Exploring core competence in first-year medical students through narrative medicine and peer-assisted learning: a mixed-methods study Authors: Peixin Lin, Xiaolin Yin, Xuankun Zheng, Sijia He, Yuexin Cai, Xiaozheng Deng, Tongtong Chen, Weiquan Liang, Suqiong Liang, Yuanke Liang, Yiwen Lu, Haoyu Lin Published: 2026 ABSTRACT BackgroundEarly development of core competencies is essential in medical education, yet traditional medical curricula often struggle to bridge the gap between biomedical knowledge and humanistic application. Drawing on social constructivism, this study examines an integrated Narrative Medicine (NM) and Peer-Assisted Learning (PAL) model designed to facilitate the transition from foundational knowledge to contextualized professional application.MethodsAn explanatory sequential mixed-methods design was employed. A 19-indicator Knowledge-Skills-Attitude (KSA) scale was developed via a Delphi process and evaluated for preliminary psychometric properties. Following a 10-month intervention with 128 first-year students (Experimental n = 64; Control n = 64), hybrid thematic analysis of 14 in-depth interviews explored the underlying pedagogical mechanisms.ResultsThe experimental group reported significantly higher self-reported core competency scores than the control group (77.42 ± 7.45 vs. 68.00 ± 7.31; p < 0.001; Cohen's d = 1.28). At the item level, critical thinking and empathy-driven awareness did not reach statistical significance. Qualitative findings suggested that participants perceived narrative features of the integrated package as a catalyst for meaning-making, while peer-assisted elements were viewed as providing social scaffolding necessary for skill development.ConclusionThis exploratory study provides preliminary evidence that the integrated NM-PAL model may support students' self-reported development of contextualized clinical competence and early professional identity. By providing a low-pressure collaborative environment, the model may support early reflective engagement that could be foundational to professional development, warranting further longitudinal and confirmatory research.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12909-026-10160-7. Introduction The transformation of professional education globally has led to the rapid adoption of Competency-Based Medical Education (CBME), shifting away from time-based training towards defining clear, outcome-based abilities for graduates [1–3]. This paradigm shift is crucial for ensuring that future healthcare professionals are equipped to respond effectively to the complex and dynamic demands of modern health systems [3, 4]. Central to medical training is the requirement for trainees to attain core competence, increasingly conceptualized as a multilayered construct that demands a holistic integration of conceptual understanding, technical proficiency, and professional values consistently demonstrated through clinical practice [3, 5]. In this context, the KSA (Knowledge, Skills, and Attitudes) framework is widely employed as a structured tool to describe and assess specific, observable dimensions of competence [6–8]. By providing a practical lens to operationalize these aspects, the KSA framework supports educators in designing effective training strategies and acknowledges that true professional competence transcends fragmented knowledge, requiring the synthesis of multiple abilities within real-world contexts [6–8]. Achieving a high level of competence necessitates a holistic integration of conceptual understanding, procedural knowledge, and essential dispositions such as ethics and values [2]. Drawing upon Ten Cate's multilayered construct of competence as a theoretical lens, this study recognizes that while the development of canonical knowledge (Layer 1), contextual application (Layer 2), and personalized approach (Layer 3) is not strictly sequential, the pursuit of these comprehensive abilities must begin early in the educational trajectory [4, 9]. Targeting first-year medical students represents a strategic developmental imperative; early immersion in patient-centred frameworks serves to consolidate a fundamental "whole-person concept" and facilitates professional identity formation before an intensive biomedical focus potentially overshadows these critical humanistic foundations [4, 10]. One promising approach to address these foundational dimensions, particularly the attitudinal and communicative elements that are frequently marginalized in traditional curricula, is Narrative Medicine (NM) [11, 12]. NM prioritizes the patient's story, leveraging hermeneutic reflection to cultivate empathy and serve as a critical bridge between technical biomedical science and the personal art of medicine [10, 12, 13]. Through narrative analysis, NM facilitates the acquisition of contextual (Layer 2) and personalized (Layer 3) competence, which are essential for clinical adaptability and relational care. Furthermore, NM provides a context-rich platform that allows students with nascent biomedical knowledge to engage with clinical scenarios through relatable human stories, fostering early clinical thinking [4, 14, 15]. To optimize the internalization of these competencies, Peer-Assisted Learning (PAL) functions as a situated learning environment that provides the social scaffolding necessary to translate foundational learning into observable professional skills within a low-pressure community [16, 17]. Rooted in social constructivist theory, this cooperative structure is essential for bridging the gap between canonical knowledge (Layer 1) and practical, observable skills (Layer 2), particularly in complex domains such as teamwork and the application of professional language [4, 5]. Despite the growing recognition of NM and PAL, empirical studies evaluating their combined impact on the development of core competencies in early medical students remain limited. Furthermore, existing research often focuses on justification rather than clarifying the underlying mechanisms of how these strategies facilitate competence evolution. Underpinning this study is social constructivist learning theory, which posits that knowledge and professional identity are co-constructed through social interaction and narrative sense-making within a community of learners [18, 19]. This study therefore addresses the following core research questions:Compared with traditional lecture-based learning, what is the effect of the integrated NM-PAL model on first-year medical students' self-reported competencies?Through what underlying mechanisms do first-year medical students perceive their learning experience and competence development within the integrated NM-PAL model? Compared with traditional lecture-based