Fang Haoran, Hong Tianqi, Liu Xinran, Luo Chang, Hou Yuanyuan, Xie Shitong
School of Pharmaceutical Science and Technology, Tianjin University, Tianjin, China.
School of Biomedical Engineering, McMaster University, Hamilton, ON, Canada.
Health Qual Life Outcomes. 2025 Apr 30;23(1):46. doi: 10.1186/s12955-025-02371-1.
The Audit of Diabetes-Dependent Quality of Life (ADDQoL) is a widely used instrument for assessing quality of life in Type 2 Diabetes Mellitus (T2DM). However, it does not directly yield health utility values essential for economic evaluations. This study developed mapping algorithms to predict EQ-5D-5L and SF-6Dv2 utility values from ADDQoL scores in T2DM patients in China.
Cross-sectional data from 800 T2DM patients in China, stratified by age, sex, and geographical region, were divided into development (80%) and validation (20%) groups. Pearson correlation analyses were conducted to assess the conceptual overlap between ADDQoL and the EQ-5D-5L and SF-6Dv2. Six predictor sets and six regression methods were explored to map ADDQoL scores to EQ-5D-5L and SF-6Dv2 utility values, respectively. Model performance was evaluated using mean absolute error (MAE), root mean square error (RMSE), and intraclass correlation coefficient (ICC).
For the development group, the mean (SD) ADDQoL Average Weighted Impact (AWI) score was - 2.426 (1.052), and the mean (SD) utility values for EQ-5D-5L and SF-6Dv2 were 0.928 (0.092) and 0.791 (0.133), respectively. Among all 36 alternative mapping models each for EQ-5D-5L and SF-6Dv2, the best performance was consistently observed in the two-part models that included the ADDQoL AWI, the first overview item, and their squared terms. For the algorithm mapping to EQ-5D-5L utility values, it achieved a MAE of 0.067, a RMSE of 0.095, and an ICC of 0.414; For the algorithm mapping to SF-6Dv2 utility values, the corresponding metrics were an MAE of 0.099, an RMSE of 0.120, and an ICC of 0.517.
This study provides a mapping framework to estimate EQ-5D-5L and SF-6Dv2 utility values from ADDQoL scores. These algorithms could be used to support economic evaluations, specifically tailored for Chinese T2DM populations.
糖尿病相关生活质量审计量表（ADDQoL）是一种广泛用于评估2型糖尿病（T2DM）患者生活质量的工具。然而，它并不能直接得出经济评估所需的健康效用值。本研究开发了映射算法，以根据中国T2DM患者的ADDQoL评分预测EQ-5D-5L和SF-6Dv2效用值。
来自中国800名T2DM患者的横断面数据，按年龄、性别和地理区域分层，分为开发组（80%）和验证组（20%）。进行Pearson相关分析，以评估ADDQoL与EQ-5D-5L和SF-6Dv2之间的概念重叠。探索了六个预测变量集和六种回归方法，分别将ADDQoL评分映射到EQ-5D-5L和SF-6Dv2效用值。使用平均绝对误差（MAE）、均方根误差（RMSE）和组内相关系数（ICC）评估模型性能。
对于开发组，ADDQoL平均加权影响（AWI）评分的平均值（标准差）为-2.426（1.052），EQ-5D-5L和SF-6Dv2的效用值平均值（标准差）分别为0.928（0.092）和0.791（0.133）。在分别用于EQ-5D-5L和SF-6Dv2的所有36个替代映射模型中，始终在包含ADDQoL AWI、第一个概述项目及其平方项的两部分模型中观察到最佳性能。对于映射到EQ-5D-5L效用值的算法，其MAE为0.067，RMSE为0.095，ICC为0.414；对于映射到SF-6Dv2效用值的算法，相应指标为MAE为0.099，RMSE为0.120，ICC为0.517。
本研究提供了一个映射框架，用于根据ADDQoL评分估计EQ-5D-5L和SF-6Dv2效用值。这些算法可用于支持经济评估，特别是为中国T2DM人群量身定制的评估。
