Measurement Properties of the EQ-5D-5L and EQ-5D-3L in Six Commonly Diagnosed Cancers.

ABSTRACT

BACKGROUND: This study aimed to compare the measurement properties of the EQ-5D-3L (3L) and EQ-5D-5L (5L) in cancer patients. METHODS: A consecutive sample of inpatients with lung, breast, colorectal, liver, gastric, or thyroid cancer were interviewed using the 3L, 5L, and Functional Assessment of Cancer Therapy-General (FACT-G) questionnaires, and a subgroup was invited to complete the 3L and 5L again. Kappa and intraclass correlation coefficient were used to assess test-retest reliability, and Spearman's correlation between the EQ-5D and FACT-G was evaluated to assess convergent validity. Comparison of subgroups defined using Eastern Cooperative Oncology Group status and cancer stage were performed to assess known-group validity and discriminatory power using the F-statistic and area under the receiver-operating characteristics curve. All analyses were also performed for each subgroup of cancer patients. RESULTS: A total of 416 cancer patients completed the baseline questionnaire and 90 patients also completed the follow-up survey after 2 days. Ceiling effects were smaller in 5L (10.1%) than in 3L (17.8%). The test-retest reliability and convergent validity of the 5L were slightly better than those of the 3L. Both the 3L and 5L showed known-group validity; however, the 5L index showed better discriminatory power. Similar trends were found in the six types of cancers. CONCLUSION: In general, 5L appears to have better measurement properties than 3L for measuring the health-related quality of life of cancer patients. While both the 3L and 5L are suitable, 5L should be preferable to 3L for use in cancer outcomes research.