|Home | About | Journals | Submit | Contact Us | Français|
Medical students are at high risk for depression and suicidal ideation. However, the prevalence estimates of these disorders vary between studies.
To estimate the prevalence of depression, depressive symptoms, and suicidal ideation in medical students.
Systematic search of EMBASE, ERIC, MEDLINE, psycARTICLES, and psycINFO without language restriction for studies on the prevalence of depression, depressive symptoms, or suicidal ideation in medical students published before September 17, 2016. Studies that were published in the peer-reviewed literature and used validated assessment methods were included.
Information on study characteristics; prevalence of depression or depressive symptoms and suicidal ideation; and whether students who screened positive for depression sought treatment was extracted independently by 3 investigators. Estimates were pooled using random-effects meta-analysis. Differences by study-level characteristics were estimated using stratified meta-analysis and meta-regression.
Point or period prevalence of depression, depressive symptoms, or suicidal ideation as assessed by validated questionnaire or structured interview.
Depression or depressive symptom prevalence data were extracted from 167 cross-sectional studies (n = 116 628) and 16 longitudinal studies (n = 5728) from 43 countries. All but 1 study used self-report instruments. The overall pooled crude prevalence of depression or depressive symptoms was 27.2% (37 933/122 356 individuals; 95% CI, 24.7% to 29.9%, I2 = 98.9%). Summary prevalence estimates ranged across assessment modalities from 9.3% to 55.9%. Depressive symptom prevalence remained relatively constant over the period studied (baseline survey year range of 1982–2015; slope, 0.2% increase per year [95% CI, −0.2% to 0.7%]). In the 9 longitudinal studies that assessed depressive symptoms before and during medical school (n = 2432), the median absolute increase in symptoms was 13.5% (range, 0.6% to 35.3%). Prevalence estimates did not significantly differ between studies of only preclinical students and studies of only clinical students (23.7% [95% CI, 19.5% to 28.5%] vs 22.4% [95% CI, 17.6% to 28.2%]; P = .72). The percentage of medical students screening positive for depression who sought psychiatric treatment was 15.7% (110/954 individuals; 95% CI, 10.2% to 23.4%, I2 = 70.1%). Suicidal ideation prevalence data were extracted from 24 cross-sectional studies (n = 21 002) from 15 countries. All but 1 study used self-report instruments. The overall pooled crude prevalence of suicidal ideation was 11.1% (2043/21 002 individuals; 95% CI, 9.0% to 13.7%, I2 = 95.8%). Summary prevalence estimates ranged across assessment modalities from 7.4% to 24.2%.
In this systematic review, the summary estimate of the prevalence of depression or depressive symptoms among medical students was 27.2% and that of suicidal ideation was 11.1%. Further research is needed to identify strategies for preventing and treating these disorders in this population.
Studies have suggested that medical students experience high rates of depression and suicidal ideation.1 However, estimates of the prevalence of depression or depressive symptoms among students vary across studies from 1.4% to 73.5%,2,3 and those of suicidal ideation vary from 4.9% to 35.6%.4,5 Studies also report conflicting findings about whether student depression and suicidality vary by undergraduate year, sex, or other characteristics.6–11
Reliable estimates of depression and suicidal ideation prevalence during medical training are important for informing efforts to prevent, treat, and identify causes of emotional distress among medical students,12 especially in light of recent work revealing a high prevalence of depression in resident physicians.13 We conducted a systematic review and meta-analysis of published studies of depression, depressive symptoms, and suicidal ideation in undergraduate medical trainees.
Two authors (M.A.R. and D.A.M.) independently identified cross-sectional and longitudinal studies published prior to September 17, 2016, that reported on the prevalence of depression, depressive symptoms, or suicidal ideation in medical students by systematically searching EMBASE, ERIC, MEDLINE, psycARTICLES, and psycINFO. In addition, the authors screened the reference lists of identified articles and corresponded with study investigators using the approaches implied by the Preferred Reporting Items for Systematic Reviews and Meta-analyses and Meta-analysis of Observational Studies in Epidemiology reporting guidelines.14,15
For the database searches, terms related to medical students and study design were combined with those related to depression and suicide without language restriction (complete details of the search strategy appear in eMethods 1 in the Supplement). Included studies (1) reported data on medical students, (2) were published in peer-reviewed journals, and (3) used a validated method to assess for depression, depressive symptoms, or suicidal ideation.16 A third author (L.S.R.) resolved discrepancies by discussion and adjudication.
