INTRODUCTION

The COVID-19 pandemic impacted all aspects of medical education in the United States (US).1,2 The Association of American Medical Colleges (AAMC) recommended suspending direct patient contact for students in March 2020, and many schools removed students from clinical rotations and transitioned to remote learning.3,4 As a result, students’ academic and social connections with their learning communities and peers were drastically altered.

Studies early on in the pandemic found high rates of burnout, anxiety, and post-traumatic stress in nurses, residents, and physicians.5,6,7 While initial studies among US and international medical students described increased feelings of stress, anxiety, fear, and uncertainty during the pandemic, none have explored burnout, loneliness, or personal experiences of students from different demographic groups.8,9,10,11,12 As a natural experiment, student response to pandemic-related stressors may help schools better understand risk factors for distress, identify protective factors, and develop targeted programming to support students in the wake of the pandemic and during other times of stress. This is important since medical students experience higher baseline burnout, stress, depression, anxiety, and other mental distress than non-medical, age-matched peers, and few studies explore protective factors for burnout and distress during times of stress.13,14,15,16,17,18,19

We aimed to conduct a national, cross-sectional survey of US medical students during the initial phase of the COVID-19 pandemic between May and June 2020 to (1) assess burnout, stress, and loneliness prevalence; (2) compare burnout and stress results to baseline pre-pandemic data; and (3) identify risk factors for distress and potential protective factors to inform support interventions.

METHODS

Survey Development

Our survey assessed student demographic characteristics (e.g., age, gender, race/ethnicity, year in school), burnout, stress, loneliness, life experiences likely impacted by the pandemic (e.g., personal experiences with COVID-19 diagnoses, racism, financial strain), and pandemic volunteer activities. Survey questions exploring students’ life experiences during the pandemic were developed after a literature review and discussion with wellness leaders at the 22 participating medical schools. Burnout and stress were measured using the most commonly utilized validated instruments in medical education literature [Maslach Burnout Inventory-Human Services Survey (MBI-HSS) and Perceived Stress Scale (PSS-10)].20,21,22,23,24,25,26,27,28,29,30,31,32,33,34 Loneliness was assessed with the University of California Los Angeles (UCLA) Loneliness Scale to assess social distancing impacts on loneliness as a contributor to student well-being. We piloted the survey with eight medical students from participating schools, and their feedback was incorporated. The final survey consisted of 87 items including Likert scale, multiple choice, and free response questions (Supplemental Digital Appendix 1). The institutional review board at the University of Chicago (IRB17-1095) and all participating schools approved the study.

Burnout

The MBI-HSS contains 22 items and three subscales: emotional exhaustion (EE) (score range 0–54), depersonalization (DP) (score range 0–30), and personal accomplishment (PA) (score range 0–48).20,21,22 Consistent with previous studies, those with high scores on the EE (≥ 27) and/or DP (≥ 10) subscales were considered to have burnout.23,24,25 Personal accomplishment scores of ≤ 33 were considered low.23,24,25

Stress

The PSS-10 is rated on a 5-point scale, ranging from 0 (never) to 4 (very often). Total possible scores range from 0 to 40, with higher scores correlating with higher stress levels.35

Loneliness

A validated, three-item version of the UCLA Loneliness Scale was included in the survey.36 While this version has not been extensively studied in medical students, it is validated against the original 20-item instrument in the general US population.36,37 Each item is rated from 1 (hardly ever) to 3 (often), for a total score of 3–9. Higher scores indicate greater loneliness, and ≥ 6 indicates high loneliness.38

Comparison with Pre-pandemic Burnout and Stress in Medical Students

To assess whether the pandemic impacted burnout and stress, we conducted a PubMed search for studies published between January 2010 and January 2020 that assessed medical student burnout and stress. The search was limited to studies using the MBI-HSS or PSS-10 to allow for direct comparison of results. Non-English studies conducted outside the US or Canada were excluded.

