Generating Real-World Utility Values for Pediatric Health States in Taiwan: A Nationwide EQ-5D-Y-3L Valuation Study Using Composite Time Trade-Off
Author(s)
Hsiang-Wen Lin, MS, DrPH, DPhil, PhD1, Yen-Ming Huang, Ph.D.2, Siang-Hao Cheng, M.S.3, Pin Hsuan Liao, M.S.4, Hao-Ping Chiang, M.S.5, Jen-Yu Amy Chang, MSc, RPh, PhD6, Chien-Ning Hsu, Ph.D.7, Chia-Ing Li, Ph.D.8.
1School of Pharmacy and Graduate Institue, China Medical University, Taichung City, Taiwan, 2National Taiwan University, Taipei, Taiwan, 3Department of Pharmaceutical Outcomes & Policy Research,School of Pharmacy, University of Pittsburgh, Pittsburgh, PA, USA, 4Health Outcomes Division, College of Pharmacy, The University of Texas at Austin, Austin, TX, USA, 5Department of Clinical Pharmacy, College of Pharmacy, University of Michigan, Ann Arbor, MI, USA, 6University of Sheffield, Sheffield, United Kingdom, 7Kaoshiung Chang Gung Memorial Hospital, Kaoshiung, Taiwan, 8China Medical University Hospital, Taichung City, Taiwan.
1School of Pharmacy and Graduate Institue, China Medical University, Taichung City, Taiwan, 2National Taiwan University, Taipei, Taiwan, 3Department of Pharmaceutical Outcomes & Policy Research,School of Pharmacy, University of Pittsburgh, Pittsburgh, PA, USA, 4Health Outcomes Division, College of Pharmacy, The University of Texas at Austin, Austin, TX, USA, 5Department of Clinical Pharmacy, College of Pharmacy, University of Michigan, Ann Arbor, MI, USA, 6University of Sheffield, Sheffield, United Kingdom, 7Kaoshiung Chang Gung Memorial Hospital, Kaoshiung, Taiwan, 8China Medical University Hospital, Taichung City, Taiwan.
OBJECTIVES: There is a growing need for real-world utility values to inform pediatric health decision-making in Taiwan, particularly in the absence of a locally derived value set for health-related quality of life (HRQoL) measures. This study presents the first Taiwan-based valuation of EQ-5D-Y-3L health states using real-world population data, following the standardized protocol of the multi-country Asian EQ-5D-Y-3L valuation initiative.
METHODS: A nationally representative sample of 200 adults, stratified by age and gender across four regions of Taiwan, participated in face-to-face composite Time Trade-Off (cTTO) interviews. Each respondent evaluated 10 randomly assigned health states (out of 28) for a hypothetical 10-year-old child from a proxy perspective. We analyzed the characteristics and distribution of the cTTO data and applied multiple statistical models—Ordinary Least Squares (OLS), Generalized Least Squares (GLS), and Tobit models (censored at -1)—to generate health utility estimates. Model performance was assessed using root mean square error (RMSE), mean absolute error (MAE), log-likelihood, AIC, and BIC criteria.
RESULTS: Respondent demographics closely matched Taiwan’s adult population profile. The cTTO data reflected real-world health valuation behaviors, with 31.2% of responses indicating negative utilities—comparable to the 2018 EQ-5D-5L study (38.5%). However, fewer respondents assigned the full health state (11111) a utility of 1. Among all models tested, the Tobit GLS model (censored at -1) provided the best predictive accuracy, with the worst health state (33333) valued at -0.1876 (without intercept) and -0.1930 (with intercept).
CONCLUSIONS: This study provides the first real-world utility evidence for EQ-5D-Y-3L in Taiwan using cTTO data alone, highlighting the feasibility of pediatric value set development without discrete choice experiments (DCE). Despite a narrower utility range compared to adult data, the consistency in negative utility proportions reinforces the robustness of our findings. These real-world results lay a critical foundation for future economic evaluations and evidence-based policy in pediatric health care.
METHODS: A nationally representative sample of 200 adults, stratified by age and gender across four regions of Taiwan, participated in face-to-face composite Time Trade-Off (cTTO) interviews. Each respondent evaluated 10 randomly assigned health states (out of 28) for a hypothetical 10-year-old child from a proxy perspective. We analyzed the characteristics and distribution of the cTTO data and applied multiple statistical models—Ordinary Least Squares (OLS), Generalized Least Squares (GLS), and Tobit models (censored at -1)—to generate health utility estimates. Model performance was assessed using root mean square error (RMSE), mean absolute error (MAE), log-likelihood, AIC, and BIC criteria.
RESULTS: Respondent demographics closely matched Taiwan’s adult population profile. The cTTO data reflected real-world health valuation behaviors, with 31.2% of responses indicating negative utilities—comparable to the 2018 EQ-5D-5L study (38.5%). However, fewer respondents assigned the full health state (11111) a utility of 1. Among all models tested, the Tobit GLS model (censored at -1) provided the best predictive accuracy, with the worst health state (33333) valued at -0.1876 (without intercept) and -0.1930 (with intercept).
CONCLUSIONS: This study provides the first real-world utility evidence for EQ-5D-Y-3L in Taiwan using cTTO data alone, highlighting the feasibility of pediatric value set development without discrete choice experiments (DCE). Despite a narrower utility range compared to adult data, the consistency in negative utility proportions reinforces the robustness of our findings. These real-world results lay a critical foundation for future economic evaluations and evidence-based policy in pediatric health care.
Conference/Value in Health Info
2025-09, ISPOR Real-World Evidence Summit 2025, Tokyo, Japan
Value in Health Regional, Volume 49S (September 2025)
Code
RWD34
Topic Subcategory
Distributed Data & Research Networks
Disease
SDC: Pediatrics