Three authors (L.S.R., M.T., and J.B.S.) independently extracted the following data from each article using a standardized form: study design; geographic location; years of survey; year in school; sample size; average age of participants; number and percentage of male participants; diagnostic or screening method used; outcome definition (ie, specific diagnostic criteria or screening instrument cutoff); and reported prevalence estimates of depression, depressive symptoms, or suicidal ideation. Whether students who screened positive for depression sought psychiatric or other mental health treatment also was extracted. When there were studies involving the same population of students, only the most comprehensive or recent publication was included.
The same 3 authors independently assessed the risk of bias of these nonrandomized studies using a modified version of the Newcastle-Ottawa scale, which assesses sample representativeness and size, comparability between respondents and nonrespondents, ascertainment of depressive or suicidal symptoms, and thoroughness of descriptive statistics reporting (complete details regarding scoring appear in eMethods 2 in the Supplement).17 Studies were judged to be at low risk of bias (≥3 points) or high risk of bias (<3 points). A fourth author (D.A.M.) resolved discrepancies through discussion and adjudication.
Prevalence estimates of depression or depressive symptoms and suicidal ideation were calculated by pooling the study-specific estimates using random-effects meta-analyses that accounted for between-study heterogeneity.18 The same approach was used to estimate the summary percentage of students screening positive for depression who sought treatment. When studies reported point prevalence estimates made at different periods within the year, the overall period prevalence was used. Standard χ2 tests and the I2 statistic (ie, the percentage of variability in prevalence estimates due to heterogeneity rather than sampling error, or chance, with values ≥75% indicating considerable heterogeneity) were used to assess between-study heterogeneity.19,20
Sensitivity analyses were performed by serially excluding each study to determine the influence of individual studies on the overall prevalence estimates. Results from studies grouped according to prespecified study-level characteristics were compared using stratified meta-analysis (for diagnostic criteria or screening instrument cutoff, study design, undergraduate level, continent or region, country, and Newcastle-Ottawa Scale components) or random-effects meta-regression (for year of baseline survey, age, and sex).21,22 To isolate associations within the medical school experience from associations with assessment tools, an analysis restricted to longitudinal studies reporting both pre- and intramedical school depressive symptom prevalence estimates was performed.
Bias secondary to small study effects was investigated using funnel plots and the Egger test.23,24 All analyses were performed using R version 3.2.3 (R Foundation for Statistical Computing).25 Statistical tests were 2-sided and used a significance threshold of P < .05.
One hundred ninety-five studies2–11,26–210 involving a total of 129 123 individuals in 47 countries were included in the analysis (Figure 1). The median number of participants per study was 336 (range, 44–10 140). One hundred sixty-seven cross-sectional studies2–4,6–9,11,26–184 (n = 116 628) and 16 longitudinal studies10,196–210 (n = 5728) in 43 countries reported on depression or depressive symptom prevalence (Table 1). Twenty-four cross-sectional studies (n = 21 002) in 15 countries reported on the prevalence of suicidal ideation (Table 2).4,5,34,62,65,73,74,79,112,160,165,167,174,185–195
Medical student training level, continent or region, country, diagnostic criteria or screening instrument cutoff, and total Newcastle-Ottawa scores for the studies appear in eTable 1 in the Supplement. Newcastle-Ottawa score components for all 195 individual studies appear in eTable 2 in the Supplement.
Meta-analytic pooling of the prevalence estimates of depression or depressive symptoms reported by 183 studies yielded a crude summary prevalence of 27.2% (37 933/122 356 individuals; 95% CI, 24.7%–29.9%), with significant evidence of between-study heterogeneity (Q = 16721.1, τ2 = 0.78, I2 = 98.9%, P < .001) (Figures 2, ,3,3, ,4,4, ,5,5, and and6).6). The prevalence estimates reported by the individual studies ranged from 1.4% to 73.5%. Sensitivity analysis, in which the meta-analysis was serially repeated after exclusion of each study, demonstrated that no individual study affected the overall prevalence estimate by more than 0.3% (eTable 3 in the Supplement).
To further characterize the range of depression or depressive symptom prevalence estimates identified by these methodologically diverse studies, meta-analyses stratified by screening instrument and cutoff score were conducted (Figure 7). Summary prevalence estimates ranged from 9.3% (157/1234 individuals [95% CI, 5.3%–15.7%]; Q = 19.7, τ2 = 0.24, I2 = 84.8%) for the Hospital Anxiety and Depression Scale with a cutoff score of 11 or greater to 55.9% (540/1039 individuals [95% CI, 45.1%–66.2%]; Q = 32.9, τ2 = 0.18, I2 = 90.9%) for the Aga Khan University Anxiety and Depression Scale with a cutoff score of 19 or greater. The median summary prevalence was 32.4% (5042/19 160 individuals [95% CI, 25.8%–39.7%]; Q = 1665.3, τ2 = 0.62, I2 = 98.6%) for the Beck Depression Inventory (BDI) with a cutoff score of 10 or greater.
