This study employed a cross-sectional, observational, complementary mixed-methods design in which quantitative and qualitative data were collected concurrently. The quantitative and qualitative strands were related but independent, and their findings were reported in parallel without integration.

The study was conducted at the College of Medicine and Health Sciences (CMHS) at UAE University, the largest medical school in the country, which graduates approximately 100 physicians annually. At CMHS, in the first 2 years, students complete professionalism and healthcare principles courses while studying general education and foundational medical sciences. In years 3 and 4, they study basic and clinical sciences through an integrated, system-based curriculum and develop clinical assessment skills through simulation-based learning. Direct clinical exposure to patients primarily begins during the final 2 years through clinical clerkships. No formal professionalism training is offered after the first two years of basic sciences.

Medical students studying at CMHS in 2023 in pre-clinical and clinical years (Years 1–6) who consented to participate in the study.

Data collection for the study was conducted using the following instruments:

All 638 currently enrolled students, comprising the accessible study population, were invited to participate using a census sampling approach. As this was an exploratory study, no formal sample size calculation was performed. One hundred eight students (16.9%) consented and completed the questionnaires.

Students were grouped based on their curriculum stage into four groups for analysis, namely Basic Sciences (Years 1–2), Preclinical (Years 3–4), Early Clinical (Year 5), and Late Clinical (Year 6).

Data were analyzed using the IBM SPSS version 22. Both the PAS and PIQ were administered in English, following the original response formats, including reverse-scored items where applicable. After descriptive analysis, bivariate comparisons were made using Pearson’s correlation coefficient and one-way ANOVA tests. Reliability analysis was carried out using Cronbach’s alpha. Principal component analysis was used for factor analysis. Factors with eigenvalues greater than one were evaluated. Multivariate analysis used linear regression with PAS and PIQ as outcome variables. An alpha of 0.05 was considered the cut-off for statistical significance. Missing values were minimal.

The qualitative data underwent inductive thematic analysis, where initial codes were developed to capture key concepts and refined iteratively into cohesive themes. Two independent experts were involved to ensure reliability and reduce individual bias. Each expert reviewed and coded the data separately, followed by iterative discussions to resolve any discrepancies. Consensus was reached through discussion until complete agreement on the final themes was achieved. Subsequently, a matrix analysis was performed to determine the frequency of themes in each question.

The Research and Graduate Studies Ethics Committee approved the study, United Arab Emirates University, Al Ain, UAE (Reference No: ERSC_2023_3069ERSC_2023_3069). Informed consent was obtained from participants regarding the publication of the findings.

The reliability coefficient (Cronbach’s alpha) was 0.763 for the PAS questionnaire and 0.767 for the PIQ instrument. The reliability coefficients remained stable if any of the items were removed from the analysis, indicating internal consistency.

The two questionnaires used Principal component analysis to discover underlying common factors. For the PAS and PIQ instruments, the Kaiser-Meyer-Olkin Measure of Sampling Adequacy was 0.656, indicating a likely utility of factor analysis. This was supported by a significant Bartlett’s Test of Sphericity (p < 0.001). The extraction communalities were greater than 0.5 for all items, indicating that all items fit with the factor solution, and none should be dropped from the analysis. The principal component analysis of PAS yielded seven factors with eigenvalues >1, accounting for 64% of the variability in the original items. The extracted and rotated factors had the same (64%) variance explained. Subscales were not extracted from the factors as a clear pattern did not emerge. The principal component analysis of PIQ yielded three factors with eigenvalues >1, accounting for 67% of the variability in the original items. The extracted and rotated factors had the same (67%) variance explained.

PAS scores correlated strongly with empathy (r = 0.804, p < 0.001, n = 108), professional relationships (r = 0.804, p < 0.001), and Responsibility (r = 0.839, p = < 0.001). Empathy was associated with Professional Relationships (r = 0.510, p < 0.001) and Responsibility (r = 0.599, p < 0.001). Responsibility was moderately correlated with Professional Relationships (r = 0.613, p = < 0.001). Age did not correlate with scores on the PAS, PIQ, Empathy, Responsibility, and Professional Relationships (all r values < 0.2). Similarly, PIQ did not correlate with PAS, Empathy, Professional Relationships, or Responsibility (all r values < 0.14). Students interested in surgical specialties had lower scores on PAS but higher scores on professional identity. Students interested in non-clinical specialties had lower scores for Professional identity than other groups (Table 2), although these differences did not reach statistical significance.

PAS (p = 0.024), Empathy (p = 0.009), Professional Relationships (p = 0.007), and age (p = < 0.001) were significantly different across the four stages of medical students’ progression. All these variables had the lowest scores in the Basic Sciences stage (Figure 1). However, the means of Responsibility (p = 0.604) and PIQ (p = 0.567) across the four stages were not significantly different. Mean scores in basic science were lower than other stages for PAS (p = 0.024), empathy (p < 0.001), and professionalism (p < 0.001) but not for PIQ (p = 0.56) (Table 2).

