Assessing the Ratio of Means as a Causal Estimand in Clinical Endpoint Bioequivalence Studies in the Presence of Intercurrent Events

• Activity Number: 507 - Non-Inferiority, Biosimilarity and Related Topics
• Type: Contributed
• Date/Time: Wednesday, July 31, 2019 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Yiyue Lou* and Michael P. Jones and Wanjie Sun
• Companies: University of Iowa College of Public Health and University of Iowa College of Public Health and FDA
• Keywords: Principal Stratification; Sensitivity Analysis; Ratio of Means; Bioequivalence; Intercurrent Events; Causal Inference

Abstract

In clinical endpoint bioequivalence (BE) studies, the ratio of means based on the observed per-protocol (PP) population (compliers and completers in general) is widely used in the primary analysis for assessing equivalence between a generic and an innovator product with a continuous endpoint. However, the occurrence of intercurrents, i.e., missingness and noncompliance, are not properly addressed in this PP analysis. The resulting estimand is not causal. In this paper, we propose an alternative estimand for the ratio of means using a principal stratification approach, which is one of the strategies recommended by the ICH E9 (R1) addendum. We identify three conditions under which the current PP estimand is equal to the proposed causal estimand, the "Survivor Average Causal Effect" (SACE) estimand. We propose a sensitivity analysis method to evaluate the robustness of the current PP estimator, which is the ratio of sample means in the observed PP population. Simulation studies demonstrate that the PP estimator is biased, and inflates Type 1 error and power when these three conditions are violated. The method is illustrated using data from a clinical study for acne vulgaris.