Among medical students who screened positive for depression,15.7%(110/954 individuals [95%CI,10.2%–23.4%]; Q = 20.1, τ2 = 0.26, I2 = 70.1%) reportedly sought psychiatric or other mental health treatment as assessed by a subset of 7 studies reporting this information (eFigure 1 in the Supplement).
No statistically significant differences in prevalence estimates were noted between cross-sectional studies (36 632/116 628 [27.3%; 95%CI, 24.7%–30.1%]) and longitudinal studies (1301/5728 [26.7%; 95%CI, 19.1%–36.1%]) (test for subgroup differences, Q = 0.02, P = .90) or studies performed in the United States (14 356/36 249 [26.7%; 95% CI, 22.5%–31.3%]) compared with those performed outside the United States (23 577/86 107 [27.4%; 95% CI, 24.5%–30.6%]) (Q = 0.08, P = .78). Studies were further stratified by continent or region in Figure 8. Prevalence estimates from studies limited to preclinical students (4866/25 462 [23.7%; 95% CI, 19.5%–28.5%]) did not significantly differ from estimates from studies limited to clinical students (2917/13 172 [22.4%; 95% CI, 17.6%–28.2%]) (Q = 0.13, P = .72).
Prevalence estimates did not significantly vary with baseline survey year (survey year range, 1982–2015; slope = 0.2% 1-year increase [95%CI, −0.2%to0.7%]; Q = 1.17, P = .28). There Were no significant associations between prevalence and mean or median age (slope = 0.2%per 1-year increase [95%CI, −1.4% to 1.8%]; Q = 0.07, P = .79) or sex (slope = −1.1% per percentage increase in male study participants [95% CI, −15.9% to 13.7%]; Q = 0.02, P = .88).
When evaluated by Newcastle-Ottawa criteria, higher prevalence estimates were found among studies with more representative participant populations (24 366/68 693; 36.3% [95%CI, 29.9%–43.3%]) compared with those with less representative participant populations (13 567/53 663; 25.4% [95% CI, 22.8%–28.2%]) (Q = 9.6, P = .002; Figure 9). There were no statistically significant differences in prevalence estimates when studies were stratified by sample size, respondent and nonrespondent comparability, validity of ascertainment of depression or depressive symptoms (details regarding determination of screening instrument validity appear in eMethods 2 in the Supplement), thoroughness of descriptive statistics reporting, or total Newcastle-Ottawa score (P > .05 for all comparisons).
To identify potential sources of heterogeneity independent of assessment modality, heterogeneity was examined within subgroups of studies using common instruments when at least 6 studies were available (complete results appear in eTable 4 in the Supplement). No significant differences between cross-sectional and longitudinal studies were observed within any instruments when at least 3 studies were in each comparator subgroup.
Heterogeneity was partially accounted for by country with US studies yielding lower depression or depressive symptom prevalence estimates than non-US studies among the 24 studies using the BDI and a cutoff score of 10 or greater (13.0% vs 37.5%, respectively; Q = 12.7, P < .001) and the 13 studies using the Center for Epidemiological Studies Depression Scale (CES-D) and a cutoff score of 16 or greater (34.4% vs 50.3%; Q = 3.8, P = .05). However, this difference was not seen among other instruments.
Level of training did not significantly contribute to between study heterogeneity among any of the examined instruments. Year of baseline survey significantly contributed to observed statistical heterogeneity among 3 instruments, although the results were inconsistent (ie, 2 analyses suggested that depression was increasing with time, whereas a third suggested it was decreasing). Age and sex were not significantly associated with depression prevalence among any instruments.
The temporal relationship between exposure to medical school and depressive symptoms was assessed in an analysis of 9 longitudinal studies that measured depressive symptoms before and during medical school (Table 3). Because studies used different assessment instruments, the relative change in depressive symptoms was calculated for each study individually (ie, follow-up prevalence divided by baseline prevalence) and then the relative changes derived from the individual studies were examined. Overall, the median absolute increase in depressive symptoms was 13.5% (range, 0.6%–35.3%) following the onset of medical training.
