See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/23231949

Burnout and Suicidal Ideation among U.S. Medical Students

Article in Annals of Internal Medicine · October 2008

Source: PubMed

CITATIONS READS
762 2,187

12 authors, including:

Liselotte Dyrbye David Power

Mayo Foundation for Medical Education and Research University of Minnesota Twin Cities
116 PUBLICATIONS 13,770 CITATIONS 33 PUBLICATIONS 3,451 CITATIONS

SEE PROFILE SEE PROFILE

Anne Eacker Steven J Durning

University of Washington Seattle Uniformed Services University of the Health Sciences
15 PUBLICATIONS 3,222 CITATIONS 395 PUBLICATIONS 10,169 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Burnout, Moral distress and Career satisfaction View project

Burnout, career satisfaction, and well-being among US neurologists in 2016 View project

All content following this page was uploaded by Anne Eacker on 30 May 2014.

The user has requested enhancement of the downloaded file.

Academia and Clinic Annals of Internal Medicine

Burnout and Suicidal Ideation among U.S. Medical Students

Liselotte N. Dyrbye, MD; Matthew R. Thomas, MD; F. Stanford Massie, MD; David V Power, MD; Anne Eacker, MD; William Harper, MD;
Steven Durning, MD; Christine Moutier, MD; Daniel W. Szydlo, BA; Paul J. Novotny, MS; Jeff A. Sloan, PhD; and Tait D. Shanafelt, MD

Background: Little is known about the prevalence of suicidal ide- quality of life (P ⬍ 0.002 for each domain), and depressive symp-
ation among U.S. medical students or how it relates to burnout. toms (P ⬍ 0.001) at baseline predicted suicidal ideation over the
following year. In multivariable analysis, burnout and low mental
Objective: To assess the frequency of suicidal ideation among quality of life at baseline were independent predictors of suicidal
medical students and explore its relationship with burnout. ideation over the following year. Of the 370 students who met
Design: Cross-sectional 2007 and longitudinal 2006 to 2007 cohort criteria for burnout in 2006, 99 (26.8%) recovered. Recovery from
study. burnout was associated with markedly less suicidal ideation, which
suggests that recovery from burnout decreased suicide risk.
Setting: 7 medical schools in the United States.
Limitation: Although response rates (52% for the cross-sectional
Participants: 4287 medical students at 7 medical schools, with study and 65% for the longitudinal cohort study) are typical of
students at 5 institutions studied longitudinally. physician surveys, nonresponse by some students reduces the pre-
cision of the estimated frequency of suicidal ideation and burnout.
Measurements: Prevalence of suicidal ideation in the past year and
its relationship to burnout, demographic characteristics, and quality Conclusion: Approximately 50% of students experience burnout
of life. and 10% experience suicidal ideation during medical school. Burn-
out seems to be associated with increased likelihood of subsequent
Results: Burnout was reported by 49.6% (95% CI, 47.5% to suicidal ideation, whereas recovery from burnout is associated with
51.8%) of students, and 11.2% (CI, 9.9% to 12.6%) reported less suicidal ideation.
suicidal ideation within the past year. In a sensitivity analysis that
assumed all nonresponders did not have suicidal ideation, the prev-
alence of suicidal ideation in the past 12 months would be 5.8%. Ann Intern Med. 2008;149:334-341. www.annals.org
In the longitudinal cohort, burnout (P ⬍ 0.001 for all domains), For author affiliations, see end of text.

D eath by suicide is a major occupational hazard for

physicians (1). The suicide rate among male physi-
cians is more than 40% higher than among men in the
ing before a tragic outcome. Several multi-institutional
studies (12–14) reveal that medical students have a sub-
stantially lower mental quality of life than similarly aged
general population, whereas that of female physicians is a individuals in the general population and that burnout af-
staggering 130% higher than among women in the general fects up to 50% of U.S. medical students.
population (1, 2). The increased risk for suicide among We hypothesized that burnout would relate to suicidal
physicians may begin during medical school (3). Available ideation among medical students. We used a mixed longi-
studies suggest that the suicide rate among medical stu- tudinal and cross-sectional study design to evaluate the
dents is higher than in the age-matched population (3–5). prevalence of suicidal ideation among U.S. medical stu-
Other small, single-institution studies (6 –10) have re- dents and to evaluate the relationship between suicidal ide-
ported that 3% to 15% of medical students have suicidal ation and burnout, symptoms of depression, and quality of
ideation during medical school training. Suicidal ideation life. Our objective was to assess the frequency of suicidal
is a well-established predictor of suicidal planning and at- ideation among medical students and explore its relation-
tempts. The National Comorbidity Survey found that ship with burnout.
34% of individuals in the general population with suicidal
ideation develop a suicide plan and, of those who plan, METHODS
more than 70% will attempt suicide (11). Notably, 26% of Participants
individuals with suicidal ideation progressed directly to an In the spring of 2006 (baseline) and 2007 (1-year fol-
unplanned suicide attempt (11). low-up), we invited all medical students at the Mayo Med-
Suicide is at the extreme end of the personal distress ical School, University of Washington School of Medicine,
continuum, and it is critical for medical schools to identify University of Chicago Pritzker School of Medicine, Uni-
students at greatest risk for suicide in the hope of interven- versity of Minnesota Medical School, and University of
Alabama School of Medicine to complete Web-based sur-
veys. Students at the University of California San Diego
See also: School of Medicine and Uniformed Services University of
the Health Sciences also participated in the 2007 survey.
Web-Only Participation was elective, and responses were anonymized.
Conversion of graphics into slides We included all students who responded to the 2007 sur-
vey in the cross-sectional analysis and those who responded
334 © 2008 American College of Physicians
 Medical Student Burnout and Suicidal Ideation Academia and Clinic

