Premium
Comparison of novel methods in two‐way enriched clinical trial design
Author(s) -
Liu Yuyin,
Rybin Denis,
Heeren Timothy C.,
Doros Gheorghe
Publication year - 2019
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.8288
Subject(s) - placebo , type i and type ii errors , sample size determination , cluster analysis , clinical trial , repeated measures design , computer science , statistics , randomized controlled trial , binary data , binary number , medicine , mathematics , artificial intelligence , alternative medicine , arithmetic , pathology
Two‐way enriched design (TED) is a novel approach addressing placebo response in clinical trials. It is a two‐stage, randomized, placebo‐controlled trial design with enrichment in placebo non‐responders and treatment responders at the second stage. All data from the first stage and data from placebo non‐responders and treatment responders in the second stage are used for the final analysis of the treatment effect. The existing methods for the analysis of TED data include score tests with one, two, and three degrees of freedom. All these methods are only applicable to binary outcomes. However, there is an interest in continuous outcomes in clinical trials in psychiatry. In this manuscript, we apply some novel methods, including a repeated measures model, a weighted repeated measures model with weights from propensity score, and weights from K‐means clustering, to analyze TED data for both binary outcomes and continuous outcomes. The simulation study indicates that the repeated measures model performs consistently well in preserving the type I error and achieving the minimum mean standard error as well as a higher power. The performance of the weighted repeated measures model with weights from K‐means clustering improves with increasing sample size. Investigators can choose from these analytic approaches under different scenarios.