OBJECTIVES : Within Europe, country-specific value sets for EQ-5D are increasingly being utilized in economic evaluations as more countries are initiating partnerships to subsidize reimbursement strategies, drug pricing, health policy making and interventions impacting health-related quality of life. In the absence of a national value set, investigators tend to use a value set from neighboring country as a proxy of the national-specific value set. The validity of such proxies can be increased if existing value sets from different countries were combined to develop a ‘pan-European’ value set. We aim to derive a methodology for estimating a ‘pan-European’ value set for the EQ-5D as a pragmatic solution for economic evaluations across Europe using EQ-5D-3L as an illustrative example.

METHODS : Using the coefficients from the published EQ-5D-3L time trade-off (TTO) valuation studies, a pooled data set from nine European countries was established: UK, Spain, Netherlands, Germany, Denmark, France, Italy, Portugal, and Poland. Different model specifications were generated using Generalized Estimating Equations (GEE) including interaction terms: N3, D1, I2, I2², I3, I3². Population weights for the included countries were applied in the analyses. Model selection was based on goodness-of-fit criteria:Root Mean Square Error(RMSE), pseudo-R² and Mean Absolute Error(MAE).

RESULTS : The pooled TTO-values from the regression coefficients of nine European valuation studies ranged from -0.711 - 1.000. The N3 and I3 models were the best-fitting models according to RMSE, MAE and pseudo R². The D1 model predicted the value 1.00 for perfect health, but the regression coefficients were two-fold higher than other models for all levels in all dimensions in the D1 model. The coefficient for moderate problems in usual activity dimension (UA2) were not significant in the model with no interaction terms, N3, I3 and I3² models.

CONCLUSIONS : In this analysis, the values from the N3 model best represent the preferences of the European population.