Psychometric Properties of the Copenhagen Burnout Inventory in a Sample of Medical Students in Kazakhstan

Background The Copenhagen Burnout Inventory (CBI) has demonstrated good psychometric properties among different populations, but there is no known data on its validity among Russian-speaking medical students. The CBI-Student Survey focuses only on fatigue, but measures exhaustion in four different life domains: Personal Burnout (PB), Studies-Related Burnout (SRB), Colleague-Related Burnout (CRB), and Teacher-Related Burnout (TRB). Objective To investigate the psychometric properties of the Russian version of the Copenhagen Burnout Inventory–Student Survey (R-CBI-S). Design A cross-sectional study was carried out among 771 medical students at Astana Medical University (Nur-Sultan, Kazakhstan). Statistical analyses included test-retest reliability, internal consistency, item analysis, convergent and concurrent validity, and confirmatory factor analysis. Concurrent validity was evaluated by bivariate correlations of R-CBI-S with anxiety, depression, and satisfaction with the study. Results Test-retest reliability showed an ICC of 0.81. All item-total correlations for the total scale were positive (range 0.31–0.76). The Cronbach’s alpha coefficient was 0.94 (0.896 for PB, 0.884 for SRB, 0.874 for CRB, and 0.926 for TRB). The Barlett’s sphericity test result was significant (p < 0.001), and the KMO measure of sampling adequacy exceeded 0.947. Convergent validity analysis results: PB (AVE = 0.52, CR = 0.87), SRB (AVE = 0.50, CR = 0.87), CRB (AVE = 0.51, CR = 0.86), TRB (AVE = 0.56, CR = 0.88). The R-CBI-S achieved good levels of goodness-of-fit indices (RMSEA = 0.0611; CFI= 0.940; TLI = 0.933). Conclusion The test results indicated that the R-CBI-S scale appears to be a reliable and valid instrument. The R-CBI-S may be a useful tool in future research to identify burnout factors based on specific life domains for developing effective prevention measures among medical students.


Introduction
Burnout is a growing epidemic among medical students, which has been shown to have psychological and performance-related detriments (Bullock et al., 2017). Medical students are not only more likely to be burned out compared to the general population, but are increasingly likely to su er burnout as they advance in their medical training (Dyrbye et al., 2014;Dyrbye et al., 2006). is o en leads to signi cant psychological changes that manifest as depression, insomnia, substance abuse disorders, poor physical health, psychosomatic conditions, relational problems, social withdrawal, and professional dysfunction (Aguiar et al., 2009;Almeida et al., 2016). Burnout can also a ect medical students' will to continue to espouse professional qualities, such as honesty, integrity, altruism, and self-regulation (Dyrbye et al., 2010). Based on these ndings, it is clear that burnout is a serious problem in the training and professional development of medical students.
Several methods have been developed to study burnout among students, namely the Maslach Burnout Inventory-General Survey for Students (MBI-SS; Maslach, Jackson, & Leiter, 2017), the Oldenburg Burnout Inventory for college students (OL-BI-S; Demerouti & Bakker, 2008), and the Copenhagen Burnout Inventory (CBI) proposed by Kristensen, Borritz, Villadsen, and Christensen (2005). e MBI-SS assesses the prevalence of burnout based on subjects' emotional exhaustion, depersonalization, and reduced professional satisfaction and e ectiveness, as captured by 22 items. e OLBI-S includes two dimensions, exhaustion and disengagement, with the distinction that it captures exhaustion across physical, a ective, and cognitive dimensions compared to the single emotional dimension measured by the MBI-SS (Demerouti & Bakker, 2008). By comparison, CBI focuses only on fatigue/emotional exhaustion, but measures the respondent's attribution of this exhaustion to three different life domains: Personal Burnout, Work-Related Burnout, and Client-Related Burnout (Molinero Ruiz, Basart Gómez-Quintero, & Moncada Lluis, 2013). e CBI measures burnout in a more straightforward way (Yeh, Cheng, Chen, Hu, & Kristensen, 2007). According to one systematic review of the CBI and the OLBI, the quality of evidence for su cient content validity was moderate, while for the MBI it was very low. Moreover, the CBI was more appropriate for valid and reliable use in medical research and practice (Shoman et al., 2021). A systematic review and meta-analysis conducted among midwives showed that the CBI addressed more realistically the levels of physical and mental exhaustion and was very useful (Suleiman-Martos et al., 2020).
In recent years, the CBI has been validated in di erent countries and study populations, such as university professors and academic sta members at Brazilian public universities (Rocha et al., 2020), an academic healthcare institution sample in the U.S. ( rush, Gathright, Atkinson, Messias, & Guise, 2020), Greek doctors (Papaefstathiou, Tsounis, Malliarou, & Sara s, 2019), Iranian nurses (Mahmoudi et al., 2017), Korean homecare workers (Jeon, You, Kim, Kim, & Cho, 2019), and U.S. nurses (Montgomery, Azuero, & Patrician, 2021). In all cases, the CBI demonstrated adequate validity and reliability for measuring burnout. Andrew Chin et al. (2018) investigated the validity of the CBI among Malaysian medical students, but using the original three-dimensional structure. Campos, Carlotto, and Marôco (2013) adapted the CBI original inventory for students as the CBI-Student Survey, and developed items measuring students' Personal Burnout, Studies-Related Burnout, Colleague-Related Burnout, and Teacher-Related Burnout.
is study aims at evaluating the reliability and validity of the Russian version of the Copenhagen Burnout Inventory-Student Survey (R-CBI-S) in a sample of medical students at the Astana Medical University, Kazakhstan.

