Use of Randomised Controlled Trials for Producing Cost-Effectiveness Evidence: Potential Impact of Design Choices on Sample Size and Study Duration

Background: A number of approaches to conducting economic evaluations could be adopted. However, some decision makers have a preference for wholly stochastic cost-effectiveness analyses, particularly if the sampled data are derived from randomised controlled trials (RCTs). Formal requirements for cost-effectiveness evidence have heightened concerns in the pharmaceutical industry that development costs and times might be increased if formal requirements increase the number, duration or costs of RCTs. Whether this proves to be the case or not will depend upon the timing, nature and extent of the cost-effectiveness evidence required. Objective: To illustrate how different requirements for wholly stochastic cost-effectiveness evidence could have a significant impact on two of the major determinants of new drug development costs and times, namely RCT sample size and study duration. Design: Using data collected prospectively in a clinical evaluation, sample sizes were calculated for a number of hypothetical cost-effectiveness study design scenarios. The results were compared with a baseline clinical trial design. Results: The sample sizes required for the cost-effectiveness study scenarios were mostly larger than those for the baseline clinical trial design. Circumstances can be such that a wholly stochastic cost-effectiveness analysis might not be a practical proposition even though its clinical counterpart is. In such situations, alternative research methodologies would be required. For wholly stochastic cost-effectiveness analyses, the importance of prior specification of the different components of study design is emphasised. However, it is doubtful whether all the information necessary for doing this will typically be available when product registration trials are being designed. Conclusions: Formal requirements for wholly stochastic cost-effectiveness evidence based on the standard frequentist paradigm have the potential to increase the size, duration and number of RCTs significantly and hence the costs and timelines associated with new product development. Moreover, it is possible to envisage situations where such an approach would be impossible to adopt. Clearly, further research is required into the issue of how to appraise the economic consequences of alternative economic evaluation research strategies.