RESUMO
OBJECTIVE: Previous research indicates that the FACIT-Sp instrument is susceptible to bias when measuring spiritual well-being in older patients. Our first focus was to evaluate the two-factor vs the three-factor model of the FACIT-Sp and our second focus was to explore how these models behave for different age groups. METHODS: We used a large national cohort of Danish cancer patients (N = 3439) which included a significant number of patients aged at least 70 years (N = 1033). Item-test correlations and factor analyses were conducted on complete cases (N = 2820). Additionally, a reliability analysis was performed using Cronbach's alpha and Guttman's lower-bound estimate. RESULTS: Factor analysis revealed a loading pattern for the oldest age group (70+) showing items on peace and meaning loading into a single factor, as originally proposed in the two-factor model. The loading estimates for the patients younger than 70 matched the three-factor model. Furthermore, item-test correlations changed as age increased. Based on Cronbach's alpha and Guttman's estimate of .83 and .89, respectively, total scores proved reliable. Items 4, 8, and 12 are discussed separately concerning their problematic influence on instrument validity in their current formulation. CONCLUSION: Overall, the three-factor model had a good fit; however, for the eldest patients a two-factor solution proved even better. Interpretation of FACIT-Sp subscale scores of older cancer patients must therefore be done with caution.