Analysis of health-related quality of life data from cancer clinical trials based on the item response theory.

摘要

The measurement of the health-related quality of life (HRQoL), which reflects patient's perception of his or her well-being, is now very frequent in cancer clinical trials (CCTs) using various types of questionnaires. By far, the approach based on the classical test theory (CTT) has been used most frequently to score these questionnaires and analyze the HRQoL data from these trials. Recently, a new approach, Item Response Theory (IRT), has been proposed to score and analyze HRQoL data with the expectation that it would be more sensitive than CTT.; The National Cancer Institute of Canada Clinical Trials Group (NCIC CTG) employed the EORTC QLQ-C30 core questionnaire and its LC13 lung-cancer module to measure HRQoL in two clinical trials (BR. 21 and BR. 18).; The purpose of this project is to explore the use of the scoring method based on IRT models in the analysis of the HRQoL data from BR.21, to compare and contrast the results of the analysis to those based on the CTT approach, and to explore whether the differential item functioning (DIF) exists in the EORTC QLQ-LC13 between non-small-cell lung-cancer (NSCLC) patients with different disease statuses (enrolled in respectively BR. 21 and BR. 18) based on parameter estimates from IRT models.; The IRT analysis of HRQoL data from BR.21 showed significant differences between two treatment arms in several domains and scales in both cross-sectional and longitudinal analyses. The comparison between CTT scores and IRT scores showed that they are highly correlated. In the cross-sectional analyses, the IRT approach is more sensitive than the CTT approach in less than half of the domains at some of the PPP assessment times. The patterns over time of the change scores are different between these two approaches. The longitudinal analysis showed the IRT approach had relative validity (RV) gains over the CTT approach demonstrating increased sensitivity in only a few domains. We did not find any DIF in the EORTC QLQ-LC13 with the exception of question 35 (dyspnea) between patients with different initial performance statuses and stages of disease.