Invariance Testing of the SF-36 Health Survey in Women Breast Cancer Survivors: Do Personal and Cancer-related Variables Influence the Meaning of Quality of Life Items?

Quality of life (QoL) is affected by issues specific to illness trajectory and thus, may differ, and potentially take on different meanings, at different stages in the cancer process. A widely used measure of QoL is the SF-36 Health Survey (SF-36; Ware 1993); therefore, support for its appropriateness in a given population is imperative. The current study aimed to examine the conceptual (measurement) model of the SF-36, as well as closely related models, and test the measurement invariance of the SF-36 to determine if meaningful comparisons could be made among three groups of breast cancer survivors (N = 358 [data collected in 2007–2008]; divided on time since treatment, type of treatment, and age). Good model fit was found for one of three models based on the original design of this instrument—the items to subscales model. Two models were considered for measurement invariance testing: (a) items to physical health/mental health and (b) items to subscales. Strong invariance was found for time since treatment and type of treatment for both models. Weak invariance was found for age in the first model, while strict invariance was confirmed for the subscales model. Group comparisons in QoL were made where justified. Significant differences were found only on age for physical functioning, role limitations due to emotional problems, vitality, mental health, and social functioning. Overall, results suggest that while the SF-36 can be used to examine differences in QoL for various breast cancer survivors, some conceptual issues with this instrument need to be further examined.