Medical students experience an immense amount of stress during training, and the Liaison Committee of Medical Education (LCME) standard on this issue is another impetus for the creation and study of wellness programs for medical students (1). Studies have shown that there is a higher prevalence of psychological distress among medical students than in the general population and age-matched peers (2, 3). The implications for psychological distress of medical students include lowered academic performance (4), increased professional misconduct (5), decreased empathy (6), increased substance abuse (7–9), and suicide (10). Medical students often avoid seeking help because they believe it might affect their future career plans or fear that they will be seen as “weak” by others (11). Given the high rates of distress among medical students, it is imperative that efficacious health-promoting interventions are offered for all students in order to limit distress. According to the standards of best-evidence medical education (12, 13), medical school educators and administrators should take account of published information on stress-management programs in order to guide the development of efficacious stress-management programs at individual medical schools.
There is a great variety in the types of health-promotion programs currently offered, including self-development groups (14), yoga interventions (15), mindfulness and lifestyle programs (16–20), self-hypnosis (21), small group stress-management programs (22), time management programs (23), reflective writing sessions (24, 25), curriculum changes (26–28), grading changes (29–31), and educational electives (32). Although there has been one notable systematic review of stress-management programs for medical students (33), this review has not been updated, nor did it identify any search terms used. Moreover, this review had wide inclusion criteria that included not only medical students, but residents, nurses, and premedical students.
Because of the dearth of systematic reviews on this wide array of interventions for reducing medical student distress, the aims of this review are to identify all randomized controlled trials and controlled non randomized trials for stress-management programs, to identify the efficacy of these interventions for reducing distress, and to identify the strengths and weaknesses of the available studies concerning allopathic and osteopathic medical students alone. We aim to describe how stress-management programs were incorporated into the medical education curriculum for medical students and their impact on psychological distress, if any, by comparing intervention-group findings with comparison-group findings.
We used standard methodologies (34) for conducting systematic reviews, searching PsycINFO and PubMed databases for peer-reviewed articles reporting on primary studies of stress-management programs integrated into allopathic and osteopathic medical schools for medical students. Also, we performed directed searches of primary publications referenced in other articles, including review articles. The search was conducted over a 22-month period (February 2010–November 2011). Combinations of the following key words were used: stress management, distress, burnout, coping, medical student, and wellness. Inclusion criteria included English-language articles published after 1995 and the use of a separate control or comparison group in the research design. The authors arbitrarily started at 1995. The major reason for exclusion was the absence of a control group; other reasons for exclusion included those programs aimed to decrease substance abuse or increase professionalism and programs conducted within residency training. We focused the review on medical school programs so that these programs could serve as a model for the development of medical school stress-management programs.
After initial searches, the team met regularly to discuss potentially relevant articles and whether these articles met the inclusion criteria. All articles that met the inclusion criteria were read by all team members. Articles were scored on criteria for establishing validity of a study, using and extending the standards developed by the Criteria from the Evidence-Based Medicine Working Group (35). These items included the presence of randomization, adequacy of methods of concealment, randomization, whether differences at baseline were identified, the presence of blinding, follow-up and intention to treat, and validity and reliability of outcome measures used. Each item was dichotomously scored by three members of the team (JC, TH, FE), with a maximum score of 7 points. Any differences in ratings were resolved by consensus. Two members (TH and MS) checked for reliability and validity of psychometric scales for those publications that did not make note of it. These included the Brief Symptom Inventory, Profile of Mood States, UCLA Loneliness Scale, Hopkins Symptom Checklist, and the Perceived Stress Scale. In presenting the results of the individual papers, we only comment on similarities or differences that were identified as being of importance between experimental and comparison groups. If sufficient data were available, the numbers needed to teach (NNT) or the numbers needed to harm (NNH) were calculated.