In an analysis of 24 studies, the crude summary prevalence of suicidal ideation, variably reported as having occurred over the past 2 weeks to the past 12 months, was 11.1% (2043/21 002 individuals; 95%CI, 9.0%–13.7%), with significant evidence of between-study heterogeneity (Q = 547.1, τ2 = 0.32, I2 = 95.8%, P < .001) (Figure 10). The prevalence estimates reported by the individual studies ranged from 4.9% to 35.6%. Sensitivity analysis showed that no individual study affected the overall pooled estimate by more than 1.9%(eTable 5 in the Supplement).
To further characterize the range of the suicidal ideation prevalence estimates identified, stratified meta-analyses were performed by screening instrument and cutoff score. Summary prevalence estimates ranged from 7.4% (69/938 individuals [95% CI, 5.9%–9.2%]; Q = 0.01, τ2 = 0, I2 = 0%) over the past 2 weeks for studies using the 9-item Patient Health Questionnaire (PHQ-9) to 24.2% (208/754 individuals [95% CI, 13.0%–40.5%]; Q = 37.2, τ2 = 0.42, I2 = 94.6%) over the past 12 months for studies using the 28-item General Health Questionnaire.
The median prevalence of suicidal ideation over the past 12 months reported by 7 studies using variably worded short-form screening instruments was 10.2% (723/8636 individuals [95% CI, 6.8%–15.0%]; Q = 176.5, τ2 = 0.33, I2 = 96.6%). Among the full set of studies, no statistically significant differences in prevalence estimates were noted by country (United States vs other countries), continent or region, level of training, baseline survey year, average age, proportion of male study participants, or total Newcastle-Ottawa score (P > .05 for all comparisons). Within-instrument heterogeneity was not examined because there were not enough studies using identical screening instruments (≤4 for each assessment modality), precluding meaningful analysis.
Visual inspection of the funnel plot of studies reporting on depression or depressive symptoms revealed significant asymmetry (eFigure 2 in the Supplement). There was evidence of publication bias, with smaller studies yielding more extreme prevalence estimates (P = .001 using the Egger test). The funnel plot of studies reporting on suicidal ideation revealed minimal asymmetry (eFigure 3 in the Supplement), suggesting the absence of significant publication bias (P = .49 using the Egger test).
This systematic review and meta-analysis of 195 studies involving 129 123 medical students in 47 countries demonstrated that 27.2% (range, 9.3%–55.9%) of students screened positive for depression and that 11.1% (range, 7.4%–24.2%) reported suicidal ideation during medical school. Only 15.7% of students who screened positive for depression reportedly sought treatment. These findings are concerning given that the development of depression and suicidality has been linked to an increased short-term risk of suicide as well as a higher long-term risk of future depressive episodes and morbidity.211,212
The present analysis builds on recent work demonstrating a high prevalence of depression among resident physicians, and the concordance between the summary prevalence estimates (27.2% in students vs 28.8% in residents) suggests that depression is a problem affecting all levels of medical training.13,213 Taken together, these data suggest that depressive and suicidal symptoms in medical trainees may adversely affect the long-term health of physicians as well as the quality of care delivered in academic medical centers.214–216
When interpreting these findings, it is important to recognize that the data synthesized in this study were almost exclusively derived from self-report inventories of depressive symptoms that varied substantially in their sensitivity and specificity for diagnosing major depressive disorder (eTable 6 in the Supplement).217 Instruments such as the PHQ-9 have high sensitivity and specificity for diagnosing major depression, whereas others such as the Primary Care Evaluation of Mental Disorders (PRIME-MD) have low specificity and should be viewed as screening tools. Although these self-report measures of depressive symptoms have limitations, they are essential tools for accurately measuring depression in medical trainees because they protect anonymity in a manner that is not possible through formal diagnostic interviews.218 To control for the differences in these inventories, we stratified our analyses by survey instrument and cutoff score, identifying a range of estimates not captured in another evidence synthesis.219
The prevalence of depressive symptoms among medical students in this study was higher than that reported in the general population.220–222 For example, the National Institute of Mental Health study of behavioral health trends in the United States, including 67 500 nationally representative participants, found that the 12-month prevalence of a major depressive episode was 9.3% among 18- to 25-yearolds and 7.2% among 26- to 49-year-olds.220 In contrast, the BDI, CES-D, and PHQ-9 summary estimates obtained in the present study were between 2.2 and 5.2 times higher than these estimates. These findings suggest that depressive symptom prevalence is substantially higher among medical students than among individuals of similar age in the general population.
