My Worst Nightmare in Randomization and Trial Supply Management: Accidental Unblinding
This blog was written by Sheri Lillis, RTSM Solution Services Principal at Medidata.
What keeps you up at night? Or prevents you from falling back asleep?
As a Medidata Rave RTSM subject matter expert (SME), there is one nightmare that can keep me up at night, and it can take several forms:
Unblinding a patient in error during a troubleshooting exercise with the study monitor by screen sharing patient details that include the study drug assignment.
Or accidentally unblinding part of a study’s inventory by sending a screenshot to the project manager of numbered inventory on site filtered for only one kit type.
Or worse yet, accidentally sending a complete patient list report containing all their unblinded treatment assignments to the blinded study team and unblinding the entire study!
It’s accidental unblinding!
Fortunately, this is a healthy fear as it means I understand the inherent risks of the work we do in the RTSM (Randomization and Trial Supply Management), or IRT (Interactive Response Technologies), space of Clinical Trials. I work with my team to do everything in our power to mitigate unblinding potential whenever possible. This blog is meant to share some of these risks with you and what we do to avoid unintentional unblinding.
Beginning with the basics, blinding is an approach in which one or more parties in a clinical trial are unaware of patients’ treatment arm assignments to minimize any potential for bias during study execution. Normally, in a double-blind study, the blind is preserved until all study procedures have been completed and patient data is cleaned, locked, and authorized to be unblinded.
Unblinding occurs when an investigator, patient, or study team member accesses information that, either directly or indirectly, identifies a patient’s treatment assignment. Unintentional unblinding can result in the removal of a study team member from the study, the exclusion of a patient’s data from final analysis, or even invalidate the outcome of an entire study. Study managers, supply managers, biostatisticians, study designers, and clinicians take great effort to protect the blind and ensure that accidental unblinding does not occur.
How Does Accidental Unblinding Happen in Clinical Trials?
Accidental unblinding is rare, but it can happen at any time. Below are a few examples of how unblinding happens and how to best limit risk with your randomization and trial supply management (RTSM) system and procedures.
Study Design and Operational Issues
The study has two treatment arms—active and placebo. Study drug is supplied as unnumbered inventory rather than uniquely numbered items. The RTSM system clearly identifies inventory as either active or placebo.
To protect against accidental unblinding, the RTSM system must not display inventory types to blinded users. Depending on how the drug is labeled and identified, lot number and/or expiry dates may also need to be hidden.
The study has an investigational arm and a comparator arm. The investigational product and the comparator product are packaged differently, making it easy for the site staff and study monitors to identify a difference. This could be due to factors such as size, temperature controls, unique contents within the kits, or existing labeling.
Ideally, this would be considered and addressed during packaging. But if this is not feasible, then it must be handled procedurally. Receipt of the inventory at site, including confirmation within the RTSM system, preparation of the treatment for patient administration, and supply accountability and return processes must be performed by unblinded site staff and monitored by unblinded study personnel.
The study’s randomization scheme is 1:1 active:placebo and stratified by site. Study documentation specifies a randomization block size of 2. If a patient has an adverse event and is unblinded for safety reasons, the treatment arm of the other patient randomized in the same block is also revealed.
It is best practice to double the minimum size of the randomization block required for balance, especially for studies with only two or three treatment arms. Using multiple block sizes and distributing the larger blocks randomly within the randomization list can also protect against this unblinding possibility.
The study design dictates that different quantities of items are assigned to patients based upon the treatment arm. As a result, some patients are assigned 1 item at randomization while other patients are assigned 2 items, for example. This may occur in situations such as site-supplied placebo or comparator product, combination therapies, or dose packaging limitations.
A double-dummy design would be the best way to address this issue. If this is not feasible then information about the items assigned and dispensed to a patient must be hidden from blinded study personnel, even when inventory is packaged and numbered in a blinded fashion. This includes also hiding or separating such information in the EDC database. Preparation and administration of treatment should not be performed by study staff who are collecting outcome data. If the patient is aware of any difference in the treatments between one arm and the other, the patient should be instructed not to discuss it with any blinded study personnel.
Patients move through two phases in the study. The first phase uses a single inventory type. The second phase uses the same inventory type plus two additional inventory types. If some patients in the second phase are assigned inventory items that have been at site since the first phase, and other patients are assigned newly received inventory items, then site staff may speculate that these patients are on different treatments.
In this circumstance RTSM can be configured to differentiate items by phase even though the contents may be the same. Alternatively, items for the second phase can be partially seeded during the first phase to prevent the potential partial unblinding scenario.