Participants
All medical students at any stage of their medical education at Astana Medical University were eligible to participate. Participants were invited via the "messenger" app and the university's information portal, Sirius, to ll out an online questionnaire created on the 1ka platform (www.1ka.si) during the period October-December 2019. e questionnaire was completed by 771 students (response rate 40%). Of the participants, 25.0% were male. Academic year distribution among students was 1 year (218), 2 year (137), 3 year (125), 4 year (62), 5 year (60), and 6 year (169). e average age of the respondent was 20.7 years (ranged in age from 18 to 33).

Procedure
e Copenhagen Burnout Inventory-Student Survey was converted into the Russian language from the original English version using a forward-backward translation process performed by specialists in the eld of psychology and language. e nal questionnaire was revised based on feedback from a sample of 20 participants through a pilot study.
Data analysis was conducted using Microso Excel 2007, SPSS version 20.0, and Jamovi version 1.2.17. A statistically signi cant di erence was accepted at a p-value of less than 5%. e reliability of the scale (performed on a sample of 20 subjects during a twoweek interval) was evaluated using the intraclass correlation coe cient (ICC). According to Koo and Li (2016) ICC values between 0.75 and 0.9 indicate good reliability, and values greater than 0.90 indicate excellent reliability.
Internal consistency was evaluated by the total scale and subscales reliability analysis re ected by Cronbach's alpha coe cient. A Cronbach's alpha coe cient with a value of ≥ 0.7 is acceptable (Taber, 2018). Corrected item-total correlation was carried out.
Convergent validity was checked with average variance extracted (AVE) and composite reliability (CR). Values of 0.5 or more for AVE and 0.6 or more for CR were considered as having signi cant convergent validity (Kline, 2011). Concurrent validity was evaluated by bivariate correlations of R-CBI-S with anxiety (GAD-7; Spitzer, Kroenke, Williams, & Lowe, 2006), depression (PHQ-9; Kroenke & Spitzer, 2002), and satisfaction with the study.
Construct validity was established by the con rmatory factor analysis (CFA) technique, with Bartlett's test of sphericity and the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy used to test the dataset for factor analysis suitability. e CFA is used to assess the overall goodness of t: the Root Mean Square of Error Approximation RMSEA (< 0.08); the Comparative Fit Index CFI (> 0.9); and the Tucker-Lewis Index TLI (> 0.9) (Xia & Yang, 2019).

