All Times EDT
Keywords: Abuse potential, randomized crossover, interpretation
Drug abuse potential is determined from multiple data sources and factors, and the human abuse potential (HAP) study is one component of this evaluation. HAP studies may be required for CNS-active new molecular entities (NME) or reformulations of existing products (eg, abuse-deterrent formulations [ADFs]); however, the goals of these studies differ. A HAP study with an NME is considered an exploratory safety study, in which the abuse potential of a new product is evaluated relative to a positive control and placebo. In these studies, it is unknown whether the product will differ from either comparator and the findings are used to inform FDA’s recommendation for drug scheduling. In the case of ADFs/other reformulations, the product is part of a drug class that is already known to have abuse potential; therefore, the question being asked is whether the product has less risk of abuse compared with the original formulation(s) and the overall goal is to inform product labeling.
Because the question being asked in each type of study is different, the corresponding statistical analysis must be different as well. Although final FDA guidances for each type of study outline specific recommendations on the data analysis, including the hypotheses and relevant pre-defined margins that must be tested, there are inherent complexities with each type of study that have become apparent since the guidances were released. The complexity of the study designs, including numerous measures, endpoints, and treatment comparisons, as well as data-related issues such as distributional violations and potential for outliers, can result in data that is challenging to interpret. This presentation will introduce some key concepts in the design of HAP studies, and discuss some of the inherent issues associated with hypothesis–testing, identification of relevant margins, and interpretation of these data from a non-statistician's perspective, including potential underlying causees of these issues.