Concerns of first-year medical students regarding their future profession: an international study

Abstract

Background

Medical students, often high-achieving and motivated, face unique challenges as they transition to clinical practice. This professional identity formation process requires adapting to patient contact, clinical reasoning, and high-stakes assessments such as OSCEs. Alongside these demands, students struggle with perfectionist expectations, heavy workloads, and patient responsibilities, leading to fears of failure, inadequacy, or professional unpreparedness. International research shows recurring concerns, including breaking bad news, managing patients, and feelings of exclusion among international students. However, existing studies are limited by their cross-sectional focus and site-specific scope, lacking longitudinal data or cross-country comparisons. Addressing this gap is crucial for understanding how concerns evolve over time and for shaping unified, high-quality medical education. This study aims to explore medical students’ fears across different countries.

Methods

In this study, we assessed the answers of medical students to the question, “What are you not looking forward to in your future job as a doctor?” This question was part of four questions of a longitudinal international survey aiming at students at the beginning and again towards the end of their studies. We analyzed responses given by medical students at the start of their studies at Ludwig-Maximilians University in Germany (LMU), Utrecht University in the Netherlands (UU), Karolinska Institutet in Sweden (KI) and the University of Nicosia in Cyprus (UNIC). A combination of qualitative analysis and hierarchical cluster analysis was used to identify recurring concerns, assess their importance, and examine differences based on university, gender, and societal context.

Results

A total of 2048 responses were collected between 2017 and 2021. The analysis identified two predominant categories: “stress” and “working conditions”. The category “failure” ranked third overall, though it did not appear among the top four response categories at LMU. Ranks 3 and 4 consisted of the categories “healthcare system”, “hierarchy”, “bad news” and “work-life balance”. The distribution of these categories varied between countries.

Conclusions

Quantifying students’ concerns and their prevalence enables international comparisons and highlights critical factors to address through systemic reforms targeted interventions. These findings provide valuable insights for improving students’ preparedness for medical practice.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12909-025-08273-6.

Background

Successful medical students are usually highly motivated and conscientious personalities [1–4]. As the entrance qualification for medical school is high, students entering medical school have already proven their academic performance thoroughly and lack experience with failure [5–7]. These students then undergo medical training with the ultimate goal of successfully transforming them from laypersons to competent medical professionals [8], which is commonly referred to as the concept of professional identity formation (PIF) [9, 10]. Even though this goal is clearly defined as enabling students to “think, act, and feel like a physician” [11–13], medical students still encounter different challenges throughout their training. Foremost, the transition from medical school to clinical practice represents a major change for graduating students as they enter unfamiliar territory [7, 14], which exemplifies the lack of junior doctors’ preparedness for the demands of their new roles [14]. High preparedness requires a thoughtful transition from the certainties of the classroom to the bedside [7], which involves the experience of early patient contact [15] and the art of clinical reasoning [16, 17], paired with novel assessment methods such as Objective Structured Clinical Exams (OSCEs) [18, 19]. Beyond that, medical students are exposed to high expectations coming from a perfectionist self [20, 21], the academic workload [22] and patients [23], resulting in a considerable amount of pressure.

Where there is uncertainty and pressure, there are also fears [24, 25]. These can range from fear of failure [26], to fear of not being excellent [15, 27] or being wrong [28] to more specific concerns regarding their own professional future life [29]. Even though they may be hidden in everyday practice, concerns are well recognized by both students and medical educators, and have therefore become the object of international research [30]. To address a person’s fears and mitigate the impact of shame [31], it is essential first to be well informed about their fears. It is known that first-year students encountering patient contact for the first time are afraid of lack of skill or intimidation by patients or staff [15]. Breaking bad news seems to be a major source of distress for medical students at start and end of study [32]. There is also worry about the stark contrast between medical school and the realities of patient care, especially regarding the degree and novelty of taking responsibility and managing own patients [33]. Furthermore, fears of international students seem to also cover the feeling of exclusion and marginalization [34]. Attempts to cluster the variety of concerns into different categories have been conducted in Sweden and Poland [35, 36].

