Mitigating medical student stress and anxiety: Should schools mandate participation in wellness intervention programs?

Abstract

Purpose

Medical students are at increased risk of poor mental health and need to regularly engage in preventive programs to maintain well-being. However, many do not and it remains an open question whether these programs should be mandatory. We implemented a RCT to examine the effectiveness of assigning medical students to a wellness intervention on adherence to engagement in the assigned intervention and on psychological and academic outcomes.

Method

Medical students participated in a 12-week randomized controlled intervention involving one-hour wellness sessions of either (1) yoga; (2) mindfulness; or (3) walking, held twice-weekly. Students completed standardized psychological assessments at baseline and following the intervention.

Results

Students randomized to the wellness intervention group engaged in more minutes of assigned activities than students randomized to the control. There was a significant difference in the change from pre- to post- intervention on measures of state anxiety and perceived stress, with better outcomes for the intervention group.

Conclusions

The assignment of twice-weekly wellness intervention sessions protects medical students from state anxiety and perceived stress with no negative impact on academic performance. Students adhered to the sessions and reported enjoying the sessions once trying them. Actual engagement is more important than wellness activity type.

Keywords:

• Medical student stress
• medical education
• stress and academic success
• psychological distress

Background

Attending medical school is a major life choice, given the volume of knowledge to be acquired, training intensity, career commitment to serve others, and high financial cost. Medical students experience elevated rates of reported anxiety, emotional stress, depression, and suicidal ideation compared to other populations. (Dahlin et al. 2005; Dyrbye et al. 2006). Medical students with higher levels of stress tend to have lower academic scores due to the negative effect of stress on working memory and decision making (Schiller et al. 2018). Distress and burnout are also more common among medical students than other populations and are associated with higher rates of substance abuse, increased medical errors and a greater likelihood of unethical conduct upon graduation (West et al. 2006; Fahrenkopf et al. 2008). Thus, reducing perceived stress, depression, burnout, anxiety, and psychological distress among medical students is a critical priority as it may increase medical students’ quality of life, long-term knowledge retention and development as future health care professionals.

Practice points

• Distress and burnout are more common among medical students than other populations and are associated with higher rates of substance abuse, increased medical errors and a greater likelihood of unethical conduct upon graduation.

• Medical educators should aim to reduce distress and burnout by altering the curriculum to make medical education less stressful and more supportive.

• If curriculum alteration is not possible, medical students should engage in preventive programs to maintain health and wellness during their studies, though many do not voluntarily do so.

• Medical educators should consider mandatory preventive programs to reduce anxiety and perceived stress among students. Evidence indicates students enjoy these programs once they are engaged.

• Students should be offered a range of activities from which to select, be informed about the benefits of the various activities, and be encouraged to try activities they have not yet experienced.

Most medical schools provide academic support services to help students improve study techniques and time management, as well as psychological services and recreational facilities to promote well-being (Yiu 2005). Medical students are advised to maintain good sleeping habits, eat properly, and remain physically active throughout their medical education (Stern et al. 1993). Though these remain a top priority in managing stress, optimal lifestyle choices can often be difficult to maintain (Ball and Bax 2002). In response, medical schools have begun implementing mandatory and/or optional programming to protect and promote the well-being of doctors in training (Yiu 2005). Though counseling programs have been widely instituted and reviewed at many medical schools in the past few decades, only recently have preventive wellness lifestyle programs such as yoga, mindfulness, and exercise been closely examined (Shapiro et al. 1998; Rosenzweig et al. 2003; Warnecke et al. 2011; Morris et al. 2012; Weight et al. 2013; Aherne et al. 2016). Further, there is little evidence for whether these programs should be voluntary or mandatory as part of the medical curriculum for students. One study demonstrated that levels of satisfaction were found to be significantly higher for an optional 7-week mindfulness-based stress reduction (MBSR) course for year two students compared to a mandatory 7-week MBSR course for year one students (Aherne et al.2016). Another study found significant improvements in emotional exhaustion, depression, self-compassion and mindfulness among fourth year medical students who took a 4-week elective on mindful medical practice (Garneau et al. 2013). On the contrary, Dyrbye et al. (2017) did not find significant improvement in empathy or well-being of first-year medical students as a result of a mandatory longitudinal MBSR course (Dyrbye et al. 2017). Given the many demands on medical students’ time, it may be that motivation is essential for a positive effect and, therefore, optional programming may be more likely to result in positive outcomes (Dobkin and Hutchinson 2013; Aherne et al. 2016; Dyrbye et al. 2017).

For this study, we selected the following three intervention types based on evidence that supports their efficacy in similar contexts:

Yoga intervention

Yoga has been widely used to decrease stress and anxiety and increase wellness in students. It is a collective term for physical, mental, and spiritual disciplines for transforming body and mind, but in its common Western usage refers primarily to the physical exercises involved (Kirkwood et al. 2005; Li and Goldsmith 2012). Malathi and Damodaran (1999) implemented a randomized controlled study of Indian medical students and found that a yoga class three times a week for three months significantly reduced reported anxiety levels, blood pressure and heart rate. A concurrent study found that pre-exam anxiety was markedly lower among those participating in the three-month yoga training (Malathi et al.1998). Significant improvements in various cardiovascular health measures were found in a study of the effects of one hour of yoga per week over five weeks on Indian medical students (Parshad et al. 2011). Other studies have demonstrated decreased rates of depression and perceived stress and increased sense of well-being among medical students who engaged in yoga (Simard and Henry 2009; Prasad et al. 2016).

