Dyrbye, Liselotte N. MD, MHPE; Thomas, Matthew R. MD; Power, David V. MD, MPH; Durning, Steven MD; Moutier, Christine MD; Massie, F Stanford Jr MD; Harper, William MD; Eacker, Anne MD; Szydlo, Daniel W.; Sloan, Jeff A. PhD; Shanafelt, Tait D. MD
In response to the projected 55,000 to 85,000 shortfall of U.S. physicians expected by the year 2020, the Association of American Medical Colleges (AAMC) has called for a 30% increase in medical school enrollment.1 To meet this demand, medical schools are increasing their numbers of matriculants, expanding existing campuses, and building new clinical alliances. Although the graduation of every matriculant is critical to ensuring an adequate physician workforce, a paucity of information is available regarding medical student attrition. The limited existing data suggest that approximately 3% of matriculating students do not graduate from medical school.2 At a rate of 3%, 533 of the 17,759 matriculants to U.S. medical schools in 20073 will fail to graduate, which is equal to four graduating classes from the average-sized U.S. medical school.4 The personal, institutional, and societal financial costs of attrition are staggering. For students, the cost of attendance can reach $68,000 per year. In 2007–2008, the average annual in-state tuition, fees, and health insurance for public medical schools totaled $22,261 per year; this total equaled $39,608 for private schools.5 Institutions, themselves supported by state funds and faculty clinical practice,6 spend an estimated $70,000 to $90,000 per student per year.7
Although approximately half of the students who drop out of U.S. medical school before graduation do so for nonacademic reasons,2 we found only four publications in MEDLINE that described nonacademic reasons for attrition among students attending U.S. medical schools.8–11 Two of these publications mentioned “personal reasons related to family and health”8,9; one separated family and personal reasons11; two included change in career interests9,11; and one demonstrated that high scores on the Beck Depression Inventory were associated with attrition.10
In light of the high prevalence of psychological distress among U.S. medical students,12–17 we hypothesized that distressed students (defined here as having symptoms of burnout, depression, and/or low quality of life [QOL]) were more likely to seriously contemplate dropping out of medical school than their nondistressed peers. We also hypothesized that experiencing a stressful negative personal life event within the previous year would increase the likelihood of serious thoughts of dropping out. To test our hypotheses, we performed a prospective, multicenter study (1) to explore the severity of thoughts of dropping out among U.S. medical students and (2) to identify whether distress, burnout, and negative personal life events can predict subsequent serious thoughts of dropping out.
We have described the methods of data collection for this large, multisite study in detail elsewhere.18 Briefly, after receiving IRB approval from each participating institution, we invited all medical students at the Mayo Medical School (Rochester, Minnesota), University of Washington School of Medicine (Seattle, Washington), University of Chicago Pritzker School of Medicine (Chicago, Illinois), University of Minnesota Medical School (Minneapolis and Duluth campuses), and University of Alabama School of Medicine (Birmingham, Alabama) to complete Web-based surveys in the spring of 2006 (for the baseline) and in the spring of 2007 (as a one-year follow-up). Students at the University of California San Diego School of Medicine (San Diego, California) and the Uniformed Services University of the Health Sciences (Bethesda, Maryland) also participated in the 2007 survey. Participation was elective, and responses were anonymized. Students did not receive any incentives to participate. Nonresponders received up to three reminders during a one-month period of time. We included all students who responded to the 2007 survey in the cross-sectional analysis, and we included those who responded in both 2006 and 2007 in the prospective cohort study analysis.