However, the landscape of studies addressing medical students’ concerns lacks two essential things. First, there are hardly any studies that follow a longitudinal approach to monitor concerns early on and examine how they will change over time. Secondly, nearly every study focuses solely on one site or university, which makes it even harder to gain reliable insight into the comparability of students’ fears. Especially the lack of cross-regional evidence stands against growing efforts towards a more unified, standardized and high-quality medical training across countries [37, 38]. To effectively close this gap in literature, we sought to understand what students in different countries are concerned about.

The presented work is part of a longitudinal study called “Medical Students’ expectations of the future: a multi-site, longitudinal study.” The primary aim of this study is to assess medical students’ expectations and concerns about their future professional lives as medical doctors at the beginning and, again, towards the end of medical school. To date, the study has been carried out and continues to be conducted at the Ludwig-Maximilians University (LMU) in Germany, Utrecht University in the Netherlands (UU), the Karolinska Institutet in Sweden (KI) and the University of Nicosia (UNIC) in Cyprus.

The questionnaire of the longitudinal study covers four questions:

1. Why do you want to become a doctor?

2. What are your expectations of your future profession as a doctor?

3. What are you looking forward to in your future profession as a doctor?

4. What are you not looking forward to in your future job as a doctor?

In this work, only the answers of the first-semester students to the last question were analyzed and described.

Research aim of the present study

To guide medical education changes and preventive measures towards an environment with less fear more preparedness, it is important to gain a comprehensive understanding of the students’ expectations and concerns regarding their future professional lives. Therefore, the primary aim of this study was to investigate key areas of these aspects by addressing the following research questions:

1. To which categories can the student’s responses to the question “What are you not looking forward to in your future job as a doctor?” be assigned?

2. With what frequency are the respective categories mentioned by students?

3. Can students’ responses be categorized and further dissected based on variables such as university, age and gender?

Based on these findings, future improvement measures and follow-up actions can be developed aimed at addressing challenges deemed most relevant by new medical students.

Methods

Data collection

The study is designed as a longitudinal study, however, the specific timepoints depend on the duration of medical education at the participating institutions. Assessing participants at two distinct stages - at the beginning and toward the end of their medical studies - enables a time-dependent evaluation of the participants’ responses, allowing for the identification of individual and standard-specific changes over time [39, 40].

Participation in the study is voluntary and takes place annually with each incoming student cohort. Data collection begins by contacting first-semester students via e-mail, providing them with detailed information about the study’s methodology and purpose, along with the questionnaire. The questionnaire was developed between 2016 and 2017 by representatives (MWM) and piloted with students from the universities involved. The questionnaire takes approximately 15 min to complete and is presented in the working language of each respective university: Dutch (Utrecht), Swedish (Stockholm), German (Munich), or English (Cyprus). This exact procedure is repeated at the end of the study. Both the questions and the answers have been translated by certified translators.

All participating medical schools are European, with comparable programme lengths (5.5–6 years). While there are differences in structure and content across the curricula, these cannot yet influence the responses, as students are surveyed at the very beginning of their studies.

Data collection is qualitative in nature, as students provide text-based, open-ended responses. Before the data is stored in password-protected tables, it is pseudonymized, meaning the students either receive (Cyprus) or select a code themselves, making it possible to compare the results of the questionnaires at the two points in time. This procedure is in accordance with the ethical guidelines of the Declaration of Helsinki. Only demographic data, such as age and gender of the respondents, are recorded. The data analyzed in this work correspond to the initial survey results covering only students at the beginning of their studies; a time-dependent analysis is therefore yet to be possible.

Categorization

As mentioned before, the responses to the fourth question, “What are you not looking forward to in your future job as a doctor?” were in text form and thus qualitative. After the responses were translated into German, except for the UNIC data set, which was in English, we followed an inductive thematic coding approach, meaning that they were categorized based on key terms selected by one of the authors (JB) and checked again by the last author (MWM).

The aim is to achieve high coverage of the questions by common categories, which results from an iterative search for key terms and phrases within the data sets. This allowed the qualitative data to be converted into quantitative form and assigned with high coverage of 96.7% to 100% to 17 different categories. The four most frequently mentioned categories at each university are presented in Table 1, along with exemplary key terms and phrases given by the students. 100% coverage means that all answers can be assigned to a category and that no answers remain unassigned.

