Sensitivity Analysis for Clinical Trials with Missing Outcome Data Using Repeated Measures: A Simulation Study Design

• Activity Number: 121
• Type: Topic Contributed
• Date/Time: Monday, August 10, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Section on Medical Devices and Diagnostics
• Author(s): Terry Liao* and Ying Yang
• Companies: Boston Scientific and FDA
• Keywords: sensitivity analysis ; missing outcome data ; repeated measures

Abstract

Missing outcome data are unavoidable in almost all medical device and pharmaceutical trials. The choices of sensitivity analyses should depend on the missing data pattern. In a recent repeated measures design consideration for a pilot study, the primary interest is the treatment effect between a new device and a masked procedure and the outcomes are the blood pressure change from baseline in two consequent follow-up visits. The missing outcome data are expected to come from two reasons: an ignorable small portion of true missing at random and the "pre-specified exclusion" due to medication rescue plan. To handle the above-described specific missing data, a few study designs including different statistical testing methods and missing data imputation will be provided using a simulation dataset.