Mindfulness intervention

Mindfulness is a practice involving attention to thoughts, feelings, and sensations that arise in moment-to-moment awareness and which emphasizes a non-judgmental attitude towards one’s experience (Grossman et al. 2004; Hofmann et al. 2010). Numerous studies have examined mindfulness meditation as an effective strategy for enhancing emotional regulation and reducing symptoms of stress and anxiety (Grossman et al. 2004; Hofmann et al. 2010). Studies have reported statistically significant improvements in mood, reductions in depression and/or anxiety, enhanced self-regulation, and a positive impact in physiological functions such as markers of immune response and stress biomarkers (Whitehouse et al. 1996; Shapiro et al. 1998; Gruzelier et al. 2001; Kiecolt-Glaser et al. 2001; Rosenzweig et al. 2003; Jain et al. 2007; Koh et al. 2008; Carmody and Baer 2009; Dayalan et al. 2010; Bhimani et al. 2011; Maclaughlin et al. 2011; Warnecke et al. 2011; Bond et al. 2013; Erogul et al. 2014). Mindfulness-based stress reduction (MBSR) is one structured intervention combining mindfulness, yoga, and other stress reduction techniques that has been studied with medical students and reported positive effects (McGrady et al. 2012; Garneau et al. 2013; Erogul et al. 2014; Brennan et al. 2016; van Dijk et al. 2017).

Walking intervention

A number of studies have demonstrated that exercise – particularly walking or light jogging – can alleviate symptoms of anxiety as effectively as anti-anxiety medications (Broocks et al. 1998; Morris et al. 2012; Aylett et al. 2018; Kelly et al. 2018). Dyrbye et al. (2017) surveyed 4,402 medical students and found that burnout was significantly lower and quality of life (QOL) was significantly higher for students who followed Centers for Disease Control (CDC) aerobic exercise or strength training recommendations. An earlier study of 2,316 medical students found that students who followed CDC guidelines had fewer days of ‘bad mental health’ and reported less stress (Frank et al. 2008).

Purpose

We implemented a randomized study in which students were assigned to a wellness intervention group that contained three sub-groups (yoga, mindfulness, walking) or a control group. We measured the amount of time students spent in assigned wellness activities and the impact of group assignment on perceived stress, depression, burnout, anxiety, and psychological distress. These criteria were selected given their predictive importance for substance use, medical errors, and unethical behaviors in later professional practice (West et al. 2006; Fahrenkopf et al. 2008). We hypothesized that students assigned to a wellness intervention group would engage in more minutes of assigned activities than students allocated to a control group and that this increased engagement would be associated with decreased levels of perceived stress, depression, burnout, anxiety, and psychological distress, and increased academic performance over the course of a term of study in the medical basic sciences.

Method

Participants

All participants provided informed consent before participating and the study received approval from the St. George’s University (SGU) Institutional Review Board (application # 13020), which is registered with the United States Department of Health and Human Services (USDHHS).

All 1,777 students enrolled in the first year of basic science studies in the School of Medicine at SGU were invited by email sent by a research assistant working on the study to participate in the study, which was conducted the following academic term. The e-mail informed students that, if they chose to participate, they would be randomly assigned to one of three wellness intervention groups (i.e. yoga, mindfulness, walking) or a control group. If assigned to an intervention group, students were also told that they would be required to attend two one-hour sessions per week over twelve weeks from January through April. Students were also told that, regardless of group assignment (including controls), they would need to complete a weekly log recording minutes of yoga, mindfulness, walking and Non-study-based exercise activities, and to complete a one-hour battery of assessments and surveys pre-intervention and post-intervention. The e-mail included a web link to click on and register if interested in participating in the study. Students were offered the incentive if they completed the study of inclusion in a lottery to win one of three full scholarships for a commercial United States Medical Licensing Exam (USMLE) step prep course. The study adhered to CONSORT guidelines for randomized control trial design and reporting.

Sample size

The sample size was estimated based on detecting a difference with an effect size of d = 0.80 at a level of significance of 5% and 90% power for a one-sided t-test. The minimum total sample size was calculated to be 62 participants with 41 in the combined wellness intervention group and 21 in the control group. Allowing for attrition, the total sample size for recruitment was set at 100 participants. Since the final sample sizes were 45 in the treatment group and 25 in the control group, the study achieved 90% power.

Procedures

The study utilized a parallel equal ratio randomized controlled-trial design with pre-intervention and post-intervention measures. After completing the pre-intervention questionnaires and tests, participants were assigned by the lead author (RW) via simple randomization to the yoga group, mindfulness group, walking group, or control group using an online random numbers generator. Depending on group random assignment, participants were asked to attend two 60-minute yoga, mindfulness or walking sessions out of three which were offered every week. All of the sessions were held on the main campus of SGU and after academic classes (i.e. 5:30 pm or later or on weekends). Attendance was taken for each intervention to ensure participation by students. No significant changes were made to the trial design after the initiation of the study and no stopping guidelines or interim data analysis were implemented in the study: The design called for participants to complete the full 12-week intervention. Neither participants, intervention instructors, nor research investigators were blinded to the random assignment given the active nature of the intervention program.