The survey included the Maslach Burnout Inventory (MBI),19 the Primary Care Evaluation of Mental Disorders (PRIME MD),20 and the Medical Outcomes Study Short Form (SF-8)21 to identify, respectively, burnout, symptoms of depression, and QOL. The MBI is a 22-item instrument considered the gold standard for measuring burnout.19,22,23 This instrument has separate subscales to evaluate each domain of burnout: emotional exhaustion (EE), depersonalization (DP), and low sense of personal accomplishment (PA). We considered students with a high score for medical professionals on either the DP and/or EE subscales as having at least one manifestation of professional burnout.19,23 We used standard categorical thresholds to categorize domain scores into low, intermediate, and high categories.19 The PRIME MD is a two-item tool to screen for symptoms of depression. Screening positive for symptoms of depression is defined as answering “yes” to at least one of the two questions: “During the past month, have you often been bothered by feeling down, depressed, or hopeless?” and “During the past month, have you often been bothered by little interest or pleasure in doing things?” The PRIME MD has a sensitivity of 86% to 96% and a specificity of 57% to 75% for major depressive disorder.20,24 The SF-8 is a shorter, eight-item, alternate form of the SF-36. Previous research has demonstrated acceptable reliability.21 Several studies have demonstrated content, construct, and criterion-related validity for the SF-8.21,25 When scoring the SF-8, norm-based scoring methods are used to calculate mental and physical QOL summary scores.21 The average mental and physical QOL summary scores for the U.S. population are, respectively, 49.2 ± 9.46 and 49.2 ± 9.07.21 The SF-8 has been used previously in samples of residents26 and medical students.13,27
Using an iterative process, two authors (L.N.D. and T.D.S.) developed study-specific questions to explore students' thoughts of dropping out of medical school and the severity of those thoughts. Survey experts in the Mayo Survey Research Center reviewed these questions, and all authors approved them, verifying that the items were clear and that they adequately evaluated students' intent to drop out of medical school. First, we asked students if they had had any thoughts of dropping out of medical school in the past 12 months. Next, we asked students who responded “yes” to this question to indicate the seriousness of these thoughts using the following options:
* not seriously (i.e., “I thought about it but not seriously”),
* somewhat seriously (i.e., “I seriously considered dropping out but never took any actions”),
* seriously (i.e., “I seriously considered dropping out, and I met with officials at my school to discuss my options”),
* very seriously (i.e., “I seriously considered dropping out and actually took time off from medical school to consider options”), and
* extremely seriously (i.e., “I am dropping out of medical school and am in the final phase of this process”).
Students who rated their thoughts of dropping out as somewhat serious, serious, very serious, or extremely serious were considered a priori to have had serious thoughts of dropping out.
Our survey also included questions about race and ethnicity as the AAMC's Student Record System suggests that people who identify as American Indian/Alaska Natives, black/African American, and Hispanic/Latino have higher rates of attrition than people who identify as white (10-year completion rates of 90%–94.1% versus 96%).2 Additional questions asked about recent life events previously shown to affect student distress.13 We asked students if they had personally experienced any of the following life events within the prior year: marriage, divorce, the birth or adoption of a child, a major illness, a major illness of a significant other or close family member, or the death of a close family member. We considered personal illness and illness in a close family member or significant other as negative life events. For the remainder of the life events, we asked students to indicate whether they viewed the event as a positive, negative, or neutral event in their life.
All seven schools also provided information on the total number of students enrolled and the number of students who had dropped out for each of the five years prior to the survey administration (i.e., academic years 2002–2006).
The primary analysis involved descriptive summary statistics (means, proportions) for (1) estimating the prevalence of serious thoughts of dropping out, burnout, and positive screens for depression, (2) calculating mental and physical QOL scores, and (3) determining the actual attrition rate. We evaluated differences between those who reported serious thoughts of dropping out within the previous year versus those who did not using the Wilcoxon-Mann-Whitney test (for continuous variables) or the Fisher exact test (for categorical variables). We used the Wilcoxon rank sum test rather than parametric tests to account for the interval-level nature of the psychological tests. All tests were two-sided with comparison-wise type I error rates of .05. We employed forward stepwise logistic regression modeling to evaluate associations of the independent variables with serious thoughts of dropping out. We did not include race in the model because there were only 149 (17.4%) non-Caucasians in our prospective cohort sample. We performed collinearity testing before we initiated the stepwise process to determine whether multiple-way collinearity existed among the independent variables.28 No variables had achieved a level of collinearity that would bias the modeling process, so we did not remove any independent variables. We used a saturated model and backward stepping to confirm results of the initial forward stepwise regression. The prospective cohort model backwards stepping produced the same model as the stepwise approach. We performed all analyses using the most current version of SAS software (version 9.1.3, SAS Institute, Inc., Cary, North Carolina).
