Materials and methods

Study design, sample, and procedures

This study was designed as an observational, cross-sectional study conducted among physicians in Montenegro. The study population included both specialist and non-specialist physicians employed at tertiary, secondary, and primary healthcare institutions.

The required sample size was initially estimated using G*Power 3.1.2 (developed by Faul et al.; Heinrich Heine University Düsseldorf, Düsseldorf, Germany) [23] based on the prevalence objective of the study. Additionally, an a priori power analysis was conducted for the multiple regression models using a fixed model (R² deviation from zero), assuming a medium effect size (f² = 0.15), a significance level of α = 0.05, statistical power of 0.95, and 11 predictors. The analysis indicated a minimum required sample size of 178 participants. Therefore, the final sample of 235 physicians was considered sufficient for the planned regression analyses.

This study was part of a larger self-funded research project entitled “Factors associated with mental health, work performance, and well-being of healthcare professionals”. The aim of the project was to investigate the mental health of healthcare professionals, with a particular focus on alexithymia, mentalizing capacity, and resilience as key psychological factors influencing professional burnout, work performance, and physician well-being. The project was approved by the Institutional Review Board for Ethical Evaluation of the University of Belgrade, Faculty of Philosophy, Department of Psychology, Serbia (approval number: #2024-70, date of approval: 18.11.2024). Ethical approval for data collection in Montenegro was obtained from all relevant institutions prior to the study, and the study was conducted in accordance with the principles of the Declaration of Helsinki [24] and Good Clinical Practice guidelines. Participants provided informed consent after receiving detailed explanations regarding the study purpose, procedures, potential risks and benefits, voluntary participation, and confidentiality of responses.

Eligible participants were licensed physicians actively engaged in clinical practice at the primary, secondary, or tertiary healthcare level who provided written informed consent. Healthcare professionals not directly involved in patient care and non-physician staff were excluded from the study. All participants were informed of their right to withdraw from the study at any time without any consequences.

The study was conducted using a paper-and-pencil survey method and included 235 participants. A total of 270 questionnaires were distributed, of which 235 were completed fully and correctly and were included in the final analysis. Ten participants completed the questionnaire incorrectly, while the remaining individuals declined to participate. The study was conducted between November 3 and November 28, 2025.

All participants provided written informed consent prior to participation. Confidentiality and anonymity were ensured throughout the study.

Instruments

Since validated Montenegrin versions of the study instruments are not available, validated Serbian versions were used. Serbian is widely understood and routinely used in healthcare and academic settings in Montenegro, making these questionnaires linguistically appropriate for physicians in the present study.

Burnout was assessed using the Serbian version of the Maslach Burnout Inventory-Human Services Survey (MBI-HSS) [4]. The instrument consists of 22 items scored on a 7-point Likert scale ranging from 0 (never) to 6 (every day) and assesses three dimensions of burnout: emotional exhaustion (nine items, score range 0-54), depersonalization (five items, score range 0-30), and personal accomplishment (eight items, score range 0-48). Higher scores on the emotional exhaustion and depersonalization subscales indicate higher levels of burnout, whereas lower scores on the personal accomplishment subscale indicate greater burnout. Continuous scores for each dimension were used in the analyses.

Alexithymia was measured using the 20-item Toronto Alexithymia Scale (TAS-20), a widely used instrument in both clinical and non-clinical research [9,10]. The TAS-20 comprises three subscales: difficulty identifying feelings, difficulty describing feelings, and externally oriented thinking. A total score ranging from 20 to 100 is obtained by summing all items. Higher scores indicate greater difficulty in identifying, describing, and processing emotions.

Mentalization capacity was assessed using the RFQ-8, a short version of the original 46-item instrument [14]. The RFQ-8 consists of eight items rated on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree) and assesses the self-reported ability to understand one’s own and others’ mental states. The instrument comprises two four-item subscales: Certainty about Mental States (RFQ-C; 4 items) and Uncertainty about Mental States (RFQ-U; 4 items). Following the standard RFQ scoring procedure described by Fonagy et al. [14], item responses were recoded to generate transformed mean subscale scores ranging from 0 to 3. Higher RFQ-U scores indicate greater uncertainty about mental states (hypomentalizing), whereas higher RFQ-C scores indicate greater certainty about mental states [25].

Sociodemographic and work-related variables were assessed using a specifically designed section of the questionnaire. Sociodemographic data included gender, age, marital status, and number of children. Work-related variables included profession (non-specialist physician, specialist physician), years of work experience, and type of healthcare institution (primary, secondary, tertiary level). The English translation of the questionnaire is presented in Appendix A.

Data analyses

Statistical analyses were performed using IBM SPSS Statistics, version 21.0 (IBM Corp., Armonk, NY, USA). Descriptive statistics were used to summarize the data, including frequencies and percentages for categorical variables and means, standard deviations, minimum and maximum values, skewness, and kurtosis for continuous variables.

The distribution of variables was evaluated using the Kolmogorov-Smirnov and Shapiro-Wilk tests. As the data did not meet the assumptions of normality, nonparametric statistical methods were employed. Group differences were examined using the Mann-Whitney U test and the Kruskal-Wallis test, followed by Bonferroni-adjusted post hoc comparisons when applicable. Relationships between continuous variables were analyzed using Spearman’s rank-order correlation coefficient.

To examine the independent contributions of sociodemographic, work-related, and psychological variables to burnout dimensions (emotional exhaustion, depersonalization, and personal accomplishment), hierarchical linear regression analyses were conducted. In the first step, sociodemographic and work-related variables were entered as control variables. In the second step, psychological variables (alexithymia, hypomentalizing, and hypermentalizing) were added. Although means and standard deviations are presented for descriptive purposes, given the observed deviations from normality, subsequent analyses and group comparisons were conducted using nonparametric methods and are therefore reported using medians and interquartile ranges where appropriate. Despite these deviations, hierarchical linear regression was applied to assess the independent contributions of predictor variables. Regression assumptions were evaluated using residual diagnostics, including inspection of standardized residuals, normal probability plots, residual scatterplots, and multicollinearity statistics. Potential outliers were examined using standard diagnostic procedures, and no evidence of influential cases affecting model stability was identified. To further evaluate the robustness of the regression estimates in the presence of non-normality, bootstrap analyses with 5,000 resamples and bias-corrected accelerated (BCa) 95% confidence intervals were additionally performed for all regression models.

Reliability of the instruments was assessed using Cronbach’s alpha coefficient. All statistical tests were two-tailed, and a p-value < 0.05 was considered statistically significant.