HOW THE EQ-5D UTILITIES ARE DERIVED MATTER. COMPARISONS BETWEEN TWO EQ-5D-3L INDICES, AND AMONG THREE EQ-5D-5L INDICES IN DIABETES PATIENTS
Author(s)
ABSTRACT WITHDRAWN
OBJECTIVES : China currently has two EQ-5D-3L tariffs (developed in 2014 and 2018), and one EQ-5D-5L tariff developed in 2017. The EQ-5D-5L index score can also be mapped from the EQ-5D-3L index score using the crosswalk approach. The study compared the EQ-5D-3L index scores derived from the two EQ-5D-3L tariffs (3L2014 and 3L2018); and to compare the EQ-5D-5L index score from the EQ-5D-5L tariff (5L) with the two mapped index scores (c5L2014 and c5L2018). METHODS : The data were from a cross-sectional study of 289 Chinese type 2 diabetes (T2D) patients who self-completed both the two versions of EQ-5D. Agreement between the two EQ-5D-3L indices, and the three EQ-5D-5L indices were examined using intraclass correlation coefficient (ICC) and Bland-Altman plots. Their ability in differentiating T2D patients with or without each of the ten clinical conditions was compared using F-statistic values. RESULTS : The mean values for the two EQ-5D-3L indices were 0.898 (3L2014), and 0.959 (3L2018), with the mean absolute difference (MAD) and ICC being 0.061 and 0.881. The values for the three EQ-5D-5L indices were 0.891 (5L), 0.876 (c5L2014), and 0.951(c5L2018), with the MADs and ICCs being 0.015 and 0.958 (5L and c5L2014); and 0.060 and 0.806 (5L and c5L2014), respectively. The 95% limits of agreement was wide: 0.34 for 3L2014 and 3L2018, 0.25 for 5L and c5L2014, and 0.44 for 5L and c5L2018, respectively, which was mainly because the poor agreement at the lower end of utility scale. In comparing T2D patients with or without a certain condition, higher F-statistic values were generally observed for 3L2014 in comparison with 3L2018; and for the two mapped EQ-5D-5L indices compared to the 5L. CONCLUSIONS : The two EQ-5L-3L indices and the three EQ-5D-5L indices can not be used interchangeably. The 3L2014 and the two mapped EQ-5D-5L indices are more discriminative than their respective counterparts.
Conference/Value in Health Info
2019-11, ISPOR Europe 2019, Copenhagen, Denmark
Code
PDB87
Disease
Diabetes/Endocrine/Metabolic Disorders