Participants

In April 2020, two investigators (W.W.L., a student affairs dean, and M.A., a pediatric clerkship director) sent recruitment emails and reminders at weeks 2 and 3 to national electronic listservs for medical student well-being professionals, student affairs deans, and pediatric clerkship directors to invite Liaison Committee on Medical Education (LCME)–accredited schools to join our study. Of the 34 schools who responded, 22 obtained IRB and leadership approval to participate. A faculty leader was designated at each institution. School-related data (e.g., school size, type (public or private), and AAMC geographic region) was obtained through publicly available resources.39,40,41,42,43,44,45,46

Survey Distribution

The survey was open for 4 weeks at each school between May and June 2020. Faculty site leaders emailed all enrolled students to participate and sent weekly reminders. Participation was voluntary, and the anonymous survey was administered using REDCap, a secure application for managing online surveys (v8.9.2 C 2019 Vanderbilt University) (Supplemental Appendix 1). Data was stored on REDCap and only accessible to University of Chicago. Participants were eligible to enter a raffle for one of ten $50 gift cards if permissible by school policy.

Data Analysis

Standard descriptive statistics were used for analysis, and a two-sided significance level of 0.05 was set for all analyses. T tests, chi-square tests, one-way ANOVA, and unadjusted odds ratios examined associations between student experiences, burnout, stress, and loneliness with covariates such as class year, race, gender, AAMC region (e.g., Northeast), school type (public or private), and size. T tests and chi-square tests compared burnout rates and stress scores between pre-pandemic and pandemic study data. All analyses were performed in R 4.0.2.

Results

Of 12,389 students, 3826 (31%) responded to the survey. While age, race, year in medical school, and geographic location were representative of national medical school enrollment data, we had more female students in our study (62%, 2327/3762).41,42,43,44,45 Table 1 summarizes respondent characteristics.

Table 1 Characteristics of 3762 US Medical Student Respondents to a Survey on Medical Student Well-Being During the COVID-19 Pandemic, 2020
Full size table

Of 22 participating schools, 13 (59%) were public and 9 (41%) were private. Thirteen (59%) schools reported > 500 students enrolled and nine (41%) reported < 500 enrolled. Our sample was representative of enrollment and school-type data for all US medical schools.42,46 Most participating schools (8, 36%) were from the AAMC Northeast region, with 6 (27%) from the Southern region, 5 (23%) from the Central region, and 3 (14%) from the Western region. In comparison to the national distribution of AAMC schools, our study overrepresented the Northeast and underrepresented the Southern region.41

Student Experiences

Thirty percent (1145/3756) of students reported financial strain during the pandemic. Seventeen percent (624/3756) reported family members being diagnosed with COVID-19, 4% (159/3756) had COVID-19 themselves, and 4% (147/3756) experienced a COVID-19–related death of a loved one. COVID-19–related racism or bias was reported by 8% (319/3756) of students. There were differences across race for students reporting personal experiences with COVID-19 (defined as COVID-19 diagnoses in themselves or a family member, or death of loved one from COVID-19), and those who reported racism due to COVID-19 and financial strain. More Black (35%, 83/234) and Hispanic (26%, 101/395) students had personal experiences with COVID-19 diagnoses, more Asian (26%, 209/811) and Black (14%, 32/234) students experienced COVID-19–related racism, and almost half of Hispanic (47%, 184/395) and Black (45%, 106/234) students experienced financial strain during the pandemic (Table 2). Compared to White students, Black students were more likely to have personal experiences with COVID-19 (OR 2.27, 95% CI 1.70–3.03), experience financial strain (OR 2.31, 95% CI 1.75–3.04), and experience COVID-19–related racism (OR 15.18, 95% CI 8.63–27.27). Compared to White students, Hispanic students were more likely to have personal experience with COVID-19 (OR 1.42, 95% CI 1.10–1.82), experience financial strain (OR 2.43, 95% CI 1.95–3.04), and experience COVID-19–related racism (OR 8.17, 95% CI 4.66–14.61). Compared to White students, Asian students were more likely to experience COVID-19–related racism (OR 33.17, 95% CI 21.49–53.96).

