Identity Security In Clinical Trial Supply: Who Has Access, And Why?
A Conversation Between Shubham Anandani And Elizabeth Urbanek, Executive Editor, Clinical Supply Leader
Clinical trial supply systems control decisions that can affect investigational product allocation, treatment blinding, and shipment release. Yet access security is not just a matter of preventing unauthorized logins. A person with valid credentials but outdated permissions can still make a change that disrupts supply operations or raises questions about data integrity and accountability.
As generative AI makes identity fraud more convincing, sponsors and their partners face added pressure to verify who is accessing critical systems, what they are authorized to do, and whether their actions can be traced. In this Q&A, Elizabeth Urbanek, executive editor of Clinical Supply Leader, speaks with Shubham Anandani, an engineering manager who builds digital identity and trust systems, about where identity and access risks intersect with clinical trial supply, how organizations can strengthen controls without slowing execution, and which workflows deserve attention first.
1. Clinical trial supply systems support activities such as randomization and investigational product allocation. Where do identity verification and access controls matter most in these workflows, and what could go wrong if they fail?
I'd look at the moments where one click changes what a patient receives or what the study knows. Randomization, unblinding, releasing or reallocating investigational product, and changing who has permission to do any of those things. Those actions are small on screen, but they carry a lot of weight. If the wrong person does them, a patient can get the wrong kit, a blind can be broken, and months later someone has to explain to a regulator why the data can still be trusted. A bad login is usually found quickly. A bad decision made with a valid login can sit unnoticed for a long time, and that's the case I worry about most.
2. Clinical supply operations involve sponsors, sites, depots, and other partners working across systems. How can organizations ensure that each person has the appropriate access for their role, particularly as responsibilities change throughout a trial?
Access should follow the person's current job, not what they were given on day one. In practice, that means connecting access to the source of truth: the delegation log at a site, the contract with a depot, HR and credentialing records on the sponsor side. When someone leaves or moves, the access should change without anyone having to remember to ask. I'd also give access an expiry date, so it has to be renewed instead of lingering forever. And I'd re-check identity and standing regularly, not only at onboarding. A license or a role that was valid in March may not be valid in October.
3. As generative AI makes it easier to create convincing fake identities, what risks could this introduce for clinical trial supply systems? What safeguards could help prevent an unauthorized person from accessing a system or attempting to alter a supply-related decision?
It makes the old social-engineering tricks cheaper and more convincing. A fake onboarding document, a cloned voice calling a help desk to reset a password, a polished email that looks like it came from a site coordinator. None of that needs a sophisticated attacker anymore. My view is that no single check should be treated as proof, because a face, a voice, or a document can now be faked well. What works better is layering. Verify the person properly at onboarding against an authoritative source, ask for extra proof only on the high-risk actions, and watch whether behavior matches the role over time. In my own research on professional identity, the pattern that kept holding was that trust has to come from several signals agreeing, not from one signal looking good.
4. When an action in a randomization or supply system needs to be investigated, what information should be available to establish who performed it, what they were authorized to do, and what changed? How can teams identify gaps in that traceability?
You want to be able to answer five questions without a scramble: who did it, what they were allowed to do at that moment, what exactly changed, when and from where, and why. "Who" has to be a real, individual identity, not a shared login or a generic admin account. "What they were allowed to do" is the one most teams miss, because they keep today's permissions but not the history of them. And "what changed" should show the before and after values. Regulations on electronic records, such as 21 CFR Part 11, already expect time-stamped audit trails. A simple test for gaps is to pick a real change from last month and try to reconstruct it in 15 minutes. Wherever you get stuck is where your gap is. In my early work on randomization software, the lesson was to make every decision traceable and every rule readable by an auditor, and that has stayed with me.
5. Stronger security controls can create additional steps for users who need to keep trials running. How can organizations strengthen identity verification and access controls without creating unnecessary delays for sites or clinical supply teams?
Don't treat every action the same. Reading a shipment status and releasing a shipment are not equally risky, so they shouldn't carry the same friction. Keep the routine steps quick, and save the stronger verification for the few actions where a mistake is expensive. Verify someone well once, then reuse that confidence instead of asking again for everything. When the system does stop someone, it should say why and send them to a fast human review, with a clear promise on how long it takes. If you ignore the review path, people find workarounds, and workarounds are worse than the original problem. I'd also measure the delay honestly in a pilot with one site before rolling anything out.
6. If a clinical supply organization wants to assess its identity and access risks over the next six to 12 months, what workflows or systems should it examine first? What practical steps should it prioritize, and which teams need to be involved?
I'd start small and with the highest stakes: unblinding, randomization overrides, shipment release and reallocation, and anyone with administrator or support access. Then look at third parties, because contractors and application programming interface (API) connections often have more access than anyone remembers. First steps are simple: list who can perform those actions, find shared or dormant accounts, and try reconstructing one recent change from the logs. Then pilot stronger checks in a single workflow. Someone has to own the problem, and the right group is a mix of clinical supply, quality, IT security, vendor management, and the site-facing teams. Meet on a regular schedule — quarterly is fine — so it doesn't turn into a one-time project.
About The Expert:
Shubham Anandani is an Engineering Manager (LinkedIn | Personal Website) who builds digital identity and trust systems and serves as secretary of the IEEE Seattle Joint Chapter. He began his career at Oracle Health Sciences building clinical trial randomization software, and later led engineering teams on identity and verification experiences used at scale. He holds a Master of Software Engineering from Carnegie Mellon University. His research on trust-aware professional identity, signal theory, and secure software architecture has been published at IEEE SoutheastCon 2026 and the IEEE World AI IoT Congress 2026, and presented at IEEE ISEC 2026. He spoke on securing Healthcare Identity in the age of AI at SecureWorld Healthcare 2026 and has judged several AI agent hackathons.