Workflows to automate covariate-adaptive randomization in REDCap via data entry triggers

Area of Science:

• Clinical Trials Methodology
• Health Informatics
• Biostatistics

Background:

• Covariate-adaptive randomization algorithms (CARAs) are effective in minimizing covariate imbalance in randomized controlled trials (RCTs).
• Limited integration of CARAs into electronic data capture systems like REDCap has hindered their widespread adoption.
• The all2GETHER study, an RCT focused on HIV prevention, sought to address this integration gap.

Purpose of the Study:

• To develop and implement a software pipeline for seamless integration of CARAs into REDCap.
• To automate the randomization process within the REDCap environment.
• To facilitate the use of advanced randomization techniques in clinical research.

Main Methods:

• A software pipeline was developed using PHP and R, leveraging REDCap's Data Entry Trigger functionality.
• Automated randomizations were triggered upon saving a specific REDCap form.
• Study personnel received automated notifications of randomization assignments.

Main Results:

• The implemented pipeline successfully reduced covariate imbalance in the all2GETHER study.
• Observed differences between study arms were minimal (Cohen's d = 0.003 for continuous variables, risk differences <2.4% for categorical/binary variables).
• The integration required minimal additional effort from study personnel.

Conclusions:

• The developed software pipeline effectively reduces covariate imbalance in RCTs conducted using REDCap.
• The pipeline is reproducible and offers a practical solution for other research studies.
• This integration enhances the efficiency and comparability of treatment arms in clinical trials.