The Analysis of Multinational Cost-Effectiveness Data for Reimbursement Decisions

Here, we use the word ‘jurisdictions’ to mean circumstances where more than one decision maker exists within a country. A typical example is that of Canada where different provinces may impose different requirements for reimbursement submissions.

This is typically the case in multinational RCTs investigating the clinical effectiveness of drug therapies, where the treatment protocol is being rigidly followed in each treatment centre. There are a number of reasons why the absolute benefit of treatment may vary across countries participating in the trial. In particular, variation in baseline (i.e. control group) events may exist because of differences in demographics, case mix (severity), clinical practice and resource constraints,[8] leading to between-location variability in the observed clinical effectiveness and resource use. In principle, such differences may also generate important variability in the relative treatment effect of the new intervention relative to control. Although patients treated within the same hospital may share treatment protocols, these could differ between treatment centres and countries, leading to potential between-country variation in clinical outcome[9,10] and use of healthcare resources. Similarly, the availability of specific resources (e.g. beds in high-dependency units, outpatient facilities, etc.) will influence the length of stay in hospital, and possibly the clinical outcome.[11] Finally, the relative costs (or relative pricesfftariffs) of healthcare resources at the country level could influence the way these factors are combined in the process of care.[12]

Methods to transfer the study results to countries beyond the trial participants are outside the scope of this article, and although some of the statistical techniques discussed herein might prove to be useful to assess such a type of transferability, this is often more appropriately done using decision models.[2]

Researchers need to be aware of the role and limitations of the alternative analytical methods to conduct multinational assessment of cost-effectiveness data in order to use the most appropriate vehicle to fulfill the information requirements relevant for jurisdiction-specific decision-making activity. At the same time, decision makers need to be aware of the limitations of the simpler approaches to trial-based CEA, to be able to interpret the information provided to them by the research community.

In some RCTs, patients receive treatment from several healthcare professionals (at once or at different points in time); similarly, patients may attend different treatment centres during the course of their therapy. This feature of the data creates particular non-strictly hierarchical structures, known as multiple membership or cross-classified. Variability by location in the results of the analysis can be ascribed to one or more of these centres or healthcare professionals, and multilevel models can be used to analyse this type of data structure to quantify the extent of impact of contextual factors, such as treatment centre or healthcare professional.[16]