Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

PURPOSE: To determine the external validity of existing mapping algorithms for predicting EQ-5D-3L utility values from EORTC QLQ-C30 responses and to establish their generalizability in different types of cancer. METHODS: A main analysis (pooled) sample of 3560 observations (1727 patients) and two disease severity patient samples (496 and 93 patients) with repeated observations over time from Cancer 2015 were used to validate the existing algorithms. Errors were calculated between observed and predicted EQ-5D-3L utility values using a single pooled sample and ten pooled tumour type-specific samples. Predictive accuracy was assessed using mean absolute error (MAE) and standardized root-mean-squared error (RMSE). The association between observed and predicted EQ-5D utility values and other covariates across the distribution was tested using quantile regression. Quality-adjusted life years (QALYs) were calculated using observed and predicted values to test responsiveness. RESULTS: Ten 'preferred' mapping algorithms were identified. Two algorithms estimated via response mapping and ordinary least-squares regression using dummy variables performed well on number of validation criteria, including accurate prediction of the best and worst QLQ-C30 health states, predicted values within the EQ-5D tariff range, relatively small MAEs and RMSEs, and minimal differences between estimated QALYs. Comparison of predictive accuracy across ten tumour type-specific samples highlighted that algorithms are relatively insensitive to grouping by tumour type and affected more by differences in disease severity. CONCLUSIONS: Two of the 'preferred' mapping algorithms suggest more accurate predictions, but limitations exist. We recommend extensive scenario analyses if mapped utilities are used in cost-utility analyses.

Original publication





Qual Life Res

Publication Date





891 - 911


Cancer, Condition-specific non-preference-based measures, External validation, Generic preference-based measures, Mapping, Quality of life, Adolescent, Adult, Aged, Aged, 80 and over, Algorithms, Cost-Benefit Analysis, Female, Humans, Least-Squares Analysis, Male, Middle Aged, Neoplasms, Quality of Life, Quality-Adjusted Life Years, Surveys and Questionnaires, Young Adult