Prioritized and Non-Prioritized Net Benefit Statistics for the Assessment of Benefit-Risk

Abstract

For a given therapy, patients, health technology assessment agencies and regulatory agencies are interested in evaluating whether the benefit outweighs the associated risks. Existing work [1] uses prioritized composite outcomes to assess benefit and risk; this manuscript extends this work. The assessment of risk is based on a derived score obtained from predefined adverse events of interest. This derived safety score incorporates full aspects of adverse events of interest. After prioritizing the components of the composite outcomes consisting of benefit and risk outcomes, the net benefit or the Wilcoxon-Mann-Whitney statistic is used to assess the composite benefit-risk outcomes. If there is no prioritization, average net benefit (also called the O’Brien method) can be used to assess the treatment differences in the composite outcomes. Via simulation, we evaluate the characteristics of these measures of benefit-risk. An existing sample size derivation is extended to the case where there is no prioritization (average net benefit). Finally, an example motivated by a plaque psoriasis study is presented.