Health-promoting factors in medical students and students of science, technology, engineering, and mathematics: design and baseline results of a comparative longitudinal study

Study design and setting

Participants

We invited all Freshmen from the 2011 and 2012 cohorts (two consecutive classes) at the University of Lübeck by advertising during the central inauguration ceremony. The baseline surveys were taken in October 2011 and October 2012 during the pre-course week. There were no exclusion criteria.

Data collection/instruments

The questionnaires comprise sociodemographic questions, health outcomes (self-rated general health and mental health as defined above) as well as potential predictors of the outcomes (dimensions of study-related behaviour and experience, personality traits and self-developed, pre-tested items covering leisure activity). The pre-tested, paper-based baseline surveys consisted of 150 items and took about 30 minutes to complete. All follow-up surveys will be websurveys and are planned to be shorter.

Data management

From the paper-based baseline surveys, data was input to a Microsoft Access database via an entry mask resembling the original questionnaire. Following a random quality control by a second researcher, data was exported to IBM SPSS statistics, version 20. From the websurvey, data is exported directly into an SPSS format file.

Data from the follow-up surveys will be matched by the pseudonyms explained below.

Data analysis

After a plausibility check, data from the baseline questionnaires was merged and summarised using descriptive statistics to provide a comprehensive profile of this cohort. Where possible, we substitute missing values following the rules provided in the handbooks for the instruments (see above). We then exclude incomplete data-sets. Data analyses are conducted with IBM SPSS statistics, version 20. We use two-tailed t-tests to compare means of continuous variables and report results as means ± standard deviation. For categorical variables, data are analysed using chi-square-tests and results reported as percentages. In order to express bivariate correlations, we use Spearman’s rho. We use binary logistic regression analyses in order to assess correlations between potential predictors and outcomes. We conduct separate regression analyses for both groups of students (medical and STEM students). In order to remove variables in the logistic regression analyses, we employ stepwise backwards elimination with p > 0.05 as the level for removing effects, while keeping age and gender as important sociodemographic variables and the other outcome (general or mental health) in order to control for them. Additional statistical methods used for the longitudinal analyses will be detailed in the respective publications.

Ethical issues

This study is being conducted in accordance with the guidelines provided by the Declaration of Helsinki. The study has been approved by the University’s Deans and President and the Student Council and we have obtained informed consent from each participant. All data are collected de-identified. The de-identification is stressed in both the advertisements for the baseline and follow-up surveys as well as for the informed consent sheet. A self-generated code and the student’s matriculation number serve as pseudonyms, both of which cannot be linked to the real name by the investigators without further investigation. The study protocol was approved by the ethical committee of the University of Lübeck (file reference: 11-010).

Participants

In 2011 and 2012, 351 medical students and 540 STEM students were recruited to participate in our study. Data from 350 medical students and 531 STEM students were included in the statistical analyses. After exclusion of incomplete data-sets, the response rates were 93.0% for medical students and 60.8% for STEM students. The average age of the study participants at baseline was 20.43 ± 2.65 years with a range of 16 to 41. The majority of the medical students (66.3%; whole student population: 65.8%) and 46.1% of the STEM students (whole student population: 41,1%) were female. Compared to STEM students, a higher percentage of medical students exercised regularly (78.2% vs 65.4%) or engaged in musical acticity (59.7% vs 48.8%). Only a few were engaged in a religious community or practiced a relaxation technique regularly.

Further demographic data are shown in Table 1.

Dimensions of study-related behaviour and experience

Medical students showed statistically significant higher sum-scores in the dimensions of subjective significance of work, career ambition, tendency to exert, striving for perfection, offensive coping with problems, satisfaction with work, satisfaction with life and experience of social support. STEM students, in contrast, showed statistically significant higher sum-scores in the dimension of emotional distancing. We observed no statistically significant difference between the two groups regarding the dimensions of resignation tendency and balance and mental stability.

Personality traits

General and mental health

Self-rated general health was slightly, but significantly, different between STEM and medical students. Medical students rated their general health better (1.73 ± 0.64) than STEM students (1.98 ± 0.76) (p < 0.01; Cohen’s d -0.36). 90.9% of the medical and 79.7% of the STEM students were in good general health (as defined above) before the beginning of the courses (p < 0.01).

