Perceived Stress, Grit, and Self-Care Behaviors in First-Year Medical Students

Abstract

Medical students experience more stress than the general population, which over time can cause mental and physical disease, including burnout. Identifying factors impacting stress during early medical training could inform strategies to minimize its impacts throughout training and in clinical practice. This study surveyed 238 first-year osteopathic medical students to assess stress (Perceived Stress Scale; PSS), grit, sleep quality (Pittsburgh Sleep Quality Index; PSQI), physical activity (Godin-Shephard Leisure-Time Physical Activity Score; LTPA), and nutrition habits (Rapid Eating Assessment for Participants; REAP) within the first 2 weeks of starting medical school and again 10 weeks later. Incomplete responses were removed, leaving 204 study participants. We observed statistically significant decreases in grittiness (∆grit = −2.230%, P = .002) and physical activity (∆LTPA = −22.147%, P < .0001), while perceived stress (∆PSS = 34.548%, P < .0001) and poor sleep quality (∆PSQI% = 19.853, P < .0001) increased. Correlation analyses identified the strongest relationships were between ∆PSS vs ∆PSQI (r = .47, P < .0001) and ∆PSS vs ∆LTPA (r = −.20, P < .01). Multivariable linear regression analysis isolated ∆PSQI (P < .0001) and ∆LTPA (P = .012) as statistically significant predictors of ∆PSS. These results suggest early, repeated curricular interventions focused on physical activity and sleep hygiene may help students better manage stress during medical education.

“Poor sleep quality and decreased physical activity were significantly related to increased perceived stress.”

Introduction

Medical students report more stress than the general population.^1,2 The first year of medical school has unique stressors, often being reported as the most stressful year of training.^2,3 These stressors can include moving to a new city, region, or country; meeting a new group of people; learning multiple new technologies and techniques; and adapting to the rigorous medical school curriculum. Increased stress during preclinical medical education can contribute to symptoms of anxiety, depression, fatigue, and academic burnout. ¹ Left unaddressed, these symptoms persist and affect student health and performance during clerkship and residency, and ultimately contribute to physician burnout and attrition from the profession. ⁴ Physician burnout, a syndrome of emotional exhaustion, depersonalization, and reduced personal accomplishment, is of growing concern among medical educators.^5,6

Grit is defined as passion and perseverance for long-term goals, ⁷ and it has been identified as a protective character trait against burnout. ⁸ Research regarding the psychology of grit aids in understanding a noncognitive characteristic that bolsters an individual’s ability to achieve higher levels of education than peers of equal intelligence. ⁷ Grittier students may be better equipped to combat stressors associated with the medical training experience, and have improved longevity in the field. ⁹ For this reason, grit is of interest to medical schools, residencies, physicians, and administrators within healthcare systems hoping to maintain high-quality and long-serving staff.

In addition to grit, employing recognized self-care strategies can mitigate stress and improve medical student well-being. Self-care is typically defined as the conscious act of preserving or improving one’s own mental, physical, and/or spiritual well-being, and is associated with reduced perceived stress levels and incidence of burnout in medical students and health professionals. Self-care habits including practicing good sleep hygiene, maintaining nutritious eating habits, and engaging in physical activity have all been proposed as protective personal strategies for managing stress. ¹⁰ Unfortunately, self-care behaviors frequently worsen among medical students when they begin to feel stressed and adapt to their busy study schedules.^11-13 Compared to the general population, US medical students report worse sleep quality, and this has been associated with increased levels of perceived stress and poor academic performance.^13,14 Medical students tend to have healthy eating habits at matriculation and are more physically active when compared to their age related peer groups.^15,16 However, rigorous schedules and high stress associated with medical education fuel worsening eating and physical activity habits over time. ¹⁷ For example, medical students are more prone to skipping meals and relying on convenient, processed, sweet, or calorie-dense fast foods.^18,19 A survey of 16 US medical schools showed that fewer than 60% of students were able to meet CDC exercise recommendations. ²⁰ All of these factors together contribute to increased risk for future challenges in maintaining healthy professional and personal life balance for physicians.