in both 2006 and 2007 in the longitudinal analysis. Each disorder (29, 30). With a reported positive likelihood ratio
institution’s institutional review board approved the study of up to 3.42 for the diagnosis of major depression (30)
before participation of their students. and an estimated 25% prevalence of depression among
Data Collection medical students (12), a positive result implies a 50%
Participants returned the surveys electronically. Pre- probability of current major depression.
serving student confidentiality was an essential feature. We We measured mental and physical quality of life by
linked individual responses on the 2006 and 2007 surveys using the Medical Outcomes Study Short Form-8 (SF-8)
for longitudinal analysis by using unique identifiers and (31, 32), an alternate version of the SF-36. Norm-based
stripped all data of identifiers before forwarding them to scoring methods of responses on this instrument are used
study statisticians for analysis. to calculate mental and physical quality of life summary
scores (31). The mean mental and physical quality of life
Study Measures summary scores for the U.S. population are 49.2 (SD,
We used established instruments to measure burnout, 9.46) and 49.2 (SD, 9.07), respectively (31). Previous re-
symptoms of depression, and quality of life on both the search has demonstrated acceptable reliability and test–
2006 and 2007 surveys. These surveys also included ques- retest reliability (31). Several studies (31) have demon-
tions about demographic characteristics, and the 2007 sur- strated content, construct, and criterion-related validity for
vey included questions about suicidal ideation. the SF-8, and other studies (33) have demonstrated high
Suicidal Ideation convergent validity and good discriminate validity. The
We assessed suicidal ideation by asking students: SF-8 has also been used in samples of residents (32) and
“Have you ever had thoughts of taking your own life, even medical students (14, 27, 28).
if you would not really do it?,” “During the past 12 Statistical Analysis
months have you had thoughts of taking your own life?,” Our primary analysis involved descriptive summary
and “Have you ever made an attempt to take your own statistics for estimating the prevalence of suicidal ideation,
life?” These questions, which originated from an inventory burnout, a positive depression screen, and mental and
developed by Meehan and colleagues (15) that has been physical quality of life for medical students. We evaluated
used to assess suicidal ideation among medical students (9), differences by reported suicidal ideation in the previous
are similar to questions used in large U.S. epidemiologic year by using the Wilcoxon–Mann–Whitney test (for con-
studies intended to assess suicidality (11, 16 –18). tinuous variables) and the Fisher exact test (for categorical
Burnout, Symptoms of Depression, and Quality of Life
variables). We used the Wilcoxon rank-sum test rather
than parametric tests to account for the interval level na-
The Maslach Burnout Inventory is a 22-item instru-
ture of the psychological tests. All tests were 2-sided, with
ment that is considered the gold standard for measuring
a type I error rate of 0.05. We performed collinearity test-
burnout (19 –21). This instrument has separate subscales
ing to determine whether multiple-way collinearity existed
to evaluate each domain of burnout: emotional exhaustion,
among the independent variables (34). No variables had
depersonalization, and low personal accomplishment. Tests
achieved a level of collinearity that would bias the model-
of discriminant and convergent validity have been accept-
ing process. We performed forward stepwise logistic regres-
able, construct validity of the 3 dimensions has been dem-
sion to evaluate independent associations of the indepen-
onstrated (22, 23), and predictive validity has been sug-
dent variables with suicidal ideation. We used a saturated
gested by burnout score predicting risk for future sick leave
model and backward stepping to confirm results of the
absences (24). The Maslach Burnout Inventory has also
initial stepwise regression. In all cases, backward stepping
been used extensively in studies of both physicians (20,21,
produced the same model as the stepwise approach. All
25) and medical students (13, 14, 26 –28). According to
analyses were done by using SAS, version 9 (SAS Institute,
convention, a score of 27 or higher on the emotional ex-
Cary, North Carolina).
haustion subscale or 10 or higher on the depersonalization
subscale was considered an indicator of professional burnout Role of the Funding Source
for medical professionals (19). Health professionals are con- This work was supported by an Education Innovation
sidered to have a low score on the personal accomplishment award from the Mayo Clinic. The Mayo Clinic played no role
scale if their score is 33 or less. Normal scores for health care in the design and conduct of the study; collection, manage-
professionals, including physicians, are 22.19, 7.12, and 36.53 ment, analysis, and interpretation of the data; or preparation
on the emotional exhaustion, depersonalization, and personal of the manuscript or decision to publish the manuscript.
accomplishment subscales, respectively (19).
We identified symptoms of depression by using the
2-item Primary Care Evaluation of Mental Disorders (29), RESULTS
a screening tool which performs as well as longer instru- Cross-Sectional Survey
ments (30). This instrument has a sensitivity of 86% to A total of 2248 (52.4%) of 4287 medical students re-
96% and a specificity of 57% to 75% for major depressive sponded to the 2007 survey. Responders were less likely than
www.annals.org 2 September 2008 Annals of Internal Medicine Volume 149 • Number 5 335
Academia and Clinic Medical Student Burnout and Suicidal Ideation