When compared to STEM students, a significantly higher proportion of medical students was free of clinically-relevant depression (as defined above) (96.6% vs 92.2%; p = 0.02). The proportions for absence of clinically-relevant anxiety (as defined above) didn’t differ significantly between medical and STEM students (89.7% vs 90.0%; p = 0.91).

A higher proportion of the medical students (88.3% vs 86.3% of the STEM students) was in good mental health (as defined above) at baseline (p = 0.41).

The correlation between general and mental health was weak, but statistically significant (Spearman’s rho 0.19, p < 0.01).

Predictors of general and mental health

General and mental health

In the baseline survey of our longitudinal study on medical students’ health and its predictors, we found that most medical students were in good general and mental health. Their fellow students from other, mainly science and technology subjects, were in similar good mental health but less often in good general health. In both groups, students with higher levels of satisfaction with life were more likely to be in good general health. Those medical and STEM students with lower levels of neuroticism were more likely to be in good mental health.

The proportion of individuals rating their general health as good in our cohort of medical students is comparable to that in the similar age group (18-29) of a recent German general population sample [44]. In this sample, 89.1% of the female and 92.3% of the male respondents rated their health as good or very good. Medical students in our cohort matched these porportions quite well (90.7% of the male and 90.9% of the female students were in good health), whereas only 79.6% of the male and 79.9% of the female STEM students were in good general health.

The proportion of medical students free of clinically-relevant depression (96.6%) tends to be higher than proportions derived in other studies on medical students in their first years of medical school using the same cut-off (70.7% and 94.3%) [24, 45] and the proportion of a similar age group (18-29) of a German general population sample (90.1%) [46]. This might be due to the very early timepoint of the baseline survey before the beginning of courses and, thus, before exposure to medical-school-associated stress. Another explanation might be that young people with clinically-relevant depression are less likely to qualify for and to start medical school. The proportion of medical students free of clinically-relevant anxiety in our cohort (89.7%) lies in between scores reported in comparable studies on medical students (79.7% and 93.2%) [23, 45]. The high proportions of students with clinically relevant depression and anxiety scores in the study of Karaoglu et al. [45] might reflect cultural differences as it was conducted in Turkey.

Our results are also in accordance with the results Leahy et al. derived from a cross-sectional study comparing medical students’ distress levels with those of students of non-medical subjects [47]. In this study, students from non-health disciplines were significantly more distressed than students from health disciplines and 11% had overall been treated for a mental health problem, compared to 12% (medical students) and 14% (STEM students) showing clinically-relevant anxiety and/or depression (as defined above) in our study.

Study-related behaviour and experience

When compared to the STEM students in our cohort and German early-year medical students surveyed in other studies [30, 31], students in our cohort showed consistently higher sum-scores for the dimensions that form the domains of professional commitment and emotional well-being. With the exception of “offensive coping with problems”, they showed lower to equal sum-scores in the dimensions that form the domain of resistance towards stress. As neither the STEM students in our cohort nor the medical students in the other studies were surveyed before the beginning of any relevant courses (see “Strengths and limitations”), this difference might again be related to the timepoint of the baseline survey, indicating that resistance towards stress is a trait which can be considered as “improving” while being exposed to stress.

Personality traits

Consistent with earlier results on personality traits in medical students [48, 49], we found low scores for neuroticism and high scores for agreeableness. Jerant et al. [48] described the medical students in their sample as having higher levels of conscientiousness and agreeableness, similar levels of openness to experience and extraversion and lower levels of neuroticism, when compared to a general population sample. This is consistent with our results in comparison to a general population sample from German-speaking countries [50]. As Jerant et al. conclude, high levels of conscientiousness and agreeableness might be expected in medical students. The observed low neuroticism scores might also reflect some social desirability in the responses, as Jerant et al. propose.

Predictors of general and mental health

Consistent with earlier research [5, 51], lower neuroticism scores predicted better mental health in both groups of students. In a study by Tyssen et al. [52] of Norwegian medical students (n = 421), neuroticism was an independent predictor of perceived stress. Hochwälder [53] reports that low neuroticism scores were correlated with low scores of emotional exhaustion in nurses (n = 659).

