Addressing quality of life in survivors of pediatric bone marrow transplantation: Practice patterns across centers

Abstract

general cancer and randomly selected prostate cancer patients (n = 295). All items together demonstrated acceptable MnSq values, high Cronbach a (a = 0.97), and reasonable item-total correlation (range, 0.30-0.90). A bell-shape test information function curve was found, which met our expectation. When we divided all patients into 5 different groups based on their response to the item “I have pain,” significant mean differences across groups were found (F [4, 2881 = 198.98; P < 0.001). All but 1 pairwise comparison were statistically significant in post hoc analysis (Tikey honest significant difference). Patients who chose “very much” did not differ significantly from patients who chose “quite a bit.” DISCUSSION: This study illustrated a structured process for building an item bank. This bank demonstrated good psychometric properties and could differentiate patients with different degrees of pain. This bank could serve as the foundation for the development of a CAT platform that would allow easy, individualized pain measurement amenable to integration into a clinical setting. ACKNOWLEDGMENT: This study was supported by the National Cancer Institute (CA60068) and the American Cancer Society (RSG-Ol-221-Ol-PBP).