1945 (30.6%) had a low sense of personal accomplish-

Table 1. Participant Characteristics, 2007
ment. Compared with age-comparable individuals and the
general U.S. population, medical students had lower men-
Variable Participants, n (%)
(n ⴝ 2248) tal quality-of-life scores (mean, 43.5 [SD, 11.0] vs. 47.2
Sex
[SD, 9.9] for age-comparable individuals [P ⬍ 0.001] and
Male 1159 (51.8) 49.2 [SD, 9.5] for the general U.S. population [P ⬍
Female 1077 (48.2) 0.001]) but higher physical quality-of-life scores (mean,
Missing* 12
52.2 [6.9] vs. 51.4 [SD, 7.9] for age-comparable individ-
Age, y uals [P ⬍ 0.001] and 49.2 [SD, 9.1] for the general U.S.
⬍25 779 (34.8) population [P ⬍ 0.001]) on the SF-8. More than one third of
25–30 1229 (54.9)
⬎30 229 (10.2)
students (742 of 2178 [34.1%]) had mental quality-of-life
Missing* 11 scores less than 40, a score that is nearly a full standard devi-
ation below the normal score for the general U.S. population.
Relationship status
Married 731 (32.6)
Table 3 shows the likelihood of reporting suicidal ide-
Nonmarried partner 262 (11.7) ation in the previous year, by demographic characteristics,
Single 1217 (54.3) burnout, depressive symptoms, and quality of life. Age, mar-
Divorced 31 (1.4)
Missing* 7
ital status, year in school, and debt were significantly associ-

Have children 266 (11.9)

Table 2. Burnout, Quality of Life, and Depression
Year in medical school
1st 623 (27.8) Symptoms, 2007
2nd 578 (25.8)
3rd 494 (22.1) Variable Participants (n ⴝ 2248)
4th 477 (21.3) and Scores
Other† 66 (2.9)
Missing* 10 Suicidality
Ever considered suicide, n (%)
Debt Yes 561 (25.1)
⬍$50 000 1268 (56.8) No 1673 (74.9)
$50 000–100 000 524 (23.5) Missing* 14
⬎$100 000 439 (19.7) Ever made suicide attempt, n (%)
Missing* 17 Yes 43 (1.9)
No 2184 (98.1)
Missing* 21
* The number of individuals who returned surveys but did not provide an answer
to this specific question. Missing responses were excluded from the total before Considered suicide last year, n (%)
percentages were calculated. Yes 249 (11.2)
† Students who took a break from medical school to pursue enrichment activities, No 1981 (88.8)
such as research projects or graduate work. Missing* 18

Burnout†
the overall population to be male (51.6% vs. 54.9%), between Mean emotional exhaustion score (SD) 24.0 (10.9) [0–52]
[range]
25 and 30 years of age (55.4% vs. 62.4%), or nonwhite Mean depersonalization score (SD) [range] 7.3 (5.9) [0–28]
(25.8% vs. 31.0%) (all P ⬍ 0.02). Table 1 shows the demo- Mean personal accomplishment score (SD) 36.2 (7.7) [0–48]
graphic characteristics of cross-sectional survey respondents. [range]‡
Burned out 1069 (49.6)
Table 2 shows the percentage of responders reporting
suicidal ideation, burnout, and symptoms of depression Quality of life
and the mean quality-of-life scores. Of the 2230 respond- Mean mental score (SD)§ 43.5 (11.0) [7.7–65.8]
Mean physical score (SD) 52.2 (6.9) [20.3–66.6]
ers on the 2007 survey who responded to questions on
suicidal ideation, 249 (11.2% [CI, 9.9% to 12.6%]) re- Depression symptoms, n (%)
ported considering suicide in the previous year and 43 Screen positive 1037 (46.5)
Screen negative 1191 (53.5)
(1.9% [CI, 1.4% to 2.6%]) had made a suicide attempt at Missing* 20
some point in the past. On sensitivity analysis, assuming
that all nonresponders did not have suicidal ideation, the * The number of individuals who returned surveys but did not provide an answer
to this specific question. Missing responses were excluded from the total before
prevalence of suicidal ideation in the past 12 months percentages were calculated.
would be 249 (5.8%) of 4287 students. † Burnout was measured by using the Maslach Burnout Inventory (19), whose 3
subscales evaluate each of the domains of burnout, characterized as emotional
Overall, 1069 (49.6%, [CI, 47.5%–51.8%]) of 2154 exhaustion, depersonalization, and low sense of personal accomplishment. A high
students met the criteria for burnout (94 students did not score on either the emotional exhaustion (ⱖ27) or depersonalization (ⱖ10) sub-
scale indicates professional burnout.
answer enough Maslach Burnout Inventory questions to be ‡ A higher score is desirable and indicates greater sense of personal accomplish-
included in this analysis). Among these students, 860 ment.
§ The mean mental quality-of-life score for students was more than one-half stan-
(40.1%) of 2142 had high emotional exhaustion, 648 dard deviation below the population norm, a difference that has been considered
(31.8%) of 2037 had high depersonalization, and 595 of clinically significant.

336 2 September 2008 Annals of Internal Medicine Volume 149 • Number 5 www.annals.org
 Medical Student Burnout and Suicidal Ideation Academia and Clinic

Table 3. Demographic Characteristics and Distress among Medical Students with and without Suicidal Ideation in the Previous
Year*

Variable Suicidal No Suicidal Unadjusted Odds Ratio

Ideation Ideation (95% CI) or Other Data
Sex, n (%)
Male 122 (10.6) 1030 (89.4) 1.13 (0.87 to 1.47)†
Female 126 (11.8) 943 (88.2)
Missing‡ 1 8

Age, n (%)
⬍25 y 64 (8.3) 709 (91.7) Reference†
25–30 y 156 (12.7) 1069 (87.3) 1.62 (1.19 to 2.20)†
⬎30 y 26 (11.6) 199 (88.4) 1.45 (0.89 to 2.34)†
Missing‡ 3 4

Relationship status, n (%)

Single or divorced 153 (12.3) 1088 (87.7) 1.32 (1.01 to 1.73)†
Married or partner 95 (9.6) 891 (90.4)
Missing‡ 1 2

