Assessing the Use of a Feedback Module to Model EQ-5D-5L Health States Values in Hong Kong

• Published in: The patient : patient-centered outcomes research 2018-04, Vol.11 (2), p.235-247
• Main Authors: Wong, Eliza L. Y., Ramos-Goñi, Juan Manuel, Cheung, Annie W. L., Wong, Amy Y. K., Rivero-Arias, Oliver
• Format: Article
• Language: English
• Subjects: Adolescent; Adult; Aged; Behavior; Censuses; Choice Behavior; Cognition & reasoning; Cognitive ability; Economics; Feedback; Female; Formative Feedback; Health Administration; Health Economics; Health Status; Humans; Male; Medicine; Medicine & Public Health; Middle Aged; Models, Theoretical; Original; Original Research Article; Patient Satisfaction; Pharmacoeconomics and Health Outcomes; Population; Public Health; Quality of Life; Quality of Life Research; Research Design; Socioeconomic Factors; Surveys and Questionnaires - standards; Tradeoff analysis; Young Adult
• ISSN: 1178-1653; 1178-1661
• DOI: 10.1007/s40271-017-0278-0

Background An international valuation protocol exists for obtaining societal values for each of the 3125 health states of the five-level EuroQol-five dimensions (EQ-5D-5L) questionnaire. A feedback module (FM) that can be related to theoretical models used in behavioral economics was recently included in this protocol. Objectives Our objective was to assess the impact of using an FM to estimate an EQ-5D-5L value set in Hong Kong. Methods EQ-5D-5L health states were elicited using a composite time trade-off (C-TTO) and a discrete-choice (DC) experiment. Use of the FM according to participant characteristics and the impact of the FM on the number of inconsistent C-TTO responses were assessed. We employed a main-effects hybrid model that combined data from both elicitation techniques. Results In total, 1014 individuals completed the survey. The sample was representative of the general Chinese Hong Kong population in terms of sex, educational attainment, marital status, and most age groups but not for employment status. The use of the FM reduced the number of C-TTO inconsistencies. Participant characteristics differed significantly between those who used and did not use the FM. The model without a constant resulted in logical consistent coefficients and was therefore selected as the model to produce the value set. The predicted EQ-5D-5L values ranged from −0.864 to 1. Conclusions The use of an FM to allow participants to exclude C-TTO responses reduced the number of inconsistent responses and improved the quality of the data when estimating an EQ-5D-5L value set in Hong Kong.