On linear regression with PAS as the outcome, none of the explanatory variables (age, sex, intended specialty, stage of medical student progression) were statistically significant predictors (adjusted R-squared = 0.375).

Qualitative analysis indicated that students learned professionalism through dedicated courses, reflection, observation of practice, and emulation of role models. Some viewed professionalism as an innate trait. Matrix analysis highlighted observation and experiential learning as prevalent modes, with less emphasis on self-reflection. Regarding professionalism, assessment, observation, and feedback-based methods were most common, while self-reflection and multisource evaluation were less cited. Tables 3, 4 present major themes and sub-themes based on frequency.

Professionalism and professional identity are overlapping and mutually reinforcing; discussing professionalism through the lens of identity can deepen understanding and promote professional behavior (22). In our study, significant associations were found between PAS scores and the constructs of empathy, professional relationships, and responsibility. Bivariate analyses revealed strong correlations between empathy and professional relationships, as well as between empathy and responsibility. Responsibility, in turn, showed a moderate correlation with professional relationships.

A recent cross-sectional study found a moderate positive correlation between medical students’ perceptions of their professional identity and their level of professionalism (23). However, our data showed that PIQ scores had low correlation with empathy, PAS, professional relationships, and responsibility (all r values < 0.14). Combined with the prolonged plateau in PIQ scores, our findings suggest that professional identity may be conceptually distinct from professionalism than previously assumed. Although prior literature has highlighted the two as distinct but interrelated constructs (5), our findings argue for separate, targeted strategies in curriculum design to address each area. This distinction warrants further exploration to refine professional development frameworks in medical education.

Our students’ overall professionalism and professional identity scores align with previous research. The mean professional identity score was 38.5 out of 50, mirroring results from a Dutch validation study (mean score: 38.7) (21). Similarly, the average professionalism score of 101.4 out of 110 was slightly higher than the 96.4 reported in a Turkish validation study (20). Despite a higher proportion of female participants in our study (66%) than in Tanrıverdi et al. (51%), we did not observe gender-based differences in empathy or professional relationship scores. This contrasts with earlier findings suggesting gender influences these domains (24), highlighting potential contextual or cultural factors that affect gender differences.

A key finding was the increase in PAS, empathy, and professional relationships scores from basic sciences to pre-clinical years, followed by a plateau through clinical years. In contrast, responsibility and professional identity scores remained consistent across all stages. This pattern differs from a cross-sectional Irish study, which found random variations in professional identity across student cohorts without a clear trajectory (25). The observed increase in early years may be attributed to structured professionalism courses and coping with academic transitions, supporting existing evidence that such transitions can enhance identity and professionalism development (26).

The lack of significant change in scores during the pre-clinical to clinical transition indicates that neither formal instruction nor the hidden curriculum substantially impacts these constructs in later years. This finding is explained by the literature that highlights the benefits of longitudinal and early clinical exposure on professional identity development (27). A strong preexisting cultural and religious identity, coupled with a relative lack of experientially immersive curriculum, may also be contributing to the plateau. These results underline the importance of continuous and integrated approaches in medical curricula, especially in the UAE and similar educational contexts, to foster sustained growth in professionalism and identity.

Students aspiring to surgical careers demonstrated slightly lower professionalism scores but higher professional identity scores, although these differences were not statistically significant. This aligns with a Danish multi-institutional study where surgical-track students had notably lower empathy scores (28). Similarly, students inclined toward non-clinical specialties in our study reported lower professional identity scores, reaffirming patterns seen in earlier studies (29). These insights may inform guidance and mentorship programs tailored to students’ career aspirations.

Qualitative responses highlighted two primary sources of professionalism development: formal education in the early years and role modeling during clinical exposure. However, consistently integrating these methods across all years remains challenging (30). Some students described professionalism as innate, and few explicitly mentioned reflection or self-awareness. The recurring emphasis on observation and role modeling underscores the hidden curriculum’s role, consistent with previous research findings (31, 32). In terms of assessment, students favored observational and multisource feedback, which aligns with the broader literature on best practices in professionalism assessment (33). These preferences should be considered when designing evaluation methods to ensure alignment with learner expectations and educational goals.

The 17% participation rate and the study’s cross-sectional nature limit generalizability and preclude longitudinal analysis. However, despite this limitation, the stagnant scores of professional identity across clinical years highlight a significant finding of concern.

The study occurred at a medical college that predominantly admitted students of one nationality. While this may restrict the generalizability of the findings, it aligns with the demographic composition of many medical colleges globally.