Have children, n (%)

Yes 38 (14.5) 225 (85.5) 1.41 (0.97 to 2.04)†
No 211 (10.7) 1754 (89.3)

Year in medical school, n (%)

First 50 (8.1) 568 (91.9) Reference†
Second 67 (11.6) 509 (88.4) 1.50 (1.02 to 2.20)†
Third 67 (13.7) 423 (86.3) 1.80 (1.22 to 2.65)†
Fourth 61 (12.9) 413 (87.1) 1.68 (1.13 to 2.49)†
Other§ 4 (6.1) 62 (93.9) 0.73 (0.26 to 2.10)†
Missing‡ 0 6

Debt, n (%)
⬍49 999 125 (9.9) 1134 (90.1) Reference†
$50 000–99 999 62 (11.9) 459 (88.1) 1.23 (0.89 to 1.69)†
⬎$100 000 61 (14.0) 376 (86.0) 1.47 (1.06 to 2.04)†
Missing‡ 1 12

Burnout
Mean emotional exhaustion score (SD) 30.9 (10.54) 23.1 (10.55) 7.84 (6.42 to 9.26)㛳
Mean depersonalization score (SD) 9.9 (6.61) 6.9 (5.74) 3.00 (2.10 to 3.89)㛳
Mean personal accomplishment score (SD) 33.4 (8.24) 36.6 (7.49) ⫺3.23 (⫺4.35 to ⫺2.10)㛳
Have burnout, n (%)
Yes 180 (16.9) 883 (83.1) 3.46 (2.55 to 4.69)
No 60 (5.6) 1017 (94.4)

Quality of life
Mean mental score (SD) 35.4 (11.30) 44.5 (10.50) ⫺9.09 (⫺10.62 to ⫺7.57)㛳
Mean physical score (SD) 50.9 (7.71) 52.4 (6.83) ⫺1.49 (⫺2.52 to ⫺0.45)㛳

Depression symptoms, n (%)

Screen positive 203 (19.7) 828 (80.3) 6.51 (4.63 to 9.15)
Screen negative 43 (3.6) 1141 (96.4)

* Of 2230 participants.
† Odds ratio for risk for suicidal ideation in the categorical group relative to the reference group. If there was ⬎1 comparison group (for example, year in school), a reference
group was selected to which all other groups were compared.
‡ The number of individuals who returned surveys but did not provide an answer to this specific question. Missing responses were excluded from the total before percentages
were calculated.
§ Students who took a break from medical school to pursue enrichment activities, such as research projects or graduate work.
㛳 Mean difference.

ated with suicidal ideation in the previous year. We observed high emotional exhaustion (odds ratio, 3.17 [CI, 2.39 to
no association between sex and suicidal ideation. 4.19]; P ⬍ 0.001), high depersonalization (odds ratio, 2.10
Suicidal ideation was also strongly correlated with [CI, 1.59 to 2.77]; P ⬍ 0.001), or a low sense of personal
measures of distress and quality of life. Students were 2- to accomplishment (odds ratio, 2.03 [CI, 1.53 to 2.68]; P ⬍
3-fold more likely to report suicidal ideation if they had 0.001). Students with suicidal ideation had statistically sig-
burnout (odds ratio, 3.46 [CI, 2.55 to 4.69]; P ⬍ 0.001), nificantly lower mean mental (⫺9.09; P ⬍ 0.001) and
www.annals.org 2 September 2008 Annals of Internal Medicine Volume 149 • Number 5 337
Academia and Clinic Medical Student Burnout and Suicidal Ideation

1184; P ⬍ 0.001). Because of the strong association of

Figure 1. Prevalence of suicidal ideation, by degree of
both burnout and depression with suicidal ideation, we
depersonalization and emotional exhaustion and depressive
explored the interactions between these variables. The
symptoms.
prevalence of suicidal ideation increased by severity of
burnout independent of symptoms of depression (Figure 1).
30
Longitudinal Cohort
The response rate for the 2006 survey was 55% (13).
25 Among the 1321 students who responded to the 2006
Students with Suicidal Ideation, %

survey and were still enrolled in the same medical school,

20 858 (65%) also responded to the 2007 survey. The demo-
graphic characteristics of the individuals providing longitu-
dinal data were similar to those for second- through
15
fourth-year students in the 2007 cross-sectional cohort
with respect to relationship status, parental status, year in
10 school, and debt, although they were slightly more likely to
be women (54% vs. 49%, P ⫽ 0.006) and be younger (9%
5
vs. 12% older than 30 years; P ⫽ 0.023).
Table 4 shows the relationship among burnout, qual-
ity of life, and depressive symptoms at baseline with sui-
0
cidal ideation over the ensuing year (spring 2006 to spring
Low Medium High
Depersonalization
2007). We observed a strong dose–response relationship
between burnout and mental quality of life at baseline and
Negative depression screen
suicidal ideation in the subsequent year. Each 1-point in-
Positive depression screen
crease in emotional exhaustion and depersonalization
scores and 1-point decrease in personal accomplishment
scores was associated with a respective 5%, 10%, and 6%
25 increase in the odds of suicidal ideation in the following
year. Similarly, each 1-point decrease in mental quality of
life was associated with a 7% increase in the odds of sui-
20
Students with Suicidal Ideation, %

cidal ideation over the subsequent year. When we classified

burnout according to the standard categorical thresholds
(low, intermediate, high) (19), high emotional exhaustion,
15
high depersonalization, and low personal accomplishment
at baseline were associated with a 1.83, 3.38, and 2.01
10