Physician burnout has been shown to increase medical expenditures, decrease patient safety, worsen patient outcomes, and reduce physician satisfaction and career longevity. ⁶ Symptoms of burnout have been observed in medical students even during the preclinical years of training. ⁴ Decreasing risk of physician burnout is of interest to healthcare organizations and organizations involved with the training of medical students such as medical schools.^5,6 Increased and prolonged occupational stress is known to be a contributing factor to physician burnout. ²¹ Though burnout is not directly measured in this study, the strong connection between stress and burnout makes understanding this potential outcome pertinent to understanding the context of this study. Decreasing stress in preclinical education would be a logical target to decrease burnout in medical students and ultimately practicing physicians.

Although increased perceived stress and worsening self-care behaviors in medical students have been well-established in previous studies, the impact of medical students’ individual grittiness on these variables has not been well studied. The purpose of this study was to evaluate how perceived stress relates to individual character traits like grit, and to self-care behaviors like nutrition quality, physical activity, and sleep quality in medical students. We hypothesized that medical students with higher grit scores would show evidence of healthier self-care behaviors at baseline and follow-up, be better able to cope with the rigors of medical school, and experience lower levels of perceived stress over time.

IRB Statement

This study was reviewed and approved by the University’s Institutional Review Board—approval number: 2018-011-PNWU. All medical student participants consented to sharing their data (Age, MCAT, GPA, survey results) and to contributing to the study. Investigators were blinded as to which students chose to participate in the study. Consent to participate in the study was not linked to course credit and there were no incentives offered.

Subjects

Pacific Northwest University—College of Osteopathic Medicine (PNWU-COM) has a curriculum focused on training the next generation of physicians with a specific mission emphasizing service among rural and medically underserved communities throughout the northwest region of the U.S. The 4-year curriculum at PNWU-COM consists of 2 years of preclinical education and 2 years of clinical rotations. This study was conducted with first-year osteopathic medical students (OMS1s) during their first semester of medical school. The first course of the preclinical curriculum for OMS1 students is Scientific Foundations of Medicine (SciFOM), a course that focuses heavily on biochemistry, cellular biology and histology, genetics, and molecular physiology. OMS1 students are also simultaneously enrolled in gross anatomy, clinical skills, osteopathic principles and practice, and pharmacology. PNWU-COM has no formalized self-care curriculum, and no on-campus student activity center or gym but does provide students a limited fitness reimbursement for those services off-campus. There are on-campus self-serve markets and microwaves available to heat food, but no cafeteria offering freshly prepared meals for purchase.

Methods

Surveys were administered electronically through REDCap™ to 2 OMS1 classes during the fall semesters of 2018 and 2019. Baseline surveys were collected within the first 2-weeks of the SciFOM course and follow-up surveys were collected within the last 2 weeks of the SciFOM course. Data collected for this study were survey-based, using 5 previously validated survey instruments: the Grit Scale, Perceived Stress Scale (PSS), Pittsburgh Sleep Quality Instrument (PSQI), Godin-Shephard Leisure-Time Physical Activity questionnaire (LTPA), and Rapid Eating Assessment for Participants (REAP) questionnaire.^{7,15,16,22,23} From the full REAP questionnaire, we selected 12 items highly correlated to chronic disease health outcomes to generate a novel REAP Nutrition Quality Score (REAP-NQS) for this study, which was based on the validated Healthy Eating Index (HEI). ¹⁶

To evaluate stress, we employed the most widely used instrument for measuring perceived stress, validated across multiple populations, the PSS.^15,24 The PSS is a set of 10 generalized questions scored between 0 (Never) and 4 (Very Often) that gauge feelings and thoughts related to events that occurred over the last month. Higher scores indicated higher levels of perceived stress.

