IRT/RTSM UAT: A Sponsor's Guide to Testing Your Randomization System Before Go-Live

Your IRT system will randomize patients, assign kits, protect the blind, and trigger every resupply shipment for the life of the trial. It is, functionally, the brain of your clinical supply chain.

And User Acceptance Testing is the last point at which you can find out it doesn't do what you assumed.

UAT is routinely under-resourced. It lands during study start-up when everyone is overloaded, it looks like a technical exercise best left to the vendor, and the temptation is to click through the happy path, confirm nothing errors, and sign off. Then the trial goes live and a site in a country you tested lightly gets shipped the wrong kit type, or the resupply trigger fires two weeks later than anyone expected.

Here's how to do UAT properly.

What is IRT/RTSM, and what is UAT?

IRT (Interactive Response Technology), also called RTSM (Randomization and Trial Supply Management), is the system that manages patient randomization and drug supply logistics during a trial. It assigns treatment, allocates specific kit numbers to patients, tracks inventory at each site and depot, and automatically triggers resupply shipments when stock falls below defined thresholds.

UAT — User Acceptance Testing — is the structured process of verifying, before go-live, that the configured system behaves the way your protocol requires. It is not the vendor testing that their software works; that's their internal validation. UAT is you confirming that the system, as configured for your study, does what your protocol says.

That distinction is the whole point. Most UAT failures are configuration failures, not software defects.

Why UAT matters more than it seems

The consequences of IRT errors are unusually severe because they compound silently.

• Randomization errors compromise data integrity. A misconfigured stratification or allocation ratio can undermine the statistical validity of the study — and may not be noticed for months.
• Unblinding risk. If the system reveals treatment assignment through kit numbering, screen display, or a report, the blind is broken.
• Supply failures. Wrong resupply thresholds cause stockouts at sites or floods of unnecessary shipments and waste.
• Regulatory exposure. IRT is a GxP system. Inspectors expect documented evidence that it was tested and fit for purpose before use.

None of these are recoverable cheaply once patients are enrolled.

What to test: the core areas

A thorough UAT covers these areas at minimum. Test each against the protocol, not against the specification document — specifications can contain the same misunderstanding the configuration does.

Randomization and allocation

Verify the randomization scheme, allocation ratio, stratification factors, and block structure behave correctly. Test that assignments distribute as the protocol requires across strata and sites.

Blinding integrity

Actively try to break the blind. Check screens, reports, emails, kit numbering, and data exports for anything that reveals treatment assignment to someone who shouldn't see it. This is where an adversarial mindset pays off.

Kit assignment and dispensing

Confirm the right kit types are assigned for the right visit, dose, and treatment arm — including dose titration and re-supply of replacement kits for damaged or lost product.

Supply thresholds and resupply triggers

Test that resupply fires at the correct inventory level, orders the correct quantity, and routes from the correct depot. Then test the edge cases: a site enrolling far faster than planned, a shipment delayed in transit, a depot running low.

Site and depot setup

Verify each country and site is configured with the correct depot, shipping lead times, temperature requirements, and local rules. Countries configured late or in bulk are a common source of errors.

Expiry management

Confirm the system handles expiry dates correctly — that it won't allocate kits expiring before the patient's next visit, and that expiry extensions propagate properly.

Emergency unblinding

Test the emergency unblinding workflow end to end, including who is notified and what is recorded. This must work perfectly the first time it's used in anger.

Notifications and reports

Check that alerts reach the right people and that inventory and enrollment reports reconcile against what the system actually did.

Who should be in the room

UAT done only by the vendor and one project manager will miss things. The people who should be testing are the ones who'll use the system: clinical supply managers who understand the resupply logic, clinical operations staff who know how sites actually behave, a QA representative who understands the compliance requirements, and — critically — someone who knows the protocol deeply enough to spot when the configuration reflects a misreading of it.

How to run it well: practical guidance

1. Write test scripts from the protocol. Derive scenarios from the source document, not from the vendor's specification.
2. Test the unhappy paths. Happy-path testing proves very little. Test dropouts, dose changes, mid-study amendments, lost shipments, sites enrolling out of sequence, and patients screened but not randomized.
3. Cover every country configuration. If you have twelve countries, don't test three and assume the rest are identical.
4. Document evidence as you go. Screenshots, expected vs. actual results, and a resolution trail for every issue. This documentation is your inspection evidence.
5. Don't accept "that's a training issue." If the system permits a wrong action easily, that's a design problem, not a user problem.
6. Re-test after fixes. A fix in one area can break another. Regression testing is not optional.
7. Leave real time for it. Compressing UAT to fit a slipping timeline is borrowing against the trial.