Yoga

A certified yoga instructor with 11 years of experience led the Hatha yoga sessions for the current study. Hatha yoga is a practice of gentle stretching through yoga poses, combined with attention to breath and to align breathing with movement through the postures. Each session was an hour long (Woodyard 2011).

Mindfulness

In the present study, a mindfulness practitioner with 25 years of training led the mindfulness sessions. The practitioner established and followed a set curriculum for the sessions. The sessions followed a three-step process outlined by Tibetan teacher Yongey Mingyur: training for relaxation, training to stabilize the attention, and training for clarity (Yongey and Swanson 2009). Sessions also incorporated teachings from MBSR (Kabat-Zinn 2013), Tonglen teachings (Chodron 1997), Metta teachings (Salzberg 1995), and Zen and mindfulness teachings (Young 2015). There was no requirement for home practice. Mingyur’s approach was selected over the relatively better researched (e.g. van Dijk et al. 2017) MBSR to better align with the time commitment required of participants in the yoga and walking groups. A copy of the mindfulness curriculum used in the present study is available upon request from the corresponding author.

Walking

A United States-registered fitness/cross-fit instructor with 9 years of experience led the walking sessions in the current study, which consisted of an hour-long brisk walk, as a group, around the SGU 42-acre campus.

Activity log

Study participants, including those in the control group, completed a weekly activity log over the 12 weeks of the intervention. Every week they were sent a link to a survey which prompted them to record any yoga, mindfulness, walking or exercise they had engaged in outside of the intervention sessions, specifying the number of minutes spent on each. Examples of activities that students recorded participating in outside of the intervention included, but were not limited to, lifting weights, treadmill, cycling, swimming, playing basketball, volleyball, soccer, and martial arts. After completion of the 12-week intervention, study participants scheduled a post-intervention appointment and completed the same questionnaires and tests that were completed pre-intervention, as well as a survey designed to capture their subjective impressions of their experience as a participant in the study.

Outcome measures

Studies in this field have not been consistent in the choice of outcome measures. The research team, including two investigators with 10+ years’ experience in neuropsychological assessment (RW and BL), held multiple meetings to select the most commonly used and validated measures administered to similar populations. Survey data was collected on paper in private offices on the main campus of SGU. The following symptom clusters were assessed given their association with higher rates of substance abuse, increased medical errors and a greater likelihood of unethical conduct upon graduation (West et al. 2006; Fahrenkopf et al. 2008).

1. Perceived Stress: The Perceived Stress Scale (PSS) (Cohen et al. 1983) is a widely used psychological instrument for measuring the perception of stress. Items ask respondents how often they felt a certain way in the past month to assess how unpredictable, uncontrollable, and overloaded respondents find their lives. Cohen and Williamson (1988) show correlations with PSS and measures of stress, self-reported health and health services, health behavior, smoking status, and help seeking behavior.

2. Depression: The Center for Epidemiologic Studies Depression scale (CES-D) is a 20-item self-report adult instrument designed to measure common symptoms of depression that have occurred over the past week, such as poor appetite, hopelessness, pessimism, and fatigue (Radloff 1977). CES-D scores range from 0 to 60 with higher scores indicating more severe depressive symptoms. For more information on CES-D, see Radloff (1977) and Naughton and Wiklund (1993).

3. Burnout: The Maslach Burnout Inventory (MBI) is the most universally utilized instrument to screen for the risk factors associated with burnout, most commonly found in service-oriented professions (Galán et al. 2011). For this study, the MBI-SS, an adaptation of the General MBI Survey designed to assess burnout in college and university students, was selected (Schaufeli et al. 2002). The MBI-SS consists of 16 items that constitute three scales: Exhaustion (EX; 5 items), Cynicism (CY; 5 items), and Efficacy (EF; 6 items). All items are scored on a 7-point frequency rating scale ranging from 0 (never) to 6 (always).

4. State Anxiety: The State-Trait Anxiety Inventory (STAI) (Spielberger and Gorsuch 1983). has been used extensively in research and clinical practice with high school, college students, and adults. The STAI scale consists of twenty statements that evaluate how respondents feel ‘right now, at this moment’ that evaluate feelings of apprehension, tension, nervousness, and worry.

5. Psychological Distress: The General Health Questionnaire (GHQ-12) is a screening instrument used to detect current psychological distress/mental health, not including psychotic psychiatric disorders (Goldberg and Hillier 1979). It uses twelve items to screen for three overarching problems: depressed mood, anxiety, and problems with social functioning. Respondents rate how much they have recently experienced each of the problems on a 4-point scale from ‘not at all’ or ‘much less than usual’ to ‘much more than usual’. Since its development by Goldberg in the 1970s it has been extensively used in different settings and different cultures. This self-administered questionnaire focuses on two major areas: 1) the inability to carry out normal functions and 2) the appearance of new and distressing phenomena.

Academic performance was measured by examining each participant’s Weighted Mean Percent Grade (WMPG) across all courses completed during the term in which the study was carried out. Baseline academic performance was each participant’s cumulative WMPG in basic sciences coursework before the term during which the study was conducted.

