6%; P = .001). Women were also overrepresented compared with all US medical students. In 2006-2007, women represented 48.3% of US medical student matriculants31 but 58.4% of respondents (P < .001). Demographic characteristics of respondents are shown in Table 1.
A total of 72 students (14.3%; 95% confidence interval [CI], 11.3%-17.3%) scored in the moderate to severe depression range on the PHQ-9 (Table 2). First- and second-year medical students were no more likely than third- and fourth-year students to report moderate to severe depression (13.4% [37 of 277] vs 15.4% [35 of 227]; 95% CI for difference, −8.2% to 4.1%; P = .14). A significantly greater percentage of women than men scored in the moderate to severe range (18.0% vs 9.0%; 95% CI for difference, −14.8 to −3.1%; P < .001).
Twenty-two students (4.4%; 95% CI 2.6%-6.1%) reported suicidal ideation at some point during medical school, with the proportion of moderate to severely depressed participants varying significantly by level of suicidal ideation (χ2 = 13.88; 95% CI for difference, 7.1%-48.6%; P = .001; Table 2). First- and second-year students less frequently reported suicidal ideation than did third- and fourth-year students (1.4% [4 of 277] vs 7.9% [18 of 227]; 95% CI for difference, 2.7%-10.3%; P = .001). The largest difference between consecutive years in the percentage of students reporting suicidal ideation occurred from the second to the third year. Suicidal ideation showed a nonsignificant difference by sex (women, 5.1% [15 of 295] vs men, 3.3% [7 of 210]; 95% CI for difference, −1.7 to 5.2%; P = .39). Thoughts of dropping out of school differed significantly across all levels of depression and suicidal ideation. The frequency with which students considered dropping out of school was nearly 8 times greater in students with moderate to severe depression (31 of 72, 43.1%) than in students with no to minimal depression (16 of 284, 5.6%). Of 77 students who reported considering dropping out of school during the past month, 8 (10.4%) reported suicidal ideation, whereas only 3.3% (14 of 428) of those who did not contemplate dropping out of school reported suicidal ideation (95% CI for difference, −14.1% to −0.1%; P = .01).
Depression scores differed significantly according to depression diagnosis and treatment history (P < .001 for most comparisons; Table 3), with higher current depression scores associated with a current or past diagnosis or treatment. However, approximately 70% to 80% of students with moderate to severe depression by PHQ-9 score reported no history of depression diagnosis or treatment. Furthermore, 77.8% of those with moderate to severe depression scores (n = 56) reported having felt seriously depressed even if not formally diagnosed. The frequency with which students reported ever feeling depressed, even if not diagnosed, varied significantly according to depression level (95% CI for difference, 11.9%-24.2%; P < .001), with the frequency of moderate to severe depression 4 times higher among students who endorsed having felt depressed even if not diagnosed than among those who reported not feeling depressed (23.9% [56 of 234] vs 5.9% [16 of 271]). Similarly, of students who had considered suicide, almost all (20 of 22, 90.9%) reported having felt seriously depressed but most had never been diagnosed or treated for depression.
Stigma perceptions varied by depression score (Table 4). Of the 27 stigma items assessed, 9 items showed significant differences by PHQ-9 score after Bonferroni adjustment. Students with higher depression scores felt more strongly than did those with no to minimal depression that telling a counselor would be risky (53.3% vs 16.7%; 95% CI for difference, 23.2%-50.1%) and that asking for help would mean the student's coping skills were inadequate (61.7% vs 33.5%; 95% CI for difference, 14.4%-42.0%). Those with moderate to severe depression scores also agreed more strongly that, if depressed, others would find them unable to handle medical school responsibilities (83.1% vs 55.1%; 95% CI for difference, 16.1%-39.8%). Medical students with moderate to severe depression scores more frequently reported feeling that, if depressed, fellow medical students would respect their opinions less than did those with no to minimal depression (56.0% vs 23.7%; 95% CI for difference, 17.3%-47.3%). (P < .001 for all preceding comparisons.)
Male and female medical students responded differently to 4 stigma items (Figure 1). A significantly larger percentage of male students agreed that most medical students would not want to work with a depressed medical student (49.0% vs 29.4%; 95% CI for difference, 9.51%-29.7%; P < .001) and that depressed medical students are viewed as dangerous to their patients (36.3% vs 20.1; 95% CI for difference, 6.1%-26.3%; P = .002). Perceptions also varied according to class year for 7 stigma items, with first- and second-year students more frequently agreeing than third- and fourth-year students that depressed medical students would provide inferior patient care (79.3% vs 66.9%; 95% CI for difference, 3.3%-21.5%; P = .007), and that seeking help for depression would make students feel that they were less intelligent than other students (34.1% vs 22.9%; 95% CI for difference, 2.3%-20.1%; P = .01; Figure 2).
This study provides detailed characterization of the stigma perceived by medical students reporting depression and the differences between depressed and nondepressed students in their beliefs about the stigma of depression. Compared with students with low self-identified depression, students with high scores more frequently agreed that the opinions of depressed medical students would be less respected, that the coping skills of depressed medical students would be viewed as less adequate, that they would be viewed as less able to handle their responsibilities by faculty members, and that telling a counselor about depression would be risky. Students with high scores would also be less likely to seek treatment if depressed than would students with low scores.