learning, what is the effect of the integrated NM-PAL model on first-year medical students' self-reported competencies? Compared with traditional lecture-based learning, what is the effect of the integrated NM-PAL model on first-year medical students' self-reported competencies? Through what underlying mechanisms do first-year medical students perceive their learning experience and competence development within the integrated NM-PAL model? Through what underlying mechanisms do first-year medical students perceive their learning experience and competence development within the integrated NM-PAL model? To answer these questions, an explanatory sequential mixed-methods approach was employed, integrating a quantitative KSA-based assessment framework with qualitative in-depth interviews to explore students' perceptions of the model's developmental mechanisms. Methods Study design and theoretical framing This study employed an explanatory sequential mixed-methods design at Shantou University Medical College (SUMC) over a 10-month period from October 2024 to July 2025. Underpinning this research is the social constructivist learning theory and Ten Cate's multilayered construct of competence, which provide the theoretical lens for examining the developmental transition from foundational knowledge (Layer 1) to contextualized clinical application (Layer 2). As illustrated in Fig. 1, the study proceeded through four chronological phases: pre-intervention instrument development, intervention implementation, quantitative outcome assessment, and qualitative explanatory exploration to clarify the underlying pedagogical mechanisms [20]. This qualitative phase aimed to explore the quantitative findings in greater depth, including both significant and non-significant outcomes across competency dimensions.Fig. 1Flowchart of the four-phase explanatory sequential mixed-methods design Flowchart of the four-phase explanatory sequential mixed-methods design Ethical approval was reviewed by the Ethics Committee of the First Affiliated Hospital of SUMC, which granted a waiver of full ethical review on 4 July 2024, as the research consisted of standard educational practices with minimal risk. Written informed consent was obtained from all participants prior to the intervention, with assurance of voluntary participation and data anonymisation. All identifiable information was replaced with unique study codes prior to analysis, and signed consent forms were stored separately from the research data. Research team and reflexivity The research team (P.X.L., X.L.Y., X.K.Z., S.J.H., Y.X.C., T.T.C., and X.Z.D.) from SUMC and other institutions collaborated under the supervision of H.Y.L. (who holds a Ph.D. in mixed methods), Y.W.L., and Y.K.L. Within this framework, W.Q.L. and S.Q.L. served as the lecturers for the experimental group under the guidance of H.Y.L. (as illustrated in Fig. 2), where they oversaw the implementation of the NM-PAL curriculum and provided faculty supervision for the peer mentors. The other research team members were involved primarily as researchers and did not hold formal teaching, grading, or supervisory roles for the participating students. To ensure methodological integrity, all team members received over six months of targeted training in mixed-methods techniques and in-depth interview strategies. Throughout the study, the team engaged in reflexive practice, acknowledging how their backgrounds in medical education and clinical research might influence data collection and the thematic interpretation of learner experiences.Fig. 2Flowchart comparing the LBL and NM-PAL teaching models Flowchart comparing the LBL and NM-PAL teaching models NM-PAL intervention model and participants The participants were first-year undergraduate students (Class of 2024) from Clinical Medicine, Stomatology, Optometry, and Psychiatry majors at SUMC. The ethics application reflected the originally planned sample size, while the final analysed sample was based on the power calculation described below. Sample size was determined using G*Power software (version 3.1). Assuming a moderate effect size (Cohen's d = 0.5), α = 0.05, and statistical power of 0.80, the required sample size was calculated as 64 participants per group, yielding a total of 128 participants. Of the approximately 568 first-year students across the participating majors, 128 were randomly selected. All selected students provided informed consent and completed the study without attrition. Participants were randomly allocated to the experimental and control groups at a 1:1 ratio using a computer-generated random number sequence, with allocation performed by an independent research assistant not involved in teaching or assessment. Randomization was conducted at the individual student level upon enrolment. The experimental group (n = 64) utilized the NM-PAL model in small classes of 6–8, while the control group (n = 64) followed traditional lecture-based learning (LBL) in larger classes of 30–40 (Fig. 2). The NM-PAL model integrated senior medical students as peer mentors within a reciprocal learning community. Eight senior medical students were recruited as peer mentors based on their academic performance (ranked in the top 30% of their class), prior teaching experience, and communication skills. They completed a 10-h training program covering NM principles, PAL facilitation techniques, case discussion leadership, and feedback provision. A standardized teaching protocol, including session objectives, suggested discussion prompts, and timing guidelines, was provided to all mentors. Throughout the course, a faculty supervisor conducted weekly debriefing sessions with the mentors, observed one session per mentor, and reviewed their prepared materials to ensure pedagogical consistency. The curriculum consisted of five synergistic modules: in-depth NM-case reading, case discussions, medical terminology acquisition, doctor-patient role-playing, and academic presentations (Fig. 3), totalling 30 instruction hours for both groups. All cases used in the NM-PAL curriculum were sourced from publicly available published resources and have been anonymised for educational use. For sensitive case discussions, peer mentors and course instructors were available for debriefing, and students were informed of the university's counselling services at the outset of the course. Both groups completed the same total formal contact hours and were expected to achieve the same curriculum objectives.Fig. 3Demonstrate a Case Used in the NM-PAL Model Demonstrate a Case Used in the NM-PAL Model Quantitative component: KSA scale and survey This Delphi-informed instrument was designed to capture early-stage competency development across the three