Data analysis

Descriptive statistics and the Kruskal Wallis H test were used to characterize the groups and assess demographic comparability of the randomized groups. Any participant who dropped out of the study was not imputed into the analysis; Stayers and Leavers were compared in terms of performance on baseline measures using ANOVA or Mann-Whitney U depending on normality of the data as detemined by Shapiro Wilk test. The pre-intervention scores on the six target outcome variables for each of the wellness intervention groups and for the control group were compared to identify possible baseline differences between the randomly assigned groups (i.e. cross-sectional analysis via ANOVA for parametric analysis, Welch ANOVA in the event of violation of homogeneity of variance, and Kruskal Wallis H-test for nonparametric analysis). Independent t-test was used to compare of minutes of engagement in wellness activities as a function of assignment to treatment or control group. Due to non-normal distribution of the data, and to account for baseline differences, Mann-Whitney U-test was used to measure differences in the median change scores from pre-to post of the control and intervention group on each of the six measures. All analyses were performed using IBM SPSS 24 OSX (Armonk, New York).

Results

No harms or unintended effects were observed or reported for any of the study participants in any of the wellness intervention or control groups.

Demographics

The full study was run as planned and stopped after the 12-week intervention was completed. A total of 101 students (41% male) signed up for the study and completed the pre-intervention questionnaires and tests. The mean age of participants was 25.7 (SD = 3.43) and 24 (23.8%) of the participants had a post-baccalaureate degree. Ethnic breakdown of the sample can be found in Table 1. There were no significant differences between the randomly-assigned groups across gender (p = .671), age (p = .338), ethnicity (p = .245), or level of education (p = .773) based on Kruskal-Wallis H test.

Table 1. Baseline (n = 101) and Post-Intervention (n = 70) demographic characteristics (means, SDs, counts) of medical students who were randomized to yoga, mindfulness, walking or control groups at St. George’s University School of Medicine as part of the wellness intervention study in 2014.

		Mindfulness Pre	Mindfulness Post	Yoga Pre	Yoga Post	Walking Pre	Walking Post	Control Pre	Control Post
	N	24	17	26	19	24	9	27	25
Gender	Male	10	9	13	8	9	5	9	8
Age	Mean (SD)	26.21 (4.42)	25.18 (3.28)	26.35 (3.95)	26.47 (4.39)	24.71 (1.60)	24.88 (2.23)	25.52 (3.03)	26.12 (3.06)
	Range	22–31	23–35	22–42	22–42	22–29	23–30	19–32	20–33
^†Ethnicity	White	15	9	11	6	10	5	^12	10
	Black	5	4	3	3	4	1	5	5
	Hispanic	0	0	2	2	1	0	2	2
	Asian	2	2	4	4	3	0	1	1
	Southeast Asian	2	2	4	3	5	3	1	1
	Middle East	0	0	2	1	1	0	5	5
Education	Undergrad	19	16	19	15	19	7	20	19
	Masters	5	1	7	4	5	2	7	6
Overall Health	Poor	0	0	1	0	0	0	0	0
	Below Average	0	0	0	0	1	0	1	0
	Average	4	0	11	0	4	1	4	0
	Above Average	6	4	4	6	9	1	2	2
	Good	7	1	6	3	5	1	14	1
	Very Good	7	7	4	4	5	3	6	17
Smoke Cigarettes	Yes	0	5	1	6	0	3	1	5
Diagnosed Mental Illness	Yes	3	0	2	0	2	0	5	1
Taking Medication	Yes	9	1	3	1	7	1	11	3

^†One participant in the control group did not report ethnicity.

SD: Standard Deviation

Stayers and leavers

A total of 70 students (N = 30 male) engaged in the intervention, completed the weekly activity logs and completed the post-intervention assessments and surveys, representing a 69.3% follow-through rate (Table 1). We compared the pre-intervention means of students who completed the study (‘Stayers’) and students who dropped out of the study (‘Leavers’) across all six outcome variables. Data was not normally distributed in the pre-intervention measures of depression, burnout and anxiety as assessed by Shapiro-Wilk test (p < .05); Mann-Whitney U test was used to compare Stayers and Leavers for these variables. ANOVA was used to compare pre-intervention scores of Stayers and Leavers for the other three variables. Although values were higher for the Leavers than the Stayers on every measure (See Table 2), the difference between the Stayers and Leavers was statistically significant only for the pre-intervention measure of anxiety: A Mann-Whitney U test indicated that anxiety was greater for Leavers (Mdn = 46) than for Stayers (Mdn = 39), U(N_leavers = 31, N_stayers = 70) = 808.50, z = −2.04, p = .042, effect size d = .41.

Table 2. Baseline mean and SD scores across outcomes measures of n = 31 participants who left the wellness intervention study (Leavers) and n = 70 Participants who completed the wellness intervention study (Stayers) at St. George’s University School of Medicine in 2014.

