The utility of the index to stratify the risk of Y4 depression symptoms developed in the discovery cohort was subsequently evaluated in the students in the replication cohort. Among students in the replication dataset, the prevalence of Y4 depression symptoms of the four risk groups was, respectively, 14.9%, 24.1%, 44.4%, and 64.2% (P < .0001; Table 4). The OR for risk associated with each incremental increase in risk category in the replication cohort was similar to that of each risk category in the discovery cohort. With respect to discrimination, c = 0.72.
Results from the multivariate logistic analysis using bootstrapping methods can be found in Supplemental Digital Table 1 at http://links.lww.com/ACADMED/A593. Supplemental Digital Table 2 shows the four risk categories, distribution of scores, and prevalence of Y4 depression symptoms. The proposed risk categories stratify the prevalence of Y4 depression rates from 20.2% (low risk) to 62.9% (very high risk) with c = 0.65 in the discovery cohort, and from 17.7% (low risk) to 58.5% (very high risk) with c = 0.67 in the replication cohort.
As mentioned, medical students with depression are at increased risk for adverse professional and personal consequences, including declines in academic performance, thoughts of dropping out of medical school, and suicide.1–3 We found a high prevalence (31.2%; 1,167/3,743) of depression symptoms among fourth-year medical students, consistent with the rate reported in the literature.4 For medical students in this prospective longitudinal cohort study, demographics (e.g., age, race, ethnicity), tuition, and depression symptoms, stress, coping behaviors, and social support at baseline were independently associated with the risk of developing depression symptoms at Y4. These findings are consistent with findings from previous cross-sectional and single-institution longitudinal studies.1 , 13 , 14 , 16 , 17 , 19 , 27 , 46–51 Information on most of these factors is available or obtainable for all students, and each factor has independent prognostic value. We determined how these factors can be combined into a Depression-PI that stratifies risk of developing depression symptoms by Y4.
We demonstrated in the discovery dataset—and further confirmed in the replication dataset—that the Depression-PI defines four prognostic subgroups at low, intermediate, high, and very high risk, therefore providing prognostic information regarding risk of depression symptoms by Y4 of medical school. When baseline depression symptoms were included in the Depression-PI, the c statistic was 0.71 and 0.72, signifying potential prognostic utility at the individual level (as this typically requires a minimum threshold level above 0.70).42 , 43 We identified high risk and very high risk groups of medical students who had four- to ninefold increased odds of demonstrating depression symptoms at Y4. These two highest-risk groups represented approximately a third of the students in the prospective longitudinal cohort.
Importantly, the Depression-PI has the potential to support medical school efforts to promote student wellness. Student wellness programming should be offered to all students. Doing so is an LCME requirement.9 More importantly, wellness initiatives help to fulfill the medical education community’s moral imperative to address medical student depression, especially given its high prevalence. A risk-stratified approach, however, allows delivery of core material to all students and additional support to those at greatest risk, an approach that could optimize support for students despite finite resources. Such tailoring does not undermine the need for or value of wellness programming offered to all students. For example, students in the intermediate risk group may benefit from added support, whereas students in the high risk groups may warrant allocation of further resources including individualized support, active screening, and early intervention. Implementing interventions in a way that protects confidentiality and avoids stigmatizing individuals in the higher-risk groups is and will continue to be vital. How best to engage students in a tailored primary prevention strategy, and whether doing so leads to less suffering and depression and to fewer adverse consequences, warrants study. These efforts should occur in conjunction with intentional steps to improve the work and learning environment and reduce drivers of distress. We do not recommend that the Depression-PI be used in any capacity to screen applicants or violate privacy.
To our knowledge, this is the largest longitudinal study to explore predictors of development of depression symptoms among U.S. medical students. Of the students in our cohort, those who were nonwhite and non-Hispanic/Latino were at increased odds of developing depression symptoms, even after controlling for baseline depression. The 2009–2012 National Health and Nutritional Examination Survey found non-Hispanic blacks to have higher rates of severe depression than non-Hispanic whites.52 Previous studies of medical students suggest that the mental health of medical students from racial and ethnic minorities may depend on whether or not their race or ethnicity has negatively affected their experience.53 Similarly, other studies have found that students from racial and ethnic minorities experience more microaggressions that they attribute to their race.54–56 Together, these studies suggest that the higher prevalence of depression symptoms among this subgroup of students is likely driven by factors within the learning environment rather than individual traits. Medical schools need to do more to improve the learning environment for nonwhite students.