KSA dimensions. A 19-item KSA scale was developed to operationalize core competence. Preliminary indicators were extracted via a scoping review of international and Chinese databases (PubMed, Embase, CNKI, Wanfang). Subsequently, two rounds of Delphi consultations were performed with 18 experts to reach consensus. The experts achieved a robust authority coefficient (CR) of 0.74, with Kendall's W increasing to 0.21 (p < 0.01) and the coefficient of variation (CV) for all indicators falling below 0.16 [21]. While the final scale is presented in Table 1 and Supplementary Materials (KSA Scale and Survey), the detailed per-round expert ratings and consensus criteria are provided in the Supplementary Data (Detailed Delphi Method) to focus the manuscript on theoretical analysis. Following the intervention, the questionnaire was distributed to all participants to assess their self-reported competency development.Table 1Mean expert ratings on the importance and feasibility of 19 indicators (5-point Likert scale)DimensionsIndicatorsImportanceFeasibilityK (Knowledge)K1-Understanding healthcare system4.44.47K2-Utilizing medical resources4.84.73K3-Mastery of medical English terminology4.44.53K4-Acquiring and applying medical knowledge (training clinical reasoning)4.874.47S (Skills)S1-Application of critical thinking4.474.27S2-Ability to use medical English in clinical scenarios4.474.33S3-Capacity for systematic collection of patient disease information4.734.67S4-Proficiency in selecting laboratory tests4.64.67S5-Retrieving, organizing, and analysing medical information4.674.67S6-Applying information technology4.734.73S7-Structured oral presentation skills for clinical cases4.734.8S8-Narrative competence in Narrative Medicine4.84.67S9-Leadership4.44.33A (Attitudes)A1-Empathy-driven awareness of doctor-patient communication4.674.6A2-Commitment to patient-centred ethical decision-making4.64.53A3-Adherence to teamwork principles4.84.73A4-Commitment to lifelong learning4.874.8A5-Maintenance of integrity and self-discipline4.874.47A6-Maintenance of psychological well-being4.674.53 Table 1Mean expert ratings on the importance and feasibility of 19 indicators (5-point Likert scale)DimensionsIndicatorsImportanceFeasibilityK (Knowledge)K1-Understanding healthcare system4.44.47K2-Utilizing medical resources4.84.73K3-Mastery of medical English terminology4.44.53K4-Acquiring and applying medical knowledge (training clinical reasoning)4.874.47S (Skills)S1-Application of critical thinking4.474.27S2-Ability to use medical English in clinical scenarios4.474.33S3-Capacity for systematic collection of patient disease information4.734.67S4-Proficiency in selecting laboratory tests4.64.67S5-Retrieving, organizing, and analysing medical information4.674.67S6-Applying information technology4.734.73S7-Structured oral presentation skills for clinical cases4.734.8S8-Narrative competence in Narrative Medicine4.84.67S9-Leadership4.44.33A (Attitudes)A1-Empathy-driven awareness of doctor-patient communication4.674.6A2-Commitment to patient-centred ethical decision-making4.64.53A3-Adherence to teamwork principles4.84.73A4-Commitment to lifelong learning4.874.8A5-Maintenance of integrity and self-discipline4.874.47A6-Maintenance of psychological well-being4.674.53 Mean expert ratings on the importance and feasibility of 19 indicators (5-point Likert scale) We assessed the reliability and preliminary validity of the scale using IBM SPSS version 26.0. Subscale reliabilities and exploratory factor analysis (maximum likelihood extraction with Promax rotation, with factor retention determined by parallel analysis) were also conducted to assess the psychometric properties of the scale. Descriptive statistics (mean, median, standard deviation, frequency, and percentage) were applied to the questionnaire data. Data normality was checked using the Shapiro–Wilk and Kolmogorov–Smirnov tests, and variance homogeneity was assessed with Levene's test. The primary quantitative outcome was defined as the total KSA score, with subscale scores examined as secondary outcomes. Independent T-tests were used for normally distributed data, while the Mann–Whitney U test was applied for non-normally distributed data. A p-value < 0.05 was considered statistically significant. For normally distributed data, 95% confidence intervals for mean differences were calculated using the t-distribution. For non-normally distributed data, confidence intervals for the Hodges-Lehmann location-shift estimate (the pseudo median of pairwise differences) were calculated. For categorical baseline variables, chi-square tests were used for gender comparisons, and Fisher's exact test was applied for major distribution due to small expected cell counts in some categories. Item-level comparisons were exploratory and were not pre-specified as primary outcomes. Therefore, no adjustment for multiple comparisons was applied; interpretation focused on total and subscale-level findings. Qualitative component In-depth interviews (60–90 min) were conducted with 14 purposively sampled students (6 males, 8 females) from the experimental group, using an open-ended interview guide (Supplementary Materials—STable 1) to minimize power dynamics. Participants were purposively selected to include a range of questionnaire performance levels: based on total scores, four students from the high-scoring group (top 25%), six from the medium-scoring group (middle 50%), and four from the low-scoring group (bottom 25%) were invited for interviews. Selection also aimed to ensure balanced representation across gender and major (Clinical Medicine, Stomatology, Optometry, and Psychiatry). This strategy was designed to capture a broad spectrum of experiences and perceptions, thereby enhancing the credibility and transferability of the qualitative findings [22]. This phase was executed by a coordinated team: P.X.L. and X.L.Y. led the sessions as primary facilitators, while X.Z.D. managed audio recordings and Z.X.K., Y.X.C., T.T.C., and S.J.H. served as co-facilitators responsible for detailed field notes. Under the supervision of H.Y.L., the team maintained a neutral, non-participatory stance to ensure the authenticity of the learners' self-reported experiences and perceptions. To minimise social desirability bias, the interviewer had no prior teaching, mentoring, grading, or supervisory relationship with any participant. Students were informed that their responses would remain confidential and would not affect their course performance or academic standing. Data analysis followed a hybrid thematic analysis approach using NVivo 15 integrating deductive themes derived from the theoretical framework with inductive themes emerging from the data [23, 24]. Audio recordings were