The Faculty of Medicine Ramathibodi hospital has provided a medical degree programme for 58 years. The curriculum consists of pre-clinical (Year 1-3) and clinical (Year 4-6) phases. Currently, this programme admits 212 new students annually and provides longitudinal mentoring for every student. Volunteer mentors were randomly assigned to groups of five students, starting in Year 2 and continuing their mentorship with the same mentees until Year 6. All mentors receive training in essential mentoring skills, including effective communication, active listening, feedback, and goal-setting. While fulfilling their roles as mentors, participants in the programme engaged in regular meetings designed to provide support and instruction prior to each activity, as well as to facilitate the assessment and reflection upon their previous mentoring activities. In the pre-clinical years, initiatives promoting Professional Identity Formation (PIF) encompass early clinical exposure, which includes visits to patient wards and operating rooms, practicing communication skills during interactions with actual patients, and conducting physical examinations under the supervision of mentors. Psychological support and personal growth activities take place within mentor-led retreat sessions, fostering meaningful mentor-mentee interactions. These retreats cultivate an environment in which mentors engage in active listening, offer constructive feedback, and facilitate goal-setting within a supportive atmosphere. After the end of each academic year, this mentor program evaluation was conducted to assess the quality of mentorship program.

This cross-sectional study was conducted during the 2021 academic year at Faculty of Medicine Ramathibodi Hospital, Mahidol University. Participants completed online questionnaires using convenience sampling. Sample size for estimating a finite population proportion from n4Studies Plus resulted in a minimum requirement of 205 participants out of a total of 593 students.¹⁵ An additional 20% was included in the sample size to accommodate potential missing data, thus a total sample size was 257. Third-year medical students responded to two questionnaires: the Mentor Behavior Scale (MBS) and Maslach Burnout Inventory-Student Survey (MBI-SS). Fifth- and sixth-year medical students, who had experienced real workplace environments, completed these two questionnaires along with the Professional Self-Identity Questionnaire (PSIQ). The inclusion of the PSIQ in the fifth- and sixth-year aimed to assess their evolving professional self-identity due to their exposure to real workplace environments. Participants with incomplete responses to required questions were excluded from the study.

In this study, informed consent was obtained from all third-, fifth-, and sixth-year medical students prior to their participation. We employed convenience sampling, representing 51.8% (n=307) of the total students in the three classes (n=593). A total of 307 participants voluntarily completed the questionnaires and were included in the analysis. Among the participants, 47.9% (n=147) were male, and 52.1% (n=160) were female. Specifically, the sample comprised 165 third-year medical students, 43 participants from the fifth-year, and 99 from the sixth-year.

Our study was approved by Ethics Committee of the Institutional Research Committee of Ramathibodi (COA. MURA2021/558). All participants volunteered willingly, and strict measures were in place to ensure the anonymity of all participants throughout the research process. Consequently, the researchers were unable to identify respondents, minimizing the potential for selection bias.

Baseline characteristics, including gender, underlying disease, medication usage, Grade Point Average (GPA), duration of sleep, duration of exercise, extracurricular activities, and club activities, were collected. GPA was categorized into two groups for ease of interpretation: the low GPA group (GPA < 3.00) and the high GPA group (GPA ≥ 3.00). Detailed baseline characteristics of the participants are presented in Table 1.

Data for this study was collected through online questionnaires administered from January to April 2022. Participants were selected using convenience sampling, and prior to completing the questionnaires, all participants were provided with detailed information about the research objectives and assured of their voluntary participation rights. In order to mitigate non-response bias, various methods were employed to encourage participation in this study, including sending information via emails, promoting engagement in well-attended classes and post-examination sessions.

All questionnaires used in this study were authorized for use by their respective developers.

In this study, we employed the Mentor Behavior Scale (MBS) questionnaire, which comprises 15 Likert-scale items assessing four domains: Mentor Relationship Structure, Engagement, Autonomy Support, and Competence Support. The MBS exhibited robust psychometric properties, having undergone validation and reliability testing, with an RMSEA of 0.066 at Time 1 and 0.079 at Time 2, and an ordinal coefficient alpha exceeding 0.7 in all domains, except the autonomy support domain.^16-17 The English version of this questionnaire was suitable for our research. We adopted the mentor behaviour profile category from a previous study.² After reversing autonomy support scores, the optimal mentor behaviour profile consisted of high scores in all domains, and the controlling mentor behaviour profile had high scores in three domains, except for low scores in the autonomy support domain.

The Maslach Burnout Inventory survey, consisting of 15 Likert scale items, is a widely recognized tool used to assess burnout across various professions, including physicians, nurses, and medical students.  A Thai version of this questionnaire has already been translated and demonstrated acceptable interrater reliability, with a Kappa coefficient of 0.83. Confirmatory factor analysis further supported its validity, showing good fit indices.¹⁸ The identification of the burnout group was based on elevated scores in emotional exhaustion and depersonalization, alongside reduced scores in personal accomplishment, providing a comprehensive framework for assessing burnout in the studied population.

