Managing Blinding Requirements in IRT: An Overview

18th January 2019 (Last Updated December 3rd, 2018 16:01)

Francesco Spoto, Clinical Project Manager, explores the opportunities of using IRT solutions to manage blinded information in clinical trials

Managing Blinding Requirements in IRT: An Overview

Global clinical trends show an increasing level of complexity in trial design, adding critical challenges on randomization and information blinding.

IRT solutions are best suited to ensure a safe management of the latter.

Why blinding is so important

In clinical trials, the act of blinding refers to keeping one of more category of participants unaware of the assigned intervention. Keeping certain study participants “blind” is important in helping reduce or eliminate biased results.

In fact, the international standards defined in the ICH guideline on Statistical Principles for Clinical Trials (E9) states that “the most important techniques to avoid bias in clinical trials are randomization and blinding.”

What to do to avoid introducing bias

Randomized Controlled Trials (RCTs) have steadily become the new standard in the clinical world as their design support bias risk reduction. In parallel to the growing popularity – and complexity – of RCTs, the use of Interactive Response Technology (IRT) platforms have sored too.

IRT became the main system in charge of managing patient randomization and drug allocation strategies (see here), hence had a spreading impact on guaranteeing the compliance to blinding requirements. The combination of the abovementioned trends suggests the importance of focusing on IRT as the ideal place to address the risk of accidental unblinding, with the aim of reducing it to limit the potential introduction of bias.

Unblinding risk as part of IRT design

Since IRT systems are designed to randomize patients and keep study participants blinded, it is paramount to assess unblinding risks as part of both study planning and live maintenance.

Starting in chronological order, the sponsor has to define – in the design phase of the study – at least the following:

These three elements are essential to thoroughly manage blinding requirements in IRT. It can become difficult to address them depending on the complexity of study design. For instance, the introduction of adaptive randomization may imply the closure of one or multiple treatment groups, variation of cohort capping, and other parameter variations based on interim study results. IRT programming shall include these features while ensuring their occurrence does not impact the compliance to blinding requirements.

Another factor of complexity can come from multiple blinding levels set on a single study. It can happen that certain roles are blinded to one factor, but not others (i.e.: blinded to dose level, but not treatment). In such cases, the main challenge is post-IRT implementation, when communications and escalation paths must be managed with extreme care to avoid causing an accidental unblinding to a specific role.

Finally yet importantly, special attention should be paid to the study support phase in IRT. Every communication, from screen sharing to coffee chat, which includes study parameters identified as unblinding during the study setup, must be carefully managed to avoid breaking the blinding.

Unblinding risks outside IRT

After discussing the setup of IRT solutions to address the blinding risk, it is time to look at potential sources of accidentally unblinding outside of it. The number one source of unblinding risks is in the setup of the clinical supply strategy, especially for what concerns partial unblinding. Partial unblinding can occur any time one patient can be dispensed medication normally while special arrangements are required for another patient at the site. Apart from trivial mistakes, partial stock-out is the best example to show how an unplanned issue on the supply chain can lead to unblinding.

Let’s examine the following scenario: a randomized blinded trial when patient A arrives at the site where one kit type is not available. The patient is set to receive that exact kit type at the current visit but dispensation fails in IRT due to kit unavailability. In the event patient B arrives at the site, but he is randomized into a different treatment, the blinded investigator can perform the dispensation in IRT and realize the two patients are assigned to different treatment arms. This is indeed risky as it may jeopardize study results due to an unwanted bias.

Other causes of accidental partial unblinding are incorrect site resupply, small shipments or by sharing the content of a kit by mistake.

Conclusion

A modern IRT solution has the capability to ensure an accurate setup and execution of complex blinded studies. The system can also help minimize allocation bias by making the process straightforward and reliable through leveraging IRT-based randomization. Reducing the risk of accidentally unblinding can be achieved by a robust setup of the study on IRT, focusing on user roles and potentially unblinding information hosted on the system.

However, after setting up the study on IRT, additional care should be used to address the challenges of a live study. Relying on an IRT expert can ensure the creation of clear study documentation. It’s important to manage communication without risking an accidental unblinding of a recipient. Finally, even a bulletproof IRT setup cannot help in case of a faulty clinical supply strategy or incorrectly trained study personnel.

One last suggestion: if in doubt, don’t send that email!

 

References:

  1. https://ind-dev-kgi-verdict-network.pantheonsite.io/news/supply-chain/maximize-randomization-trial-management-systems-clinical-studies/
  2. http://www.clinicalinformaticsnews.com/2018/06/08/avoid-unintentional-unblinding-in-clinical-trials.aspx
  3. https://www.sciencedirect.com/science/article/pii/S2451865415300089