Zhou Huali, Liu Xianxi, Bao Rong, Qiu Liping, Zhang Yuhan, Gu Qiong, Yang Qing
Department of Gastric Surgery, Sichuan Clinical Research Center for Cancer, Sichuan Cancer Hospital & Institute, Sichuan Cancer Center, University of Electronic Science and Technology of China, Chengdu, China.
School of Nursing, Chengdu Medical College, Chengdu, 610500, China.
Qual Life Res. 2025 Sep 13. doi: 10.1007/s11136-025-04063-1.
This study aims to develop a mapping model from the EORTC QLQ-C30 and QLQ-STO22 to the EQ-5D-5L.
Data were collected from 1059 gastric cancer patients in mainland China. The conceptual overlap between the EORTC QLQ-C30/QLQ-STO22 and EQ-5D-5L was assessed using the Pearson correlation coefficient. Four regression models were applied to estimate the algorithm: ordinary least squares (OLS), Tobit regression (Tobit), ordered probit regression (Oprobit), and beta mixture regression (Betamix). The independent variables in the models comprised the scale dimension scores of the EORTC QLQ-C30 and QLQ-STO22, squared terms of relevant dimension scores, age, and gender. The generalizability of the models was assessed using random sample validation and five-fold cross-validation. Model performance was evaluated using four primary metrics: root mean squared error (RMSE), mean absolute error (MAE), intraclass correlation coefficient (ICC), and absolute error (AE).
The mean health utility value of the EQ-5D-5L was 0.853 (SD = 0.240). The Oprobit3 demonstrated the best performance among all evaluated models, with RMSE = 0.197, MAE = 0.111, AE > 0.05 (%) = 56.84, AE > 0.1 (%) = 28.61, and ICC = 0.703. The predictors included all dimensions of the EORTC QLQ-C30 and QLQ-STO22 questionnaires, as well as age and gender.
The developed algorithm enables researchers to estimate EQ-5D-5L health utilities based on EORTC QLQ-C30 and QLQ-STO22 scores. This approach facilitates cost-utility analyses in gastric cancer patients when EQ-5D-5L data are unavailable.
本研究旨在开发一种从欧洲癌症研究与治疗组织生活质量问卷核心30项（EORTC QLQ-C30）和胃癌特异模块22项问卷（QLQ-STO22）到欧洲五维度健康量表5级版本（EQ-5D-5L）的映射模型。
收集了中国大陆1059例胃癌患者的数据。使用Pearson相关系数评估EORTC QLQ-C30/QLQ-STO22与EQ-5D-5L之间的概念重叠。应用四种回归模型来估计算法：普通最小二乘法（OLS）、托比特回归（Tobit）、有序概率回归（Oprobit）和贝塔混合回归（Betamix）。模型中的自变量包括EORTC QLQ-C30和QLQ-STO22的量表维度得分、相关维度得分的平方项、年龄和性别。使用随机样本验证和五折交叉验证评估模型的可推广性。使用四个主要指标评估模型性能：均方根误差（RMSE）、平均绝对误差（MAE）、组内相关系数（ICC）和绝对误差（AE）。
EQ-5D-5L的平均健康效用值为0.853（标准差=0.240）。在所有评估模型中，Oprobit3表现最佳，RMSE=0.197，MAE=0.111，AE>0.05（%）=56.84，AE>0.1（%）=28.61，ICC=0.703。预测因素包括EORTC QLQ-C30和QLQ-STO22问卷的所有维度，以及年龄和性别。
所开发的算法使研究人员能够根据EORTC QLQ-C30和QLQ-STO22得分估计EQ-5D-5L健康效用。当没有EQ-5D-5L数据时，这种方法有助于对胃癌患者进行成本-效用分析。