Our findings—that negative coping behaviors and low social support increase the risk of subsequently developing depression symptoms—are congruent with findings from cross-sectional46–48 and one-year longitudinal studies of medical students.13 , 49–51 Independent of coping behavior, we also found that high stress levels at baseline predicted subsequent depression symptoms. Stress scores have also been found to be associated with global measures of distress in cross-sectional studies,57 and with depression13 , 14 and anxiety13 in one-year longitudinal studies of first-year medical students. In aggregate, these studies suggest that intervention studies focused on decreasing reliance on negative coping behaviors, teaching strategies to reduce stress, and promoting social support among medical students should be pursued.
Among the medical school factors we explored, only tuition was independently associated with increased risk of developing depression symptoms; that is, those who attended a medical school with middle-tertile tuition were at higher risk than students who attended a medical school with lower tuition costs. This finding persisted after adjusting for socioeconomic background. Importantly, we did not collect information on accrued undergraduate and medical school debt directly, nor on income from working while in medical school—both of which have been shown in previous studies of medical trainees to be associated with poorer mental health.50 , 58 , 59 Other studies suggest that grading structure,60–62 curricular structure and learning communities,63 and harassment/belittlement, as well as poor role-modeling behaviors by faculty,51 relate to medical student mental health.
Our study has a number of limitations. First, although the sample was drawn from a random stratified sample of students from 49 U.S. medical schools, we do not know how well the experience of participants represents the experience of U.S. medical students. We also detected some minor demographic differences between students who responded at both time points and students who responded only to the first survey. We did, however, have a robust response rate, and the demographic characteristics of participants were similar to those of all U.S. medical students who matriculated in the same year, suggesting that our results likely generalize to medical students overall.44 , 45 Second, we used the PROMIS depression short form 4a as our screening instrument for depression symptoms, and we applied a cutoff score of ≥ 8 (sensitivity 83.1%, specificity 84.3%).32 If we had used a higher cutoff score, our specificity would have improved but our sensitivity would have declined. Third, we depended on self-reported characteristics and included a limited set of individual and school characteristics. Multiple social, psychological, and biological factors contribute to depression,64 and we did not include concomitant anxiety disorders. Fourth, we assessed depression symptoms at two time points (Y1 and Y4) and know that depression symptoms may come and go at other points during the four-year interval. Fifth, we recognize that further validation of the Depression-PI in other datasets is warranted.
The rate of depression symptoms in this longitudinal, national cohort of U.S. medical students at the end of Y4 was 31.2%. Being older, nonwhite, or non-Hispanic/Latino; relying on negative coping behaviors; having high stress, low social support, and depression symptoms at Y1; and being at a medical school with middle-tertile tuition were strongly associated with an increased risk of developing depression symptoms during medical school. The Depression-PI is the first prognostic model to incorporate multiple Y1 factors to stratify the risk of Y4 depression symptoms among medical students. The Depression-PI score can identify students at highest risk for developing depression symptoms and may be useful to medical schools interested in developing a tiered approach to prevention and support for medical students in or at risk for distress.
1. Clark DC, Zeldow PB. Vicissitudes of depressed mood during four years of medical school. JAMA. 1988;260:2521–2528.
2. Dyrbye LN, Thomas MR, Power DV, et al. Burnout and serious thoughts of dropping out of medical school: A multi-institutional study. Acad Med. 2010;85:94–102.
3. Dyrbye LN, Thomas MR, Massie FS, et al. Burnout and suicidal ideation among U.S. medical students. Ann Intern Med. 2008;149:334–341.
4. Rotenstein LS, Ramos MA, Torre M, et al. Prevalence of depression, depressive symptoms, and suicidal ideation among medical students: A systematic review and meta-analysis. JAMA. 2016;316:2214–2236.
5. Brazeau CM, Shanafelt T, Durning SJ, et al. Distress among matriculating medical students relative to the general population. Acad Med. 2014;89:1520–1525.
6. Goebert D, Thompson D, Takeshita J, et al. Depressive symptoms in medical students and residents: A multischool study. Acad Med. 2009;84:236–241.
7. Dyrbye LN, Schwartz A, Downing SM, Szydlo DW, Sloan JA, Shanafelt TD. Efficacy of a brief screening tool to identify medical students in distress. Acad Med. 2011;86:907–914.
8. Moutier C, Norcross W, Jong P, et al. The suicide prevention and depression awareness program at the University of California, San Diego School of Medicine. Acad Med. 2012;87:320–326.