Grit was assessed using the previously validated Grit Scale, a 12-item questionnaire assessing an individual’s consistency, passion, and perseverance while pursuing long-term goals. ⁷ Since its development, it has been administered in many studies pertaining to academic success and personal well-being of medical students.^8,25 Each item is rated on a 5-point scale from 1 (Not at all like me) to 5 (Very much like me); grit score is an average of all 12 items with higher scores indicating a higher level of grit.

Sleep quality was assessed using the PSQI. ²³ The PSQI has been utilized in many studies concerning medical students to distinguish “good” from “poor” sleepers. ²⁶ There is evidence of reliability and validity when administering the PSQI to young adults enrolled in higher education. ²⁷ The 19 items on the PSQI questionnaire were grouped into 7 components: sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleeping medications, and daytime dysfunction. Each component score was weighted equally on a 0 to 3 scale. The sum of each component score produced a final PSQI score between 0 and 21. Higher scores indicate worse sleep quality.

Physical activity was assessed using the Godin ²² -Shephard Leisure-Time Physical Activity Questionnaire. Several studies have used the tool to assess LTPA in medical students and there is evidence supporting its validity in healthy adults. ²⁸ The LTPA score was calculated by how many times “strenuous,” “moderate,” or “mild” intensity exercise is performed for more than 15 minutes over a typical week. Each intensity level of exercise was designated a multiplier (mild = x3, moderate = x5, strenuous = x9). The final LTPA score was expressed in arbitrary units that can be compared relative to one another. Higher scores indicate more LTPA.

Eating habits were assessed using the Rapid Eating Assessment for Participants (REAP) questionnaire. The REAP was originally developed for use in clinics and physician offices, intended to facilitate nutrition-focused discussion between patients and physicians. While there are many ways to assess eating habits, most are time-intensive and laborious. ²⁹ The REAP instrument was selected for ease and efficiency, and because it was validated and based on the HEI.^16,30 From the completed data, we selected 12 high-yield items to generate a composite nutrition quality score, which we call the REAP-NQS. Items used to formulate the REAP-NQS included: eating out of home; fruit intake; vegetable intake; processed meat intake; sugar-sweetened beverage intake; high-sodium food intake; and alcohol intake. Rationale for the included items was based on their well-recognized impact on cardiometabolic risk. ³¹ Eating out of the home in the USA has been associated with higher fat and alcohol intake, lower dietary quality and micronutrient intake, and increased adverse nutritional consequences.^32,33 Suboptimal intake of fruit, vegetables, processed meat, sugar-sweetened beverages, and sodium are associated with increased estimated mortality among adults. ³¹ All exercise related items were excluded from calculating the REAP-NQS, since physical activity was assessed using the Godin-Shephard Leisure-Time Physical Activity Questionnaire described earlier. Items on the 12-item REAP-NQS were responded as Usually/Often, Sometimes and Rarely/Never. Numerical values were assigned from 1 to 3, respectively, with “Usually/Often” equal to value of 1 and “Rarely/Never” equal to 3. A fourth option “not applicable” for food items was converted to Rarely/Never and scored as a 3. An average was taken for all 12 scored items, with higher scores corresponding with healthier eating habits.

Analysis

Investigators were supplied with coded, deidentified survey and demographic data including age, gender, MCAT score, and undergraduate GPA for all consenting participants. Survey data was converted to total scores as required by each survey’s scoring methodology for further analysis. Because survey scoring relies on complete responses and total scores, any incomplete survey responses were removed from the data set prior to any analysis. Demographic and survey data approximated normal distributions when examined visually by quintile-quintile (QQ) and frequency distribution plots. Demographic data were summarized for the study cohort using means and standard deviations (SD). Mean baseline vs follow-up scores for grit, PSS, PSQI, LTPA, and Nutrition Quality Score were compared using paired t-tests for the aggregate study population and Welch’s unpaired t-tests for comparisons by gender. A correlation matrix analysis was performed between demographic characteristics and change in survey score (∆) from baseline to follow-up for perceived stress (∆PSS), grit (∆Grit), and assessed self-care behaviors (∆PSQI, ∆LTPA, and ∆REAP-NQS). Multivariable linear regression modeling was used to isolate factors contributing to change in perceived stress from baseline to follow-up (∆PSS). The multivariable linear regression model was reduced by backwards elimination until only significant (P < .05) factors were included. All statistical analyses were performed using GraphPad Prism 9™ software.