Table 2 Experiences of US medical student respondents from a survey on medical student well-being during the COVID-19 pandemic, 2020
Full size table

Burnout

Half (1635/3296) of students experienced burnout with 29% (948/3296) reporting high depersonalization, 40% (1324/3296) reporting high emotional exhaustion, and 42% (1389/3296) reporting low personal accomplishment. No significant difference was found by sex (52%, 623/1209 males vs. 48%, 987/2047 females, P = 0.07). There was a significant difference in burnout across year in school (P = 0.002) and race (P = 0.004), with higher burnout reported among second- (54%, 398/731) and third-year students (51%, 425/840) and those identifying as “other” (e.g., Native American, Pacific Islander, or “other”) racial minority (55%, 126/229), Asian (54%, 378/706), and Black (51%, 98/194).

Burnout prevalence was greater in students reporting financial strain [59% (580/989) vs. 46% (1055/2304), P < 0.001] and COVID-19–related racism [63% (168/266) vs. 48% (1467/3027), P < 0.001] during the pandemic compared to those without these experiences (Fig. 1). While there was no difference in burnout among students with personal COVID-19 experiences compared to those without these experiences [52% (342/655) vs. 49% (1293/2638), P = 0.16], students with personal COVID-19 experiences reported higher emotional exhaustion [44% (289/655) vs. 39% (1035/2639), P = 0.02]. Burnout was higher in private compared to public schools [52% (743/1433) vs. 48% (892/1862), P = 0.03]. No significant association was found between burnout and school region or size (Table 3).

Figure 1
figure 1

Burnout, stress, and loneliness among US medical student respondents with pandemic-related experiences. A Percentage of students who were burned out vs. student experiences (N = 3296). Burnout was assessed using the Maslach Burnout Inventory-Human Services Survey (MBI-HSS) which contains three subscales: emotional exhaustion (EE) (score range 0–54), depersonalization (DP) (score range 0–30), and personal accomplishment (PA) (score range 0–48).20 Those with high scores on the EE (≥ 27) and/or DP (≥ 10) subscales were considered to have burnout. Overall percent of all 3296 students who were burned out is included for reference. B Mean stress score vs. student experiences (N = 3250). Stress was measured using the 10-item Perceived Stress Scale (PSS-10), which has been used to assess stress in medical students. Total possible scores range from 0 to 40, with higher scores indicating higher levels of stress. The overall mean stress score for all 3250 students is included for reference. C Mean loneliness score vs. student experiences (N = 3247). Loneliness was measured using the 3-item University of California Los Angeles Loneliness Scale. Scores range from 3 to 9, with higher scores indicating greater loneliness. The overall mean loneliness score for all 3247 students is included for reference. * COVID-19 diagnosis in either the student or a family member, or having a loved one die from COVID-19. † P < 0.05.

Full size image
Table 3 Burnout, stress, and loneliness in US medical student respondents from a survey on medical student well-being during the COVID-19 pandemic, 2020
Full size table

Comparison of Burnout Prevalence to Pre-pandemic Data

Five pre-pandemic studies reporting MBI-HSS burnout prevalence in a total of 9493 medical students met inclusion criteria and were used in our analysis.15,17,26,27,28 Compared to pre-pandemic data, burnout in our study was significantly lower [52% (4918/9493) vs. 50% (1635/3296), respectively, P = 0.03] (Table 4).

Table 4 Burnout and stress among us medical student respondents: comparison of student respondents to a well-being survey during COVID-19 Pandemic (2020) to Pre-pandemic burnout and stress studies
Full size table

Stress

The overall mean (SD) stress score was 18.9 (6.7, n = 3250), with significantly higher mean scores reported in female than male students [19.8 (6.6) vs. 17.4 (6.6), P < 0.001]. There was a significant difference across race (P < 0.001) and medical school year (P < 0.001), with higher mean stress scores among Blacks [20.2 (6.3)] and other racial minorities [20.4 (6.8)], and second- [20.2 (6.7)] or third-year students [19.2 (6.7)] (Table 3).

