Valuations of SF-6D health states: Could United Kingdom data be used to select a smaller sample of Hong Kong states

Abstract

Background: There is interest in the extent of using the combined analysis when we have plenty of data on two countries, but few have considered the combined analysis when we have plenty of data on one country and limited data on another. This kind of analysis may produce better estimation of the second country’s population utility function than analysing its data separately.

Methods: The data set is the HK and UK SF-6D valuation studies where two samples of 197 and 249 states defined by the SF-6D were valued by representative samples of the HK and UK general populations respectively, both using the standard gamble technique. We apply a nonparameteric Bayesian method to estimate a utility function applicable across both countries to see whether with a small sample of HK health states, but also drawing extra information from the UK data, we obtain the same accuracy we would get with the full HK sample.

Results: The results suggest that with the use of 100 health states, the HK analysis gets the broad features as the full analysis with all states as far as predicted mean valuations, covariates, interactions, regression parameters and dimension-specific parameters are concerned.

Conclusions: The implications of these results will be hugely important in countries without the same capacity to run large evaluation exercises.