Results

Over the 2 years of the study, 238 OMS1 students consented to participate and completed both baseline and follow-up surveys; incomplete responses were removed prior to analysis, leaving 204 complete survey records for analysis, a completed survey response rate of 85.7%. Demographic characteristics for the aggregate study population, including mean age, MCAT score, undergraduate GPA, and self-reported gender are summarized in Table 1. Briefly, the mean, median, and interquartile range (IQR) for age of participants were 26 years, 25 years, and 3 years, respectively; the mean, median, and IQR for MCAT score were 503.1, 502.0 and 5.0, respectively; and the mean, median, and IQR for total GPA upon matriculation were 3.48, 3.52, and .40, respectively. Age, MCAT scores, and undergraduate GPA of the study population fell within the range of nationwide data collected by the American Association of Colleges of Osteopathic Medicine (AACOM) for all colleges of osteopathic medicine during the study period, though the mean age of study participants skewed slightly older and mean undergraduate GPA and MCAT scores skewed slightly lower than national data. Additionally, the proportion of female study participants (57.2%) was greater than the proportion of female osteopathic medical students who matriculated nationwide (48.6% and 50.0%) during 2018 and 2019 when this study was conducted.^34,35

Table 1.

Demographic Characteristics.

Demographic Characteristics	PNWU-COM Matriculants		AACOM Matriculants
	Mean	SD	2018 Mean	2019 Mean
Age (years)	26.01	3.61	23	23
UG GPA	3.48	.28	3.54	3.52
MCAT Score	503.13	5.85	503.83	504.04
Gender, n (%)
Female	121 (59.3)		3592 (48.6)	3868 (50.0)
Male	83 (40.7)		3805 (51.4)	3866 (50.0)

Matriculant demographic characteristics for PNWU-COM as compared to national AACOM data reported for all colleges of osteopathic medicine in 2018 and 2019.

Abbreviations: AACOM, American Association of Colleges of Osteopathic Medicine; UG GPA, undergraduate grade point average; SD, standard deviation.

Baseline survey scores were analyzed for any differences by self-reported gender (Table 2). Statistically significant differences in mean baseline survey scores for female vs male study participants were identified for PSS, (14.87 vs 10.72, P < .0001), PSQI, (5.71 vs 4.78, P = .02), LTPA, (42.37 vs 52.47, P = .01), and REAP-NQS, (2.27 vs 2.12, P = .001). Likewise, there were similar statistically significant differences for female vs male study participants in follow-up survey scores for PSS, (19.56 vs 15.02, P < .0001), PSQI, (6.80 vs 5.80, P = .03), LTPA, (31.58 vs 42.90, P = .004), and REAP-NQS, (2.25 vs 2.12, P = .004). However, there were no statistically significant differences between female vs male participants in baseline or follow-up grit score, or in the change (∆ score = baseline score – follow-up score) of any study variables over the 12-week study period during the SciFOM course (Table 2).

Table 2.

Baseline vs Follow-Up Survey Comparisons by Gender.

Measure		Female		Male		P Value
		M	SD	M	SD
PSS	Baseline	14.868	5.767	10.723	6.094	<.0001
	Follow-Up	19.595	6.252	15.024	7.522	<.0001
	∆ (BL − FU)	4.727	6.003	4.301	6.707	n.s
PSQI	Baseline	5.711	2.969	4.783	2.644	.02
	Follow-Up	6.802	3.185	5.795	3.076	.03
	∆ (BL − FU)	1.091	2.866	1.012	2.452	n.s
Grit Score	Baseline	3.948	.456	3.835	.506	n.s
	Follow-Up	3.844	.489	3.772	.500	n.s
	∆ (BL − FU)	−.103	.420	−.063	.372	n.s
LTPA Score	Baseline	42.372	21.845	52.470	27.566	.006
	Follow-Up	31.579	19.857	42.904	30.949	.004
	∆ (BL − FU)	−10.793	20.409	−9.566	33.236	n.s
REAP-NQS	Baseline	2.267	.291	2.121	.308	.001
	Follow-Up	2.250	.311	2.118	.312	.004
	∆ (BL − FU)	−.017	.254	−.003	.253	n.s