Among the 4,287 medical students surveyed (2,966 cross-sectional students and 1,321 additional prospective students), 2,248 students returned surveys for an overall response rate of 52.4%. Compared with the overall study population, nonresponders were more likely to be male (54.9% versus 51.6%), between 25 and 30 years old (62.4% versus 55.4%), and non-Caucasian (31.0% versus 25.8%; all P < .02). We have previously reported the demographic characteristics, QOL scores, burnout frequencies, and depression symptoms of participants.18
All but 26 responders (1.16%) answered the question about thoughts of dropping out of medical school. Among the 2,222 who responded to the question about thoughts of dropping out, 559 (25.2%) reported they had considered dropping out in the previous year. The majority of these 559 students (316/559, 56.5%) indicated that they had not seriously considered dropping out (see Method), whereas the remaining 243 students reported they had considered dropping out somewhat seriously (203, 36.3%), seriously (27, 4.8%), or very seriously (13, 2.3%). On sensitivity analysis—assuming all nonresponders did not have serious thoughts of dropping out—the prevalence of serious thoughts of dropping out within the previous year was 243/4,287 (5.7%). During the five years prior to the survey, 121 students actually dropped out of the seven participating medical schools (mean class size 149.89, median 173, range 42–232). The average rate of drop-out across all sites from 2002 to 2006 was 0.55% per year, suggesting the risk of a student dropping out during four years of medical school could be up to 2.2%, consistent with prior reports.2
The likelihood of reporting serious thoughts of dropping out within the previous year broken down by demographic characteristics and life events is shown in Table 1. Being older in age (over 30), having a child, being in the third year of school, and self-identifying as American Indian/Native Alaskan or Hawaiian/Pacific Islander were significantly associated with serious thoughts of dropping out within the previous year. No association was observed between ethnic groups (Hispanic or Latino), gender, or relationship status and serious thoughts of dropping out.
Approximately one third of all responding students (764/2,248, 34.0%) experienced at least one major negative life event in the previous year, and 321/2,248 (14.3%) students experienced at least one major positive life event in the previous year. Personally experiencing a major negative life event in the past year was associated with serious thoughts of dropping out, whereas personally experiencing a major positive life event was not. When each life event was examined individually, divorce (OR 2.82, CI 1.11–7.18), major personal illness (OR 2.54, CI 1.73–3.73), and a major illness in a significant other/family member (OR 1.39, CI 1.03–1.86) in the previous year were each associated with serious thoughts of dropping out.
EE, DP, and PA scores worsened incrementally as thoughts of dropping out increased in severity (all P < .001; data not shown). Students with burnout (OR 5.91, CI 4.16–8.41) or high EE (OR 6.34, CI 4.61–8.74) were five to six times more likely to report serious thoughts of dropping out within the previous year, whereas students with low PA (OR 2.46, CI 1.86–3.26) or high DP (OR 2.42, CI 1.83–3.20) were two to three times more likely. QOL scores and symptoms of depression were also related to serious thoughts of dropping out, as shown in Table 2. On the basis of the strong association between both burnout and depression with serious thoughts of dropping out, we explored interactions among these variables. The prevalence of serious thoughts of dropping out increased with the severity of burnout, independent of symptoms of depression (Figure 1).
As previously reported,18 858 of the 1,321 (64.9%) students who responded to the 2006 survey and remained enrolled at the same medical school responded to the 2007 survey. In comparison with the second-, third-, and fourth-year students in the 2007 cross-sectional cohort, students providing prospective data were similar with respect to year in school, relationship status, parental status, and debt; however, they were slightly more likely to be women (54% versus 49%, P = .01) and to be younger (9% versus 12% greater than 30 years old, P = .02).
We observed a strong dose–response relationship between burnout and lower mental QOL at baseline (spring 2006) and serious thoughts of dropping out in the subsequent year (assessed spring 2007) as shown in Table 3. Using preestablished thresholds for health professionals,19 low PA (score ≤ 33), high EE (score ≥ 27), and high DP (score ≥ 10) at baseline were associated with, respectively, a 2.28, 1.95, and 2.53 increased odds ratio of serious thoughts of dropping out during the following year (data not shown; all P < .01). A positive depression screen at baseline was also associated with a two- to threefold increase in serious thoughts of dropping out during the following year. Gender, relationship status, and year in school were not associated with future serious thoughts of dropping out, whereas age, parental status, race, and debt were associated with future serious thoughts of dropping out.
In stepwise multivariable logistic regression, controlling for variables associated with serious thoughts of dropping out on univariate analysis (P < .05), only decreased sense of PA, lower mental and physical QOL, and having children remained independently associated with serious thoughts of dropping out. For each one-point difference in PA (OR 1.04, CI 1.01–1.08), mental QOL (OR 1.05, CI 1.03–1.07), or physical QOL (OR 1.05, CI 1.02–1.08) score between two otherwise identical students at baseline, the student with the lower score is 4% to 5% more likely to experience serious thoughts of dropping out during the following year. Those with children (OR 2.53, CI 1.18–5.43) were also more likely to have thoughts of dropping out. The saturated multivariable modeling yielded similar results with respect to concordance and the amount of variability accounted for.