transcribed verbatim within 24 h by the co-facilitator team, with P.X.L. performing accuracy cross-checks against original recordings. P.X.L. and S.J.H. independently coded the transcripts, compared codes, and resolved discrepancies through consensus; H.Y.L. adjudicated any unresolved disagreements. Methodological rigor was maintained through ongoing reflexivity, with all team members documenting their assumptions throughout the analysis. Thematic saturation was determined after the 12th interview, when no new sub-themes emerged; two additional interviews were conducted to confirm saturation [24]. Interviews were conducted in Mandarin Chinese and transcribed verbatim. Representative quotations were translated into English by P.X.L. and cross-verified for semantic accuracy by X.L.Y., an English-language specialist from the Applied Foreign Languages College, Shantou Polytechnic. Any discrepancies in translation were resolved through consensus to maintain the original meaning and context. Rigor and trustworthiness In accordance with the explanatory sequential design, quantitative and qualitative data were integrated during the interpretation phase through a Joint Display. This tool allowed the team to map significant and non-significant quantitative differences against qualitative evidence to "tell the story" of how the NM-PAL model may be associated with students' perceived competence development. To ensure rigor and trustworthiness, the Delphi expert consultation method was used to refine scale indicators. In the qualitative section, the interviewees represented various medical specialties to capture broader perspectives. This study integrated both quantitative and qualitative data, allowing for mutual complementarity. Two researchers independently coded the interview data, with consensus reached through discussion [22]. Results Quantitative phase: comparison of competency development A total of 128 students participated in the survey, including 52 males (40.6%) and 76 females (59.4%), with the experimental and control groups each comprising 64 students (Table 2). The evaluation scale demonstrated acceptable internal consistency (α = 0.878) and preliminary construct validity evidence (KMO = 0.843, p < 0.001), although the factor structure requires further validation. Subscale reliabilities were acceptable for Skills (α = 0.810) and Attitudes (α = 0.703), and modest for Knowledge (α = 0.625, given its four items). All corrected item-total correlations exceeded 0.30. Intercorrelations among the three subscales ranged from 0.452 to 0.750, indicating that they measure related but distinct aspects of competence. The sample size of 128 for 19 items (ratio ≈ 6.7:1) meets the minimum recommended ratio for exploratory factor analysis, and falls within the acceptable range suggested for such analyses [25, 26]. Exploratory factor analysis was conducted using maximum likelihood extraction with Promax rotation; factor retention was determined by parallel analysis with 100 random permutations (95th percentile criterion). The analysis supported a three-factor structure, consistent with the hypothesised Knowledge-Skills-Attitudes (KSA) framework. Detailed EFA results, including the pattern matrix, eigenvalues, and variance explained by each factor, are provided in Supplementary Data (Detailed EFA Results). Further validation in larger samples is warranted. Nonetheless, content validity was rigorously established through the Delphi process.Table 2Baseline demographic characteristics of participants by groupCharacteristicExperimental Group (n = 64)Control Group (n = 64)Total (N = 128)Gender, n (%) Male22 (34.4%)30 (46.9%)52 (40.6%) Female42 (65.6%)34 (53.1%)76 (59.4%)Major, n (%) Clinical Medicine50 (78.1%)49 (76.6%)99 (77.3%) Stomatology8 (12.5%)7 (10.9%)15 (11.7%) Optometry5 (7.8%)6 (9.4%)11 (8.6%) Psychiatry1 (1.6%)2 (3.1%)3 (2.3%)No significant group differences were observed for gender (χ2 = 2.08, p = 0.15) or major distribution (Fisher's exact test, p = 0.72). No participants had prior NM exposure. The sample's major distribution was representative of the cohort of approximately 568 students across the four majors Table 2Baseline demographic characteristics of participants by groupCharacteristicExperimental Group (n = 64)Control Group (n = 64)Total (N = 128)Gender, n (%) Male22 (34.4%)30 (46.9%)52 (40.6%) Female42 (65.6%)34 (53.1%)76 (59.4%)Major, n (%) Clinical Medicine50 (78.1%)49 (76.6%)99 (77.3%) Stomatology8 (12.5%)7 (10.9%)15 (11.7%) Optometry5 (7.8%)6 (9.4%)11 (8.6%) Psychiatry1 (1.6%)2 (3.1%)3 (2.3%)No significant group differences were observed for gender (χ2 = 2.08, p = 0.15) or major distribution (Fisher's exact test, p = 0.72). No participants had prior NM exposure. The sample's major distribution was representative of the cohort of approximately 568 students across the four majors Baseline demographic characteristics of participants by group No significant group differences were observed for gender (χ2 = 2.08, p = 0.15) or major distribution (Fisher's exact test, p = 0.72). No participants had prior NM exposure. The sample's major distribution was representative of the cohort of approximately 568 students across the four majors Regarding the overall self-reported core competence scores, the experimental group reported a significantly higher mean score of 77.42 ± 7.45 compared to the control group's 68.00 ± 7.31 (mean difference = 9.42, 95% CI: 6.84–12.00, p < 0.001, Cohen's d = 1.28), suggesting that the integrated NM-PAL model was associated with significantly higher self-reported core competence scores. Notably, all item medians in the experimental group were three or higher, as shown in Fig. 4.Fig. 4Response distribution (%) across all items for experimental and control groups Response distribution (%) across all items for experimental and control groups Significantly higher self-reported scores were observed across all three dimensions of the KSA framework (Knowledge, Skills, and Attitude), as detailed in Table 3. In the Knowledge dimension, the experimental group scored significantly higher (15.00 vs. 13.00, p < 0.001, r = 0.48). Consistent with this overall pattern, exploratory item-level analysis revealed complementary patterns, such as in utilizing medical resources (K2) and applying medical knowledge (K4). The Skills dimension also showed a marked advantage for the experimental group (35.41 ± 4.42 vs. 30.58 ± 4.00, p < 0.001, d = 1.15), further supported by notable exploratory findings in retrieving and analysing medical information (S5) and structured oral presentation (S7). In the Attitude dimension, the experimental group showed superior commitment (27.00 vs. 24.50, p < 0.001, r = 0.42), especially regarding adherence to teamwork principles (A3). However, critical thinking (S1) and empathy-driven communication awareness (A1) did not reach statistical significance (p > 0.05).Table 3Comparison of knowledge, skills and attitudes scores between experimental and control groupsDimensionsIndicatorsGroup95% CIt/Z-valuep-valueCohen's d/rExperimental GroupControl Group\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\overline{\chi }$$\end{document}/M\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\overline{\chi }$$\end{document}/MK (Knowledge)K1 a3.00(2.50–3.50)3.00(2.50–3.50)0.00–1.00a−2.990a0.0030.264aK2 a4.00(3.50–4.50)3.00(2.50–3.50)0.00–1.00a−4.153a < 0.0010.367aK3 a4.00(3.00–5.00)3.50(3.00–4.00)0.00–1.00a−4.081a < 0.0010.361aK4 a4.00(3.50–4.50)4.00(3.50–4.50)0.00–0.99a−3.654a < 0.0010.323aTotal Knowledge Score a15.00(13.50–16.50)13.00(12.00–14.00)2.00–3.00a−5.477a < 0.0010.484aS (Skills)S1 a4.00(3.50–4.50)4.00(3.50–4.50)—−1.242a0.210.110aS2 a4.00(3.50–4.50)3.00(2.50–3.50)0.00–1.00a−3.616a < 0.0010.320aS3 a4.00(3.50–4.50)4.00(3.50–4.50)0.00–1.00a−3.414a0.0010.302aS4 a3.00(2.50–3.50)3.00(2.00–4.00)0.00–1.00a−2.803a0.0050.248aS5 a4.00(3.00–5.00)3.00(2.50–3.50)0.00–1.00a−4.755a < 0.0010.420aS6 a4.00(3.50–4.50)3.50(3.00–4.00)0.00–1.00a−4.809a < 0.0010.425aS7 a4.00(3.00–5.00)3.00(2.50–3.50)1.00–1.00a−4.663a < 0.0010.412aS8 a5.00(4.00–5.00)4.00(3.50–4.50)0.00–1.00a−5.115a < 0.0010.452aS9 a4.00(3.50–4.50)3.00(2.50–3.50)0.00–1.00a−2.369a0.020.209aTotal Skills Score b35.41 ± 4.42130.58 ± 4.0033.35–6.30 b6.476b < 0.0011.145bA (Attitudes)A1 a4.00(3.00–5.00)4.00(3.50–4.50)—−1.631a0.100.144aA2 a5.00(4.00–5.00)4.00(3.50–4.50)0.00–1.00a−3.280a0.0010.290aA3 a5.00(4.00–5.00)4.00(3.50–4.50)0.00–1.00a−2.662a0.0080.235aA4 a4.50(4.00–5.00)4.00(3.50–4.50)0.00–1.00a−3.807 a < 0.0010.337aA5 a5.00(4.00–5.00)4.00(3.50–4.50)0.00–1.00a−3.709a < 0.0010.328aA6 a5.00(5.00–5.00)4.00(3.50–4.50)0.00–1.00a−3.696a < 0.0010.327aTotal Attitudes Score a27.00(25.00–29.00)24.50(22.50–26.50)1.00–3.00a−4.765a < 0.0010.421aTotal Score b77.42 ± 7.44568.00 ± 7.3076.84–12.00 b7.226b < 0.0011.277bBoth experimental and control groups had 64 participantsa Data presented as median (Q1-Q3), analysed using Mann–Whitney U test (Z-value), with effect size reported as r; 95% CIs for these comparisons are for the Hodges-Lehmann location-shift estimateb Data presented as mean ± SD, analysed using t-test (t-value), with effect size reported as Cohen's d; 95% CIs for these comparisons are for mean differences. Item-level comparisons were exploratory in nature; therefore, no adjustment for multiple comparisons was applied Table 3Comparison of knowledge, skills and attitudes scores between experimental and control groupsDimensionsIndicatorsGroup95% CIt/Z-valuep-valueCohen's d/rExperimental GroupControl Group\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\overline{\chi }$$\end{document}/M\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\overline{\chi }$$\end{document}/MK (Knowledge)K1 a3.00(2.50–3.50)3.00(2.50–3.50)0.00–1.00a−2.990a0.0030.264aK2 a4.00(3.50–4.50)3.00(2.50–3.50)0.00–1.00a−4.153a < 0.0010.367aK3 a4.00(3.00–5.00)3.50(3.00–4.00)0.00–1.00a−4.081a < 0.0010.361aK4 a4.00(3.50–4.50)4.00(3.50–4.50)0.00–0.99a−3.654a < 0.0010.323aTotal Knowledge Score a15.00(13.50–16.50)13.00(12.00–14.00)2.00–3.00a−5.477a < 0.0010.484aS (Skills)S1 a4.00(3.50–4.50)4.00(3.50–4.50)—−1.242a0.210.110aS2 a4.00(3.50–4.50)3.00(2.50–3.50)0.00–1.00a−3.616a < 0.0010.320aS3 a4.00(3.50–4.50)4.00(3.50–4.50)0.00–1.00a−3.414a0.0010.302aS4 a3.00(2.50–3.50)3.00(2.00–4.00)0.00–1.00a−2.803a0.0050.248aS5 a4.00(3.00–5.00)3.00(2.50–3.50)0.00–1.00a−4.755a < 0.0010.420aS6 a4.00(3.50–4.50)3.50(3.00–4.00)0.00–1.00a−4.809a < 0.0010.425aS7 a4.00(3.00–5.00)3.00(2.50–3.50)1.00–1.00a−4.663a < 0.0010.412aS8 a5.00(4.00–5.00)4.00(3.50–4.50)0.00–1.00a−5.115a < 0.0010.452aS9 a4.00(3.50–4.50)3.00(2.50–3.50)0.00–1.00a−2.369a0.020.209aTotal Skills Score b35.41 ± 4.42130.58 ± 4.0033.35–6.30 b6.476b < 0.0011.145bA (Attitudes)A1 a4.00(3.00–5.00)4.00(3.50–4.50)—−1.631a0.100.144aA2 a5.00(4.00–5.00)4.00(3.50–4.50)0.00–1.00a−3.280a0.0010.290aA3 a5.00(4.00–5.00)4.00(3.50–4.50)0.00–1.00a−2.662a0.0080.235aA4 a4.50(4.00–5.00)4.00(3.50–4.50)0.00–1.00a−3.807 a < 0.0010.337aA5 a5.00(4.00–5.00)4.00(3.50–4.50)0.00–1.00a−3.709a < 0.0010.328aA6 a5.00(5.00–5.00)4.00(3.50–4.50)0.00–1.00a−3.696a < 0.0010.327aTotal Attitudes Score a27.00(25.00–29.00)24.50(22.50–26.50)1.00–3.00a−4.765a < 0.0010.421aTotal Score b77.42 ± 7.44568.00 ± 7.3076.84–12.00 b7.226b < 0.0011.277bBoth experimental and control groups had 64 participantsa Data presented as median (Q1-Q3), analysed using Mann–Whitney U test (Z-value), with effect size reported as r; 95% CIs for these comparisons are for the Hodges-Lehmann location-shift estimateb Data presented as mean ± SD, analysed using t-test (t-value), with effect size reported as Cohen's d; 95% CIs for these comparisons are for mean differences. Item-level comparisons were exploratory in nature; therefore, no adjustment for multiple comparisons was applied Comparison of knowledge, skills and attitudes scores between experimental and control groups Both experimental and control groups had 64 participants a Data presented as median (Q1-Q3), analysed using Mann–Whitney U test (Z-value), with effect size reported as r; 95% CIs for these comparisons are for the Hodges-Lehmann location-shift estimate b Data presented as mean ± SD, analysed using t-test (t-value), with effect size reported as Cohen's d; 95% CIs for these comparisons are for mean differences. Item-level comparisons were exploratory in nature; therefore, no adjustment for multiple comparisons was applied Qualitative phase: clarifying the mechanisms of competence development Fourteen participants (P1–P14) consented to and engaged in in-depth interviews, comprising 6 males and 8 females (Table 4). The qualitative findings clarified participants' perceptions of the underlying pedagogical mechanisms within the integrated model. Participants described the narrative components as helpful for meaning-making and the peer-assisted elements as supportive of skill practice within the overall course structure. The complete hierarchical structure of themes, sub-themes, and additional supporting verbatim quotes is provided in the Supplementary Materials (STable 2) to ensure analytical transparency.Table 4Demographic and Performance Characteristics of Qualitative Interview Participants (N = 14)GenderMajorKSA Score Category2 Male, 2 FemaleClinical Medicine (3), Stomatology (1)High (Top 25%)2 Male, 4 FemaleClinical Medicine (3), Optometry (2), Psychiatry (1)Medium (Middle 50%)2 Male, 2 FemaleClinical Medicine (2), Stomatology (1), Psychiatry (1)Low (Bottom 25%) Table 4Demographic and Performance Characteristics of Qualitative Interview