Table 1.

Study participation and evaluability

Survey	Valid absolute	Not valid	Valid relative (%)	Participation absolute	Students per annum	Participation relative (%)
KI_2019	156	4	97.5	160	360	44.4
KI_2020	150	0	100	150	360	41.7
LMU_2019	127	17	88.2	144	918	15.7
LMU_2020	243	40	85.9	283	921	30.7
LMU_2021	172	9	95.0	181	847	21.4
UNIC_2018	35	2	94.6	37	96	38.5
UNIC_2019	47	0	100	47	138	34.8
UNIC_2020	6	0	100	6	95	0.06
UNIC_2021	30	0	100	30	159	18.7
UU_2017	306	3	99.0	309	344	89.8
UU_2018	205	1	99.5	206	344	59.9
UU_2019	249	2	99.2	251	344	73.0
UU_2020	93	0	100	93	344	27.0
UU_2021	229	2	99.1	231	344	67.2
Total	2,048	80	96.2	2,128

It should be emphasized that the total number of students v aries greatly between the different universities. At LMU, it ranges from 847 to 921, and at UNIC, between 95 and 159

Data analysis approach

As an initial descriptive step, the distribution of relative frequencies of responses across the selected categories is presented by university and time point. To further explore how the large number of responses could be grouped, we performed a hierarchical cluster analysis. This method creates a hierarchical tree (dendrogram) based on the euclidean distance between the objects as a measure of similarity, representing their relationships and illustrating how they are grouped into clusters at different levels [41].

Using derived cluster dendrograms and rank cross-correlation coefficients, the given responses and their derived categories were first analyzed across gender and by university (a), then analyzed by gender (b), and university (c), and finally summarized using the cluster mean values of every category’s relative frequency to derive the four most frequent categories by gender and university.

Based on the results of the cluster analysis in Fig. 1 and the low participation rate in Table 2, the data for Cyprus 2020 were not included in the further analysis and are therefore no longer included from Fig. 2 onwards. The age distributions are unsymmetrical and show sporadic individual outliers towards higher age. That’s why we have chosen the median as a robust measure. For the rank cross-correlation in Fig. 5, we use a Spearman rank correlation. The relative frequencies of the selected answer categories, their ranks and fluctuations, and potential multiple occupancies are thus given for three analysis directions. For the statistical analysis, we used R, version 2022.02.0 Build 443 [42]. The figures and tables presented here are drawn from the doctoral thesis of JB [43].

Fig. 1.

Relative frequencies of categories among all answers. The categories are plotted on the x-axis while the corresponding relative frequencies are shown on the y-axis. Each university is represented by a unique color. Ludwig-Maximilians University in Germany (LMU), Utrecht University in the Netherlands (UU), Karolinska Institutet in Sweden (KI) and the University of Nicosia in Cyprus (UNIC). The number represents the cohort year. The grouping of data by university is evident, as graphs of the same color are almost on top of each other

Table 2.

Exemplary depictions of the four most frequently mentioned categories according to gender and university resolved

Category	Example key terms/-phrases
Stress	• “being overwhelmed”
	• “working too much”
	• “being left alone”
	• “burnout”
Example:	“That I take patient stories home into my private life.”
Working conditions	• “working hours”
	• “shift work”
	• “on-call duty”
	• “night work”
Example:	“I particularly dislike night shifts, because sleep is very important to me and I don’t yet cope well with lack of sleep.”
Failing	• “mistakes”
	• “not being able to help”
	• “helplessness”
	• “not knowing what to do anymore”
Example:	“Harming a patient because of a mistake I made.”
Health system	• “bureaucracy”
	• “formal tasks”
	• “little time for patients”
	• “insurance benefits”
Example:	“The German healthcare system, which is not ideal for either employees or patients, and the large amount of paperwork that comes with it.”
Hierarchy	• “arrogance”
	• “deliberately keeping others down”
	• “arrogant doctors”
	• “being bossed around”
Example:	“Bad atmosphere and rivalry / catfights among staff.”
Work-life balance	• “not being at home much”
	• “neglecting oneself”
	• “little time for things outside my profession”
	• “neglecting social life”
Example:	“Working hours that make it impossible to reconcile career and family (that’s partly part of the job, since you always have to be available, but surely there are things that could be improved).”
Bad news	• “delivering bad news”
	• “delivering grim news”
	• “having no good news to deliver”
Example:	“Having to deliver bad news to patients and the feeling that nothing can be done.”

