JMIR Publications

ABSTRACT

Background:

Randomized controlled trials (RCTs) are the gold standard for evaluating medical interventions because randomization reduces bias and provides high-quality evidence. To further minimize confounding, participants are often stratified by prognostic variables before random assignment to treatment groups. As clinical trials are increasingly managed digitally, integrating randomization directly into hospital information systems and electronic data capture platforms can simplify trial workflows by eliminating the need for separate software or manual randomization lists. Furthermore, the growing prevalence of multicenter studies highlights the need for interoperable and flexible randomization services.

Objective:

Our objective was to provide a web service that handles the randomized assignments in clinical trials and integrates seamlessly into the existing digital infrastructure. By considering versatility in trail designs, we intended to ensure the capability to apply our service in various real-world clinical research trials.

Methods:

Different randomization algorithms are leveraged to ensure the quality of randomized assignments, such as restricted, blocked, and dynamic randomization. Evaluation was done by surveying clinicians who use RandIMI and its REDCap integration in their real-world clinical trials using the standardized System Usability Scale.

Results:

We present RandIMI, an open-source web-based randomization service designed for both standalone use and seamless integration into existing clinical research platforms via a REST API. Existing integrations include REDCap and the hospital information system ORBIS. RandIMI supports a wide range of trial designs, including multicenter studies, an unlimited number of treatment groups with configurable allocation ratios, and stratification by study site and categorical variables. Its flexible study model enables modifications during recruitment, allowing adaptation to evolving trial requirements. Comprehensive user management and an audit trail ensure data security and traceability. RandIMI further provides configurable static and dynamic randomization algorithms to support robust allocation while minimizing selection bias. RandIMI achieved a SUS score of 77.3 from a total of 24 participants, indicating good usability. To date, the system has been successfully deployed in 15 real-world trials demonstrating its applicability across diverse study designs. RandIMI is freely available as open-source software on GitHub: https://github.com/imi-ms/RandIMI.

Conclusions:

This study shows that the presented randomization service RandIMI is capable of supporting research studies by providing a robust and user-friendly method to conduct the randomization of participants. Its API and integrations into REDCap and ORBIS offer a simple yet powerful extension to the trial infrastructure. Although audit trails, deterministic assignments, and a recruitment history are provided, RandIMI is not certified for usage in the development and approval of medical products.