Psychometric characteristics of the Functional Assessment of Cancer Therapy-General when applied to Brazilian cancer patients: A cross-cultural adaptation and validation

Abstract

Background: The psychometric properties of an instrument should be evaluated routinely when using different samples. This study evaluated the psychometric properties of the Functional Assessment of Cancer Therapy-General (FACT-G) when applied to a sample of Brazilian cancer patients. Methods: The face, content, and construct (factorial, convergent, and discriminant) validities of the FACT-G were estimated. Confirmatory factor analysis (CFA) was conducted the ratio chi-square by degrees of freedom (χ 2/df), the comparative fit index (CFI), the Tucker-Lewis index (TLI), and the root mean square error of approximation (RMSEA) as indices. The invariance of the best model was assessed with multi-group analysis using the difference of chi-squares method (∆χ2). Convergent validity was assessed using Average Variance Extracted (AVE) and discriminant validity was determined via correlational analysis. Internal consistency was assessed using the Cronbach's alpha (α) coefficient, and the Composite Reliability (CR) was estimated. Results: A total of 975 cancer patients participated in the study, with a mean age of 53.3 (SD = 13.0) years. Of these participants, 61.5 % were women. In CFA, five correlations between errors were included to fit the FACT-G to the sample (χ 2/df = 8.611, CFI = .913, TLI = .902, RMSEA = .088). The model did not indicate invariant independent samples (∆χ2: μ: p < .001, i: p < .958, Cov: p < .001, Res: p < .001). While there was adequate convergent validity for the physical well-being (AVE = .54) and social and family Well-being factors (AVE = .55), there was low convergent validity for the other factors. Reliability was adequate (CR = .76-.89 and α = .71-.82). Functional well-being, emotional well-being, and physical well-being were the factors that demonstrated a strong contribution to patients' health-related quality of life (β = -.99,.88, and.64, respectively). Conclusion: The FACT-G was found to be a valid and reliable assessment of health-related quality of life in a Brazilian sample of patients with cancer.