Clinical Trials

Biography

Biography: Wenle Zhao

Abstract

Abstract: Random treatment assignment provides the fundamental protection for the integrity of clinical trial results. Permuted block randomization and minimization method remain the most used randomization methods in clinical trial practice. In trials with unequal or response adaptive target allocations, multiple categorical or continuous baseline covariates, and large number of clinical sites, both stratified permuted block randomization and minimization are unable to meet the investigators’ needs.

Over the past few decades many advanced randomization methods have been published with superior statistical properties in treatment allocation imbalance control, baseline covariate balancing, and allocation randomness protection. However, these newer randomization methods have gained few applications in real trials, mainly due to implementation difficulties. Many interactive voice response system (IVRS) and interactive web response system (IWRS) packages and electronic data capture (EDC) systems like REDCap use pre-generated randomization lists, allowing stratified permuted block randomization only. Knowledge dissemination for implementing advanced randomization methods in clinical trial information systems is highly demanded.

The first part of this 90-minute education workshop reviews the statistical performances of stratified permuted randomization, minimization, as well as those newly published randomization methods, including a family of restricted randomization designs with maximum tolerated imbalance (MTI) and the minimal sufficient balance strategy. These randomization designs will be quantitatively compared based on the capacity of accurately achieving target allocation (two-arm or multi-arm equal or unequal allocations), effectively controlling imbalances in baseline covariates (few or many categorical or continuous covariates), and the allocation randomness measured by the proportion of deterministic assignments and correct guess probability.

 The second part of the workshop will discuss the implementation methods for these advanced randomization methods in the clinical trial information system, illustrated with real trial examples. The workshop will present a generic strategy to integrate the subject randomization program as a special case report form (CRF) into the trial’s EDC system in which eligibility and baseline covariates data are collected. Practical issues associated with subject randomization will be discussed, including the generation of randomization code in drug studies for treatment blinding and masking, randomization algorithm adjustment for site study drug availability, randomization algorithms for the burn-in period and fixed allocation period prior to the start of response adaptive randomization (RAR) phase, target allocation update in RAR trial with pre-specified time period length or pre-specified subject enrollment size, selection between balancing categorical covariates after dichotomizing vs. balancing continuous baseline covariates, handling of site operation errors in subject randomization, and emergency randomization in case technical glitches occur at site or the trial’s EDC system.       

Goal of Session: Attendees of this education workshop is expected to learn what are better alternatives to the stratified permuted block randomization and minimization, how to select a randomization method to better meet the requirement of their trial, how to implement the selected randomization algorithm into the EDC system, and how to deal with possible glitches in the randomization procedure.  The workshop is expected to be conducted in an interactive way. Attendees are encouraged to bring questions and cases for discussion. Knowledge and experiences in basic statistics and information system programming can help, but not required, to understand the contents of this workshop