Reversibility of burnout
As previously reported,18 among the 792 students in the prospective cohort who adequately completed the MBI at both time points, 271 (34.2%) had burnout at both time points (“chronic burnout”), 99 (12.5%) had burnout at time point 1 but not time point 2 (i.e., recovered from burnout), 132 (16.7%) were burned out only at time point 2 (“new burnout”), and 290 (36.6%) were free of burnout at both time points (“never burned out”). Figure 2 depicts prevalence of serious thoughts of dropping out among these groups. Students who recovered from burnout were less likely to report serious thoughts of dropping out during the previous year than students with chronic burnout (6.1% versus 17.7%, P = .005) and had a rate of serious thoughts of dropping out similar to those who either had never experienced burnout (6.1% versus 3.1%, P = .19) or had newly developed burnout (6.1% versus 10.0%, P = .28). Students who developed new burnout were less likely to report serious thoughts of dropping out than students with chronic burnout (10.0% versus 17.7%, P = .04). Students who had never been burned out were less likely to report serious thoughts of dropping out than either students with new burnout (P = .004) or those with chronic burnout (P < .0001).
In this large, multi-institutional study, 11% of medical students reported having seriously considered dropping out of medical school within the last year. Previously reported attrition rates of 3% to 6% among U.S. medical students2,8,9,11 suggest that up to half of the students who seriously contemplate leaving medical school will do so. The actual number of students (n = 121) who dropped out across five years at the seven participating schools is approximately three quarters of one average-sized class at one school.4 In our study, serious thoughts of dropping out had a strong relationship with measures of both personal distress (i.e., QOL and depressive symptoms) and professional distress (i.e., burnout) and, to a lesser degree, with self-identifying as American Indian/Native Alaskan or Hawaiian/Pacific Islander, with having children, and with experiencing a major negative personal life event. While the relationship between depression and attrition has been previously reported,10 the association between burnout and serious thoughts of dropping out has not been previously reported. Notably, while we have previously shown that the DP and PA domains of burnout are most strongly related to suicidal ideation among medical students,18 only the PA domain remained independently associated with serious thoughts of dropping out of medical school. Our findings have important implications as burnout is more prevalent than depression among medical students.
Although our study cannot confirm a causal relationship between burnout and serious thoughts of dropping out, several criteria for causality are satisfied.29 First, the association between burnout and serious thoughts of dropping out satisfied the temporal relationship (i.e., burnout preceded thoughts of dropping out). Second, the association between burnout and serious thoughts of dropping out was large (threefold increased risk of subsequent serious thoughts of dropping out). Such a large association suggests a strong relationship between burnout and serious thoughts of dropping out. Also, the domains of burnout demonstrated a dose–response relationship with serious thoughts of dropping out. Third, the criterion for reversibility (frequency of serious thoughts of dropping out returned to baseline with recovery from burnout) was met. Fourth, the causal relationship is plausible. EE, DP, and low sense of PA—all components of burnout—likely lower a student's sense of self-efficacy, motivation, and willingness to persevere through the challenges of medical school. Fifth, burnout has also been associated with turnover of personnel in health care, an event analogous to dropping out.30 Additional studies are now needed to confirm and further explore this relationship.
With respect to demographic characteristics, students with children represented a subgroup with increased likelihood of serious thoughts of dropping out of medical school. On the multivariate analysis, being a parent was associated with a threefold increase in subsequent serious thoughts of dropping out during the ensuing 12 months. Consistent with these results, a prior single-institution longitudinal study by Rosal and colleagues31 found that having a child was associated with a gender-specific negative mental health impact. Children undoubtedly add a level of complexity to students' lives,32,33 and students with children may need unique support services to facilitate adjustment to medical school and to increase their chances of successfully pursuing their career goals. The most recently available national data on the proportion of students with children prior to graduation from the AAMC's Graduation Questionnaire suggest that 13% of U.S. medical students have children by the completion of their training (1995 was the last year this question was administered).34 Given the number of students who are parents and the relationship between parental status and serious thoughts of dropping out, further studies are needed to better understand experiences of students with children and to explore how schools can best help them complete their professional training while they simultaneously attend to the needs of their children.
