Comparison of three utility measures in stroke patients using item response theory analysis

IR

Search this article

Description

Our study aimed to clarify the measurement characteristics of three utility measures by item response theory (IRT) analysis in stroke patients. A multicenter cross-sectional study was conducted on patients with stroke. The Japanese versions of the EQ-5D-3L, EQ-5D-5L and HUI3 were used for utility measurement. Twoparameter logistic models were used for the model of IRT analysis. The choice probability (item information) of patients when a patient's capability value theta (health states) is given is expressed with this model. Moreover, discrimination and difficulty, which are the parameters showing the item characteristic, can be shown. Subjects were 526 stroke patients. The maximum of the total information was 13.1 in the EQ-5D-3L, 22.5 in the EQ-5D-5L and 8.7 in the HUI3. Moreover, of the three measures, the information in the EQ-5D-5L was the highest, and information could be acquired over a wide range of health states (-1.4 < theta < 1.1, information >10.0). Discrimination was good for all of the measures, and the EQ-5D-5L especially had high discrimination. The three utility measures showed a wide range for difficulty. However, theta was not adapted when extremely high or low. These findings indicate that the EQ-5D-3L, EQ-5D-5L and HUI3 could all measure the stroke patient’s utility scores over a wide range of health states. However, in the stroke patients, the EQ- 5D-5L gave the most information and was the highest of the three utility measures in terms of discrimination.

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top