Participants (N = 14)GenderMajorKSA Score Category2 Male, 2 FemaleClinical Medicine (3), Stomatology (1)High (Top 25%)2 Male, 4 FemaleClinical Medicine (3), Optometry (2), Psychiatry (1)Medium (Middle 50%)2 Male, 2 FemaleClinical Medicine (2), Stomatology (1), Psychiatry (1)Low (Bottom 25%) Demographic and Performance Characteristics of Qualitative Interview Participants (N = 14) Perceived role of narrative features within the integrated NM-PAL model Students perceived the narrative case materials within the integrated NM-PAL model as helpful for fostering learning motivation and clinical thinking, supporting their transition from rote memorization toward contextualized (Layer 2) understanding. Participants reported that case-based storytelling made knowledge "easier to accept" (P9) and helped them transcend viewing patients as "cold cases" to seeing them as "real, living people" (P10). This shift facilitated an empathy awakening, as students spontaneously considered the "family background, profession, and financial ability" of the characters (P3). From an analytical perspective, by embedding medical terminology within human stories, the model reduced the cognitive load associated with technical language, thereby strengthening the synthesis of symptoms, tests, and treatments into a coherent clinical reasoning process. These findings align with NM's theoretical emphasis on representation and affiliation, providing a context-rich platform for students to begin constructing their professional identities. Perceived role of peer-assisted features within the integrated NM-PAL model Students described the PAL environment within the integrated model as creating a safety-participation-practice dynamic that they felt supported the development of their professional skills. Students described the peer-led environment as "very open" (P14) and "less stressful" (P7) than traditional teacher-led classes, noting that the "group format forced me to speak" (P9). This relaxed psychological space allowed participants to ask questions and receive "immediate corrections from peers" (P5) without fear of making mistakes (P1). This perceived social scaffolding appeared to correspond with the higher self-reported scores observed in oral presentation (S7) and teamwork attitudes (A3). However, the qualitative data also identified a boundary condition of this model: some participants "trusted the professionalism of the teacher more" (P3, P5) or doubted the accuracy of senior students' knowledge (P5, P14). These insights suggest that while, within the integrated NM-PAL model, peer-assisted elements appear to bridge the gap between canonical knowledge and practical performance through the Zone of Proximal Development, its efficacy is sensitive to the perceived expertise of the peer mentors. Joint display of quantitative and qualitative integration In accordance with the explanatory sequential design, the following Joint Display (Table 5) integrates significant and non-significant quantitative results with representative qualitative evidence. This integrative analysis clarifies participants' perceptions of the pedagogical mechanisms, exploring how various features within the integrated NM-PAL model may support their perceived competence, while presenting multiple plausible explanations for non-significant outcomes.Table 5Joint display integrating quantitative results, qualitative evidence, and theoretical mechanismsQuantitative Result (Significance)Representative Qualitative EvidencePedagogical Mechanism & Theory MappingIntegration StrengthS5 Information Processing & S7 Oral Presentation ↑ Significantly higher scores (p < 0.001)"Preparing the NM reports required systematically literature retrieval… presenting cases helped me clearly organize the clinical logic." (P1, P4, P5)Task-Driven Social Scaffolding: Within the integrated NM-PAL package, narrative cases provided a humanistic context while the peer-led environment offered a low-pressure platform for feedback, supporting perceived gains in information organization.StrongA3 Teamwork Principles ↑ Significantly higher scores (p = 0.008)"The group format provided constant opportunities to practice collaboration… we complement each other as peers." (P7, P10)Collaborative Engagement: Shared meaning-making within the integrated model encouraged students to perceive themselves as active collaborators rather than passive learners within a community of practice.Moderate-StrongS1 Critical Thinking ↔ No significant difference (p = 0.21)"I began to think more carefully about the reasoning behind medical information… reflecting on why a treatment might change." (P14, P8, P1)Perceived Analytical Engagement: While qualitative reports suggest an increased awareness of reasoning, the lack of quantitative divergence may reflect equally plausible factors: development timing, measurement sensitivity of single-item indicators, or no meaningful intervention impact on this domain.ModerateA1 Empathy Awareness ↔ No significant difference (p = 0.10)"I no longer view patients as cold cases; I think from their perspective… considering their financial and family background." (P10, P1, P3)Perceived Perspective-Taking: Participants perceived a shift toward considering patient storied; however, the statistical non-significance suggests these attitudes may remain foundational, potentially due to ack of authentic clinical stressors, ceiling effects, or inadequate measurement tools.Moderate Table 5Joint display integrating quantitative results, qualitative evidence, and theoretical mechanismsQuantitative Result (Significance)Representative Qualitative EvidencePedagogical Mechanism & Theory MappingIntegration StrengthS5 Information Processing & S7 Oral Presentation ↑ Significantly higher scores (p < 0.001)"Preparing the NM reports required systematically literature retrieval… presenting cases helped me clearly organize the clinical logic." (P1, P4, P5)Task-Driven Social Scaffolding: Within the integrated NM-PAL package, narrative cases provided a humanistic context while the peer-led environment offered a low-pressure platform for feedback, supporting perceived gains in information organization.StrongA3 Teamwork Principles ↑ Significantly higher scores (p = 0.008)"The group format provided constant opportunities to practice collaboration… we complement each other as peers." (P7, P10)Collaborative Engagement: Shared meaning-making within the integrated model encouraged students to perceive themselves as active collaborators rather than passive learners within a community of practice.Moderate-StrongS1 Critical Thinking ↔ No significant difference (p = 0.21)"I began to think more carefully about the reasoning behind medical information… reflecting on why a treatment