Evaluating the factor structure and measurement invariance of the Tilburg Frailty Index among community-dwelling older people in Saudi Arabia

Frailty is a multidimensional syndrome associated with adverse health outcomes in older adults and is commonly assessed using the self-reported Tilburg Frailty Indicator (TFI). Although the TFI was originally developed with physical, psychological, and social domains, evidence regarding its factor structure and measurement invariance across cultures remains mixed. To date, no large-scale study has examined the factorial validity and measurement invariance of the TFI in Saudi Arabia. A cross-sectional study was conducted among 428 community-dwelling Saudi adults aged ≥ 50 years. Exploratory factor analysis (EFA) using tetrachoric correlations and unweighted least squares was performed to examine dimensionality. Confirmatory factor analysis (CFA) using weighted least squares mean, and variance adjustment (WLSMV) initially evaluated the original three-factor TFI structure, followed by additional models informed by EFA findings. Model stability was assessed using repeated 70/30 holdout cross-validation (200 iterations). Measurement invariance across gender, age group (50-64 vs. ≥ 65 years), and marital status was examined using multiple-group CFA. Exploratory factor-retention analyses yielded partially divergent recommendations regarding dimensionality; however, comparative evaluation of competing solutions supported retention of a three-factor structure corresponding broadly to physical, psychological, and social frailty, explaining approximately 57% of the total variance. CFA indicated that an adapted three-factor model retaining all 15 items demonstrated excellent fit (CFI = 0.971, TLI = 0.966, RMSEA = 0.046) and outperformed the original specification (CFI = 0.926, TLI = 0.913, RMSEA = 0.073). A three-factor 14-item model excluding item Q6 showed comparable fit in sensitivity analyses. Cross-validation confirmed the robustness of the adapted 15-item model (median CFI = 1.00, median RMSEA = 0.00). Multiple-group CFA supported configural and threshold invariance across all examined subgroups. This study provides substantial evidence for the factorial validity, stability, and measurement invariance of an adapted three-factor 15-item TFI among community-dwelling older adults in Saudi Arabia. The findings support its use for frailty assessment and subgroup comparisons in Saudi and similar Middle Eastern populations, while highlighting the importance of culturally informed interpretation of frailty domains.