A mixed-methods study on the course of professional identity formation in undergraduate medical students | BMC Medical Education

Context

This study was conducted at the Radboud University in Nijmegen, the Netherlands, where the undergraduate medical school curriculum is structured into a 3-year Bachelor’s programme followed by a 3-year Master’s programme [19] (provided in Fig. 1). Postgraduate medical training is the same as residency in the United States. The Master’s programme integrates clerkships with occasional classroom training, creating a dynamic interplay between practical and theoretical learning.

Medical curriculum at Radboud University, Nijmegen

Participants

The target population for this study consisted of medical students enrolled in the undergraduate medical curriculum at Radboud University in Nijmegen, the Netherlands. Only students of the Master’s programme were included in this study. This programme follows a structured schedule divided into distinct”episodes,”alternating between classroom training and clinical training (clerkships). Figure 2 presents an overview of the Master’s programme and its clerkship structure. Episodes 0–3 correspond with the first year of the Master’s programme, episodes 4–6 with the second year and episodes 7–8 with the third and last year. Each month, a new cohort, referred to as a Clinical Rotation Group (CRG), comprising 26–30 medical students, starts their clinical rotations. Students remain within their assigned CRG throughout the 3-year Master’s programme, fostering continuity and collaboration within their cohort.

Clinical rotations at the Radboud University Medical Center

This study was divided into two sub-studies. In Sub-study 1, all registered students between Episode 1 and Episode 7 of the Master’s programme were included, between September 2020 – September 2022. Participants were asked to complete an online questionnaire, the extended Professional Self Identity Questionnaire (extended PSIQ), which is detailed later in this paper. The data collected from sub-study 1 was anonymized, preventing direct linkage of the information to individual participants. As a result, we undertook sub-study 2, which was designed to address this limitation. In sub-Study 2, we aimed to correlate data from the extended PSIQ with the personal data stored in participants’online portfolios, for a more comprehensive analysis of the participants’progress and performance.

Sub-study 2 employed purposive sampling to select the participants. Out of the group of students mentioned above (sub-study 1), 12 CRGs (out of 36 in total) were invited to also complete the extended PSIQ. This approach enabled the connection of individual students’ PSIQ scores to their e-portfolio data, which was not feasible in Sub-study 1 due to the anonymity of responses. Specifically, four CRGs were selected from Episode 1, four from Episode 4, and four from Episode 7, ensuring representation from different stages of the Master’s curriculum, as these Episodes represent the beginning, mid- and end of the fixed components of the studies. We selected these specific clerkships as they comprehensively reflect the entirety of the master’s curriculum. Each clerkship was carefully chosen to align with the core areas of study and practical experiences emphasized throughout the program. By integrating these diverse clerkships, we aim to ensure a well-rounded approach to the learning process, encompassing as many as possible essential facets of the curriculum.

Instrument

Sub-study 1

For Sub-study 1, we utilized the “extended PSIQ” questionnaire. The original PSIQ, developed by Crossley and Vivekananda-Schmidt [28], serves as a research instrument designed to investigate curricular elements that contribute to the development of professional self-identity. It is not intended to analyse the nature or content of professional self-identity or for individual assessment but may serve as a reflective tool for developmental purposes [28].

To adapt the PSIQ for this study, we added three questions to examine the relationship between entrustable professional activities (EPAs), transitioning between clerkships and PIF. The complete original PSIQ can be found in Appendix 1 A, while the extended version is provided in Appendix 1 C (Dutch) and Appendix 1D (English). The original PSIQ has been translated by Radboud In’to Languages, a Dutch expertise centre in language and communication and translation agency. The translation process included back translation, involving translating the questionnaire into Dutch and then having a separate translator convert it back into English, to identify and correct any discrepancies or loss of meaning.

Sub-study 2

In Sub-study 2, the extended PSIQ, as described earlier, was utilized alongside data from participants’online portfolios to obtain a more comprehensive understanding of their actual performance during clerkships. Within the Master’s programme at the Radboud University Medical Center, students systematically collect feedback forms in their online portfolios. These forms include narrative feedback from supervisors on specific EPAs, individual learning goals, and overall evaluations. By integrating the extended PSIQ responses with this portfolio data, we aimed to establish a clearer connection between self-reported PIF and actual performance during clinical training.

Data collection

The data collection process was divided into two sub-studies, which are detailed below and illustrated in Fig. 3.

Sub-study 1

In Sub-study 1, all registered medical students between episode 1 and 7 were invited to complete the extended PSIQ between September 2020 and September 2022, employing a longitudinal research design. The extended PSIQ was administered during the regular online education evaluations, taking place after each episode, which occur on a quarterly basis. Thus, in the 2 years between September 2020 and September 2022, the extended PSIQ was administered eight times. Prior to participation, informed consent was obtained via an online consent form. To ensure anonymity, no personal identifying information was collected during this phase. The longitudinal data gathered in this sub-study provides insights into how PIF develops at the group level over time.

