The Value of Information in Benefit-Risk Analysis for Regulatory Approval
In 2006, the Institute of Medicine called for the Food and Drug Administration to take a more systematic approach in their drug approval deliberations. The current process has been described as non-transparent, subjective, inconsistent, and arbitrary. Several more quantitative methods have been proposed as candidates to address this concern, including stated preference assessment, multi-criteria decision analysis, and health outcomes modeling using the quality-adjusted life year (QALY) as the outcome metric. We focus here on this last methodology. Pharmacoeconomic models generally are based on Phase III pivotal efficacy trials and attempt to assess the potential cost-effectiveness of new drugs. In addition to assessing cost impacts, these models project lifetime impacts on quality and length of life based on bioclinical markers of intermediate, shorter term effects. However, these trials are underpowered to measure safety impacts, i.e., risk. Nonetheless, regulatory authorities use this same information in their efforts to assess pre-launch benefit-risk, but they do not explicitly combine the information in a mathematical model, as pharmacoeconomists aim to do. In principle, this could be done. This paper explores the challenges in and opportunities for developing mathematical models for assessing risk-benefit in regulatory deliberations regarding drug approval. Both theoretical and practical challenges are discussed, including weighing known vs. unknown risks, the role of risk management tools, the limitations of the QALY, the definition of a level playing field across drug classes and disease areas, the role of agency staff vs. manufacturers in preparing and reviewing models, feasibility within the advisory committee process, regulatory risk aversion, and the balance between pre- and post-launch information. As an example, using diabetes drugs in general and rosiglitazone in particular, these challenges are discussed. In December, 2008, the FDA issued a guidance raising the evidence requirements for diabetes drugs. Because of concerns about the cardiovascular risks of diabetes drugs, future trials are supposed to be larger and longer in duration. Using an existing model of the benefits and risks of rosiglitazone, the implications of the new regulations are illustrated. Several issues are explored, such as the value of this additional information in terms of reducing CV deaths, the costs of gathering the additional data, the forgone health benefits of further delay, and impact on prior and subsequent compounds in this class. Finally, the implications for powering trials are considered as well as the cost-effectiveness of pre- and post-launch data collection and implications of the public goods nature of this information.