These data could reflect the cognitive distortion known to occur in patients with depression,32 such that depressed students could have an inaccurate and excessively negative view of how they are viewed by other students. The data could also indicate an accurate perception by depressed students that they are, in fact, viewed as less capable. The findings may reflect a medical school environment in which depressed students are stigmatized because of their disease rather than on the basis of performance. In such an environment, revealing depression to friends, faculty members, and residency program directors could have real and adverse consequences.
The prevalence of self-identified depression in University of Michigan medical students is consistent with that found in several prior studies, approximately 10% to 25% depending on severity and the specific instruments used.4,5,10- 12 Most students with high depression scores or who had thoughts about suicide did not report a current or past diagnosis or treatment of depression. However, the self-perception of previous depressive episodes, even if not formally diagnosed, was significantly associated with both high depression scores and the prevalence of suicidal thinking. These results suggest the importance of developing a medical school culture in which medical students have the opportunity to discuss their mental health concerns, irrespective of actual diagnosis or treatment, in a safe and confidential way. Where this discussion might best occur is unclear because potential stigma is seen as coming from several sources, including other students, faculty members, and counselors. Many medical schools have small-group settings led by faculty mentors, but approaching these issues in such a venue may have risks and unintended consequences that would need to be explored before implementation.
The prevalence of depressive symptoms is significantly higher in female than in male medical students, consistent with previous studies of medical students and physicians.3- 5,10,11 The risk of suicidal ideation was also higher in female students, although not reaching statistical significance. These findings are consistent with the known increased risk of suicidal ideation as well as suicide completion in female physicians.15 When combined with the finding that men were more likely than women to agree that depressed medical students may be dangerous in their patient care and are undesirable members of the medical care team, these results suggest potential directions for further study regarding sex differences in how medical students experience their educational environment.
There are also differences between first- and second-year (preclinical) and third- and fourth-year (clinical) medical students in their views of depression, with preclinical students more likely to endorse that depressed medical students would provide inferior care to their patients, are unable to cope with medical school stress, and are less intelligent than their peers. These results could reflect the anticipatory anxiety experienced by preclinical students as they look ahead to the clinical years or could suggest that medical students may become more accepting and supportive of depressed students as they become more clinically knowledgeable and experienced. Educational, preventive, and clinical interventions may need to be framed differently for preclinical than for clinical students.
These results suggest that new approaches may be needed to reduce the stigma of depression and to enhance its prevention, detection, and treatment. The characteristics of medical education emphasizing professional competence and outstanding performance might be explored as reinforcing, rather than potentially sabotaging, factors in the creation of a culture that promotes professional mental health. The effective care of mental illness, the maintenance of mental health and effective emotional function, and the care of professional colleagues with mental illness could be taught as part of the ethical and professional responsibilities of the outstanding physician and become a critical component of the teaching, role modeling, and professional guidance that medical students receive as part of their curriculum in professionalism.
Strengths of this study include the extension of previous work on general sources of depression stigma, with more detailed descriptions of specific stigma perceptions and the comparison of responses to stigma items by depressed and nondepressed medical students. As with any self-report cross-sectional survey of a complex and sensitive issue like depression, there are also limitations to this study. First, the results represent the views of a single medical school student population, although we are aware of no reason why University of Michigan students would be different from those at other medical schools with regard to their perceptions of depression.
Second, bias could have been introduced due to missing respondents. The relatively good response rate, based on the entire medical student sample, is somewhat reassuring. The response rate for preclinical students (80.8%) was higher than that for clinical students (53.6%), which may reflect their greater opportunity and time flexibility for responding, but also reduces the generalizability of results for clinical students. It is possible that depressed students chose not to respond because of the sensitivity of the topic. However, it is also possible that depressed students would be more willing to respond because of concerns about their medical school experience. Neither of these factors should affect the attitudes of those students who did respond.
Third, student concerns about the confidentiality of a study conducted by members of their faculty could have influenced the students' responses. However, focus group participants perceived this concern to be well-addressed by the use of a private survey research firm as a firewall to render the responses anonymous to the investigators.
In conclusion, depressed medical students more frequently endorsed several depression stigma attitudes than did nondepressed students. Stigma perceptions also differed by sex and class year.
Corresponding Authors: Thomas L. Schwenk, MD, Department of Family Medicine, 1500 E Medical Center Dr, L2003 Women's Hospital, SPC 5239, Ann Arbor, MI 48109-5239 (firstname.lastname@example.org).
Author Contributions: Drs Schwenk and Wimsatt had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Schwenk, Wimsatt, Davis.
Acquisition of data: Schwenk, Wimsatt, Davis.
Analysis and interpretation of data: Schwenk, Wimsatt.
Drafting of the manuscript: Schwenk, Wimsatt.
Critical revision of the manuscript for important intellectual content: Schwenk, Wimsatt, Davis.
Statistical analysis: Wimsatt.
Obtained funding: Schwenk.
Administrative, technical, or material support: Schwenk, Wimsatt, Davis.
Study supervision: Schwenk.
Financial Disclosures: None reported.
Funding/Support: The Department of Family Medicine, University of Michigan.
Role of the Sponsor: The sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.
Previous Presentations: A limited set of preliminary data and analysis was presented at a Colloquium of the Depression Center at the University of Michigan on February 5, 2010.
Additional Contributions: Guidance in study design and concept was provided by Daniel Eisenberg, PhD; assistance in data analysis by Ananda Sen, PhD; critical review of the manuscript by Katherine T. Gold, MD; and logistical and philosophical support by James O. Woolliscroft, MD. All are affiliated with the University of Michigan, and none received compensation for their assistance.