Sub-study 2

To address the research question at the individual level, Sub-study 2 was incorporated. For this sub-study, medical students who met the inclusion criteria were personally invited to participate by the lead researcher (AB) during a regular educational meeting. After obtaining written informed consent, students were asked to complete the extended PSIQ. The data collected were pseudonymized to protect participants’identities.

Additionally, data were gathered from the personal online portfolios of the abovementioned selected medical students to review their actual performance, including:

1. a.

   Final clerkship assessment: at the conclusion of each clerkship, students receive an assessment from their supervisor, which is based on both the collected feedback and the supervisor’s professional evaluation of the student. This assessment is classified into one of four categories: OVN (below the expected level), VN (at the expected level), BVN (above the expected level), or GU (no assessment possible).

2. b.

   Supervision levels from the previous clerkships: students receive daily feedback reports from their supervisor, which include written feedback and a supervision level. The supervision level, indicated on a scale, reflects the degree of trust the supervisor places in the student to perform specific EPAs independently. An overview of the supervision levels is provided in Appendix E.

3. c.

   The mean score on the Dutch national medicine progress test: this test, administered four times a year, measures the overall level of medical knowledge. The results contribute to a yearly summative assessment of the student’s academic progress.

4. d.

   The written feedback collected in the students’ evaluations: this includes the narrative feedback collected during both intermediate and final evaluations of the student’s performance.

In sub-study 2, we received responses from 240 medical students in 12 different CRGs. All attending students agreed to participate. We missed the students who were absent from the class at the time the questionnaire was handed out. This group consisted of students from 3 different clerkships. In order to gain more knowledge about how the highest- and the lowest scoring students score on the extended PSIQ, we selected 60 students by purposive sampling: the 10 students with the lowest mean score and the 10 students with the highest mean score on the extended PSIQ in all three clerkships. This produced a total of 30 students with the lowest scores (hereafter called the LS group) and 30 students with the highest scores (hereafter called the HS group).

Data analysis

To address our research question, we employed a mixed-methods design that integrated both statistical and qualitative analyses.

Sub-study 1: statistical analysis of the PSIQ

The data were analysed using SPSS (version 29). Initially, descriptive statistics were applied to examine the data. Subsequently, reliability and factor analyses were conducted, followed by a one-way ANOVA to assess group differences.

Sub-study 2: statistical analysis of the PSIQ & qualitative analysis of the portfolio

An independent samples t-test was conducted to compare scores between the two groups (LS and HS). Moreover, a chi-square test of independence was used to analyse the relationship between gender and group assignment (LS vs. HS). The written feedback collected from the intermediate and final evaluations was subjected to qualitative content analysis. This analysis was conducted by AB (lead author, educationalist, and researcher), SS (student assistant), CF (professor of workplace learning with expertise in both medicine and education), and MP (professor of student wellbeing and lifelong learning, and general practitioner).

A content analysis can be ‘any technique for making inferences by systematically and objectively identifying special characteristics of messages’ [31]. In this study, a deductive approach to content analysis was chosen due to its focus on testing pre-existing theories by organizing and coding data into predefined categories. This deductive approach also requires attention to the organization of categories to avoid overlooking potentially relevant themes in the data. In this study, an extensive review of the literature and relevant theories guided the creation of the coding scheme, ensuring that it was comprehensive and aligned with the research objectives.

The initial categorization process was informed by defining the research objectives and established concepts within the field, which helped ensure that all key themes related to the research focus were considered in the coding structure. Thereafter, we became more familiar with the data by reading through the data (the online portfolio) to get an initial sense of the content. During this stage, we started noting recurring themes and concepts. Based on the research objectives and the initial review of the data, AB and SS identified categories that might be relevant for the analysis, and preliminary codes based on these categories were created. These codes were also informed by the subjects of the PSIQ (e.g. professional identity formation, entrustment, collaboration, communication and reflection). All intermediate and final evaluations were then coded by AB and SS with this coding scheme. AB and SS then discussed the codes developed and formulated new codes. The coding scheme was updated until a consensus had been achieved, and a definitive coding scheme was then agreed by all researchers. All the data were coded again in the final analysis based on this coding scheme. The feedback, provided by supervisors at the end of specific clerkships, offered detailed insights into students’actual performance during the corresponding clinical rotations.

To mitigate potential researcher biases during data collection and analysis, several strategies were implemented throughout the research process. First, multiple researchers were involved in the data collection and analysis phases, allowing for a collaborative approach that promoted a diversity of perspectives and minimized individual biases. Inter-coder reliability checks were conducted during the data analysis to ensure consistency and to validate the findings across different analysts. Furthermore, we documented decisions made at each stage of the research process to demonstrate the objectivity of our approach. Also, where possible, we applied reflexivity by encouraging our research team to reflect on their own potential biases and preconceptions throughout the study. This self-awareness helped identify and control for any inadvertent influences that could have shaped the research process or outcomes.