Inference for Clinical Trials That Rely on Historical Data: a Review of Statistical Approaches for Reducing Risk of Bias

• Activity Number: 617 - Indirect Comparisons of Treatment Effects for Clinical Regulatory and Health Economic Evaluations
• Type: Topic Contributed
• Date/Time: Thursday, August 2, 2018 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Lei Nie and James Signorovitch and Rajeev Ayyagari*
• Companies: Division of Biometrics V, office of Biostatistics, CDER/FDA and Analysis Group and Analysis Group
• Keywords: indirect comparison; historical trials; single arm trials; clinical trials

Abstract

According to the code of federal regulation , clinical investigations of a drug distinguish the effect of a drug from other influences through adequate and well-controlled clinical investigation. Randomized, double blind, placebo or active controlled trials are the gold standard. However, due to ethical concern or infeasibility, non-inferiority trials or single arm historical controlled trials are conducted. Inferences from these trials involve historical trials which could be substantially different from non-inferiority trials or single arm historical controlled trials, leading to potentially biased assessment therefore inadequate conclusions. In the presentation, we review statistical approaches developed to reduce the bias.