Countries without EQ-5D general-population surveys might use valuations from another country with similar health valuations. This study aimed to identify whether there are groups of countries with similar health-state valuations. METHODS: A cluster analysis was performed for the 242 states of the EQ-5D valuations for 23 studies in 18 countries, excluding the perfect-health state. Clusters were identified by means of the Ward algorithm with the Euclidean measure and hierarchical clustering. RESULTS: 5 clusters were identified. Valuations are sensitive to the preference elicitation methodology. Some countries in the same cluster have cultural similarities but a particular cluster included otherwise dissimilar countries and methodologies. Using the states associated to diabetes and breast cancer, the choice of cluster is shown to be relevant for cost-utility analysis. CONCLUSIONS: Health-state valuations tend to be clustered in a few groups of countries. The clusters found might be useful for performing cost-utility analysis in countries without such valuations.