Results
e nal translated Russian version of the R-CBI-S is presented in Table 1. e test-retest reliability showed an ICC of 0.81 (CI 95% 0.63-0.94) for the R-CBI-S. e overall Cronbach's alpha coe cient of the R-CBI-S was 0.939 (0.896 for PB, 0.884 for SRB, 0.874 for CRB, and 0.926 for TRB), which indicates a high level of internal consistency. Corrected item-total correlation is shown in Table 1. All item-total correlations for the total scale were positive (range 0.31-0.76) within the criterion of the item-total correlation greater than 0.30 (DeVellis, 2003). e Barlett's sphericity test result was signi cant (p < 0.001), and the KMO measure of sampling adequacy exceeded 0.947. Extracted AVE and CR from convergent validity analysis showed in Table 1. According to the CFA analysis, the model t of the four-factor R-CBI-S model was con rmed by the indices: χ2/df 3.881; RM-SEA = 0.0611; CFI = 0.940; TLI = 0.933, with cumulative variance at 59.5% (by com-  Figure 1 shows the factor model. Analysis of the eigenvalues indicated that four factors extracted with values above 1.0 (9.76 for TRB, 2.33 for the CRB, 1.44 for the SRB, and 1.12 for the PB) according to Henson & Roberts (2006). Table 2 Concurrent validity of the R-CBI-S

Variables
(1)  Table 2 shows the correlations of the R-CBI-S with other variables. Weak positive correlations were found for R-CBI-S and GAD-7, PHQ-9. Satisfaction with the study was found to be negatively associated with R-CBI-S. e total R-CBI-S mean score was 39.96, and the mean subscale scores for this sample were 52.62 (PB), 50.93 (SRB), 23.50 (CRB), and 32.77 (TRB).

Discussion
Analysis of the literature showed that burnout is an important component of medical students' mental health, which can a ect the learning process and have further professional consequences. A feature of the CBI is that it divides burnout into four components. is makes it possible to identify predictors of burnout covering not only exhaustion, but dividing it into personal, studies-related, colleague-related, and teacher-related burnout, ultimately to draw up a more comprehensive approach to organization of the educational process. We agree with the opinion of Sedlar, Šprah, Tement, and Sočan (2015), that before using the scale, one must go through a validation process to obtain the most reliable results. e purpose of this study was to examine the psychometric properties of the Russian version of the CBI-S. Following adaptation and psychometric tests, this study found that the survey was reliable and valid for assessing burnout among Russianspeaking medical students in Kazakhstan. e ICC analysis showed that the R-CBI-S had high stability within 2 weeks of the test-retest (mean ICC 0.81). e internal consistencies of the four subscales were satisfactory, with all the Cronbach's alpha values ranging from 0.874 to 0.926, and Cronbach's alpha for R-CBI-S being 0.939. ese results are slightly lower than those reported by Campos et al. (2013), with Cronbach's alpha ranging from 0.875 to 0.931, and 0.957 for the CBI-S. e results of the current study present good internal consistency values. e corrected item-total correlation values obtained for the items are relatively high, which demonstrates that the items of R-CBI-S are relatively homogeneous and are measuring the same overall construct. AVE for all dimensions was equal to or more than 0.5, suggesting an adequate level of convergent validity. e CR values of the R-CBI-S constructs ranged between 0.86 and 0.88, which indicates a high level of convergent validity. e R-CBI-S was associated with anxiety, depression, and satisfaction with the study, lending support to the scale's concurrent validity. A validated Chinese version of CBI was correlated not only with anxiety and depression, but also with physical distress and social support (Fong, Ho, & Ng, 2013). e R-CBI-S demonstrated satisfactory construct validity, as tested by CFA. e results indicated that most t indices were in acceptable ranges. Su ciency of the model was demonstrated by Bartlett's test of sphericity and the KMO measure.

Conclusion
e R-CBI-S appears to be a reliable and valid instrument in measuring medical students' burnout. e instrument could be useful for future e orts to develop an e ective preventive intervention for burnout syndrome determination among Russianspeaking medical students.

Ethics Statement
e study was approved by the Local Ethics Committee of the NpJSC "Astana Medical University" (extract from protocol No. 3, dated September 20, 2018).

Con ict of Interest
e authors declare no con ict of interest.