Adaptive Partial Drug Approval (WP-07-08)
Charles F. Manski
In the United States, the drug approval process of the Food and Drug Administration (FDA) is currently the main mechanism through which the government influences the production and dissemination of information on drug treatments. To obtain approval for a new drug, a pharmaceutical firm provides evidence on treatment response in randomized clinical trials that compare the new drug with an accepted treatment or a placebo. The FDA makes a binary approval decision after reviewing these trails' empirical findings. This paper brings welfare-economic and decision-theoretic thinking to bear on drug approval. Considering the matter from the minimax-regret perspective suggests an adaptive social planning process in which treatment with a new drug would vary—instead of being either fully allowed or denied as in current practice—as empirical evidence accumulates. The stronger the evidence on identified health outcomes, then the more the drug could be used. The adaptive process would improve on the current one by stimulating production of stronger information on treatment response and by reducing the welfare losses that arise from errors in approval decisions. Manski suggests a pragmatic version of the adaptive process that the FDA could implement.