Each category shows some example key terms and phrases the students gave based on which the categorization took place

Fig. 2.

Cluster grouping of the datasets. With the help of previously defined clusters, one can further differentiate the effect of university on the given answers. The cluster average is drawn in black

Fig. 5.

Rank cross-correlation matrix and derived similarity dendrogram. It can be seen that the first grouping level links the gender-dependent data within a university with each other

Ethics approval and consent to participate

The survey was voluntary and pseudonymized. All students received information on the survey’s nature, purpose, and procedure, as well as their right to withhold or revoke their consent at any time. Participants voluntarily filled out the questionnaire and informed consent was given for anonymous use of the data. Ethical approval has been obtained from: the Cyprus National Bioethics Committee (Nicosia, Cyprus), NVMO-ERB (Utrecht, the Netherlands), Ethical Review Board of the Technical University Munich (Munich, Germany), Ethical Review Board of the Karolinska Institute (Stockholm, Sweden).

Results

Survey participation and age distribution

Table 1 compares the available survey data by the number of students per cohort, university, and year and distinguishes between evaluable and non-evaluable responses. Each empty answer field was set as not evaluable. The evaluability of the given answers ranges between 85.9% (LMU_2020) and 100% and is, therefore, high on average at 96.2%. Although the categorization approach allows for the assignment of given responses to multiple categories, the median number of categories per response was one across all universities, except for Munich, where the median was two.

The study participation differs greatly between the four universities (KI in Sweden, LMU in Germany, UU in the Netherlands and UNIC in Cyprus) in the considered time. The relative study participation varies between 15.7% in Germany in 2019 and 89.8% in the Netherlands in 2017. The number of students per year is constant in Sweden and the Netherlands, while in Germany and Cyprus, it is not. It can be seen from Table 1 that both the absolute and the relative participation of the students varies greatly depending on the university and timing of the survey. In half of the cases, the relative participation is greater than 38.5%. During the study period, more than twice as many female than male students participated in Germany, Cyprus and the Netherlands, while Sweden showed an equal distribution. This corresponds to the gender distribution in the overall population.

The mean age of all respondents is 20.4 years, with a median of 19 years and a standard deviation of 3.4 years. The highest age median was found in Sweden, with 23 or 21 and the lowest in Cyprus, with 18 in 2021. Even after stratifying the data by gender, the median and mean remained almost unchanged. However, in Germany, male students are generally older than their female peers. For all other distributions, the median age difference is below one year, with no systematic effects in favor of one gender being observed.

Data analysis across gender

Relative frequencies of categories

Figure 1 shows the distribution of all responses across the specified categories, stating survey time and university. The data are grouped by university because the graphs of the same color are almost on top of each other. Only the later discarded dataset from Cyprus 2020 stands out here, but this is likely due to the small number of participants of six, which is also displayed in Table 1.

The visual result is confirmed in the following using hierarchical cluster analysis.

Cluster grouping of responses

It can be seen that the datasets, except for Cyprus 2020, show grouping by university. The data most similar are from the Netherlands and Sweden, followed by Cyprus and Germany. Based on the clusters, the data visualization from Fig. 1 was modified and simplified using cluster averages, which can be seen in the black graphs in Fig. 2.

Summary and similarity dendrogram

The representation of the cluster properties by cluster mean values enables the direct comparison of the groups based on their mean value, as shown in Fig. 3. Each column corresponds to a different university and thus dataset, while each color corresponds to a category named in the left part of the graph. The elements of each column are sorted according to their relative frequency from top to bottom. The vertical position of the symbols within a column corresponds to the rank indicated in the right part of the plot. The highest rank, checking to the highest relative frequency, is assigned to the top row, and the lowest rank to the bottom. The associated relative frequency is indicated within the round symbols in percent, and the data categories are sorted using different colors. The same categories are connected between the columns by straight lines whose color corresponds to the respective category. Categories with the same frequency are assigned to the same rank within the cluster. Several ranks corresponding to the multiple number of ranks remain unoccupied below. This representation allows, on the one hand, the identification of the rank of a category within a university and, on the other hand, the comparison of the ranks of a category between the universities.