Clinical Supply Management Plan and Site Resupply Issues
The study is blinded and inventory is uniquely numbered. The supply plan strategy triggers a resupply shipment for a site upon assignment of a single inventory item to a patient. The new shipment contains a single item. The site notices that the newly received item is assigned to the next randomized patient and that the existing item in inventory was not used. For example, patient 102 is assigned kit 2345. The new order contains only item 5678. The newly received item 5678 is assigned to patient 103. The site deduces that these two patients, 102 and 103, are randomized to the same treatment arm. This is an example of partial unblinding.
Partial unblinding occurs when blinded staff are aware that two or more patients are on the same treatment arm but are unable to identify the specific treatment arm. This can lead to full unblinding of all impacted patients should any of these patients become unblinded. Partial unblinding also occurs when blinded staff know that one or more patients are not on a specific arm but are unable to identify the specific treatment arm. To mitigate this risk, RTSM supply plans can be configured to include a decoy item in shipments when only 1 item or 1 item type are needed. Other resupply strategies can also be used such as sending larger initial shipments or triggering resupply when only 2 or more items have been used.
The study has limited supplies of one inventory type. This can occur for numerous reasons including an expensive or difficult to obtain comparator product, minimal shelf-life or short expiry dates, or limitations with manufacturing. The site realizes that an error is triggered when attempting to assign study drug at dispensing visits for a few patients. However, there are inventory items at the site. The site staff and study monitor suspect that these patients are on the same treatment arm.
To address this issue, the clinical supply management plan should be adjusted to ensure that adequate supply is maintained for all potential upcoming patient visits at the site. If this is not practical given the inventory limitations, then it may be necessary to hide inventory information from blinded site users and utilize unblinded staff.
On a weekly study management call with both blinded and unblinded team members, the clinical supplies manager shares unblinding information with the team.
As soon as someone realizes this is occurring, they should speak up and ask the unblinded user to stop sharing the information. When unblinding information has been shared inadvertently, the study team should reach out to their Quality Assurance or Compliance liaison to determine any next steps to protect the integrity of the clinical trial and the study data. It is possible that some study staff may need to be removed and replaced.
An unblinded study monitor initiates an email conversation with other unblinded team members to discuss how to handle the misallocation of the wrong item to a patient. As the conversation progresses, it is determined that more information is needed to determine next steps. The blinded project manager and blinded biostatistician are added into the email chain.
Whenever someone notices that unblinding information has been provided to blinded personnel via email, immediately instruct the blinded personnel not to view or forward the unblinding information, to immediately delete the email, and to confirm once this is complete. The study team’s Quality Assurance or Compliance group should be consulted for any additional actions including the potential need to remove and replace study staff. Labeling unblinded emails appropriately and double-checking distribution lists can also help limit accidental distribution of sensitive data.
A site user is accidentally provided an unblinded role instead of a blinded role.
As soon as the mistake is known, access should be removed, and the study’s Quality Assurance or Compliance group should be engaged. If the user never accessed the system while improper access was granted and objective evidence can be provided, then correcting the user’s access may be sufficient. Including secondary review steps when submitting and processing user account requests can help prevent this error.
The examples above show how a lack of experience and awareness can jeopardize a study and even invalidate the entire trial! A seasoned expert can identify and mitigate unblinding risks during study build and Go Live activities. Including an RTSM SME as early as possible during study planning can help monitor for potential unblinding scenarios that are best addressed during protocol finalization, packaging and labeling, and other activities that occur before RTSM implementation. An experienced RTSM services team will use formal Standard Operational Procedures, Work Instructions, and best practices that are thoroughly vetted to ensure quality delivery in all aspects of an RTSM system and proper study conduct.
How do we do it?
- We routinely evaluate, update, and implement processes to capture and monitor risks
- We document blinded vs. unblinded users to enable proper information flow
- We review protocols and any protocol amendments for potential risks as part of our development and study management process
At Medidata, our RTSM Professional Services team (averaging over 12 years of experience per team member) works diligently to identify potential unblinding risks and ensure sensitive information is not shared with blinded study team members from study start to database lock. In addition, the Rave RTSM product contains built-in checks and balances and study configuration settings that prevent blinded users from accessing unblinded data in RTSM. This combined approach ensures success for our customers and mitigates the risks inherent in executing blinded clinical trials.
Learn more about Rave RTSM and our Professional Services capabilities.
Follow Kevin Collier on LinkedIn for the Experience Matters in RTSM Series.