might change." (P14, P8, P1)Perceived Analytical Engagement: While qualitative reports suggest an increased awareness of reasoning, the lack of quantitative divergence may reflect equally plausible factors: development timing, measurement sensitivity of single-item indicators, or no meaningful intervention impact on this domain.ModerateA1 Empathy Awareness ↔ No significant difference (p = 0.10)"I no longer view patients as cold cases; I think from their perspective… considering their financial and family background." (P10, P1, P3)Perceived Perspective-Taking: Participants perceived a shift toward considering patient storied; however, the statistical non-significance suggests these attitudes may remain foundational, potentially due to ack of authentic clinical stressors, ceiling effects, or inadequate measurement tools.Moderate Joint display integrating quantitative results, qualitative evidence, and theoretical mechanisms Discussion Mechanisms of competence development through narrative context and social scaffolding The findings of this study suggest that the integrated NM-PAL model may serve as a useful pedagogical framework to support first-year medical students' perceived transition from foundational knowledge to contextualized application. These conclusions are supported by both the quantitative self-reported competency scores and the qualitative interview data, which together provide a more complete picture of students' learning processes. Consistent with social constructivist theory, students perceived the narrative cases as supporting shared meaning-making and the peer interactions as providing the vital social scaffolding necessary for learning within the zone of proximal development [18, 19]. Our analysis suggests that the observed higher scores in overall competencies, particularly in supporting patient-centred clinical thinking, align with the transition from canonical knowledge (Layer 1) to contextualized application (Layer 2) in Ten Cate's multilayered construct of competence [4, 9]. The qualitative evidence suggests that students perceived the narrative components within the integrated model as supporting this development by embedding biomedical facts within human stories, which helps students move beyond viewing patients as "cold cases" to perceiving them as "real, living people" [4, 10, 13]. This process of narrative sense-making is essential for building empathy and clinical reasoning, as supported by previous research suggesting that narrative learning helps trainees grasp the complexity of diseases through multiple social and psychological dimensions [27, 28]. Furthermore, the potential benefits of the NM-PAL model may stem from the egalitarian and low-pressure environment it creates. By interacting with peer mentors who are closer in age and experience, first-year students experienced reduced anxiety and increased motivation to engage in complex role-playing and case discussions [29]. This safety-participation-practice mechanism allowed for immediate peer feedback and the co-construction of professional language, which were associated with stronger self-reported communication and collaboration skills compared to traditional lecture-based models [30, 31]. The magnitude of these effects was most pronounced for competencies closely aligned with the model's core activities. Information retrieval and organization (S5, r = 0.420), narrative competence (S8, r = 0.452), and structured oral presentation (S7, r = 0.412) all showed moderate effect sizes, suggesting that these score differences are not merely statistically detectable but educationally substantive in terms of students' self-reported development, as students were explicitly practicing these skills through case presentations, peer feedback, and narrative writing tasks. The qualitative data further explain these findings. Students consistently reported that preparing case presentations required systematic literature retrieval and peer consultation, which appeared to strengthen their information processing skills. Similarly, the structured oral presentation format, practiced repeatedly in small-group settings with immediate peer feedback, was frequently cited as a key driver of their growing confidence and clarity in clinical communication. These accounts suggest that the observed higher self-reported scores reflect tangible skills developed through sustained, scaffolded practice, consistent with students' perceptions of the social scaffolding they experienced within the integrated NM-PAL model. Potential explanations for observed patterns in higher-order competencies In the quantitative phase, critical thinking (S1) and empathy-driven communication awareness (A1) did not reach statistical significance between the two groups. These non-significance findings should be interpreted through several equally plausible lenses. One possibility is that higher-order competencies requiring more sustained practice and longer pedagogical reinforcement than a ten-month window can provide [32]. While quantitative divergence was absent, qualitative reports from the experimental group indicated a perceived awareness of the reasoning behind clinical decisions, similar to patterns observed in other research [30]. However, other factors must be considered, including the limitations of using single-item indicators to capture such complex constructs, potential ceiling effects, or insufficient statistical power to detect subtle differences. It is also possible that the integrated NM-PAL model had no meaningful intervention-related impact on these specific higher-order dimensions at this early stage of medical education. These observations may support initial cognitive engagement, the measurable manifestation of these competencies on a self-reported scale may be influenced by both developmental timing and measurement sensitivity. Similarly, while students reported a shift toward perspective-taking, this remained a foundational cognitive attitude rather than a fully matured clinical skill [4, 33]. Empathy and communication awareness often face a significant stress test during the transition to actual patient-care activities, where they may even experience a decline due to time pressure and emotional detachment [34]. Since first-year students lack authentic clinical encounters, their humanistic awareness is still maturing and has not yet been tested against real-world clinical stressors [4]. Therefore, while the integrated NM-PAL model appears to support the cognitive components of empathy, such as perspective-taking, during the pre-clinical stage, these traits are still in a developmental phase that requires long-term reinforcement [34, 35]. We interpret this pattern as suggestive of early-stage