Table 4. Relationship of Burnout, Quality-of-Life Score, and

Depression Symptoms at Baseline with Suicidal Ideation in
5
the Following Year*

Variable Univariate Odds Ratio P Value

0
(95% CI)†
Low Medium High
Burnout
Emotional Exhaustion
Emotional exhaustion 1.05 (1.03–1.08) ⬍0.001
Negative depression screen Depersonalization 1.10 (1.06–1.15) ⬍0.001
Positive depression screen Personal accomplishment 1.06 (1.03–1.09) ⬍0.001
Burned out 2.33 (1.47–3.70) ⬍0.001

Relationship among depression screening result, degree of depersonaliza- Quality of life

tion or emotional exhaustion, and prevalence of suicidal ideation in the Mental score 1.07 (1.05–1.09) ⬍0.001
previous year (n ⫽ 2248). The prevalence of suicidal ideation increases as Physical score 1.02 (0.99–1.05) 0.192
burnout increases (all P ⬍ 0.001), regardless of whether individuals
screened positive for depression. Positive for symptoms on 3.08 (1.91–4.97) ⬍0.001
depression screen

physical (⫺1.49; P ⫽ 0.002) quality-of-life scores on the * Of 858 participants.

† Odds ratios indicate the incremental increase in the likelihood of suicidal ide-
SF-8. Students with depressive symptoms were more likely ation associated with each 1-unit increase in emotional exhaustion and deperson-
alization scores or each 1-unit decrease in personal accomplishment or mental or
than those without depressive symptoms to endorse sui- physical quality-of-life score. The odds ratio of suicidal ideation for those with a
cidal ideation (203 [19.7%] of 1031 vs. 43 [3.6%] of positive depression screen is relative to those with a negative depression screen.

338 2 September 2008 Annals of Internal Medicine Volume 149 • Number 5 www.annals.org
 Medical Student Burnout and Suicidal Ideation Academia and Clinic

increased odds of suicidal ideation over the following year (all

Figure 2. Prevalence of suicidal ideation, by burnout at 2
P ⬍ 0.01). Screening positive for depression at baseline was
time points.
also associated with a 3-fold increase in suicidal ideation in the
following year (odds ratio, 3.08 [CI, 1.91 to 4.97]; P ⬍
20
0.001). No demographic characteristics (sex, age, relationship
status, parental status, year in school, debt) were associated 18

Students with Suicidal Ideation in the Past Year, %

with future suicidal ideation (data not shown).
In stepwise multivariable logistic regression that con- 16

trolled for variables associated with suicidal ideation on

14
univariate analysis (P ⬍ 0.100) and medical school site,
only the domains of burnout and mental quality of life 12
remained independently associated with suicidal ideation.
10
For each 1-point increase in depersonalization (odds ratio,
1.05; P ⫽ 0.032) or each 1-point decrease in personal ac- 8
complishment (odds ratio, 1.04; P ⫽ 0.008) or mental
quality of life (odds ratio, 1.06; P ⬍ 0.001) at baseline, 6

students were 4% to 6% more likely to experience suicidal

4
ideation in the following year. Because a low score for
depersonalization ranges from 0 to 5 (whereas a high score 2
ranges from 10 to 30), the odds of suicidal ideation could
0
increase between 25% and 150% as individuals move from
Never Recovered New Chronic
low to high depersonalization. Similarly, because high per- Burned out from Burnout Burnout Burnout
sonal accomplishment is characterized by a score from 40 (n = 290) (n = 99) (n = 132) (n = 271)
to 48 and a low personal accomplishment score from 0 to
33, the odds of suicidal ideation could increase between We found differences in suicidal ideation between groups (P ⬍ 0.001).
Pairwise comparisons showed statistically significant differences between
25% and 192% as individuals move from a high to low students who recovered versus students with chronic burnout (P ⫽
degree of personal accomplishment. The saturated multi- 0.001) and among students who never experienced burnout versus those
variable modeling yielded similar results with respect to with new burnout (P ⫽ 0.007) or chronic burnout (P ⬍ 0.001).
concordance and the amount of variability accounted.

Reversibility of Burnout DISCUSSION

To further evaluate the relationship between burnout and Our large, multi-institutional study demonstrates a
suicidal ideation, we evaluated how changes in burnout be- high prevalence of recent suicidal ideation among U.S.
tween 2006 and 2007 related to suicidal ideation. At baseline, medical students, with approximately 1 of 9 students hav-
370 (46.7%) of 792 students met criteria for burnout. (Seven ing thoughts of suicide in the past year. The rate of suicidal
hundred ninety-two of 858 students in the longitudinal co- ideation among medical students in our study (11.2%) is
hort completed the Maslach Burnout Inventory adequately to higher than for individuals of similar age in the general
be included in this analysis.) Of these 370 individuals, 271 U.S. population (6.9% among 25- to 34-year-olds) (35).
(73.2%) remained burned out at 1 year follow-up (chronic In our cohort, suicidal ideation had a strong relationship
burnout), whereas 99 (26.8%) were no longer burned out at 1 with both personal distress (quality of life and depressive
year follow-up (recovered from burnout). Among the 422 stu- symptoms) and professional distress (burnout). Although
dents who were not burned out at baseline, 132 (31.3%) the relationship between depression and suicidal ideation is
experienced burnout at 1 year follow-up (new burnout), well recognized, the association between burnout and sui-
whereas the remaining 290 (68.7%) were not burned out at cidal ideation has not been previously reported. This rela-
either time point (never burned out). Figure 2 shows the tionship is notable because burnout seems to be a much
prevalence of suicidal ideation among these groups. Students more common form of distress among medical students
who recovered from burnout were less likely than students (12–14, 36).
with chronic burnout to report suicidal ideation over the pre- Although we cannot confirm a causal relationship
vious year (7.2% vs. 18.2%; P ⫽ 0.001) and had a rate of between burnout and suicidal ideation, several criteria
suicidal ideation similar to those who had never experienced for causality are satisfied (37). The association is bio-
burnout (7.2% vs. 4.9%; P ⫽ 0.38). Students who developed logically plausible and analogous to the association be-
new burnout were as likely to report suicidal ideation in 2007 tween suicidal ideation and depression. Burnout dem-
as students with chronic burnout (12.1% vs. 18.2%, P ⫽ onstrated a strong dose–response relationship with
0.120). Students who never had burned out were less likely to suicidal ideation that persisted on multivariable analysis
report suicidal ideation than students with new burnout (P ⫽ that controlled for other factors, including symptoms of
0.007) and chronic burnout (P ⬍ 0.001). depression. The association between burnout and sui-
www.annals.org 2 September 2008 Annals of Internal Medicine Volume 149 • Number 5 339
Academia and Clinic Medical Student Burnout and Suicidal Ideation