Our initial search yielded 735 primary articles. All abstracts were scanned, and the abstracts of articles pertaining to distress in medical students were read. Also, a secondary screening was done, in which the reference lists of these articles were read, to include articles that may not have shown up on the primary search. Furthermore, other articles relevant to managing stress in allied health students or physicians were identified, and their reference sections were searched for any articles pertaining to medical students. Our search of the databases plus our additional searches yielded 13 articles that met our inclusion criteria. These included five randomized, controlled trials (RCTs; 17, 20, 21, 25, 32) and eight controlled, non-randomized trials (CNRTs; 18, 19, 26–31). As shown in Table 1, eight of the studies examined one medical school year class; one study included premedical students and first- and second-year medical students, and three studies included two medical school class years. Interventions included self-hypnosis (21), changes in the length and type of preclinical curriculum (19, 26–28), meditation (17), feedback on various health habits (32), educational discussion groups on self-care (32), classes focusing on mind–body medicine (19), mindfulness-based stress-reduction (18, 20), reflective writing seminars (25), and changes in basic sciences curriculum grading (29–31). All of the studies aimed to decrease medical students’ emotional distress; however, one study also addressed the effects of the intervention on health habits (32); a second study examined the effects of the intervention on immune system reactivity (21); and a third study examined the effects of the intervention on United States Medical Licensing Examination (USMLE) scores, residency placement, and attendance (29). Assessment methods were heterogeneous; none used all of the same outcome measures.
One study (20) scored a full 7 on the validity scale, whereas the validity assessments of the other RCTs and CNRTs fell within a narrow range of 3 to 4. All of the studies had high follow-up rates of at least 70%, used outcome measures that were identified as valid, and assessed the reliability of the outcome measures. Moreover, all studies except one (27) identified some similarities and differences between groups at baseline before implementation. Only one study (20) described a method of concealment of randomization or noted using “blind” raters. This study was also the only one that used a power analysis.
There were five RCTs. The first RCT (20) was a three-centered, single-blinded, RCT. The intervention consisted of an audio compact disc that contained 30 minutes of spoken, mindfulness-guided practice. Participants were asked to practice this independently every day for 8 weeks. Validated measures of stress and anxiety were completed at baseline and at the end of the trial by both groups of participants. A primary outcome of perceived stress was significantly lower in the intervention group than the control group. There were no reported adverse effects of the intervention. An NNT or NNH could not be calculated.
In the second RCT (32), first-year medical students completed questionnaires about their health habits. In this study, the two experimental groups included a self-awareness intervention of educational written feedback on scores alone and a self-care intervention that included a lecture plus a group discussion plus written information about self-care. The self-awareness group was given individual written feedback on questionnaire scores as compared with the norm and their peers. The self-care intervention involved a lecture, written information about self-care habits, and a group discussion. Although there were differences in outcomes between the groups in terms of sleep and exercise habits, the feedback and educational intervention groups did not differ in emotional or academic adjustment. Also, of the 23 students that participated in the educational discussion groups, 8 reported changed behaviors (improved sleep hygiene and getting therapy for depression), compared with 5 of the 29 students who received written feedback on their questionnaires. The NNT was 6.21.
In the third RCT (17), we assessed the effects of an 8-week, meditation-based, stress-reduction elective. The course consisted of seven sessions of 2.5 hours each week, weekly home practice assignments, and daily journaling. The exercises were aimed to increase mindful listening skills and empathy. The intervention group reported significantly reduced anxiety, reduced depression, reduced overall psychological distress, increased empathy, and increased spirituality. In the fourth RCT (21), volunteers at an incoming medical school class were randomly selected to receive self-hypnosis training as a means of coping with stress, as compared with a no-treatment control condition . The self-hypnosis intervention did not result in any significant immune changes relative to controls; however, subjects in the self-hypnosis intervention assigned significantly lower stress ratings during the exam period than did the subjects in the no-treatment control condition. In the final RCT (25), one group wrote about their most emotional and traumatic experiences; a second group wrote about their future as if their personal goals had been achieved, including how they had overcome at least one obstacle; and a third control group described everything they had done in the past 24 hours, without expressing emotions or opinions. There were no significant between-group differences in depressive symptoms or physical health reported at 3 months.
Two CNRTs addressed the impact of interventions that included mindfulness on mood and anxiety states. In one study (18), medical students participated in a 10-week, mindfulness-based stress-reduction seminar. Their mood states were compared with controls who participated in a didactic seminar on complementary medicine. The Total Mood Disturbance was significantly lower in the intervention group than in the control group. In the second CNRT (19), 30 first-year medical student enrollees in an experimental elective on mind–body medicine were compared with second-year medical students who had not enrolled in the elective. At the end of the elective, there were no significant differences in mood states, perceived stress, or anxiety.
One CNRT (27) evaluated a change in curriculum from a traditional track (TT) to a reformed track (RT). In the TT, teaching and learning were based on lectures and seminars organized by individual disciplines. In contrast, teaching and learning in the RT were problem-based, focused on self-directed learning in small groups, and supplemented by seminars and tutorials. In the RT, the emphasis was on providing interdisciplinary teaching and earlier contact with patients. Reformed-track students felt significantly more supported than TT students. Supports included good contact with fellow students and teachers, high-quality courses, and a curriculum that fulfilled students’ expectations.