Mean stress scores were significantly higher among students reporting personal COVID-19 experiences [20.0 (6.8) vs. 18.7 (6.7), P < 0.001], COVID-19–related racism [21.0 (6.9) vs. 18.8 (6.7), P < 0.001], and financial strain [20.8 (6.6) vs. 18.2 (6.6), P < 0.001] during the pandemic compared to those without these experiences.

Mean stress scores were significantly higher in smaller schools (< 500 enrolled students) compared with larger schools (≥ 500 enrolled students) [19.3 (6.6) vs. 18.7 (6.7), P = 0.01]. Southern [19.3 (7.0)] and Western [19.7 (6.6)] region students reported significantly higher mean stress scores (P = 0.006). There was no significant association between stress and school type.

Comparison of Mean Stress to Pre-pandemic Data

Four pre-pandemic studies reporting PSS-10 stress scores in a total of 425 medical students met inclusion criteria and were used for analysis.31,32,33,34 Students in our study reported higher mean stress scores compared to pre-pandemic data [18.9 (6.7) vs. 16.0 (6.2), P < 0.001] (Table 4).

Loneliness

Mean (SD) loneliness scores were 5.5 (1.9), and half of the students (1609/3247) reported high levels of loneliness (≥ 6/9). Females had significantly higher mean loneliness scores compared with males [5.6 (1.8) vs. 5.4 (1.8), P = 0.009]. Significant differences were found across medical school year (P < 0.001), with higher mean scores among first- [5.8 (1.8)] and second-year [5.5 (1.8)] students. No significant differences were found across race/ethnicity.

Mean loneliness scores were significantly higher among students reporting personal COVID-19 experiences [5.8 (1.9) vs. 5.5 (1.8), P = 0.001], COVID-19–related racism [5.9 (2.0) vs. 5.5 (1.8), P = 0.002], and financial strain [5.9 (1.9) vs. 5.4 (1.8), P < 0.001] during the pandemic compared to those without these experiences. Significant differences were found across AAMC regions (P = 0.01), with the highest mean loneliness scores in Western [5.8 (SD 1.9)] region students. There were no significant associations between loneliness and school size or type (Table 4).

Volunteer Activities

Those who volunteered during the pandemic were less likely to report burnout [48% (782/1639) vs. 52% (853/1656), P = 0.03], high EE [38% (616/1640) vs. 43% (708/1656), P = 0.003], and low PA [37% (606/1639) vs. 47% (783/1657), P < 0.001] than those who did not volunteer. No significant differences were found between pandemic volunteer and non-volunteer students for stress [18.8 (6.5) vs. 19.1 (6.9), P = 0.11] and loneliness [5.5 (1.8) vs. 5.5 (1.9), P = 0.89].

DISCUSSION

This is the first study to assess how the pandemic affected burnout and loneliness among US medical students and to describe differential impacts across demographic and student experiences (e.g., race/ethnicity, class year, financial strain, racism, etc.). Stress levels among medical students were higher compared to pre-pandemic data, and stress levels were higher in medical students during the pandemic compared with the general population.47 Our results were also concordant with other studies that found higher levels of medical student stress during the pandemic compared to baseline.31,32,33,34

We are the first study to report burnout rates among US medical students during the pandemic, and our finding of lower burnout suggests that higher stress may not necessarily lead to higher burnout, which differs from pre-pandemic studies.48,49,50,51 Decreased burnout may be related to removal of students from clinical responsibilities, transition to flexible curricular policies (e.g., pass-fail grading) at some schools, and increased time for self-care. Unsurprisingly, loneliness was high, with findings comparable to those in the general young adult US population during the same period.52