Fig. 3.

Comparative presentation of relative frequencies of the named categories in percent. The numbers at the bottom of the graph indicate the sample size for each university. Each color represents one category, while the steepness of the connecting lines illustrates fluctuations in ranking across universities

The four categories most frequently mentioned at the various universities are:

Germany dataset (LMU)

“health care system” with 43.3%, “working conditions” with 40.4%, “stress” with 28.2% and “hierarchy” with 26.3%.

Netherlands dataset (UU)

“failure” with 40.6%, “stress” with 36.8%, “bad news” with 19.2% and “working conditions” with 18.1%.

Sweden dataset (KI)

“stress” at 50.9%, “failure” at 40.7%, “health care system” at 20.0%, and “working conditions” at 17.0%.

Cyprus dataset (UNIC)

“failure” with 45.2%, “work-life balance” with 25.6%, “stress” with 21.7%, and “working conditions” with 20.2%.

The categories “working conditions” and “stress” are within the top four at all universities. “Failure” takes one of the top two ranks in the Netherlands, Sweden, and Cyprus but is only ranked sixth in Germany. “Health care system” is often mentioned in Germany, ranks third in Sweden, and fifth for the Netherlands and Cyprus. Concerns about “hierarchy” are mentioned in Germany within the top four ranks but play a minor role at the other universities, ranging from sixth in Cyprus to eighth in the Netherlands and Sweden. Other major differences in ranking the categories mentioned can be observed in the “bad news” and “work-life balance” categories. The delivery of “bad news” ranks third in the Netherlands, in the middle range in Sweden and Cyprus, and in the lower third of the rankings in Germany. The category “work-life balance” is ranked second in Cyprus and is in the middle of the ranking at the other universities. Less than 6% of all students ranked the categories “merit”, “no concerns”, “everyday monotony”, “undesirable field of activity”, “genderism”, “litigation”, and “risk of contagion”, meaning that these categories play only a minor role.

Figure 3 can also be used to find categories whose rankings differ greatly between universities, as evidenced by steep connecting lines. A quantitative measure of rank variation is the distance between a category’s maximum and minimum rank across all universities. An overview of rank variation is derived from Fig. 3. In 50% of the cases, the rank variation is between two and five ranks. Thus, the most rank-stable category is “risk of contagion”, with a rank of 16 or 17 across all data sets. “Shortage of skilled workers” and “bad news”, however, show high-rank fluctuations of more than eight ranks.

Based on a Spearman rank correlation, a rank cross-correlation matrix was created in Fig. 5 using a simple dendrogram and Fig. 3. Again, the responses in Sweden are most similar to those in the Netherlands, followed by Cyprus. Responses from Germany differ the most from those of the rest of the universities.

As the following analysis by gender uses the same procedure described here, only a summary based on a similarity dendrogram will be given.

Data analysis by gender

Male students

As already in the analysis across-gender, the data was grouped by university (Fig. 4). The ranking of the categories of students and their linkage between the different universities mentioned by the male students were also examined.

Fig. 4.

Rank correlation plots were made university-specifically to compare cross-gender and gender-specific evaluations. Each row corresponds to a university; the columns are assigned to gender. The rank of the gender-independent categories is on the y-axis, and that of the gender-dependent ones is on the x-axis in descending order. The four most frequently mentioned categories are shown at the top right, separated by lines

Comparing this evaluation with the cross-gender analysis, no difference between the rank assignments is noticeable. The greatest frequency differences are found in the Germany dataset in the category of working conditions and Sweden in the categories of stress and failure, each of which is more than 5% lower difference in relative frequency.

Large differences in ranking between different universities can be seen in the “bad news” category. It is ranked third in the Netherlands and fifth in Sweden, but in Germany, on the other hand, only in 16th place, and in Cyprus in tenth place.