Baseline vs follow-up survey results for perceived stress (PSS), sleep quality (PSQI), grit, physical activity (LTPA Score), and nutrition quality (REAP-NQS) were compared by gender using Welch’s unpaired t-tests.

Abbreviations: M, mean; SD, standard deviation; ns, not significant.

For the aggregate study population, there were no statistically significant differences in nutrition quality score (REAP-NQS) from baseline to follow-up surveys. However, over the study period, there were statistically significant decreases in grittiness (∆% Grit Score = −2.230, P = .002), and physical activity level (∆% LTPA Score = −22.147, P < .0001), while poor sleep quality (∆% PSQI = 19.853, P < .0001) and perceived stress (∆% PSS = 34.548, P < .0001) significantly increased (Table 3). These data suggested a possible relationship between declining grit and worsening self-care behaviors and increasing perceived stress after medical school matriculation. Therefore, a correlation matrix analysis was performed to identify any significant relationships between study variables (Figure 1). There were statistically significant negative correlations between ∆Grit vs ∆PSS (r = −.17, P < .05) and ∆Grit vs ∆PSQI (r = −.16, P < .05), and statistically significant positive correlations between ∆Grit vs ∆REAP-NQS (r = .16, P < .05) and ∆Grit vs ∆LTPA (r = .14, P < .05). The strongest statistically significant correlations from this analysis were between ∆PSS vs ∆PSQI (r = .47, P < .0001) and ∆PSS vs ∆LTPA (r = −.20, P < .01).

Table 3.

Baseline vs Follow-Up Survey Comparisons: Aggregate Population.

Measure	Baseline		Follow-Up		Baseline vs. Follow-Up
	M	SD	M	SD	%∆	P Value
PSS	13.181	6.244	17.735	7.158	34.548	<.0001
PSQI	5.333	2.878	6.392	3.180	19.853	<.0001
Grit Score	3.902	.481	3.815	.495	−2.230	.002
LTPA Score	46.480	24.836	36.186	25.584	−22.147	<.0001
REAP-NQS	2.208	.307	2.196	.318	−.518	n.s

Pre-vs Post-survey results for perceived stress (PSS), sleep quality (PSQI), grit, physical activity (LTPA Score), and nutrition quality (REAP-NQS) were compared by two-tailed paired t test for the aggregate study population. Higher PSS and PSQI values reflect worse perceived stress and sleep quality, respectively. Lower grit, LTPA and REAP-NQS values reflect reduced grittiness, physical activity, and nutrition quality, respectively.

Abbreviations: M, mean; SD, standard deviation; ns, not significant.

Figure 1.

Correlation matrix. Correlation matrix depicted as heat-map for demographic variables (age, GPA, MCAT score), character traits (grit, perceived stress), and self-care behaviors (sleep, physical activity, nutrition) at the beginning and end of a required first-year medical school course; character traits and self-care behaviors are represented as ∆ (Baseline Survey Score – Follow-Up Survey Score). Reported values are statistically significant Pearson r correlation coefficients: blue intensity indicates positive correlation; red intensity indicates negative correlation: * P < .05, ** P < .01, *** P < .001, and **** P < .0001. Empty boxes were not statistically significant. Abbreviations: GPA, Grade Point Average; MCAT, Medical College Admission Test; PSQI, Pittsburgh Sleep Quality Index; PSS, Perceived Stress Score; LTPA, Leisure-Time Physical Activity; NQS, Nutrition Quality Score.