development, where statistical non-significance does not preclude educational potential. This discrepancy may reflect the developmental nature of these competencies. For empathy, interviews captured cognitive empathy (students' awareness of patients' circumstances), whereas the quantitative item may have tapped behavioural empathy, which requires clinical exposure to mature. Similarly, for critical thinking (S1), students showed nascent reflective awareness in interviews, yet the quantitative measure did not detect significant change. These findings could be interpreted in several ways. One possibility is that these competencies are in an early developmental stage that requires more time to become measurable. Alternatively, the single-item measures may have been insufficient to capture the full scope of these constructs, ceiling effects may have limited the observed variability, or the study may have had limited statistical power to detect small effects. Boundary conditions and professional identity formation Despite the positive overall outcomes, the qualitative analysis identified specific boundary conditions that may influence the perceived benefits associated with the integrated NM-PAL model. Student concerns regarding the accuracy of senior mentors' knowledge and the consistency of teaching quality suggest that the benefits of peer learning are sensitive to the perceived expertise of the mentors [36]. This highlights the necessity for institutions to standardize the recruitment and training of peer tutors to maintain pedagogical rigor. Beyond mentor-related concerns, the modest effect on leadership (S9, r = 0.209) further illustrates a boundary of the model: competencies not directly targeted by the intervention are unlikely to improve substantially within this timeframe. Nevertheless, the integrated NM-PAL model showed potential in supporting early professional identity formation. The use of narrative reflections may have helped students maintain sensitivity toward professionalism while navigating the initial uncertainties of medical education [27, 37]. By maintaining a passion for the profession through relatable human stories, students began to internalize the concept of medicine as a lifelong learning process [1]. As established in recent literature, humanistic education of this nature may help respond more compassionately to complex relationships, which may contribute to their long-term professional competence and overall capabilities [35]. Expert expectations and instrumentation challenges The nineteen assessment indicators established via the Delphi expert method reflect the high expectations that international medical education holds for early-stage students [21]. Although some higher-order competencies remained in the early stages of development, these expert-defined standards provide essential guidance for competency-based curriculum reform [6, 38]. The tension observed between expert expectations and the short-term results for critical thinking and empathy suggests that while early development should be encouraged, measurement criteria must be adjusted to account for the specific learning stage of the student and the length of the intervention [4]. The KSA scale developed for this study should be considered a preliminary instrument with emerging validity evidence. It was developed and initially validated using the same sample that was subsequently employed to evaluate the intervention effects. The exploratory factor analysis, using maximum likelihood extraction with Promax rotation and parallel analysis, supported the hypothesised three-dimensional Knowledge-Skills-Attitudes structure. However, given this sample overlap, further validation in larger, independent samples is warranted to confirm its factor structure and broader applicability. These findings offer a potentially useful framework for ongoing medical education reform by providing a practical framework that integrates humanistic education with collaborative social learning. While the KSA scale currently presents only preliminary validity evidence, its application offers important support for assessing key competencies in early-stage medical students. Future research should continue to validate this scale to further establish its psychometric properties and to enhance the quality of medical education. Limitations and future research This study has several limitations and implications for future research that should be acknowledged. First, the 10-month intervention period may have been insufficient for developing higher-order competencies such as critical thinking and empathy-driven communication, which are believed to take longer to manifest, especially in early medical education. Additionally, conducting the study at a single institution restricts the generalizability of the findings to other cultural or educational settings. Furthermore, qualitative interviews were only performed with students in the experimental group, potentially overlooking the perspectives of control-group participants. The involvement of multiple peer mentors across different small groups might also introduce clustering effects that were not considered in the analysis. A multilevel modelling approach was not pursued due to the limited number of peer mentors (n = 8), which would have yielded unstable estimates. Moreover, the primary outcome relied on students' self-reported perceptions rather than direct observations of clinical performance. To improve future studies, incorporating performance-based measures like Objective Structured Clinical Examinations (OSCEs), faculty ratings, peer assessments, or standardised patient evaluations would provide a more objective evaluation of clinical skills. Although qualitative data adds valuable context, it does not independently validate changes in competencies. Lastly, while the participants were first-year students with comparable academic backgrounds, unmeasured variables such as prior communication experience, motivation, language proficiency, or specific academic preparation may have differed between groups, potentially influencing the outcomes. Addressing these limitations in future research can enhance our understanding of the development of higher-order competencies in medical education. Conclusion This exploratory mixed-methods study suggests that the integrated NM-PAL model is associated with higher post-intervention self-reported competency scores. The non-significant patterns in higher-order competencies may be multifactorial, highlighting the need for longitudinal reinforcement and more sensitive evaluation tools. These findings offer preliminary evidence for competency-based medical education reform. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4.