In half of the instances, the rank fluctuation ranges from three to six ranks. Consequently, the category that displays the highest stability in terms of rank is “risk of contagion,” again maintaining ranks 16 or 17 across all datasets. However, the category “bad news” exhibits significant fluctuations in rank, exceeding up to 13 levels. In contrast to Fig. 3, the categories “health care system,” “studying,” and “everyday monotony” among male respondents are subject to higher fluctuations between the universities than when all data are included.

Female students

Analysis of female students’ responses also confirmed data grouping by university and followed the same procedure as in the male cohort.

Compared to the male-specific and the across-gender-evaluation, the categories in this approach have a partially different ranking, whilst the four most frequent ranks are still congruent. In Germany, the categories “working conditions” and “health care system” as well as “hierarchy” and “stress” changed ranks, indicating a different perception of importance among female students. This also applies to the categories “working conditions” and “bad news” at Utrecht University. It is worth noting that the categories “failure” and “stress” are mentioned 5% or higher in absolute importance in the Netherlands and Sweden compared to the cross-gender analysis.

There are substantial discrepancies in ranking the category “bad news.” It ranks fourth in the Netherlands and fifth in Sweden but only 15th in Germany and 11th in Cyprus.

In half of the cases, the rank fluctuation ranges from three to six ranks. The categories demonstrating the highest rank stability are “exhausting patients” and “litigation”, maintaining levels of six or eight and 15 to 17 across all datasets. However, the category “bad news” exhibits significant rank fluctuations, spanning up to 11 ranks. Contrary to Fig. 3, the categories “hierarchy” and “genderism” among female participants experience greater fluctuations between universities than when all data are considered.

The university-specific data analysis

The analysis to date does not allow for direct conclusions about correlations between gender within a university, even though it is important for gender-specific improvement measures. Due to this, rank correlation plots were made university-specifically between the cross-gender and gender-specific evaluation, as seen in Fig. 4. All data sets show a high rank correlation in the range of ρ = 0.943 to ρ = 0.994. The rank of the four categories most frequently mentioned by male participants is always identical within the university to the rank of the cross-gender analysis. Within the top four levels, the data sets of female students have a slightly lower correlation. In this figure, the exceptionally high rank correlation between the cross-gender and gender-specific analysis of the mentioned categories for each university is striking. A derived rank cross-correlation is shown in Fig. 5 and demonstrates that the answers given by both genders within a university are most similar.

Discussion

Cross-gender analysis of the distribution of responses across categories using cluster analysis shows that the data can be successfully grouped by university and their change over time within a university is less than the differences that occur between universities. Since the data for a university is thus not significantly dependent on the time of the survey, we were able to represent it by its mean value over time. From the rankings of the categories in Fig. 3, it could be deduced that “stress” and “working conditions” were among the four most frequently mentioned categories at all universities and hold first international rank in Table 3. In addition, the categories “failure”, “healthcare system”, “hierarchy”, “bad news” and “work–life balance” were frequently mentioned across all universities. This grouping in categories closes an important gap in literature, as they represent the key areas of concerns of first-year medical students on an international level and appear to be of comparable relevance across different universities. Table 2 provides valuable insight into the concrete fears expressed by medical students, which they already hold at the beginning of their studies and thus without having much prior experience. The analogous pattern of the cluster grouping in Fig. 2 and the overall high rank cross-correlation values in Fig. 5 suggest that the challenges in medical training are fairly similar in different parts of Europe and may reveal structural problems of the healthcare system that should be of primary interest for future educational changes. Interestingly, despite minor differences in relative frequencies of the four highest ranked categories, no direct influence of respondents’ gender on the frequency of the categories mentioned can be observed, which could also be confirmed by the cluster analysis of the cross-university rank cross-correlation in Fig. 5.

Table 3.

