Improving precision in cost-effectiveness analysis using copulas.

A copula is best described, as in Joe (1997), as a multivariate distribution function that is used to bind each marginal distribution function to form the joint. The copula parameterises the dependence between the margins, while the parameters of each marginal distribution function can be estimated separately. This is a brief introduction to copulas and multivariate dependence issues within a health economics context. The research presented here will make its own contributions to the development of copulas as a methodology, but more importantly will make deliberate inroads into health economic applications of copulas. To do this, common analytic problems faced by health economists are considered. Some of the differences between the copula methodology and existing alternatives are discussed, and a generalisable, systematic approach to estimation is provided.