Measure		Mean (SD)	Mdn
Perceived stress	Stayers (N = 70)	21.7 (7.46)	21	Mean Difference = −3.27, 95% CI (−6.61, .07), t(99) = −1.94, p = .055, effect size d=.41
	Leavers (N = 31)	24.97 (8.53)	26
Depression	Stayers (N = 70)	10.83 (9.81)	7.5	U = 949.50, z = .999, p = .318, effect size d = .2
	Leavers (N = 31)	12.68 (10.51)	8.0
Burnout	Stayers (n = 70)	27.24 (14.74)	25	U = 968.50, z = −.858, p = .391, effect size d=.17
	Leavers (N = 31)	29.16 (12.43)	30
State anxiety*	Stayers (N = 70)	40.3 (9.7)	39	U = 808.50, z = −2.04, p = .042, effect size d = .41
	Leavers (N = 31)	44.29 (10.63)	46
Psychological distress	Stayers (N = 70)	16.81 (4.65)	17	Mean Difference = −1.86, 95% CI (−3.95, .23), t(99) = −1.77, p = .08, effect size d = .38
	Leavers (N = 31)	18.68 (5.38)	18
Academic performance	Stayers (N = 66)	82.76 (5.29)	82.64	Mean Difference = −.67, 95% CI (−3.00, 1.67), t(94) = −.57, p = .57, effect size d = .12
	Leavers (N = 30)	83.43 (5.45)	82.32

*Statistically significant at p = .042.

Comparison of randomized groups on outcome measures at baseline

The three intervention sub-groups were compared with the control group across the six target outcome variables at pre-intervention to determine the comparability of the randomly assigned groups on the outcome measures at baseline. Data was not normally distributed for the pre-intervention depression and anxiety scores (Shapiro-Wilk test: p < .0001). Therefore, Independent Samples Kruskal-Wallis test was used for these measures; ANOVA and Welch ANOVA in the event of violation of homogeneity of variance was used to compare the groups on the other measures). None of the comparisons were significant; participants who completed the intervention and were randomly assigned to intervention groups and to the control group (Stayers) did not differ in perceived stress, depression, burnout, anxiety, psychological distress, or academic performance at the start of the study.

Comparison of minutes of participation in assigned and monitored activities across wellness intervention and control groups

Descriptive statistics were calculated for number of minutes that wellness intervention and control participants engaged in each of the intervention activities (i.e. yoga, mindfulness, walking) as well as the number of minutes of exercise and of all monitored activities combined (See Table 3). All intervention participants who were included in the study completed at least 11 of the 24 sessions of assigned activity. Participants varied in the number of minutes of participation in their own assigned activity from 660 min to 3,760 min (control group participants were recorded as having zero minutes of an assigned activity). Minutes spent in assigned activity for each participant was subtracted from total minutes of assigned activity and other activity outside of the intervention combined, giving minutes of ‘non-study-based exercise’. Independent t-test indicated no significant difference in minutes of non-study-based exercise between wellness intervention and control groups; Intervention (M = 1,388.89, SD = 1165.90) and Control (M = 1,349.92, SD = 993.09), Mean Difference = 38.97, 95% CI (−512.55, 590.49), t(68) =.14, p =.89, effect size d = .036. As expected, independent t-test indicated a significant difference between the wellness intervention and control groups in minutes of participation in assigned wellness activities; Intervention (M = 2,696.11, SD = 1287.7) and Control (M = 1,770.40, SD = 1196.7), Mean Difference = 925.71, 95% CI (300.36, 1551.06), t(68) = 2.95, p = .004, effect size d =.745. This confirms adherence to the assigned yoga, mindfulness, and walking sessions by the students assigned to the wellness intervention, and that they attended these sessions over and above their regular exercise routine (i.e. non-study-based exercise).

Table 3. Minutes of yoga, mindfulness, walking, non-study-based exercise and total minutes of assigned intervention and non-study-based exercise (mean and SD) engaged in by each group in the wellness intervention study at St. George’s University School of Medicine in 2014.

Group	Yoga minutes (SD) min–max	Mindfulness minutes (SD) min–max	Walking minutes (SD) min–max	Non-study-based exercise minutes (SD) min–max	Total minutes of assigned intervention and non-study-based exercise combined (SD) min–max
Yoga N = 19	1181.1 (269.1) 840–1860	212.1 (339.0) 0–1270	N/A	914.73 (730.9) 85–3000	2307.9 (1014.3) 1170–5170
Mindfulness N = 17	49.4 (102.3) 0–360	1739.1 (804.9) 720–3760	N/A	1484.12 (1467.8) 0–4410	3272.7 (1524.8) 1320–6330
Walking N = 9	96.7 (268.2) 0–810	120.0 (226.5) 0–660	1256.7 (402.1) 660–1980	2210.00 (808.7) 1275–3450	2426.7 (1004.5) 1275–3750
All interventions combined N = 45	536.7 (597.6) 0–1860	770.6 (935.8) 0–3760	251.3 (536.5) 0–1980	1388.9 (1165.9) 0–4410	2696.1 (1287.7) 1170–6330
Control N = 25	129.7 (296.6) 0–1200	290.8 (471.4) 0–1520	N/A	1349.9 (993.09) 340–4620	1770.4 (1196.7) 375–4620

Impact of group assignment on outcome measures (control vs. wellness intervention)

