Melissa Peirsel is the Manager of IRT Services at Sharp and brings over 13 years of experience in clinical trial operations. Her career began in the United States Air Force, where she specialized in managing electronic technology systems. Today, she leads the successful delivery of IRT solutions from Sharp’s Bethlehem, Pennsylvania office, helping sponsors navigate the complexities of modern trial design with practical, system-driven insights.

What’s the first thing I should know before choosing an IRT or RTSM for my trial?

Melissa: The most important starting point is understanding your specific trial design demands from an IRT system. Too often, sponsors default to a single vendor for all their studies, regardless of indication, complexity, or phase, which can lead to misalignment between system capabilities and trial needs.

Think of it like buying a car: every IRT solution comes with core features, but also optional modules and customizations. You can customize heavily, but that often comes with added configuration effort, longer timelines, and future data maintenance challenges. Matching the platform’s strengths to your protocol upfront is what sets the trial up for success.  

Review how similar studies were supported by the vendor by asking for case studies, lessons learned, or even a system demo tailored to your protocol. Don’t be afraid to ask the hard questions: How much of this is truly configurable? What will require custom development? How does that impact study timelines and data integrity later on?

Do I choose bespoke or out of the box?

Out-of-the-box solutions are pre-validated and typically require minimal configuration. They’re efficient and cost-effective when your trial design aligns with the core system’s existing capabilities. Any configuration must operate within those predefined system parameters.

Bespoke solutions, on the other hand, involve developing custom code that extends the platform beyond its standard features. This approach is ideal when your protocol includes unique requirements that can’t be met with standard configurations.

Take oncology trials, for example—these often benefit from features like cyclic visit design, which provides flexibility beyond a fixed schedule of events. This is particularly useful when delays between treatment cycles are expected. Some vendors offer this as an out-of-the-box solution; others only as a bespoke enhancement. Knowing that distinction early can save both time and budget. 

Out-of-the-box IRT builds offer several key advantages: shorter timelines, reduced risk, and easier future changes, all supported by a validated framework. In contrast, bespoke solutions require deeper scrutiny during amendments. Each change must be assessed not only for its impact on custom code—potentially extending timelines and increasing validation workload—but also for how it affects existing study data. These added layers of complexity can significantly slow down the process and introduce additional risk.

Randomization methodology: Which one is right for me?

The most common randomization methodology is the permuted block design (PBD), in which subjects are randomized into a predetermined block. When completed, that block is guaranteed to be balanced. The PBD is a sound and true methodology that is utilized by most randomized controlled trials, but it has some limitations. These include block dependency, whereby balance is only guaranteed once the block is completed, covariate imbalance, where involving more covariates makes it harder to maintain statistical balance across arms, and selection bias.

Open-label trials are especially vulnerable to bias. For instance, if the first two subjects in a block are assigned to the investigational arm, site personnel may begin to predict future assignments. Even without confirmation, that perception alone can influence how subsequent subjects are screened and enrolled.

Blinding helps, but isn’t always feasible and even then, bias can sneak in. I once supported a trial where label bands differed in hue between placebo and active product under certain lighting—another subtle way that site staff could deduce treatment. 

An alternative worth serious consideration is covariate adaptive randomization (CAR). CAR aims to minimize imbalance across predefined covariates and is especially helpful in complex studies with multiple subject characteristics. While many assume CAR requires explicit stratification factors, that’s not always the case. Some IRT platforms can implement CAR logic without them. If needed, stratifying by site is often a simple and effective option to preserve balance both within and across sites.

Understanding source data: How can I reduce reconciliations?

One common source of reconciliation headaches in clinical trials is the misinterpretation of dispensation dates—specifically, the assumption that the date recorded in an IRT system represents the actual moment a drug was dispensed to a subject. In reality, the IRT system logs the date when material is assigned or allocated to a subject, not when it is physically handed over at the site.

In many trial designs, especially blinded studies, materials are assigned in advance so pharmacists have enough time to prepare for dispensation. However, when the IRT assignment date is sent to the EDC and treated as the dispensation date, discrepancies arise. Data management teams or CROs then compare site records to the EDC and raise queries when the dates don’t match—leading to avoidable change requests, unnecessary database updates, and added burden on site staff and IRT providers.

The root issue is a lack of distinction between two valid but different data points:

  • IRT is the system of record for assignment/allocation dates
  • EDC is the system of record for actual visit or dispensation dates

The best solution is to treat each system as the source of truth for the data it’s designed to capture. If both dates are needed in either system, they should be labeled and handled as separate fields. For example, drug accountability logs in the EDC could include both the IRT assignment date (captured automatically) and the actual dispensation date (entered by site staff).

Designing trials with this distinction in mind will reduce unnecessary reconciliation effort, minimize confusion, and help prevent the cascade of manual corrections that frustrate sites and delay data lock.

To learn more about best IRT practices for streamlined processes, improved visibility, and faster trials, as well as modern approaches to reducing reconciliation burden in IRT-EDC integrations, why not register for the webinar below?