top of page

Randomization in clinical trials ensures unbiased assignment of participants to treatment groups. It minimizes confounding variables and increases statistical validity, enhancing the reliability of results. By randomly allocating interventions, researchers can assess the true effects of treatments, leading to evidence-based medical decisions and improved patient care.

To learn more on this topic I interviewed Paul Vanbosterhaut, Senior RTMS System Specialist, & Elisabeth Coart, Principal Biostatistician, Director of Consulting Services and Biomarker Development at IDDI.

Can you explain the importance of randomization in clinical trials and its role in ensuring unbiased and reliable results?

Randomization is the process of allocating one of the treatments under investigation to a patient, using a mechanism that involves some random element. Random allocation of patients guarantees that the treatment groups are comparable with respect to all known and, importantly, also unknown prognostic factors. In this way, randomization between treatment groups is the paramount statistical element that allows to claim that a study is unbiased. However, the treatment allocation procedure is also expected to achieve the planned allocation ratio and often prognostic balance is also aimed for. The latter is achieved when the observed distributions of prognostic factors are similar across the randomized treatment groups, like a similar age distribution in the treatment groups.

It is clear that planning and implementing the randomization algorithm should be handled with care: One Chance, No Regrets.

Many controversies exist about treatment allocation with minimization. In which type of trial or situation would you recommend this procedure?

These controversies were mainly caused by former regulatory guidance documents that incorrectly cautioned against minimization. It is true that minimization was sometimes not appropriately implemented by trialists, especially for trials with an unequal allocation ratio. At present, all theoretical properties of minimization are known, and the procedure is well accepted, both by FDA and EMA, if a random element is included (stochastic minimization).

At IDDI, we have implemented stochastic minimization for more than 50 trials, with 2 or more treatment arms, with equal and unequal randomization ratios, with a few and many prognostic factors for balancing, small and large trials, and in many different indications. For all trials, the overall imbalances (deviation from the allocation ratio) and marginal imbalances (imbalance within prognostic factor levels) were minor. This illustrates that minimization is a flexible procedure that can be adapted to most clinical settings. However, minimization is especially recommended for smaller trials in settings with many known prognostic factors. In contrast to e.g., stratified permuted block randomization, more prognostic factors can be included in the allocation procedure.

How do you ensure adequate trial supply management to support randomization in large-scale clinical trials with multiple study sites? What challenges might arise in this process and how do you address them?

RTSM solutions must include efficient and accurate control of patient randomization, inventory management and drug distribution to depots, sites or even directly to patients.

The choice of the randomization method can be a challenge on its own. For example, randomization designs can involve multiple strata combinations but a limited number of patients. Others require the implementation of dynamic allocation methods such as minimization for large-scale trials.

At IDDI we rely on an experienced statistical team that can advise on both the randomization method to be chosen and its implementation, supported by a flexible RTSM team and platform. Clinical trial supply management solutions ensure that sites have just enough inventory to treat subjects on time without having to waste unnecessary products.

If the study has many kit types and dosing regimens, and is spread across many sites globally, the ability to organize trial supply efficiently might become a real challenge.

A team that has extensive industry experience in RTSM to design, implement and support the execution of the study with the sponsor is key to the trial success as well as an RTSM solution that can predict and automate the ideal resupply processes considering the sites particularities (enrollment rates, shipping distances, labelling…) and IP requirements (temperature control, ancillary products, expiry management…), and control those centrally.

IDDI offers the bridge between the drug supply and the clinical operations teams, helping them find the optimal balance between their needs.

What strategies or technologies have you employed to optimize randomization and trial supply management processes, ensuring efficient allocation of investigational products while maintaining blinding and minimizing errors?

Today’s trials involve more patients, sites, countries, and more uncertainty than ever before.

The number and complexity of issues affecting randomization and supply chain trial management has also grown to include costly comparator drugs, intricate protocols, delicate investigative compounds, adaptive clinical trials, and variable dosing schemes. Added to the mix are increased regulation and country-specific approvals surrounding temperature-controlled drugs, packaging, shipping, and labelling.

At IDDI, we collaborate with our in-house biostatistical experts to determine the best randomization strategy for a specific trial. We then implement the proposed randomization method in our RTSM system that supports all types of randomization scenarios, including completely random allocation, (Stratified) Permuted block randomization and Minimization (Pocock & Simon (1975).

The RTSM system strictly controls sensitive information such as treatment arm and treatment assignments to maintain study blinding. Access to functions in the system is controlled based on user privileges, so only authorized users can have access to unblinded data in the system. Similarly, controls are put in place so that unblinded data can only be sent securely to the intended authorized recipients.

The RTSM system includes an emergency unblinding functionality. This can be setup so that Principal Investigators are able to unblind patients at their sites in case of an emergency. Often, that patient who was unblinded by the site is then automatically discontinued from the study, preventing further drug assignments. The RTSM system can also provide access to Safety Officers who can unblind any patient at any site without impacting their ongoing participation in the study.

Finally to minimize errors, at IDDI we use an independent biostatistics team that regularly investigate the randomization lists for inconsistencies such as the current allocation ratio and correctness of received treatment. This allows for in-time remediating actions while the study is still randomizing patients.

Could you share an example of a situation where randomization posed unique challenges in a clinical trial you were involved with? How did you overcome those challenges and what lessons did you learn?

For a Phase III study, two post-chemotherapy drugs were compared in a relatively large sample of patients with newly diagnosed FLT3-mutated acute myeloid leukemia, with randomization using a 1:1 minimization algorithm based on six stratification factors.

Although the randomization was nicely balanced during the initial review, 10 events occurred in 27 patients with an important biomarker that could potentially impact the outcome. In this subgroup, the treatment allocation was imbalanced.

The sponsor wanted to include this additional biomarker as a stratification factor, and the question was whether to replace one or add it as the seventh factor.

The decision was a joint effort between the sponsor, the IDDI biostatistics team and the IDDI RTSM team. The IDDI biostatisticians recommended adding a seventh factor, and the RTSM team imported the collected biomarker data for the already randomized patients. This meant that the updated randomization algorithm could consider all previous patients during future randomization, eliminating the need to start a new cohort, risking imbalance across the entire sample.

Although identified while the study was already underway, the sponsor was able to introduce an important biomarker into the stratification factors while maintaining the necessary overall balance. This was possible because the minimization was used as randomization procedure.

Although it is best to carefully plan the randomization process, if needed the RTSM system has the flexibility to accommodate mid-study changes to the stratification factors and can import additional data for the future randomizations.

Elisabeth Coart will be presenting on ‘Randomization: Handle with Care – One Chance, No Regrets’ at COG Europe, taking place on the 10th & 11th October 2023, in Milan, Italy. More information can be found here:

** Disclaimer: the views shared here are authors alone, and not of the Pharmaceutical Business Conference Group, Greater Gift or any other organisations mentioned **


bottom of page