Summary of the four most frequently mentioned categories according to gender and university resolved

University	Gender	Age (a)	Health system	Working conditions	Stress	Hierarchy	Failing	Bad news	Work-life balance	No. Answers
LMU	M	≤ 19	•	•	•	•				69
	M	>19	•	•	•	•				75
	W	≤ 19	•	•	•	•				194
	W	>19	•	•	•	•				172
UU	M	≤ 19		•	•		•	•		490
	M	>19		•	•		•	•		246
	W	≤ 19		•	•		•	•		218
	W	>19		•	•		•			123
KI	M	≤ 19		•	•		•			25
	M	>19	•	•	•		•			116
	W	≤ 19	•		•		•			32
	W	>19	•	•	•		•			130
UNIC	M	≤ 19			•		•		•	24
	M	>19					•		•	12
	W	≤ 19		•			•		•	43
	W	>19			•		•		•	33
International rank			3	1	1	4	2	4	4

The lower part of the figure indicates the international rank of the respective category

From the ranking scatter in Fig. 3, we can see that the delivery of “bad news” and “shortage of skilled workers” show the greatest variation and are therefore rated very differently between the universities in terms of their importance. In the Netherlands, the delivery of “bad news” is ranked fourth and is also generally considered one of the most challenging tasks for physicians [44], as it often involves emotional insecurity and thus disengagement and can involve psychological as well as physical illness [45]. It is noticeable that medical students in Munich care less about the delivery of bad news or failure, because they are more concerned about their financial situation, hierarchy or demanding patients compared to the other universities. Even though previous research on this data identified hierarchy and work-life balance as key concerns, work-life balance is of even greater importance to the other universities [29].

On the other hand, fear of “risk of infection”, for example, has a stable ranking and is together with “lawsuits” in the last two ranks at all universities with a ranking fluctuation of one. It is also notable that sexism is placed between rank 11 and 15, which seems to be underestimating the existing effect of gender-bias in medical education and medical practice [46, 47]. The grouping of the data by universities could be confirmed by the rank cross-correlation of the university-specific categories in Fig. 5.

The category “no concerns” was selected only rarely across universities (Fig. 3), indicating that the data reflect the presence of concerns. Care should be taken when interpreting the data, as it only includes students at the very beginning of medical school. Results from previous analyses revealed that concerns expressed by students have a dynamic component and change over the course of study [29]. A particularly insightful follow-up study would be to examine how more advanced students respond to this. Importantly, many first-year medical students have limited clinical exposure and, as several authors note, have often not experienced personal failure in an academic or clinical context [48], which may hamper the interpretability of the given data [7]. This means that the presented data should be treated with care in terms of its generalizability on the whole body of medical students, especially final-year medical students [14].

To date, no such international study has been conducted to identify the concerns and fears of students about their professional future. Comparability with other publications is limited by the fact that most of these surveys were conducted under different conditions and with different questions. Nevertheless, the following section will point out commonalities with other studies and address differences in the answers given by the survey universities.

In 2011, Diderichsen et al. took a similar approach at a Swedish university, surveying first- and last-semester students over two years about their future expectations [35]. In a gender-specific evaluation methodology similar to our work, categorization was used. The 34 categories were divided into four main themes, with the main themes of “work” and “family” ranked by most students as crucial. At the same time, “leisure” and “quality of personal life” were cited less frequently. Almost one-third of the respondents had concerns regarding “work-life balance” at the end of their studies, which could only be indirectly confirmed by our research, as at the beginning of studies, the cross-gender ranking of the category “work-life balance” in Sweden was only ranked sixth. One possible explanation is that first-year medical students have not yet experienced the adverse effects of long working hours and work-life interference, which contribute to their different perceptions of working life [49].

A study by Gąsiorowski and Rudowicz in 2014 captured the change in attitude of Polish students during their studies at the Pomeranian Medical University [36]. Questionnaire surveys showed that mistrust and pessimism increase during the course of studies and that attitudes toward the profession change negatively. Thus, at the end of their studies, almost eight times more students than at the beginning of their studies stated that they would not decide to study medicine again. At the start of their studies, students criticized teaching quality and lack of support, whereas, by the end, their primary concern shifted to a lack of practical skills, aligning with findings from other studies [50].