To further isolate factors predictive of increased stress at the onset of medical education, and to control for potentially confounding factors, a multivariable linear regression model was employed (Table 4). After adjusting for differences in demographic factors in the aggregate study population, this analysis identified age (P = .04), worsening sleep quality (∆PSQI, P < .0001) during the study period, and declining physical activity level (∆LTPA, P = .01) during the study period as statistically significant predictive factors for increased perceived stress (∆PSS).

Table 4.

Multivariable Linear Regression Model for Factors Influencing Perceived Stress.

Outcome Factors	Factor Effect – Full Model			Factor Effect – Reduced Model
	Beta (SE)	95% CI	P Value	Beta (SE)	95% CI	P Value
∆ Perceived Stress Score
Intercept	13.27 (35.22)	−56.19-82.72	n.s	−2.542 (2.831)z	−8.152-3.042	n.s
GPA	.091 (1.442)	−2.752-2.935	n.s
MCAT	−.035 (.068)	−.169-.099	n.s
Age	.253 (.118)	.021-.485	.03	.214 (.107)	.003-.425	.05
Gender	.717 (.830)	−.912-2.353	n.s
∆ Grit	−1.478 (.994)	−3.438-.483	n.s
∆ PSQI	1.056 (.146)	.767-1.344	<.0001	1.082 (.143)	.799-1.365	<.0001
∆ LTPA	−.034 (.015)	−.063-−.004	.03	−.037 (.015)	−.658-−.008	.01
∆ NQS	1.268 (1.553)	−1.795-4.331	n.s

Multivariable linear regression modeling was used to identify factors predictive for increased perceived stress during the SciFOM course. The model was reduced through a backwards elimination procedure with P < .05 required for reduced model inclusion with the exception of the intercept.

Abbreviations: ns, not significant; PSS, perceived stress; PSQI, sleep quality; LTPA, physical activity; NQS, nutrition quality.

Discussion

Overall, our findings indicate that medical students at PNWU-COM struggled to maintain self-care behaviors, all of which are well-recognized to mitigate stress, over the first 12 weeks of their medical school training, regardless of their gender. Poor sleep quality and decreased physical activity were significantly related to increased perceived stress. Multivariable regression modeling identified sleep quality and physical activity, along with age, as predictive factors for increased perceived stress during the study period. These findings were consistent with other studies of medical students and support existing literature suggesting better maintenance of these specific self-care behaviors may be protective against increased perceived stress during medical school.^3,10,11 This is of clinical significance because a high level of perceived stress for a sustained period is a major risk factor for both academic and physician burnout. ²¹ This study adds to a growing body of literature suggesting mitigation of stress in medical students should be prioritized to prevent future physician attrition, especially in the face of current physician shortages. Furthermore, previous findings suggest that physicians tend to counsel patients based on their own activities.^20,36,37 Therefore, interventions targeted at ensuring medical students don’t lose already established healthy self-care behaviors during medical school could improve not only their own health, but also health outcomes of their future patients.^20,37

We also observed a statistically significant and somewhat surprising decrease in grittiness in our study population during the SciFOM course. Grit has traditionally been considered to be a stable trait, ⁷ but currently, there are few studies looking at how grit scores might change acutely while facing challenges. A recent study found decreased grit scores in orthopedic residents compared to applicants suggesting the acute and additional stressors associated with residency training may have a negative impact on self-perception of grittiness. ³⁸ Additionally, there is some evidence that changes in grit can occur over long periods of time; for example, grit tends to increase with age when education is controlled for. ⁷ Some possibilities for the observed decrease in grittiness among our study participants are recall and self-desirability bias. ⁷ Students upon matriculation to medical school may desire traits of grittiness in themselves, which could artificially inflate baseline grit scores. The same students re-evaluating their follow-up grit score during an acute challenge such as a difficult medical school course may report reduced grittiness due to fatigue or more realistic perceptions of their ability to deal with adversity.

