Performance of the Family Satisfaction with the End-of-Life Care (FAMCARE) measure in an ethnically diverse cohort: psychometric analyses using item response theory

Abstract

The Family Satisfaction with End-of-Life Care (FAMCARE) has been used widely among caregivers to individuals with cancer. The aim of this study was to evaluate the psychometric properties of this measure using item response theory (IRT). The analytic sample was comprised of caregivers to 1,983 patients with advanced cancer. Among the patients, 56% were females, with mean age 59.9years (s.d. = 11.8), 20% were non-Hispanic Black. The majority were family members either living with (44%) or not living with (35%) the patient. Factor analyses and IRT were used to examine the dimensionality, information, and reliability of the FAMCARE. Although a bi-factor model fit the data slightly better than did a unidimensional model, the loadings on the group factors were very low. Thus, a unidimensional model appears to provide adequate representation for the item set. The reliability estimates, calculated along the satisfaction (theta) continuum, were adequate (>0.80) for all levels of theta for which subjects had scores. Examination of the category response functions from IRT showed overlap in the lower categories with little unique information provided; moreover, the categories were not observed to be interval. Based on these analyses, a three-response category format was recommended: very satisfied, satisfied, and not satisfied. Most information was provided in the range indicative of either dissatisfaction or high satisfaction. These analyses support the use of fewer response categories and provide item parameters that form a basis for developing shorter-form scales. Such a revision has the potential to reduce respondent burden.