Table 4 shows the pre-test and post-test means and mean changes from pre- to post-test for the control group, for the intervention groups combined and for each of the three wellness intervention groups separately on each outcome measure. There is a trend for reduced scores (i.e. improved psychological outcomes) across all outcome measures except burnout and academic performance for the intervention group compared to the control group. Given deviations from normality for the measures, Mann-Whitney tests were used to determine whether differences between control and intervention groups in changes from baseline scores to post-test scores were statistically significant and Cohen’s d was calculated to evaluate effect sizes. Table 5 shows the pre and post mean ranks and the median change from pre-to post-test on each measure and the results of the Mann-Whitney tests and Cohen’s d comparing the wellness intervention group and the control group. There is a significant statistical difference with a moderate effect size, in scores, as measured by change from pre-to post intervention, for anxiety (U = 741, z = 1.84, p = .02, d = .54) and perceived stress (U = 769, z = 2.53, p = 0.01, d = .63). Specifically, the wellness intervention group showed a significant decrease in scores (i.e. improved outcome) from pre- to post-intervention for anxiety while the control group showed no change. The wellness intervention group showed no change in perceived stress from pre- to post-intervention while the control group showed an increase (i.e. poorer outcome). Furthermore, there is a trend for a difference in scores, with low to moderate effect, as measured by change from pre-to post intervention, for psychological distress (U = 741, z = 1.84, p = .065, d =.45) and depression (U = 710.50, z = 1.82, p = 0.069, d =.44). Specifically, the wellness intervention group showed a decrease in scores (i.e. improved outcome) from pre- to post-intervention for psychological distress and for depression, though at a trend level.

Table 4. Baseline, post-intervention, and change scores (mean and SD) for perceived stress, depression, burnout, state anxiety, psychological distress, and academic performance for each randomized group and for the intervention groups combined following the wellness intervention study at St. George’s University School of Medicine in 2014.

Outcome measures		N	Mean change from baseline to post-intervention	Baseline mean (SD)	Post-intervention mean (SD)
Perceived stress	Yoga	19	0.84 (7.86)	22.74 (7.74)	23.58 (5.7)
	Mindfulness	17	−1.12 (8.18)	21.77 (5.04)	20.65 (7.67)
	Walking	9	−1.89 (8.68)	23.78 (9.44)	21.89 (5.42)
	All interventions	45	−0.44 (8.04)	22.58 (7.11)	22.13 (6.47)
	Control	25	4.92 (8.10)	20.12 (7.94)	25.04 (10.14)
Depression	Yoga	19	0.37 (10.54)	11.9 (10.17)	12.26 (7.66)
	Mindfulness	17	−1.29 (8.59)	11.35 (11.58)	10.06 (9.41)
	Walking	9	−3.33 (8.22)	14.44 (11.76)	11.11 (5.44)
	All interventions	45	−1.0 (9.30)	12.20 (10.84)	11.20 (7.91)
	Control	25	4.32 (8.72)	8.36 (7.17)	12.68 (10.75)
Burnout	Yoga	19	1.74 (12.25)	29.84 (17.41)	31.58 (16.92)
	Mindfulness	17	4.18 (12.89)	28.29 (13.34)	32.47 (17.93)
	Walking	9	−4.11 (9.73)	31.22 (14.16)	27.11 (13.27)
	All interventions	45	1.49 (12.17)	29.53 (15.04)	31.02 (16.43)
	Control	25	7.44 (15.86)	23.12 (13.49)	30.56 (20.55)
State anxiety	Yoga	19	−2.79 (11.51)	39.84 (10.33)	37.05 (7.37)
	Mindfulness	17	−4.94 (10.34)	39.59 (8.47)	34.65 (8.28)
	Walking	9	−6.67 (5.77)	44.44 (10.6)	37.78 (9.3)
	All interventions	45	−4.38 (10.07)	40.67 (9.69)	36.29 (8.04)
	Control	25	1.6 (10.05)	39.64 (9.89)	41.24 (12.77)
Psychological distress	Yoga	19	−1.42 (6.45)	16.74 (5.73)	15.32 (4.87)
	Mindfulness	17	−1.94 (8.93)	15.47 (5.23)	13.53 (7.42)
	Walking	9	−4.22 (3.8)	18.44 (4.8)	14.22 (5.02)
	All interventions	45	−2.18 (7.06)	16.60 (5.37)	14.42 (5.91)
	Control	25	0.64 (5.04)	17.2 (3.0)	17.84 (5.23)
Academic performance	Yoga	19	0.31 (2.42)	81.94 (5.45)	82.25 (4.91)
	Mindfulness	17	0.39 (2.29)	83.52 (5.87)	83.91 (6.95)
	Walking	8	−0.34 (2.49)	84.61 (4.19)	84.41 (4.02)
	All interventions	44	0.28 (2.33)	83.04 (5.41)	83.31 (5.59)
	Control	23	0.42 (1.95)	82.12 (5.02)	82.78 (5.58)

Table 5. Statistical analysis of changes from baseline to post-intervention in perceived stress, depression, burnout, state anxiety, psychological distress, and academic performance between intervention (n = 45) and control groups (n = 25) as part of the wellness intervention study at St. George’s University School of Medicine in 2014.

