Estimating an EQ-5D population value set: the case of Japan

Quality adjustment weights for quality-adjusted life years (QALYs) are available with the EQ-5D Instrument, which are based on a survey that quantified the preferences of the British public. How ever, the extent to which this British value set is applicable to other, especially non-European, countries is yet unclear. The objectives of this study are (a) to compare the valuations obtained in Japan and Britain, and (b) to explore a local Japanese value set. A diminished Study design is employed, where 17 hypothetical EQ-5D health states are evaluated as opposed to 42 in the British study. The official Japanese version of the instrument and the Time Trade-Off method are used to interview 543 members of the public. The results are: firstly, the evaluations obtained in Japan and those from Britain differ by 0.24 on average on a [-1, +1] scale, and mean absolute error (MAE) in predicting the Japanese preferences with the British value set is 0.23. Secondly, comparable regressions suggest that the two peoples have systematically different preference structures (p<0.001 for 8 of 12 coefficients; F-test). Thirdly, using alternative models, the predictions are improved so that the local Japanese value set achieves MAE in the order of 0.01. Copyright (C) 2002 John Wiley Sons, Ltd.