How depression levels in medical students compare with those in nonmedical undergraduate students and professional students is unclear. One review concluded that depressive symptom prevalence did not statistically differ between medical students and nonmedical undergraduate students.223 However, this conclusion may be confounded because the analysis did not control for assessment modality and did not include a comprehensive or representative set of studies (only 12 studies and 4 studies exclusively composed of medical students and nonmedical students, respectively). Two large, representative epidemiological studies have estimated that depressive symptom prevalence in nonmedical students ranges from 13.8% to 21.0%, lower than the estimates reported by many studies of medical students in the present meta-analysis.224,225
Some professional students, such as law students, may not markedly differ from medical students in their susceptibility to depression, although firm conclusions cannot be drawn from the currently available data.226,227 Together, these findings suggest that factors responsible for depression in medical students may also be operative in other undergraduate and professional schools. The finding in the longitudinal analysis of an increase in depressive symptom prevalence with the onset of medical school suggests that it is not just that medical students (and other students) are prone to depression, but that the school experience may be a causal factor.
This analysis identified a pooled prevalence of suicidal ideation of 11.1%. Endorsement of suicidal ideation as assessed by the PHQ-9 or other similar instruments increases the cumulative risk of a suicide attempt or completion over the next year by 10- and 100-fold, respectively.228 Combined with the finding that only 15.7%of medical students who screened positive for depression sought treatment, the high prevalence of suicidal ideation underscores the need for effective preventive efforts and increased access to care that accommodate the needs of medical students and the demands of their training.
This study has important limitations. First, the data were derived from studies that had different designs, screening instruments, and trainee demographics. The substantial heterogeneity among the studies remained largely unexplained by the variables inspected. Second, many subgroup analyses relied on unpaired cross-sectional data collected at different medical schools, which may cause confounding. Third, because the studies were heterogeneous with respect to screening inventories and student populations, the prevalence of major depression could not be determined. Fourth, the analysis relied on aggregated published data. A multicenter, prospective study using a single validated measure of depression and suicidal ideation with structured diagnostic interviews in a random subset of participants would provide a more accurate estimate of the prevalence of depression and suicidal ideation among medical students.
Because of the high prevalence of depressive and suicidal symptomatology in medical students, there is a need for additional studies to identify the root causes of emotional distress in this population. To provide more relevant information, future epidemiological studies should consider adopting prospective study designs so that the same individuals can be assessed over time, use commonly used screening instruments with valid cutoffs for assessing depression in the community (eg, the BDI, CES-D, or PHQ-9), screen for comorbid anxiety disorders, and completely and accurately report their data, for example, by closely following the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.229
Possible causes of depressive and suicidal symptomatology in medical students likely include stress and anxiety secondary to the competitiveness of medical school.62 Restructuring medical school curricula and student evaluations (such as using a pass-fail grading schema rather than a tiered grading schema and fostering collaborative group learning through a “flipped-classroom” education model) might ameliorate these stresses.230,231 Future research should also determine how strongly depression in medical school predicts depression during residency and whether interventions that reduce depression in medical students carry over in their effectiveness when those students transition to residency.232 Furthermore, efforts are continually needed to reduce barriers to mental health services, including addressing the stigma of depression.146,233
In this systematic review, the summary estimate of the prevalence of depression or depressive symptoms among medical studentswas27.2%andthatof suicidal ideationwas11.1%. Further research is needed to identify strategies for preventing and treating these disorders in this population.
Are medical students at high risk for depression and suicidal ideation?
In this meta-analysis, the overall prevalence of depression or depressive symptoms among medical students was 27.2%, and the overall prevalence of suicidal ideation was 11.1%. Among medical students who screened positive for depression, 15.7% sought psychiatric treatment.
The overall prevalence of depressive symptoms among medical students in this study was higher than that reported in the general population, which underscores the need for effective preventive efforts and increased access to care for medical students.
Funding/Support: Funding was provided by the National Institutes of Health (MSTP TG 2T32GM07205 awarded to Mr Ramos and grant R01MH101459 awarded to Dr Sen) and the US Department of State (Fulbright Scholarship awarded to Dr Mata).
Conflict of Interest Disclosures: The authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Role of the Funder/Sponsor: The National Institutes of Health and the US Department of State had no role in the design and conduct of the study; the collection, management, analysis, or interpretation of the data; the preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication.
Disclaimer: The opinions, results, and conclusions reported in this article are those of the authors and are independent from the National Institutes of Health and the US Department of State.
Author Contributions: Dr Mata had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the analysis. Ms Rotenstein, Messrs Ramos and Segal, and Dr Torre are equal contributors.Concept and design: Mata.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Mata.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Mata.
Obtained funding: Guille, Sen, Mata.
Administrative, technical, or material support: Guille, Sen, Mata.
Study supervision: Guille, Sen, Mata.