Outcome measures	Group	N	Mean rank	Baseline median	Post-intervention median	Median change pre-post	Mann-Whitney comparing median change between interventions and control
Perceived stress	Combined intervention	45	30.92	22.00	23.00	0.00	U = 768.5, z = 2.530 p = .011 effect size d = .63
	Control	25	43.74	20.00	24.00	4.00
Depression	Combined intervention	45	32.21	8.00	10.00	−1.00	U = 710.5, z = 1.817 p = .069 effect size d = .44
	Control	25	41.42	7.00	9.00	2.00
Burnout	Combined intervention	45	33.40	28.00	31.00	2.00	U = 657, z = 1.159, p = .246 effect size = .28
	Control	25	39.28	20.00	28.00	4.00
State anxiety	Combined intervention	45	31.53	39.00	36.00	−5.00	U = 741, z = 2.191 p = .028 effect size d = .54
	Control	25	42.64	39.00	40.00	0.00
Psychological distress	Combined intervention	45	32.17	17.00	14.00	−1.00	U = 712.5, z = 1.842, p = .065 effect size =.45
	Control	25	41.50	17.00	18.00	1.00
Academic performance	Combined intervention	44	33.49	83.44	83.51	0.40	U = 528.5, z=.297, p = .766 effect size d = .07
	Control	23	34.98	81.56	81.74	0.82

Subjective experience results

As part of the post-study survey, participants were asked to provide feedback about their involvement in the study. Intervention participants responded positively to the question: ‘What effect did you feel this experience had on you, personally?’ with 38 of 43 (88%) of the wellness intervention group versus 6 of 25 (24%) of the control group responding that the experience had a positive or very positive effect on them personally (19 of 25 [76%] of the control group indicated that the experience had a neutral effect). Intervention participants also responded positively to the question: ‘What effect did you feel this experience had on you, academically?’ with 27 of 43 (63%) of the wellness intervention group responding that the experience had a positive or very positive effect on them academically versus 2 of 25 (8%) of the control group. With the exception of two students in the control group, all participants said they would recommend study participation to others. With the exception of one student in the intervention group and four in the control group, all participants indicated that they felt the time allocated to the study was worth it. Of the participants assigned to a wellness intervention, 18 (95%) of those in the yoga group, all 17 (100%) of those in the mindfulness group, and 6 (80%) of those in the walking group said they intended to continue engaging in mindfulness, yoga or walking following the study. Participants had an initial preference for yoga over mindfulness or walking; Only 1 participant (5%) in the yoga group responded ‘Yes’ to the question ‘Were you hoping to be assigned to a different group?’. However, 10 (59%) participants in the mindfulness and 4 (50%) students in the walking group responded ‘Yes’ to this same question. These results suggest a ‘yoga bias’ among the participants going into the study. Despite this initial preference for yoga and an expressed disappointment regarding being assigned to mindfulness or walking, 16 (94%) of the students assigned to mindfulness indicated that they liked it or liked it very much. For students assigned to yoga, 16 (84%) liked it or liked it very much and 5 (63%) of students assigned to walking liked it or liked it very much. Note however, that 15 (63%) of the students assigned to the walking intervention group left the study, compared to 7 (29%) assigned to the mindfulness group, 7 (27%) assigned to the yoga group, and 2 (7%) assigned to the control group.

Discussion

We hypothesized that students allocated to a wellness intervention would engage in more minutes of assigned activities than students allocated to control, demonstrating adherence to the program, and that this increased activity would be associated with decreased levels of perceived stress, depression, burnout, anxiety, and psychological distress, and enhanced academic performance. The results partially confirmed our hypothesis: Students assigned to the wellness intervention group, whether yoga, mindfulness, or walking, engaged in more minutes of assigned activities than students assigned to a control, confirming adherence to the program. The intervention group experienced a decrease in state anxiety from pre- to post-intervention compared to the control group. The intervention group experienced a stable level of perceived stress from pre- to post-intervention compared to the control group which experienced an increase. The intervention group also experienced a decrease in psychological distress from pre- to post-intervention compared to the control group, which experienced an increase in psychological distress. While the post-intervention comparison between the wellness intervention and control group was only statistically significant for state anxiety and perceived stress, the wellness intervention might also influence psychological distress and depression, as seen by the moderate effect size in scores between the wellness and control groups, which merits further investigation. The difference between the groups was sustained even though the groups engaged in a similar amount of exercise/wellness activities outside the assigned yoga, mindfulness, or walking interventions. Our results indicate that participation in a wellness intervention program does not have a discernable impact on academic performance in the short term but does assist psychological well-being of medical students during their basic sciences studies. These results are consistent with previous studies demonstrating the benefits of yoga, mindfulness, and/or exercise on the well-being of medical students (Petruzzello et al. 1991; Malathi et al. 1998; Malathi and Damodaran 1999; Simard and Henry 2009; McGrady et al. 2012; Morris et al 2012; Dobkin and Hutchinson 2013; Weight et al. 2013; Erogul et al. 2014; Prasad et al. 2016; Dyrbye et al. 2017; van Dijk et al 2017). Further, while we examined the impact of yoga, exercise and mindfulness in this study, these are not the only interventions that could support medical student wellness. For example, Tai Chi has been shown to provide health benefits for students in higher education (Webster et al. 2016).

Subjective responses illustrate that students valued the intervention even though it added additional weekly time commitments to their busy schedules. The change in student opinions regarding the mindfulness activity from pre- to post-intervention demonstrates the value of including a mandatory experiential component to education on well-being. However, the attrition of participants assigned to walking highlights the danger of not providing options if activity is to be mandated.