With regard to the focus of our work on health promotion, it was important to see that regular physical activity was a significant predictor for general health, at least in medical students. The beneficial effects of physical activiy on health and well-being have been reported extensively [54, 55]. Especially if follow-up results reveal a health-promoting effect of regular physical activity, leaving space in the curriculum for regular physical activity for students from the beginning of their courses onwards could be a promising, albeit difficult to enforce, health-promoting intervention [38].

Despite a growing body of literature that has also shown the positive effects of regular exercise on mental health [56, 57], we were not able to show such an effect in our regression analyses. This is consistent with the findings of Buchmann et al. [58] who reported that depression and anxiety were not correlated to exercise in Freshmen medical students. Instead, we found that lower levels of resignation tendencies and higher levels of emotional distancing were correlated to being mentally healthy in medical students. The ability to distance oneself from work has been described as an important factor to avoid workaholism and burnout [59].

Surprisingly, a higher level of striving for perfection was also a predictor of being mentally healthy in medical students. This is in contrast to a number of studies that describe perfectionism as a major influence for medical students’ or physicians’ stress and burnout [60], and which interpret a relative decrease in the course of study as a sign of a healthy adaptation [30]. On the other hand, Enns et al. [36] discern between an adaptive and a maladaptive form of perfectionism and report that only the latter was associated with higher levels of depression in a sample of first to third year medical students. Likewise, perfectionism as imposed by others and not by oneself was found to cause negative mental health effects [61]. Fry [62] even described perfectionism together with optimism and humour as positive moderators of the impact of daily hassles on self-esteem and burnout in female executives. However, none of these studies examined the association between striving for perfection and mental health as early in the students’ course as ours did. It will therefore be of interest as to how perfectionism will develop and how it will predict mental health during the course of our study.

Strengths and limitations

The major strength of this study is its longitudinal design and the high number of participants, this latter due to a high response rate, especially among medical students. We will be following a cohort of medical students (two consecutive classes) throughout their entire study period.

The generalisability of our results may be limited by the single-centred nature of the study. However, apart from special study courses at a few medical schools, medical education in Germany follows a very similar curriculum nationwide. The age and gender distribution of our sample resembles the nationwide distribution [63]. Our findings may thus be generalised beyond Lübeck University Medical School at a national level.

The annual presentation of results to the cohort may have an influence on the subsequent surveys, but we consider them an indispensable measure to keep participants motivated to respond to the yearly surveys.

One limitation of our study is the lower than expected initial response rate among the STEM students. Due to the fact that we conducted the baseline survey in the pre-course week, which mainly consists of Mathematics lectures, we may have missed those students feeling well enough prepared in Mathematics to skip this course. Despite the fact that we conducted two subsequent baseline data collection waves in the first week of courses, this may be a potential source of bias, as these students could have estimated themselves healthier than those feeling the need to take this course in order to pass the exams. Also, a higher share of female STEM students participated in our study, when compared to the whole student population. As gender did not predict any of the health outcomes in STEM students, the risk of bias resulting from this is probably low.

The response rate of over 90% among the medical students makes any selection bias very unlikely. In contrast to the STEM counterpart, the medical students’ pre-course week doesn’t contain any subject matter relevant for examinations, which may have led to a different answering behaviour.

Implications for research and practice

The high percentages of medical students in good self-rated general and/or mental health at the beginning of their undergraduate education and the knowledge from earlier research, indicate that the health of these students is likely to decline throughout medical school, highlighting the urgent need for health-promoting and preventive interventions at a very early point in time. The resilience of students should be fostered by, for example, teaching stress-reduction skills and promoting social and personal well-being activities. Compared to empirical studies describing medical school stress and related symptoms, there are only a few articles describing health-promoting interventions, and only a small proportion of these have been evaluated using scientifically-rigorous methods [64]. Thus, further high quality evaluation studies are urgently needed. Since we found a particularly low prevalence of regular practice of a relaxation technique among the study participants at baseline, and because there is evidence that learning and practising a relaxation technique might have a positive influence on health outcomes [64, 65], this might be a potential starting point for a health-promoting intervention.

Also, the finding of the relatively bad health of STEM subject Freshmen should be further investigated in larger, multi-center trials. This group of students could require special interventions to promote their health and well-being and keep the rate of premature drop-outs low.