OBJECTIVES

Generic preference-based measures (GPBM), such as the EQ-5D, are used to derive utility scores. In oncology, GPBMs may lack sensitivity to aspects of quality of life (QoL). A solution is to derive a disease-specific PBM from a disease-specific measure. This study compared utilities generated by the EQ-5D-5L, EQ-5D-3L (cross-walked), EQ-5D visual analogue scale (VAS) and the QLU-C10D - a cancer-specific PBM.

METHODS

The EQ-5D-5L and EORTC QLQ-C30 were administered to people who had completed treatment for breast, prostate or bowel cancer in England. EQ-5D-derived utilities were calculated using: the UK tariff, EQ-5D-3L cross-walked value set and VAS. The QLU-C10D was derived from the EORTC QLQ-C30. Mean utilities were compared for the whole sample, by sex and by time since completing treatment (<8.5 months ago, ≥8.5 to 13 months ago, and >13 months to 23 months ago). The QLU-C10D domain scores were regressed onto the QLU-C10D to identify domains driving utilities.

RESULTS

N=1384 responders. The highest utility score was for the EQ-5D-5L (0.863), followed by the cross-walked value (0.801), and the VAS (0.787). The lowest score was for the QLU-C10D (0.773); significantly lower than both EQ-5D utilities. Standardised beta coefficients were highest for the QLU-C10D physical functioning and pain domains, but all domains were significant. Men scored significantly higher than women for all measures except the VAS. Only the QLU-C10D detected significant differences in utility by time since completing treatment: patients who completed ≥8.5 to 13 months ago had higher utilities than those who completed treatment more or less recently.

CONCLUSIONS

The disease specific QLU-C10D derived utilities was significantly lower than the EQ5D derived utilities, with the EQ5D-5L utility score exceeding UK general population norms. The EQ5D may overestimate utilities for post-treatment cancer patients compared with the QLU-C10D. Physical functioning and pain were the best predictors of QLU C10D utilities.