Correlation matrix analysis suggested possible relationships between grittiness, perceived stress, and ability to maintain good self-care behaviors. Participants experiencing declining grit scores during this study experienced increased perceived stress and reduced sleep and nutrition quality; whereas those that maintained higher grit scores experienced less perceived stress and had smaller declines in sleep quality, suggesting that grit may be somewhat protective against increased perceived stress and declining self-care behaviors during medical school. These findings were consistent with our hypothesis that gritty students would be better equipped to manage stress and maintain health self-care habits during medical training. However, grittiness was not identified as a predictor of perceived stress in our multivariable linear regression model when other factors were controlled for, suggesting the impact of grit on stress in our participants is weak. Future studies could evaluate grit and its relation to more multifaceted measures of self-care, such as interpersonal relationships, spiritual growth, and stress management. Interventions aimed at improving self-care behaviors could look at changes in grit to see if improvement in self-care behaviors allow medical students to maintain their grittiness when faced with acute challenges.

Although our study participants experienced a modest decline in nutrition quality during this study, it was not statistically significant. This was somewhat unexpected based on previous studies that reported worsening eating habits in medical students, presumably due to increased stress and time constraints. ³⁹ Students at PNWU received nutrition education during the study period from a faculty member who is also a registered dietitian nutritionist, so it is possible this made them more aware of the importance maintaining healthy eating habits and this confounded their follow-up survey responses. Previous studies have demonstrated the validity of using the REAP tool with medical students, but medical students likely have higher levels of nutrition and nutrition literacy compared to the general patient population for which this tool was developed.^16,39 It’s also possible that this novel use of the REAP-NQS could not detect significant changes over the relatively short 12 week duration of this study.

Limitations

Our study identified a number of significant correlations between the character traits grit and perceived stress, and self-care behaviors including sleep hygiene, physical activity, and healthy nutrition, but we cannot assert causation in these relationships. This study relied on self-reported survey data and grit scores, which are vulnerable to self-desirability bias, and all the instruments we used are subject to recall bias. We did not measure other character and psychological traits such as optimism, resilience, self-control and conscientiousness, ³⁸ and those could impact academic outcomes, perceived stress, self-care behaviors, and reduce burnout equal to or more than grit. The data in this study also only represents findings from two cohorts of students at one osteopathic medical school and therefore, are not generalizable to all medical students.

Conclusions

The findings of this study add to previously identified and growing concerns medical educators and student groups have expressed about the intensity of medical education and its potential negative impacts. Even students with good self-care habits upon beginning their medical training are at risk of declining self-care behaviors and increased stress as medical school progresses.^2,3,21,40-42 Stress is expected when undertaking any big life challenge, but what amount of stress is necessary or appropriate for preparing medical students for a successful career in medicine? What more can the medical education community do to prepare students for the career risks of psychological distress and burnout facing them as future residents and physicians?

Early, repeated interventions, focused on reinforcing healthy habits and self-care behaviors during medical school, may help medical students manage stress and make healthier choices. ⁴² We suggest constructing interventions focused on sleep hygiene and physical activity as one approach to effectively reduce perceived stress in medical students. Decreased sleep duration and quality are associated with increased stress, worse mental and physical health, reduced physical activity, and declines in healthy eating behaviors.^11,13,14 Low physical activity levels among medical students are associated with less happiness, increased cardiovascular health risks and perceived stress, and higher rates of depression.^43-45 Given the predictive nature of sleep quality and physical activity for perceived stress in medical students identified in this study, we suggest implementing this intervention early and often, possibly starting during new student orientation or in the summer prior to starting medical school and continuing throughout the program.^13,42 The demands of medical school have been shown to worsen sleep habits and physical activity levels over time, therefore repeated intervention during the first, and possibly each of the medical education years, would likely be needed to sustain change among the medical student population.^11,46

ORCID iDs

Alan Boyd https://orcid.org/0000-0002-2259-4095

Kathaleen Briggs Early https://orcid.org/0000-0003-2224-2548

Emily Oestreich https://orcid.org/0000-0002-8256-0647