Objective Develop a general patient-reported outcome(PRO)measure for cancer patients,which will be suitable for Chinese demographic characteristics.It can reflect disease burden from patients' perspective.Methods Theoretical framework of the PRO measure was constructed according to the PRO guide draft.The initial version of the PRO measure was formed after literature search,experts' consultation,and recognition tests of patients.The pre-survey and formal survey were conducted respectively in provincial,municipal,county level hospitals.Items were screened by the combination of classical test theory and item response theory.Its reliability,validity and feasibility were analyzed.The minimal detectable change of three domains were estimated using a distribution-based methods.Results A total of 2213 valid questionnaires were collected.The demographic data of cancer patients and non-cancer patients were comparable(P＞0.05).19 items were deleted after classical test theory.Item responsiveness of four disease systems were analyzed by the Bayesian item response theory.Four items were deleted.The formal PRO measure included 4 domains,13 dimensions and 49 items.And its reliability,validity and feasibility results met the expected criteria.The minimal detectable change in physiological,psychological and social domains were 5.63,3.42 and 4.16,respectively.Conclusion The general PRO measure for cancer patients was reliable and feasible.The tool can effectively measure and evaluate the quality of life for cancer patients.

Objective Develop a general patient-reported outcome(PRO)measure for cancer patients,which will be suitable for Chinese demographic characteristics.It can reflect disease burden from patients' perspective.Methods Theoretical framework of the PRO measure was constructed according to the PRO guide draft.The initial version of the PRO measure was formed after literature search,experts' consultation,and recognition tests of patients.The pre-survey and formal survey were conducted respectively in provincial,municipal,county level hospitals.Items were screened by the combination of classical test theory and item response theory.Its reliability,validity and feasibility were analyzed.The minimal detectable change of three domains were estimated using a distribution-based methods.Results A total of 2213 valid questionnaires were collected.The demographic data of cancer patients and non-cancer patients were comparable(P＞0.05).19 items were deleted after classical test theory.Item responsiveness of four disease systems were analyzed by the Bayesian item response theory.Four items were deleted.The formal PRO measure included 4 domains,13 dimensions and 49 items.And its reliability,validity and feasibility results met the expected criteria.The minimal detectable change in physiological,psychological and social domains were 5.63,3.42 and 4.16,respectively.Conclusion The general PRO measure for cancer patients was reliable and feasible.The tool can effectively measure and evaluate the quality of life for cancer patients.