9. Liasion Committee on Medical Education; Association of American Medical Colleges; American Medical Association. Standards, publications, and notification forms. http://lcme.org/publications
. Updated March 2018. Accessed August 7, 2018.
10. Camp DL, Hollingsworth MA, Zaccaro DJ, Cariaga-Lo LD, Richards BF. Does a problem-based learning curriculum affect depression in medical students? Acad Med. 1994;69(10 suppl):S25–S27.
11. Vitaliano PP, Maiuro RD, Russo J, Mitchell ES, Carr JE, Van Citters RL. A biopsychosocial model of medical student distress. J Behav Med. 1988;11:311–331.
12. Ball S, Bax A. Self-care in medical education: Effectiveness of health-habits interventions for first-year medical students. Acad Med. 2002;77:911–917.
13. Vitaliano PP, Maiuro RD, Russo J, Mitchell ES. Medical student distress. A longitudinal study. J Nerv Ment Dis. 1989;177:70–76.
14. Richman JA, Flaherty JA. Gender differences in medical student distress: Contributions of prior socialization and current role-related stress. Soc Sci Med. 1990;30:777–787.
15. Pyskoty CE, Richman JA, Flaherty JA. Psychosocial assets and mental health of minority medical students. Acad Med. 1990;65:581–585.
16. Lloyd C, Gartrell NK. Sex differences in medical student mental health. Am J Psychiatry. 1981;138:1346–1351.
17. Zoccolillo M, Murphy GE, Wetzel RD. Depression among medical students. J Affect Disord. 1986;11:91–96.
18. Levine RE, Litwins SD, Frye AW. An evaluation of depressed mood in two classes of medical students. Acad Psychiatry. 2006;30:235–237.
19. Rosal MC, Ockene IS, Ockene JK, Barrett SV, Ma Y, Hebert JR. A longitudinal study of students’ depression at one medical school. Acad Med. 1997;72:542–546.
20. Hardeman RR, Burgess D, Phelan S, Yeazel M, Nelson D, van Ryn M. Medical student socio-demographic characteristics and attitudes toward patient centered care: Do race, socioeconomic status and gender matter? A report from the Medical Student CHANGES study. Patient Educ Couns. 2015;98:350–355.
21. Hardeman RR, Przedworski JM, Burke SE, et al. Mental well-being in first year medical students: A comparison by race and gender: A report from the Medical Student CHANGE study. J Racial Ethn Health Disparities. 2015;2:403–413.
22. Phelan SM, Burgess DJ, Puhl R, et al. The adverse effect of weight stigma on the well-being of medical students with overweight or obesity: Findings from a national survey. J Gen Intern Med. 2015;30:1251–1258.
23. Phelan SM, Dovidio JF, Puhl RM, et al. Implicit and explicit weight bias in a national sample of 4,732 medical students: The medical student CHANGES study. Obesity (Silver Spring). 2014;22:1201–1208.
24. Phelan SM, Puhl RM, Burke SE, et al. The mixed impact of medical school on medical students’ implicit and explicit weight bias. Med Educ. 2015;49:983–992.
25. van Ryn M, Hardeman R, Phelan SM, et al. Medical school experiences associated with change in implicit racial bias among 3547 students: A Medical Student CHANGES study report. J Gen Intern Med. 2015;30:1748–1756.
26. van Ryn M, Hardeman RR, Phelan SM, et al. Psychosocial predictors of attitudes toward physician empathy in clinical encounters among 4732 1st year medical students: A report from the CHANGES study. Patient Educ Couns. 2014;96:367–375.
27. Dyrbye LN, Thomas MR, Shanafelt TD. Systematic review of depression, anxiety, and other indicators of psychological distress among U.S. and Canadian medical students. Acad Med. 2006;81:354–373.
28. Dyrbye LN, Harper W, Moutier C, et al. A multi-institutional study exploring the impact of positive mental health on medical students’ professionalism in an era of high burnout. Acad Med. 2012;87:1024–1031.
29. Dyrbye L, Shanafelt T. A narrative review on burnout experienced by medical students and residents. Med Educ. 2016;50:132–149.
31. Pilkonis PA, Choi SW, Reise SP, Stover AM, Riley WT, Cella D; PROMIS Cooperative Group. Item banks for measuring emotional distress from the Patient-Reported Outcomes Measurement Information System (PROMIS®): Depression, anxiety, and anger. Assessment. 2011;18:263–283.
