Mapping EORTC-QLQ-C30 to EQ-5D-3L in patients with colorectal cancer

The primary aim of this study was to perform a mapping of the EORTC-QLQ-C30 scores to EQ-5D-3L for the SIRFLOX study; a large dataset of patients with previously untreated liver-only or liver-dominant metastatic colorectal cancer (mCRC). A secondary aim was to compare the predictive validity of existing mappings from EORTC-QLQ-C30 to EQ-5D-3L conducted in other cancers.

Questionnaires (completed within 529 patients) were used in a linear mixed regression to model EQ-5D-3L utility values (scored using the UK tariff) as a function of the five function scores, nine symptom scores, and the global score from the EORTC-QLQ-C30 questionnaire. A Tobit regression was also performed. The mean EQ-5D-3L values for the SIRFLOX trial were calculated and compared with predicted EQ-5D-3L values derived using published mapping algorithms.

The linear mixed regression model provided a satisfactory mapping between the EORTC-QLQ-C30 and the EQ-5D-3L, whilst the Tobit model did not perform as well. When utilities from the SIRFLOX data were calculated with previously published mapping studies, three out of five studies performed well (< 10% mean difference).

The main limitation of the study was the lack of meaningful observations post-progression (67 paired observations). For this reason, this study was unable to test whether the mapping holds by disease stage. Additionally, although the study adds to the literature of mappings to the EQ-5D-3L, it is not known how results would differ using the EQ-5D-5L.

This study is the first of its kind in liver-only or liver-dominant mCRC, and mCRC in general. The mapping constructed showed a good fit to the data and provides practitioners with an additional mapping between EORTC-QLQ-C30 to EQ-5D-3L using a large dataset (529 patients, 707 paired observations). The study also confirmed the generalizability of mappings published by Proskorovsky, Kontodimopoulos, and Longworth to liver-only or liver-dominant mCRC.