Are Preference-Weighted Cancer-Specific Measures More Sensitive Than Generic Measures? A Comparison of EQ-5D-5L, SF-6Dv2, FACT-8D, and QLU-C10D in Gastrointestinal Cancers.

OBJECTIVES

To compare the psychometric performance of generic EuroQol 5-Dimension 5-Level instrument (EQ-5D-5L), Short Form 6-Dimension version 2 (SF-6Dv2), cancer-specific Functional Assessment of Cancer Therapy Eight Dimension (FACT-8D) and the European Organization for Research and Treatment of Cancer Quality of Life Utility - Core 10 Dimensions (QLU-C10D) in gastric, pancreatic, and colorectal cancers.

METHODS

Patients recruited from a tertiary hospital in Shanghai, China, assessed their health using preference-weighted measures (PWMs) during 2 consecutive visits. Clinical data were extracted from medical records. Known-group validity, responsiveness, and agreement were assessed. PWMs' index scores were computed using Chinese value sets.

RESULTS

The analysis included 170 gastric, 142 pancreatic, and 207 colorectal cancer patients. Eastern Cooperative Oncology Group (ECOG) status of 1 was most common (82.4%-83.5%), and stage-IV predominated (60.0%-65.5%) across 3 cancers. EQ-5D-5L showed strongest known-group validity for ECOG status (Cohen's d: 1.87, 1.01, 1.73 for gastric, pancreatic, and colorectal cancers, respectively), followed by QLU-C10D (1.33, 1.22, 1.44), SF-6Dv2 (1.37, 1.00, 1.11), and FACT-8D (0.71, 0.72, 1.25). PWMs generally discriminated between early (0-III) and advanced (IV) stages; effect sizes were small or negligible, though SF-6Dv2 and FACT-8D failed for colorectal cancer. PWMs showed similar responsiveness to improvement across cancers, with standardized response means (SRMs) larger in pancreatic (range: 0.32 to 0.81) than colorectal (-0.07 to 0.45) or gastric cancers (0.00 to 0.51). Their responsiveness to deterioration varied by cancer type: in colorectal cancer, FACT-8D produced larger SRMs (-0.53 to -0.83) and QLU-C10D the greatest variation (-0.34 to -0.92); in gastric cancer, SF-6Dv2 had higher SRMs (-0.45 to -1.22) than others (-0.29 to -0.74); in pancreatic cancer, SRMs were comparable across PWMs. Agreement between PWMs was poor to moderate (intraclass correlation coefficient: 0.42 to 0.75).

CONCLUSIONS

This study suggests that cancer-specific PWMs are not necessarily more sensitive than generic PWMs. Psychometric performance of these PWMs varies by cancer type. Their suboptimal agreement limits interchangeability.