32. Kroenke K, Yu Z, Wu J, Kean J, Monahan PO. Operating characteristics of PROMIS four-item depression and anxiety scales in primary care patients with chronic pain. Pain Med. 2014;15:1892–1901.
33. Cohen S, Williamson GM. Oskamp S, Spacapan S. Perceived stress in a probability sample of the United States. In: The Social Psychology of Health. 1988:Newbury Park, CA: Sage; 31–67.
34. Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health Soc Behav. 1983;24:385–396.
35. Norman GR, Sloan JA, Wyrwich KW. The truly remarkable universality of half a standard deviation: Confirmation through another look. Expert Rev Pharmacoecon Outcomes Res. 2004;4:581–585.
36. Carver CS. You want to measure coping but your protocol’s too long: Consider the brief COPE. Int J Behav Med. 1997;4:92–100.
38. Sherbourne CD, Stewart AL. The MOS social support survey. Soc Sci Med. 1991;32:705–714.
42. Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: Issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15:361–387.
43. Hosmer DW, Lemeshow S. Applied Logistic Regression. 2000.2nd ed. New York, NY: John Wiley & Sons, Inc..
44. Przedworski JM, Dovidio JF, Hardeman RR, et al. A comparison of the mental health and well-being of sexual minority and heterosexual first-year medical students: A report from the Medical Student CHANGE study. Acad Med. 2015;90:652–659.
46. Mosley TH Jr, Perrin SG, Neral SM, Dubbert PM, Grothues CA, Pinto BM. Stress, coping, and well-being among third-year medical students. Acad Med. 1994;69:765–767.
47. Vitaliano PP, Maiuro RD, Russo J, Becker J. Raw versus relative scores in the assessment of coping strategies. J Behav Med. 1987;10:1–18.
48. Vitaliano PP, Russo J, Carr JE, Heerwagen JH. Medical school pressures and their relationship to anxiety. J Nerv Ment Dis. 1984;172:730–736.
49. Vitaliano PP, Maiuro RD, Mitchell E, Russo J. Perceived stress in medical school: Resistors, persistors, adaptors and maladaptors. Soc Sci Med. 1989;28:1321–1329.
50. Dyrbye LN, Power DV, Massie FS, et al. Factors associated with resilience to and recovery from burnout: A prospective, multi-institutional study of US medical students. Med Educ. 2010;44:1016–1026.
51. Haglund ME, aan het Rot M, Cooper NS, et al. Resilience in the third year of medical school: A prospective study of the associations between stressful events occurring during clinical rotations and student well-being. Acad Med. 2009;84:258–268.
53. Dyrbye LN, Thomas MR, Eacker A, et al. Race, ethnicity, and medical student well-being in the United States. Arch Intern Med. 2007;167:2103–2109.
54. Montenegro RE. A piece of my mind. My name is not “interpreter.” JAMA. 2016;315:2071–2072.
55. Ulloa JG, Talamantes E, Moreno G. Microaggressions during medical training. JAMA. 2016;316:1113–1114.
56. Beagan BL. “Is this worth getting into a big fuss over?” Everyday racism in medical school. Med Educ. 2003;37:852–860.
57. Notman MT, Salt P, Nadelson CC. Stress and adaptation in medical students: Who is most vulnerable? Compr Psychiatry. 1984;25:355–366.
58. West CP, Shanafelt TD, Kolars JC. Quality of life, burnout, educational debt, and medical knowledge among internal medicine residents. JAMA. 2011;306:952–960.
59. Jackson ER, Shanafelt TD, Hasan O, Satele DV, Dyrbye LN. Burnout and alcohol abuse/dependence among U.S. medical students. Acad Med. 2016;91:1251–1256.
60. Reed DA, Shanafelt TD, Satele DW, et al. Relationship of pass/fail grading and curriculum structure with well-being among preclinical medical students: A multi-institutional study. Acad Med. 2011;86:1367–1373.
61. Rohe DE, Barrier PA, Clark MM, Cook DA, Vickers KS, Decker PA. The benefits of pass–fail grading on stress, mood, and group cohesion in medical students. Mayo Clin Proc. 2006;81:1443–1448.
62. Bloodgood RA, Short JG, Jackson JM, Martindale JR. A change to pass/fail grading in the first two years at one medical school results in improved psychological well-being. Acad Med. 2009;84:655–662.
63. Slavin SJ, Schindler DL, Chibnall JT. Medical student mental health 3.0: Improving student wellness through curricular changes. Acad Med. 2014;89:573–577.