Clinical trials of cost effectiveness in technology evaluation.

This article discusses models of efficacy, design of clinical trials and the role of mathematical modeling in diagnostic technology evaluation and determination of cost-effectiveness. A multi-tiered model of efficacy, which views diagnostic imaging as part of a global process of patient management and outcome, has been described. The first tier involves imaging efficacy, which must be determined by clinical trial. Direct comparison of new and established modalities in a single study population has major advantages over randomized controlled trials, which are extremely costly and time-consuming and are not appropriate for most evaluations of diagnostic modalities. Selection of patients for inclusion in the trial, interpretation and verification of results, and determination of a reference standard are all possible sources of bias, which need to be identified and controlled. Decision analysis modeling can be used to assess diagnostic, therapeutic, patient-outcome and cost efficacy, once imaging efficacy has been evaluated by clinical trial. Decision analysis is easier and less expensive to perform than clinical trials, and results are easily generalizable to other settings. Disadvantages arise from the nondescriptive nature of modeling and lack of transparency, which make it difficult to evaluate the appropriateness of decision tree structures and input data. Modeling is an unavoidable fact of life in technology evaluation, since the resources that would be required for full evaluation of imaging modalities by clinical trial are not available.