cidal ideation was large (2- to 3-fold increased risk) and schools should be aware of students experiencing such
satisfied both the temporality requirement (burnout pre- events and provide counseling support, confidential mental
ceded suicidal ideation) and the criteria for reversibility health services, and flexibility in curricular scheduling. Ed-
(risk for suicidal ideation returned to baseline with re- ucating students about the frequency of depression and
covery from burnout). Additional studies are now suicidal ideation and frequently making them aware of
needed to confirm and further explore this relationship. available resources seem to be sensible, worthy, and helpful
Identifying risk factors for suicidal ideation provides (1, 39).
the opportunity for interventions to prevent an adverse Our study has several limitations. First, although our
outcome. Of note, the trajectory for students with burnout response rate is typical of physician (40) and medical stu-
is not an inevitable worsening of distress. In our study, dent (12) surveys, response bias is a possibility. The effect
26% of students who were burned out at baseline recov- of professional and personal distress on response rate is
ered within the following year, indicating that burnout is unknown. One could hypothesize that distressed students
reversible. Recovery from burnout was associated with a may be less motivated to fill out a survey or that they
dramatic decrease in the likelihood of suicidal ideation, would be more likely to participate because the topic is
which suggests that identifying and treating burnout may relevant to them. Regardless, even if one assumes that none
provide an opportunity for medical schools to reduce sui- of the nonresponders had suicidal ideation within the past
cide risk. The factors and experiences that help students year, the prevalence of suicidal ideation in our study would
recover from burnout are unknown, which indicates the still be 249 (5.8%) of 4287 students. Second, our cross-
need for further study. Such information could make the sectional sample was slightly biased toward female and
services provided by the student wellness programs now white students, as well as those outside the 25- to 30-year
mandated by the Liaison Committee on Medical Educa- age range. Whether the level of distress or experience of
tion (38) more valuable. suicidal thoughts among nonresponders differs by these
How does the rate of suicidal ideation in our study demographic characteristics is unknown.
compare with other studies? Differences among studies in Our study has several important strengths. First, it is a
how the suicidal ideation question was asked and what large, prospective multi-institutional study. Second, stu-
population was studied limit our ability to compare the dents in this study attended diverse private and public
rate of suicidal ideation across studies. Nonetheless, preva- medical schools geographically distributed across the U.S
lence of suicidal ideation within the previous year in our and respondents were representative of medical students in
study is similar to that in a national sample of 522 gradu- the U.S. with respect to sex, relationship status, and par-
ating medical students in Norway (14%) (10). In a longi- enting status (41, 42). The prevalence of depressive symp-
tudinal study of medical students at a Midwestern medical toms among students in this survey is similar to that in
school, Clark and colleagues (6) reported that 15% to 20% other studies of medical students (12), which suggests that
of students had persistent suicidal ideation at every assess- the distress we observed is typical for students in the
ment after the first year. More recently, Givens and col- United States. Third, we used validated metrics to measure
leagues (7) surveyed all medical students at a private U.S. burnout, depressive symptoms, and quality of life, which
medical school and found that 6.2% of surveyed students allowed comparison with the general population and with
reported contemplating suicide during their medical school other samples of medical students, residents, and physi-
training. Other small, single-institution studies have re- cians. Fourth, we asked about suicidal ideation by using
ported rates of suicidal ideation during medical school questions from an existing inventory (15), one previously
ranging from 3% to 6% (8, 9). used to assess suicidal ideation among medical students (9),
How should medical schools respond to our findings? and the questions are similar to those assessing suicidality
First, medical schools should have a system in place to in large U.S. epidemiologic surveys (11, 16 –18), all of
identify students who are currently suicidal. Second, which lend credence and content validity. Finally, we con-
schools should work to identify students at risk for future tained measurement error by using established instruments
suicidal thoughts. In the longitudinal component of our and multiple measures of distress (burnout, quality of life,
study, burnout (depersonalization and personal accom- and symptoms of depression), which allowed triangulation
plishment scores) and mental quality-of-life scores pre- of results.
