Li Cheng, Qing He, Xiaoyu Fang, Qiang Sun, Jia Yin
Both instruments showed acceptable measurement properties among Chinese PLHIV, but systematic utility differences suggest they should not be used interchangeably. These findings support objective-driven instrument selection and may inform cost-utility analyses and utility-improvement targets.
PURPOSE: To compare health utilities and measurement properties of the EQ-5D-5L and SF-6Dv2 among people living with HIV (PLHIV) in China, and to identify factors associated with utility values.
METHODS: A multicenter cross-sectional survey was conducted among PLHIV in Shandong, Guangdong, and Sichuan provinces. Utilities were calculated using Chinese value sets. Ceiling/floor effects were evaluated. Convergent validity and agreement were assessed using Spearman's rank correlation, intraclass correlation coefficients (ICC), and Bland-Altman (B-A) plots. Known-groups validity and discriminative performance were evaluated using effect sizes and relative efficiency. Tobit models identified factors associated with utilities.
RESULTS: Among 705 respondents, mean (SD) utilities were 0.93 (0.15) for EQ-5D-5L and 0.80 (0.17) for SF-6Dv2. Ceiling effects were higher for EQ-5D-5L than SF-6Dv2 (48.2 vs. 14.3%). The instruments showed good convergent validity (r = 0.719) and agreement (ICC = 0.729), but the B-A limits of agreement spanned 0.48, exceeding the instrument-defined minimally important difference (MID) of 0.06. Both instruments discriminated across known groups. The EQ-5D-5L had higher relative efficiency for comorbid chronic disease, CD4 + T-cell counts, and prior HIV self-testing experience, whereas the SF-6Dv2 had higher relative efficiency for duration of ART, economic burden, and self-rated health. Comorbid chronic disease, no prior HIV self-testing experience, and heavy economic burden were associated with lower utility values.
CONCLUSIONS: Both instruments showed acceptable measurement properties among Chinese PLHIV, but systematic utility differences suggest they should not be used interchangeably. These findings support objective-driven instrument selection and may inform cost-utility analyses and utility-improvement targets.