Our results highlight the inequities of the pandemic, which disproportionately impacted stress and burnout among Black and other racial minority students, and those experiencing financial strain or COVID-19–related racism.38 Second- and third-year students also experienced high levels of burnout and stress, which may be related to testing center closures for the United States Medical Licensing Examination Step 1 exams and clerkship students’ concerns about disruptions to their clinical training and career planning.1,5 Additionally, first- and second-year students had higher levels of loneliness, which could be related to isolation due to remote learning. Strikingly, Black and Hispanic students experienced markedly higher rates of financial strain and COVID-19 diagnoses in themselves or loved ones, and Asian students experienced the highest rates of pandemic-related racism. Together, these findings reinforce the need for proactive and tailored support for students during times of distress. For example, moving forward, schools should proactively support students with financial stressors, which is often a hidden identity and one heavily impacted by the pandemic. Additionally, in light of national events centered on racially motivated violence, schools should initiate targeted support for vulnerable student groups, with specific attention to understanding Black and Asian student experiences when developing anti-racism initiatives.

Medical students volunteered in large numbers and in a variety of roles, such as helping to secure masks, staffing call centers, and conducting contact tracing.53 Students who volunteered had lower burnout and emotional exhaustion, and greater levels of personal accomplishment compared to non-volunteers. While volunteers may be a self-selected group of students with lower baseline burnout, prior studies suggest volunteering may promote feelings of purpose, value, personal achievement, and empathy, which may contribute to the lower burnout among volunteers in our study.54,55,56,57 Similarly, a recent survey of US healthcare workers conducted in the same time period as our study found that feeling valued was associated with lower burnout and enhanced meaning and purpose.58 At a time when students were pulled from clinical duties, volunteering may have been an adaptive coping mechanism that helped them feel more accomplished and connected to their communities which may have contributed to lower burnout.

Additionally, some students may have experienced professional growth and heightened motivation as a direct consequence of the stressful and disorienting effects of the pandemic. Such development is consistent with the concept of transformative learning where disorienting experiences prompt critical reflection and favorably change an individual’s perspective of themselves and their place in the world.59 Post-traumatic growth documented in Japanese medical students after the Fukushima nuclear disaster is an example of this possible effect.54,56 These findings suggest further research is needed to identify specific personal attributes, support systems, and protective factors that promote growth and resilience, as opposed to burnout, as trainees face the inherent challenges of medical training and practice.

Our study has several limitations. While we report on experiences from a large group of diverse students around the country, we had a low overall response rate. The survey was administered between May and June of 2020 which may have limited fourth-year participation due to graduation. In addition, concurrent national stressors of racial injustice following George Floyd’s death and the polarized presidential election may have influenced responses. We did not assess timelines for the implementation of pandemic-related curricular changes (e.g., transition to virtual learning) at each school and were unable to assess whether specific curricular changes impacted student well-being. Lastly, students who were more distressed may not have participated, resulting in non-response bias, and our study represented a snapshot in time in the early phase of the pandemic. Longitudinal data is needed to assess ongoing impacts on student well-being and identify areas for ongoing support.

As the pandemic continues to evolve, medical schools should acknowledge the high levels of stress and burnout in their students and proactively support vulnerable students. This is particularly important since studies after the 2003 SARS outbreak found stress and burnout can persist long term among healthcare workers.5,60 Understanding and addressing sustained impacts of COVID-19–related distress on the well-being of trainees and physicians are critical to the development of a healthy healthcare workforce in the years ahead. Recognizing both drivers and protective factors for burnout and stress can inform programming that supports trainees through the rigors of medical training. Our data suggests that program leadership should promote initiatives to foster community and connectedness. Educational programs may also consider augmenting volunteer opportunities to promote engagement and professional fulfillment to help protect against burnout. In addition, institutional wellness and professional development programs should proactively support vulnerable trainees and develop tailored interventions for specific stressors. Given our finding that higher levels of stress do not necessarily correlate with higher burnout, medical educators should design interventions to help trainees develop adaptive coping skills in the face of stress, uncertainty, and loss to promote growth from stressful experiences.