dicted suicidal thoughts over the following year. Third, In summary, our results indicate a high prevalence of
schools should implement student support and wellness suicidal ideation among U.S. medical students and suggest
programs and optimize the learning environment, the or- that the increased risk for suicide among physicians may
ganization of clinical rotations, and the diversity of clinical begin in medical school (1, 3). Burnout among medical
experiences. Schools should also be aware that negative life students seems to be an important predictor of subsequent
events, such as a serious personal illness or the death of a suicidal ideation even without symptoms of depression.
close family member, occur frequently among the medical Future research is needed to develop practical ways to iden-
student population and increase the risk for burnout (14). tify students at risk and strategies to reduce student distress
Although these events are outside their control, medical in the hope of improving the well-being of all students.
340 2 September 2008 Annals of Internal Medicine Volume 149 • Number 5 www.annals.org
 Medical Student Burnout and Suicidal Ideation Academia and Clinic
From the Mayo Clinic, Rochester, Minnesota; University of Alabama 20. West CP, Huschka MM, Novotny PJ, Sloan JA, Kolars JC, Habermann
School of Medicine, Birmingham, Alabama; University of Minnesota TM, et al. Association of perceived medical errors with resident distress and
Medical School, Minneapolis, Minnesota; University of Washington empathy: a prospective longitudinal study. JAMA. 2006;296:1071-8. [PMID:
School of Medicine, Seattle, Washington; University of Chicago Pritzker 16954486]
21. Thomas NK. Resident burnout. JAMA. 2004;292:2880-9. [PMID:
School of Medicine, Chicago, Illinois; Uniformed Services University of
15598920]
the Health Sciences, Bethesda, Maryland; and University of California,
22. Arthur NM. The assessment of burnout: A review of three inventories useful
San Diego, San Diego, California. for research and counseling. J Couns Dev. 1990;69:186-89.
23. Söderfeldt M, Söderfeldt B, Warg LE, Ohlson CG. The factor structure of
Grant Support: By an Education Innovation award from the Mayo the Maslach Burnout Inventory in two Swedish human service organizations.
Clinic. Scand J Psychol. 1996;37:437-43. [PMID: 8931397]
24. Toppinen-Tanner S, Ojajärvi A, Väänänen A, Kalimo R, Jäppinen P. Burn-
Potential Financial Conflicts of Interest: None disclosed. out as a predictor of medically certified sick-leave absences and their diagnosed
causes. Behav Med. 2005;31:18-27. [PMID: 16078523]
Requests for Single Reprints: Liselotte N. Dyrbye, MD, 200 First 25. Shanafelt TD, Bradley KA, Wipf JE, Back AL. Burnout and self-reported
Street Southwest, Rochester, MN 55905. patient care in an internal medicine residency program. Ann Intern Med. 2002;
136:358-67. [PMID: 11874308]
Current author addresses are available at www.annals.org. 26. Guthrie E, Black D, Bagalkote H, Shaw C, Campbell M, Creed F. Psycho-
logical stress and burnout in medical students: a five-year prospective longitudinal
study. J R Soc Med. 1998;91:237-43. [PMID: 9764076]
References 27. Thomas MR, Dyrbye LN, Huntington JL, Lawson KL, Novotny PJ, Sloan
1. Center C, Davis M, Detre T, Ford DE, Hansbrough W, Hendin H, et al. JA, et al. How do distress and well-being relate to medical student empathy? A
Confronting depression and suicide in physicians: a consensus statement. JAMA. multicenter study. J Gen Intern Med. 2007;22:177-83. [PMID: 17356983]
2003;289:3161-6. [PMID: 12813122] 28. Dyrbye LN, Thomas MR, Huschka MM, Lawson KL, Novotny PJ, Sloan
2. Schernhammer ES, Colditz GA. Suicide rates among physicians: a quantita- JA, et al. A multicenter study of burnout, depression, and quality of life in
tive and gender assessment (meta-analysis). Am J Psychiatry. 2004;161:2295-302. minority and nonminority US medical students. Mayo Clin Proc. 2006;81:1435-
[PMID: 15569903] 42. [PMID: 17120398]
3. Schernhammer E. Taking their own lives—the high rate of physician suicide. 29. Spitzer RL, Williams JB, Kroenke K, Linzer M, deGruy FV 3rd, Hahn SR,
N Engl J Med. 2005;352:2473-6. [PMID: 15958803] et al. Utility of a new procedure for diagnosing mental disorders in primary care.
4. Hays LR, Cheever T, Patel P. Medical student suicide, 1989-1994. Am J The PRIME-MD 1000 study. JAMA. 1994;272:1749-56. [PMID: 7966923]
Psychiatry. 1996;153:553-5. [PMID: 8599405] 30. Whooley MA, Avins AL, Miranda J, Browner WS. Case-finding instru-
5. Pepitone-Arreola-Rockwell F, Rockwell D, Core N. Fifty-two medical stu- ments for depression. Two questions are as good as many. J Gen Intern Med.
dent suicides. Am J Psychiatry. 1981;138:198-201. [PMID: 7457640] 1997;12:439-45. [PMID: 9229283]
6. Clark DC, Zeldow PB. Vicissitudes of depressed mood during four years of 31. Ware JE, Kosinski M, Dewey JE, Gandek B. How to Score and Interpret
medical school. JAMA. 1988;260:2521-8. [PMID: 3172426] Single-Item Health Status Measures: A Manual for Users of the SF-8 Health
7. Givens JL, Tjia J. Depressed medical students’ use of mental health services Survey. Lincoln, RI: QualityMetric Incorporated; 2001.