Professional self-identity questionnaire (PSIQ), English version, was employed to determine the professional development in fifth-and sixth-year medical students. This questionnaire comprises nine Likert scale items and has undergone content validity and reliability testing, yielding a Cronbach's alpha of 0.93.¹⁴ These nine items of PSIQ reflected professional self-identity formation in 9 domains. It included teamwork (PSIQ1), communication (PSIQ2), conducting assessment (PSIQ3), cultural awareness (PSIQ4), ethical awareness (PSIQ5), using records (PSIQ6), dealing with emergencies (PSIQ7), reflection (PSIQ8), and teaching (PSIQ9). Each item is scored on a scale ranging from 1 to 6. Participants were classified into two groups based on their scores: the low professional group (mean score ≤4) and the high professional group (mean score >4).

The data analysis was performed using Stata Version 17. Categorical data were reported in terms of frequency and percentage, while continuous data were presented as either mean and standard deviation or median and range, depending on the data distribution characteristics. The comparison of baseline characteristics and study factors between groups of interest (burnout versus non-burnout groups and low versus high professional groups) was carried out using the Chi-square or Fisher’s exact test for categorical variables and the T-test or Mann-Whitney U test for continuous variables. Variables with p-values less than 0.1 in the univariate analysis were considered for inclusion in subsequent steps. The forward selection method was employed to select the appropriate model, and multiple logistic regression analysis was used to determine the associations between mentor behaviour domains and both burnout and each domain of professional self-identity formation in medical students. Odds ratios with 95% confidence intervals (CI) were estimated in the multiple logistic regression analysis. All statistical analyses were two-tailed, and a p-value of less than 0.05 was considered statistically significant.

The scores were classified into three categories: high, moderate, and low, with the cut-off points for these categories referenced from a prior study.¹⁹ According to medical students' perspectives, the overall mean scores (sum of scores) were high in three out of four domains: mentor relationship structure, engagement, and competency support (Table 2).

Out of the total 307 participants, 80 individuals (26%) were identified as experiencing burnout according to the Maslach Burnout Inventory-Student Survey. Multivariate analysis revealed several factors significantly associated with medical students' burnout. Participants reporting medication usage were found to be associated with burnout (OR = 2.1, 95%CI: 1.1-4.0, p = 0.029). Medical students with a low GPA (GPA<3.00) were also significantly associated with burnout (OR = 3.3, 95%CI: 1.6-6.9, p = 0.001). Additionally, low to moderate levels of competency support from mentors were significantly associated with burnout in medical students (OR = 2.0, 95%CI: 1.1-3.5, p = 0.016).

Given potential variations in curricula and experiences among medical students in different years, a subgroup analysis was conducted, focusing on participants from the pre-clinical year (third-year medical students) and those in the clinical years (fifth- to sixth-year medical students). Factors associated with burnout in pre-clinical year medical students included a low to moderate level of competency support from mentors (OR = 2.4, 95% CI: 1.1-5.0, p = 0.023) and a low GPA (OR = 10.0, 95% CI: 3.3-30.2, p < 0.001). However, in clinical years medical students, only the low to moderate level of mentor relationship structure was significantly associated with burnout (OR = 3, 95% CI: 1.3 to 7.2, p = 0.014) (Table 3).

A total of 142 participants, consisting of fifth- and sixth-year medical students, completed the nine-item Professional Self-Identity Questionnaire (PSIQ). In the multivariate analysis, a high level of mentor relationship structure was significantly associated with high professional scores in the teamwork, communication, ethical awareness, and the use of records domains (Table 4). Additionally, a high level of autonomy support was significantly associated with a high professional score in the communication domain (OR = 4.3, 95% CI: 1.1-16.0, p = 0.031). Furthermore, a high level of competency support was significantly associated with a high professional score in the conducting assessment domain (OR = 5.8, 95% CI: 2.1-16.4, p = 0.001). However, no mentor behaviours were found to be significantly associated with the high professional group in the domains of cultural awareness, dealing with emergencies, reflection, and teaching (p > 0.05).

This study makes a substantial contribution to the exploration of the relationship between mentor behaviours, as assessed through the MBS questionnaire, and their impact on the mental well-being and professional self-identity formation of medical students. The findings offer valuable insights into specific mentor behaviours that can benefit medical students and provide guidance for mentors looking to enhance their mentoring skills. However, it is essential to acknowledge and address several limitations in this study. Firstly, the absence of a standardized evaluation of mentors' capabilities in guiding and supporting medical students may limit the generalizability of the study's findings to other mentorship programs. However, it is worth noting that our mentorship program did include training sessions, feedback from student satisfaction, and senior consultants for all participating faculty members. Secondly, the use of self-administered questionnaires may introduce recall bias and social desirability bias, and the assessment of mentor behaviours relied solely on the perspectives of medical students. Thirdly, important external factors, such as curriculum design, the learning environment, and workplace-based learning, were not considered in this study. These factors could potentially influence the formation of professional self-identity of medical students. Lastly, this study employed a cross-sectional design, which implies that the results only establish associations between mentor behaviours, medical student burnout, and professional identity formation. Causality cannot be determined based on this single-point data collection. Future research should consider conducting a longitudinal prospective cohort study that incorporates standardized evaluations of mentor abilities to further investigate these relationships.

The authors would like to express sincere gratitude to the Faculty of Medicine at Ramathibodi Hospital, Mahidol University, for generously providing the opportunity and funding to undertake this research project. Special appreciation is extended to Associate Professor Sasivimol Rattanasiri and Mr Chanatpon Aonnuam from the Department of Clinical Epidemiology and Biostatistics for their invaluable contributions to the statistical analysis. Furthermore, the authors wish to convey heartfelt thanks to all the participants in this study.

The authors declare that they have no conflict of interest.