Recently, several researchers have recommended that the medical curriculum include research findings on various stress reducing activities but that the experiential components remain optional (Slavin et al. 2014; Aherne et al. 2016; Brennan et al. 2016). Medical schools have an opportunity to instill behaviors in students that will influence life-long choices and well-being (Ripp et al. 2017). The results of this study indicate that there may be value in mandating students to participate in a regular wellness intervention program, and that this participation may have a protective effect on key measures of psychological well-being. However, not all students enjoy all activities, and a student who does not want to spend the time, but is mandated to participate, may sabotage the experience for other students (Aherne et al. 2016). Ripp et al. (2017) recommend that programs be mandated to include a minimum number of wellness-focused curricular hours which would replace rather than add to current curricular offerings and that they be required to provide students a selection of research-based wellness interventions, such as mindfulness, to choose from. They propose that these interventions be standardized, and the outcomes monitored for effectiveness (Ripp et al. 2017). It was found that US medical students identify a wide range of non-programmed activities when prompted to describe their own wellness activities (Ayala et al. 2017). Rather than focusing on identifying the ‘right’ self-care activities, medical educators should contribute to a culture of prioritizing self-care in the long run because wellness is a highly individualized process with unique and changing solutions (Ayala et al. 2017). In support of this effort, guides have recently been published to support physicians and medical educators in striving for mindful medical practice (Dobkin and Hassed 2016; Liben and Hutchinson 2020)

Our findings support the recommendation that activities to promote well-being may be beneficial if incorporated into the medical curriculum. However, students should be offered a variety of wellness program options from which to select. Practically, this means medical schools must provide multiple programs for students with the knowledge that some will be more popular than others and preferences might change over time. The results reported here indicate that the type of wellness activity is not as important as requiring engagement in some form of wellness activity. Further studies are required to measure the impact of various models of integration into the curriculum on minutes of activity and on student satisfaction, mental health, well-being and academic performance over time.

Limitations

Given the demand on participants’ time, a low response rate was observed (101 (6%) of the possible 1777 students participated) and therefore, the study sample may not be representative of the overall medical student population. The attrition of participants from pre- to post- assessment and particularly of participants assigned to the walking intervention is a limitation.

To address the potential of self-selection bias, we compared the students who remained in the study (i.e. ‘Stayers’) to students who dropped out of the study (i.e. ‘Leavers’) against key pre-intervention variables. There were no statistically significant differences between Stayers and the Leavers on any of the variables except anxiety. However, it should be noted that, while not statistically significant, the differences between Stayers and Leavers on perceived stress and psychological distress showed effect sizes in the moderate range. That the students who left the study were initially more anxious than those who stayed is a limitation and highlights a problem with optional well-being initiatives; The students most in need are least likely to feel that they have time for the activity. This further supports the concept of attendance at mandatory intervention sessions.

Ethical approval

All participants provided signed informed consent before participating in the study and ethical approval was granted by the St. George’s University institutional review board, October 18, 2013 (application # 13020).

Acknowledgements

The authors wish to thank the following individuals for their contribution to the delivery of the interventions or the collection of data: Rebecca Phillip and Lexi Fisher. The authors also gratefully acknowledge all the students who participated.

Disclosure statement

The authors declare that they have no competing interests.

Data availability statement

The data generated and analyzed during the study are available from the corresponding author on reasonable request.

Additional information

Funding

Funding for this study was provided by St. George’s University Small Research Grants Initiative, which provides up to $5,000 USD for faculty members to carry out research in their fields of expertise. In-kind support for this project was provided by Kaplan Test Prep and The Institute of Medical Boards in the form free tuition for enrollment for one student (each institution) in the United States Medical Licensing Exam (USMLE) Step 1 preparation course. A draw was held for all students who completed the post-intervention assessment to determine which two students would receive the USMLE Step 1 prep course free tuition.

Notes on contributors

Randall Waechter

Randall Waechter, MA, PhD, is Associate Director of Research and Assistant Dean in the School of Graduate Studies and Associate Professor in the School of Medicine at St. George’s University.

Gabriel Stahl

Gabriel Stahl, MPH, MD, is Emergency Medicine Resident Physician PGY3 at the Brookdale University Hospital and Medical Center in Brooklyn, NY.

Sara Rabie

Sara Rabie, MA, PhD, is Chair and Professor, Department of Educational Services at St. George’s University, St. George’s, Grenada.

Bora Colak

Bora Colak, MPH, MD, is a Psychiatry Resident at Westchester Medical Center/New York Medical College program.

Debbi Johnson-Rais

Debbi Johnson-Rais, MA, EdD, is the Director of the Scholars Program, Division of Student Affairs and Enrollment Management, and Adjunct Professor in the College of Science at Florida Atlantic University, Boca Raton, Florida.

Barbara Landon

Barbara Landon, PsyD, is Director of the Psychological Services Center and Professor in the School of Medicine at St. George’s University.

Kristen Petersen

Kristen Petersen, BA, is a Clinical Evaluation Report Specialist at Global Medical Writing and Translation, Kent, Washington.

Shirin Davari

Shirin Davari, MPH, is a student in the School of Medicine at St. George’s University.

Thinn Zaw

Thinn Zaw, MPH, is a student in the School of Medicine at St. George’s University.

Kesava Mandalaneni

Kesava Mandalaneni, MBBS, is Assistant Professor in the School of Medicine at St. George’s University.

Bianca Punch

Bianca Punch, MPH, is a research scientist at the Windward Islands Research and Education Foundation.