and barriers to use. Acad Med. 2002;77:918-21. [PMID: 12228091] 32. Shanafelt TD, West C, Zhao X, Novotny P, Kolars J, Habermann T, et al.
8. Tjia J, Givens JL, Shea JA. Factors associated with undertreatment of medical Relationship between increased personal well-being and enhanced empathy
student depression. J Am Coll Health. 2005;53:219-24. [PMID: 15813232] among internal medicine residents. J Gen Intern Med. 2005;20:559-64. [PMID:
9. Dahlin M, Joneborg N, Runeson B. Stress and depression among medical 16050855]
students: a cross-sectional study. Med Educ. 2005;39:594-604. [PMID: 33. Turner-Bowker DM, Bayliss MS, Ware JE Jr, Kosinski M. Usefulness of
15910436] the SF-8 Health Survey for comparing the impact of migraine and other condi-
10. Tyssen R, Vaglum P, Grønvold NT, Ekeberg O. Suicidal ideation among tions. Qual Life Res. 2003;12:1003-12. [PMID: 14651418]
medical students and young physicians: a nationwide and prospective study of 34. Freund RJ, Littell RC. SAS System for Regression. 2nd ed. Cary, NC: SAS
prevalence and predictors. J Affect Disord. 2001;64:69-79. [PMID: 11292521] Institute; 2000.
11. Kessler RC, Borges G, Walters EE. Prevalence of and risk factors for lifetime 35. Crosby AE, Cheltenham MP, Sacks JJ. Incidence of suicidal ideation and
suicide attempts in the National Comorbidity Survey. Arch Gen Psychiatry. behavior in the United States, 1994. Suicide Life Threat Behav. 1999;29:131-40.
1999;56:617-26. [PMID: 10401507] [PMID: 10407966]
12. Dyrbye LN, Thomas MR, Shanafelt TD. Systematic review of depression, 36. Dahlin ME, Runeson B. Burnout and psychiatric morbidity among medical
anxiety, and other indicators of psychological distress among U.S. and Canadian students entering clinical training: a three year prospective questionnaire and
medical students. Acad Med. 2006;81:354-73. [PMID: 16565188] interview-based study. BMC Med Educ. 2007;7:6. [PMID: 17430583]
13. Dyrbye LN, Thomas MR, Eacker A, Harper W, Massie FS Jr, Power DV, 37. Fletcher RH, Fletcher SW. Chapter 11: Cause. In: Clinical Epidemiology:
et al. Race, ethnicity, and medical student well-being in the United States. Arch The Essentials. Philadelphia: Lippincott Williams & Wilkins; 2005.
Intern Med. 2007;167:2103-9. [PMID: 17954805] 38. Liaison Committee on Medical Education. Accreditation Standards. Chi-
14. Dyrbye LN, Thomas MR, Huntington JL, Lawson KL, Novotny PJ, Sloan cago, IL and Washington, DC: Association of American Medical Colleges and
JA, et al. Personal life events and medical student burnout: a multicenter study. American Medical Association; 2003. Accessed at www.lcme.org/standard.htm
Acad Med. 2006;81:374-84. [PMID: 16565189] on 16 July 2008.
15. Meehan PJ, Lamb JA, Saltzman LE, O’Carroll PW. Attempted suicide 39. Mann JJ, Apter A, Bertolote J, Beautrais A, Currier D, Haas A, et al.
among young adults: progress toward a meaningful estimate of prevalence. Am J Suicide prevention strategies: a systematic review. JAMA. 2005;294:2064-74.
Psychiatry. 1992;149:41-4. [PMID: 1728183] [PMID: 16249421]
16. Cooper-Patrick L, Crum RM, Ford DE. Identifying suicidal ideation in 40. Kellerman SE, Herold J. Physician response to surveys. A review of the
general medical patients. JAMA. 1994;272:1757-62. [PMID: 7966924] literature. Am J Prev Med. 2001;20:61-7. [PMID: 11137777]
17. Suicidality. In: National Comorbidity Survey, Collaborative Psychiatric 41. Association of American Medical Colleges. Medical School Graduation
Epidemiology Surveys. Accessed at www.icpsr.umich.edu/cocoon/cpes/cpes Questionnaire. Washington, DC: Assoc of American Medical Colleges; 1995.
/BLSUICIDALITY/all/section.xml on 15 May 2008. Accessed at www.aamc.org/data/gq/allschoolsreports/1995.pdf on 22 January
18. National Center for Health Statistics. Youth Risk Behavior Survey. Hyatts- 2007.
ville, MD: U.S. Department of Health and Human Services; 2007. Accessed at 42. Association of American Medical Colleges. FACTS—Applicants, Matricu-
www.cdc.gov/nchs/datawh/nchsdefs/suicidalideation.htm on 15 May 2008. lants and Graduates. Total Enrollment by Sex and School, 2001–2005. Wash-
19. Maslach C, Jackson SE, Leiter MP. Maslach Burnout Inventory Manual. ington, DC: Assoc of American Medical Colleges; 2005. Accessed at www.aamc
3rd ed. Palo Alto, CA: Consulting Psychologists Pr; 1996. .org/data/facts/2005/factsenrl.htm on 17 December 2007.

www.annals.org 2 September 2008 Annals of Internal Medicine Volume 149 • Number 5 341
 Annals of Internal Medicine
Current Author Addresses: Drs. Dyrbye, Thomas, Sloan, and Dr. Harper: 5841 South Maryland Avenue, MC 3051, Chicago, IL
Shadafelt; Mr. Szydlo; and Mr. Novotny: 200 First Street Southwest, 60637.
Rochester, MN 55905. Dr. Durning: 4301 Jones Bridge Road, Bethesda, MD 20814-4799.
Dr. Massie: 1530 Third Avenue South, FOT 720, Birmingham, AL Dr. Moutier: University of California, San Diego School of Medicine,
35294. Medical Teaching Facility Room 180, 9500 Gilman Drive, 0606, La
Dr. Power: University of Minnesota, 516 Delaware Street Southeast, Jolla, CA 92093-0606.
Minneapolis, MN 55455.
Dr. Eacker: General Internal Medicine Center, 4245 Roosevelt Way
Northeast, Seattle, WA 98105.

W-70 2 September 2008 Annals of Internal Medicine Volume 149 • Number 5 www.annals.org

View publication stats