Previous research has shown minorities (e.g., American Indian/Alaska Native, black/African American, Hispanic/Latino) to be at a higher risk for attrition than white and Asian students.2,8,9 The AAMC's Student Record System suggests that there may also be differences in reasons for attrition by race/ethnicity. White and Asian students most commonly cite nonacademic reasons for dropping out of medical school, whereas students from other racial/ethnic groups most commonly cite academic reasons.2 Among our cross-sectional cohort of students, students who identified as American Indian/Native Alaskan and Native Hawaiian/Pacific Islander were more likely than white students to have seriously considered dropping out, whereas in our prospective cohort of students, only students who identified as Native Hawaiian/Pacific Islander were more likely than white students to have seriously considered dropping out. Further research is needed to evaluate whether the relationship between race and dropping out may be mediated, in part, through differences in burnout and QOL among racial groups. We have previously demonstrated that minority students who reported that their race/ethnicity had adversely affected their medical school experience were more likely to have burnout, depressive symptoms, and low mental QOL than were fellow minority students without such experiences.12 Further research into the experiences of minority medical students and discrepancies in graduation rates by race/ethnicity is warranted.
A potential limitation of our study is response bias. We do not know how professional and personal distress affect response rate. Distressed students may be less motivated to fill out a survey; on the other hand, they may be more likely to participate because the topic of the survey is of relevance to them. Our response rate, however, is typical of physician35 and medical student14 surveys. There were few demographic differences between respondents and nonrespondents, with the exception of our cross-sectional sample being slightly biased toward female and white students as well as those outside of the 25- to 30-year-old age range. Whether the level of distress or experience of serious thoughts of dropping out among nonresponders differs by these demographic characteristics is unknown.
Strengths of our study include its size, inclusion of multiple schools, and prospective design. Our cohort of students attended diverse private and public medical schools geographically distributed across the United States. Respondents were similar to medical students in the United States with respect to gender, relationship status, parenting status,34,36 and prevalence of depressive symptoms.14 We also used instruments with documented sound psychometric properties to measure burnout, depressive symptoms, and QOL.
Intensive strategies to alleviate anticipated physician shortages need to go beyond increasing medical school class sizes and building new schools. In addition to these efforts, medical schools must work to keep their matriculants in school. Focused endeavors need to go beyond academic support because half of medical students who leave medical school before graduation do so for nonacademic reasons.2 On the basis of our findings, we have several suggestions. First, medical schools should implement a system to identify students who are seriously considering dropping out. In the prospective component of our study, burnout and mental QOL scores predicted serious thoughts of dropping out on prospective follow-up, so schools should work to identify students at risk for future serious thoughts of dropping out. Third, schools should implement and evaluate student support/counseling programs (including some tailored to students with children), wellness programs, and confidential mental health services to create a culture of wellness that equips students with the skills to promote personal well-being and resilience. Fourth, medical schools should support research endeavors exploring what personal (e.g., coping skills) and school factors (e.g., learning environment, faculty support, curricular factors) are associated with student distress, how best to identify students in need of personalized attention, and the usefulness of interventions intended to equip students with skills to promote their personal well-being.
In summary, our results indicate a high prevalence of serious thoughts of dropping out among U.S. medical students that may threaten pipeline efforts intended to bolster the physician workforce. Burnout and lower QOL among medical students seem to be important predictors of subsequent serious thoughts of dropping out even in the absence of depression. Future research is needed to develop practical ways to identify students at risk, prevent and reduce student distress, and facilitate development of focused retention programs. Such programs are vital to ensuring the well-being of students and an adequate number of physicians to provide for the nation's medical needs.
Funding/Support: This work was supported by an intramural award from the Mayo Clinic which played no role in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; or the preparation of the manuscript or decision to publish the manuscript.
Other disclosures: None.
Ethical approval: Institutional review boards at Mayo Clinic; University of Chicago Pritzker School of Medicine; Uniformed Services University of the Health Sciences; University of California, San Diego; University of Alabama School of Medicine; University of Washington School of Medicine; and University of Minnesota Medical School approved this study.
Previous presentations: An oral presentation of this study entitled “Consideration of dropping out among us medical students: A multi-institutional study” was given at the 13th Ottawa International Conference on Clinical Competence, Melbourne, Australia, March 2008.
Disclaimer: The opinions in this article are the authors' own and do not reflect those of the Department of Defense or the U.S. military.
