Cross-Cultural Adaptation and Exploratory Psychometric Evaluation of the Bilingual Cervical Cancer Screening Health Belief Model Scale (CCS-HBMS) in Kazakhstan

Background: Cervical cancer remains an important preventable cause of morbidity and mortality among women in Kazakhstan. Reliable assessment of beliefs related to cervical cancer screening requires instruments that are linguistically and culturally appropriate for the target population. This study aimed to cross-culturally adapt the CCS-HBMS for bilingual Kazakh–Russian administration and to evaluate the preliminary psychometric properties of this bilingual administration format. Methods: An exploratory cross-sectional psychometric study was conducted among 199 women aged 20–60 years attending primary healthcare facilities in Astana, Kazakhstan. The 35-item CCS-HBMS underwent forward and back translation, expert review, and pilot testing and was administered in a simultaneous Russian–Kazakh bilingual format. Psychometric analyses were performed in 196 participants with complete responses to all 35 items. Internal consistency was assessed using Cronbach’s alpha, and structural validity of the original five-factor model was evaluated using confirmatory factor analysis (CFA) with the weighted least squares mean- and variance-adjusted estimator. Exploratory factor analysis (EFA) was subsequently conducted because of inadequate CFA model fit. Composite reliability (CR), average variance extracted (AVE), and the heterotrait–monotrait ratio (HTMT) were also examined. Results: Internal consistency varied across the five original subscales, with Cronbach’s α ranging from 0.481 to 0.827. Reliability was highest for Barriers (α = 0.827) and Benefits and Health Motivation (α = 0.804), followed by Susceptibility (α = 0.725), whereas lower values were observed for Seriousness (α = 0.662) and Health Motivation (α = 0.481). The original five-factor CFA model showed inadequate fit (χ2(550) = 1952.71, p < 0.001; CFI = 0.751; TLI = 0.731; RMSEA = 0.114, 90% CI 0.109–0.120; SRMR = 0.137). Parallel analysis supported retention of six factors. The six-factor solution showed lower residual misfit than the original five-factor solution (RMSR = 0.045 vs. 0.055). Given the redistribution of items across factors and the exploratory nature of these analyses, no definitive alternative factor structure was established. Convergent validity was strongest for Susceptibility (CR = 0.766; AVE = 0.523), while weaker evidence was observed for Seriousness and Health Motivation. Conclusions: The original five-factor structure of the CCS-HBMS was not supported in the present exploratory bilingual sample, and only two of the five original subscales showed Cronbach’s alpha values above 0.80. Exploratory analyses suggested a different dimensional organization but did not establish a definitive alternative factor structure. These findings should be interpreted as preliminary evidence regarding the bilingual administration format rather than as separate validation of the Kazakh- or Russian-language versions, and further evaluation in larger, more diverse, and independent samples is required.