Economic issues in cost-effectiveness analysis alongside multinational clinical trials

This thesis is about conducting cost-effectiveness analysis when patient-level cost and outcome data are available in more than one country as part of a clinical trial. Economic evaluation alongside a multinational clinical trial poses interesting challenges to analysts. Health care systems are context-specific and as a result, estimates of incremental costs and effects can be influenced by the allocation and costs of resources within each country in the trial. This thesis uses the economic theory of production as a model to explain the behaviour of a potential provider of health care in each country participating in a hypothetical simulated multinational clinical trial. It is shown that the substitution effect of input factors measured by the elasticity of substitution plays a role in the estimation of country-specific incremental costs. Because heterogeneity in incremental costs and outcomes across countries may exist, this thesis explores different tests for heterogeneity and concludes that researchers should explore heterogeneity of treatment effect in the main categories of resource use in addition to costs and outcomes. The thesis recognises the limited power of these tests but suggests that they can be a potentially useful tool for selecting the appropriate method to generate country-specific results. The advantages of multilevel models over standard regression analysis for analysing and presenting the results from these studies are recognised within the thesis. However, analyses also demonstrate that the validity of these models depends crucially upon the number of countries and the sample size within each country. These are aspects that receive little attention in current published studies, and therefore this thesis calls for improvements to be made to the design stage of multinational clinical trials. The multinational nature of published cost-effectiveness analyses alongside multinational clinical trials is also reviewed and it is shown that the reporting of frequencies and country contribution in these studies is currently very poor. As a result, it is recommended that future guidelines should emphasise the importance of reporting the multinational nature of the study.