6 Goodwin MC, Cleason WM, Kontos HA. A pilot study of the cost of educating undergraduate medical students at Virginia Commonwealth University. Acad Med. 1997;72:211–217.
7 Jones RF, Korn D. On the cost of educating a medical student. Acad Med. 1997;72:200–210.
8 Tekian A. Attrition rates of underrepresented minority students at the University of Illinois at Chicago College of Medicine, 1993–1997. Acad Med. 1998;73:336–338.
9 Stetto JE, Gackstetter GD, Cruess DF, Hooper TI. Variables associated with attrition from Uniformed Services University of the Health Sciences Medical School. Mil Med. 2004;169:102–107.
10 Clark DC, Zeldow PB. Vicissitudes of depressed mood during four years of medical school. JAMA. 1988;260:2521–2528.
11 Fogleman BY, Vander Zwagg R. Demographic, situational, and scholastic factors in medical school attrition. South Med J. 1981;74:602–606.
12 Dyrbye LN, Thomas MR, Harper W, et al. The association of race with medical student well-being across the United States. Arch Intern Med. 2007;167:2103–2109.
13 Dyrbye LN, Thomas MR, Huntington JL, et al. Personal life events and medical student burnout: A multicenter study. Acad Med. 2006;81:374–384.
14 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.
15 Goldin SB, Wahi MM, Farooq OS, et al. Student quality-of-life declines during third year surgical clerkship. J Surg Res. 2007;143:151–157.
16 Goebert D, Thompson D, Takeshita J, et al. Depressive symptoms in medical students and residents: A multischool study. Acad Med. 2009;84:236–241.
17 Haglund ME, ann het Rot M, Cooper NS, et al. Resilience in the third year of medical school: A prospective study of the associations between stressful events occuring during clinical rotations and student well-being. Acad Med. 2009;84:258–268.
18 Dyrbye LN, Thomas MR, Massie FS, et al. Burnout and suicidal ideation among US medical students. Ann Intern Med. 2009;149:334–341.
19 Maslach C, Jackson SE, Leiter MP. Maslach Burnout Inventory Manual. 3rd ed. Palo Alto, Calif: Consulting Psychologists Press; 1996.
20 Spitzer RL, Williams JB, Kroenke K, et al. Utility of a new procedure for diagnosing mental disorders in primary care. The PRIME-MD 1000 study. JAMA. 1994;272:1749–1756.
21 Ware J, Kosinski M, Dewey J, Gandek B. How to Score and Interpret Single-Item Health Status Measures: A Manual for Users of the SF-8 Health Survey. Lincoln, RI: QualityMetric Incorporated; 2001.
22 West CP, Huschka MM, Novotny PJ, et al. Association of perceived medical errors with resident distress and empathy: A prospective longitudinal study. JAMA. 2006;296:1071–1078.
23 Thomas NK. Resident burnout. JAMA. 2004;292:2880–2889.
24 Whooley MA, Avins AL, Miranda J, Browner WS. Case-finding instruments for depression. Two questions are as good as many. J Gen Intern Med. 1997;12:439–445.
25 Turner-Bowker DM, Bayliss MS, Ware JE Jr, Kosinski M. Usefulness of the SF-8 Health Survey for comparing the impact of migraine and other conditions. Qual Life Res. 2003;12:1003–1012.
26 Shanafelt T, West C, Zhao X, et al. Relationship between increased personal well-being and enhanced empathy among internal medicine residents. J Gen Intern Med. 2005;20:559–564.
27 Thomas MR, Dyrbye LN, Huntington JL, et al. How do distress and well-being relate to medical student empathy? A multicenter study. J Gen Intern Med. 2007;22:177–183.
28 Freund RJ, Littell RC. SAS System for Regression. 2nd ed. Cary, NC: SAS Institute; 2000.
29 Fletcher RH, Fletcher SW. Chapter 11: Cause. In: Fletcher RH, Fletcher SW. Clinical Epidemiology: The Essentials. 4th ed. Philadelphia, Pa: Lippincott Williams & Wilkins; 2005.
30 Goodman EA, Boss RW. Burnout dimensions and voluntary and involuntary turnover in a health care setting. J Health Hum Serv Adm. 1999;21:462–471.
31 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.
32 Lazo R. Family matters. JAMA. 1991;265:119.
33 Law J. Starting a family in medical school. JAMA. 1997;227:767.
35 Kellerman S, Herold J. Physician response to surveys. A review of the literature. Am J Prev Med. 2001;20:61–71.