Two CNRTs evaluated the length of medical school curriculum on stress levels. One study (26) compared two medical schools with different lengths of preclinical curriculum, whereas another study (28) compared 2 final years, when 1 year had a shortened curriculum. In the first study (26), the 2- versus 3-year preclinical curriculum was associated with higher stress levels. Since a greater proportion of students in this 5-year program (23%) contemplated dropping out, as compared with the 6-year program (15.8%), the NNH was 13.9. On the Maslach Burnout Inventory, there were no significant differences on the Emotional Exhaustion, or the Depersonalization subscales between the two groups. In the remaining study (28), shortening the final academic year from 47 to 42 weeks did not significantly alter perceived stress levels.
The final three CNRTs all examined the relationship of Pass/Fail grading on student well-being. In the first study (31), students who were evaluated on the Pass/Fail grading system were compared with the previous class of students, who were evaluated on a five-interval grading system (A, B, C, D, F). At the end of the second year of medical school, students graded with the Pass/Fail system reported feeling less stress and greater group cohesion than those graded with the five-interval graded system. Also, there were no differences in USMLE Step 1 scores between the Pass/Fail and the five-interval grading groups. In the second study (29), two medical school classes, with different grading systems (Pass/Fail and five-interval) were compared. Data from each class showed that the Pass/Fail class felt a significant increase in well-being, greater satisfaction with the quality of their education, and greater satisfaction with their personal lives. Moreover, there were no differences in academic performance, USMLE scores, residency placement, or attendance between the two classes. In the third study (30), first- and second-year medical students were surveyed at seven different medical schools about stress, burnout, and quality of life. Students who were graded using three or more intervals had high levels of stress, emotional exhaustion, and depersonalization than students who were graded on a Pass/Fail system. Also, students who were graded using three or more intervals were more likely to consider dropping out of medical school and experience burnout than their Pass/Fail peers.
We found 13 studies that met the inclusion criteria. These included five randomized, controlled trials (RCTs) and eight controlled, non-randomized trials (CNRTs). Nine of the studies included one medical school class, whereas four studies included multiple class years. With the exception of one study (20), the quality of the studies varied little. Study strengths included similarities on group differences, good rates of follow-up, and the use of valid and reliable outcome measures. Only one study (20) described randomization methods, concealment of randomization, or methods for blinding raters. Designing adequate randomization methods is an especially important standard to achieve in education research (36). The scarcity of studies meeting this standard limits the rigor of data, and therefore limits the development of stress-management programs aimed to promote medical student wellness.
The available data support the efficacy of some forms of intervention in promoting health and reducing stress among medical students. A combination of lectures, discussion groups, and written educational material is more efficacious than just providing written feedback on questionnaire answers (32). Three of the four studies that included mindfulness-based stress-reduction techniques reported results supporting those techniques (17, 18, 20). Self-hypnosis appears to be efficacious in reducing stress and anxiety in the study setting (21). Also, implementing Pass/Fail grading appears to enhance student well-being (29–31). One of the two studies on curriculum length found higher stress levels associated with a shorter curriculum, although these differences were not reported to be significant (26).
One methodological concern was that there was a lack of consistency of outcome measures across studies. Standardization of outcome measures across studies constitutes an important goal for this area of research. Although medical students are well known to experience burnout, only two studies utilized the Maslach Burnout Inventory (26, 30). Researchers should consider using the Maslach Burnout Inventory and the Perceived Medical School Stress Scale (37) for this area of research. In order to compare interventions, there is a need for investigators to use consistent and important outcome measures, including outcome measures on depression and burnout, which are commonly experienced among medical students. Also, these studies did not assess the long-term effects of interventions, and most experimental interventions were conducted at one site, limiting the generalizability of the data.
There were several limitations to our methods. Although we conducted several searches and used multiple databases, we did not conduct a search of the “gray” (unpublished and not easily accessible) literature. We also only included English-language articles that were published after 1995. Nevertheless, our results suggest that there is a role for stress-management programs in reducing the psychological distress of medical students. The small number of controlled studies and the limitations of the data described above indicate a need for developing rigorously-conducted studies in this critically important area. Psychiatrists should be at the forefront of this research.