Improving physicians’ mental health throughout their careers requires individual and organizational measures from different institutions. For our cohort, especially medical schools and hospitals are key players. The initially introduced concept of professional identity formation is a means to being a good and healthy physician [12]. PIF is deeply rooted in socialization, benefiting significantly from mentoring, role models, and experiential learning, particularly when guided by a longitudinal approach [12, 13]. Integrating PIF as a core element of medical curricula forces institutions to create explicit opportunities and room for socialization and establish international standards for professionalism [10]. This can be achieved by implementing Entrustable Professional Activities (EPAs) into medical curricula [51]. Core EPAs, such as patient handover and obtaining informed consent, would strengthen essential communication skills. Other, e.g., identifying and communicating system failures, promote honesty and transparency in handling errors [52]. Mastering these core EPAs can facilitate a smooth transition into medical practice, as students will have higher preparedness for their role as residents [50, 52]. Importantly, students should be supported up until they are professionals. This means earlier and more in-depth patient contact [10, 33], strengthening critical (self-) reflection [53] as well as structured transitional roles such as FiY1 [54] or mentoring [55].

Future research should focus more on longitudinally tracking the development of professionalism in medical students. This includes addressing individual challenges, such as managing failure and delivering bad news, alongside broader systemic considerations of healthcare, such as work-life balance and stress management. How systemic changes and highly individual student needs interact is of growing interest. Moreover, efforts to implement practical changes in medical curricula, such as integrating core EPAs, warrant deeper practical exploration [52]. Such research will help shape medical education to effectively establish concrete improvements on the side of individual student needs.

Strengths and limitations

The present study is set longitudinally for ten years and already provides a data basis of more than 2000 responses, being unique in the way it evaluates international students’ concerns and fears about their future as a doctor. The most significant gain in knowledge lies in the systematic categorization and quantification of the given answers to identify transnational and national problems as a function of the evaluation parameters of university, time and gender. Even though the gender parameter has only a minor influence on the results, based on the geographical grouping, it is possible to derive improvement measures sorted according to student-assessed relevance later. In the future, the categorization can be used to develop multiple-choice questionnaires, which could improve structuring, participation and evaluation.

As this work covers only a partial aspect of the entire study, once the overall study has been completed and all the questions it contains have been evaluated longitudinally, a more extensive coverage of relevant topics can be expected, which will significantly broaden the spectrum of the current state of knowledge in the field. It is important to note that although such international comparisons are beneficial, differences in education systems, healthcare systems, and social contexts make the comparisons harder to interpret.

Conclusions

With a sample size of 2,048 responses across four different universities, our study examined international medical students’ answers to the question, “What are you not looking forward to in your future job as a doctor?” which provides direct insights into medical students’ concerns and anxieties regarding their future. After successfully converting responses into quantitative form, relative frequencies between universities were analyzed as a function of timing and gender. Cross-gender analysis of the distribution of response categories showed that the data can be grouped by university and that their change over time within a university is smaller than between them. This finding, obtained by cluster analysis, could be validated with rank cross-correlation and similarity dendrograms based on the respective university-specific temporal mean of relative frequencies in all mentioned granularities of the study.

The categories “stress” and “working conditions” are among the four most frequently mentioned categories at all universities, while breaking “bad news” and “shortage of skilled workers” show the greatest variation. The main result of this work can be seen in Table 3 and can serve as a basis for university-specific and cross-university follow-up actions.

Abbreviations

LMU

Ludwig-Maximilians University in Germany

UU

Utrecht University in the Netherlands

KI

Karolinska Institutet in Sweden

UNIC

University of Nicosia in Cyprus

FiY1

Foundation Interim Year 1

PIF

Professional Identity Formation

EPA

Entrustable Professional Activity

Authors’ contributions

JW, MR, CC, PL, AW and MWM designed and ran the study and collected the data. JB, VK and MWM analyzed the data. VN and MWM drafted the manuscript. VK, PB and MWM supervised the project. All authors contributed to the critical revision of the manuscript and read and approved the submitted version.

Funding

None.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The survey was voluntary and pseudonymized. All students received information on the survey’s nature, purpose, and procedure, as well as their right to withhold or revoke their consent at any time. Participants voluntarily filled out the questionnaire and informed consent was given for anonymous use of the data. Ethical approval has been obtained from: the Cyprus National Bioethics Committee (Nicosia, Cyprus), NVMO-ERB (Utrecht, the Netherlands), Ethical Review Board of the Technical University Munich (Munich, Germany), Ethical Review Board of the Karolinska Institute (Stockholm, Sweden).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.