PURPOSE: To develop a mapping algorithm for a conversion of the EORTC QLQ-C30 and EORTC QLQ BR-23 into the EQ-5D-derived utilities in metastatic breast cancer (MBC) patients. METHODS: We enrolled 199 patients with MBC from four leading Korean hospitals in 2009. EQ-5D utility, cancer-specific (QLQ-C30) and breast cancer-specific quality of life data (QLQ-BR23) and selected clinical and demographic information were collected from the study participants. Ordinary least squares regression models were used to model the EQ-5D using QLQ-C30 and QLQ-BR23 scale scores. To select the best model specification, six different sets of explanatory variables were compared. RESULT: Regression analysis with the multiitem scale scores of QLQ-C30 was the best-performing model, explaining for 48.7% of the observed EQ-5D variation. Its mean absolute error between the observed and predicted EQ-5D utilities (0.092) and relative prediction error (2.784%) was among the smallest. Also, this mapping model showed the least systematic errors according to disease severity. CONCLUSIONS: The mapping algorithms developed have good predictive validity, and therefore, they enable researchers to translate cancer-specific health-related quality of life measures to the preference-adjusted health status of MBC patients.
PURPOSE: To develop a mapping algorithm for a conversion of the EORTC QLQ-C30 and EORTC QLQ BR-23 into the EQ-5D-derived utilities in metastatic breast cancer (MBC) patients. METHODS: We enrolled 199 patients with MBC from four leading Korean hospitals in 2009. EQ-5D utility, cancer-specific (QLQ-C30) and breast cancer-specific quality of life data (QLQ-BR23) and selected clinical and demographic information were collected from the study participants. Ordinary least squares regression models were used to model the EQ-5D using QLQ-C30 and QLQ-BR23 scale scores. To select the best model specification, six different sets of explanatory variables were compared. RESULT: Regression analysis with the multiitem scale scores of QLQ-C30 was the best-performing model, explaining for 48.7% of the observed EQ-5D variation. Its mean absolute error between the observed and predicted EQ-5D utilities (0.092) and relative prediction error (2.784%) was among the smallest. Also, this mapping model showed the least systematic errors according to disease severity. CONCLUSIONS: The mapping algorithms developed have good predictive validity, and therefore, they enable researchers to translate cancer-specific health-related quality of life measures to the preference-adjusted health status of MBCpatients.
Authors: M A Sprangers; M Groenvold; J I Arraras; J Franklin; A te Velde; M Muller; L Franzini; A Williams; H C de Haes; P Hopwood; A Cull; N K Aaronson Journal: J Clin Oncol Date: 1996-10 Impact factor: 44.544
Authors: Stefan Sauerland; Sylvia Weiner; Karin Dolezalova; Luigi Angrisani; Carlos Masdevall Noguera; Manuel García-Caballero; Frédéric Rupprecht; Marc Immenroth Journal: Value Health Date: 2009 Mar-Apr Impact factor: 5.725
Authors: N K Aaronson; S Ahmedzai; B Bergman; M Bullinger; A Cull; N J Duez; A Filiberti; H Flechtner; S B Fleishman; J C de Haes Journal: J Natl Cancer Inst Date: 1993-03-03 Impact factor: 13.506