The Risks And Unresolved Rules Of Real-Time Clinical Trials

By John Oncea, Chief Editor, Clinical Tech Leader

The FDA’s real-time clinical trials initiative has been announced. Two proof-of-concept studies are running. An RFI is open for a summer pilot. The timeline is real.

What is not yet real is the regulatory framework that will govern everything the initiative implies. The legal and operational questions raised by RTCT, about sponsor accountability, blinded study integrity, safety reporting obligations, the relationship between FDA oversight and DSMB governance, and liability for signal detection timing, are not yet clearly defined in public guidance. The pilot is being designed, in part, to answer them. That gap between an ambitious initiative and a resolved framework carries concrete risks for sponsors, CROs, and sites that move forward without appreciating them.

It also carries real upside. Both sides of the ledger deserve honest consideration.

This is Article 3 of a three-part series. Article 1 explained what the FDA announced and why. Article 2 examined the technology and operational infrastructure RTCT demands.

What’s Not Changing, And Why That Matters

Before addressing what is unresolved, it is worth being clear about what has not changed. Existing GCP requirements, ICH E6 standards, randomized controlled trial standards, blinding requirements, and DSMB governance structures all remain in effect. The FDA has not proposed replacing formal meetings, submission processes, or independent safety oversight structures. RTCT is a signal-layer overlay on existing trial infrastructure, not a redesign of regulatory authority.

Tala Fakhouri, who led development of the FDA’s AI policy before joining Parexel as a regulatory strategy executive, put it directly: “Regulations have not changed. ICH E6, GCP requirements, RCT standards, and blinding requirements all still apply. The purpose of the pilots is to figure out how to maintain the integrity of the trial while doing things faster.” That framing is important context for every question that follows.

The Regulatory Questions The Industry Hasn’t Answered

Fakhouri describes a set of questions that every organization operating in the RTCT space should be asking, and that the FDA has not yet formally resolved. Will RTCTs be treated like seamless trials or rolling reviews? Will existing regulations need to change? What happens to data monitoring committees? Will Bayesian statistics be required for real-time signal analysis? Will the reviewer monitoring signals in real time be independent from the reviewer making approval decisions?

These are not abstract questions. Each carries regulatory consequences. And the only forum in which they can be resolved is the pilot itself, which is exactly why Fakhouri repeatedly encourages sponsors to participate in the RFI process and, if selected, in the pilot. “The non-technical aspects of the pilot are actually some of the most important issues to resolve,” she said.

Legal analysts reviewing the initiative have raised similar concerns: how can the FDA communicate with sponsors about safety or efficacy signals while sponsors remain blinded to results before completion? There is also a statistical risk in relying on data in a rolling fashion, including the potential for false-positive or false-negative conclusions, questions about multiplicity control in real-time data collection, and whether FDA analyses would be conducted in accordance with a study’s pre-specified statistical analysis plan.

Sponsor Accountability When Data Flows Toward The FDA

Under the current IND framework, the sponsor bears responsibility for safety reporting. The sponsor monitors data, identifies adverse events, and fulfills expedited reporting obligations under applicable timelines. RTCT introduces a new dynamic: in the proof-of-concept model, data flows from sites through a technical platform toward the FDA, with the sponsor not necessarily receiving every signal the agency may observe at the same moment it is transmitted.

Kent Thoelke, CEO of Paradigm Health, describes the current approach as one designed around staged release and collaborative governance rather than bypassing sponsor oversight: “Who receives a signal depends on the agreed operating model for the specific study. In the proof-of-concept work, certain predefined signals may be visible to sponsors and FDA reviewers in near real time, so both parties can evaluate the same information close to the point of collection.” The platform does not adjudicate disagreements between the sponsor and the FDA; clinical interpretation and regulatory decision-making remain the responsibility of the sponsor and FDA reviewers.

Even so, the accountability question is not fully resolved. If the FDA observes a signal before a sponsor’s medical monitor has escalated it, even by a matter of hours, does the sponsor bear liability for the gap between the agency’s awareness and the sponsor’s formal response? Does the FDA’s earlier visibility create an implied obligation to act before formal safety reporting requirements have been triggered? These questions do not yet have regulatory answers. Their absence is itself a compliance risk for sponsors entering the pilot.

Clinical trial agreements between sponsors and sites were also written for a batch-reporting world. RTCT fundamentally changes the site’s workload and liability exposure. Every sponsor entering the pilot should be reviewing and updating their CTAs before sites go live, and sites should be asking for those updates proactively.

Blinding, Integrity, And The Problem Of Partial Data

The integrity of blinded, randomized clinical trials is a foundational principle of evidence-based drug development. RTCT creates a structural challenge: the FDA will be seeing data signals in near real time, but sponsors remain blinded to outcomes pending completion. If a reviewer monitoring an ongoing blinded trial sees a signal that prompts a question, how does the FDA communicate with the sponsor without compromising the blind?

Thoelke offers important context: the current proof-of-concept studies are open-label, and the FDA has indicated that randomized, blinded Phase 3 trials are not expected to be part of the initial pilot phase. That is intentional. “Earlier visibility introduces important governance questions around interim review, multiplicity, firewalls, and unblinding risk,” he said. “For blinded studies, the workflow would need carefully designed access controls, predefined review boundaries, and clear governance around who can see what information and when.” That framework is still being developed through collaboration between sponsors, the FDA, and the technology layer.

Fakhouri acknowledged the structural complexity and suggested a practical path: the reviewer monitoring real-time signals may need to be a different individual from the reviewer making final regulatory decisions, creating a structural firewall that would need to be formalized in agency guidance before RTCT scales. “There are ways to ensure blinding with the technology. There are ways to ensure randomization. That’s why it’s very important to figure out some of these operational details, but that’s only possible via the pilots.”

“You only get one chance to make a first impression. RTCT may get data to the FDA faster, but data cannot replace dialogue. Time spent in formal FDA meetings is often time spent contextualizing data, identifying confounders, and ensuring regulators understand what the numbers mean.” – Tala Fakhouri, Parexel

There is also a deeper analytical point. Clinical trial data are rarely unambiguous, particularly in rare disease settings. The time currently allocated to formal FDA meetings is not simply administrative dead space; it is time for scientific experts on both sides to contextualize results, identify confounders, and explain how sponsors propose to meet regulatory standards. RTCT may deliver pre-agreed signals faster, but it does not automatically supply the analytical context that makes those signals interpretable.

Where RTCT Starts To Resemble DSMB Territory

Data Safety Monitoring Boards exist to provide independent oversight of ongoing trials, to monitor accumulating data, and make recommendations, including trial modification or termination, based on what they observe. RTCT puts the FDA in a position of observing many of the same signals that DSMBs are constituted to review. When the FDA can see safety signals in near real time, the question is how DSMB recommendations intersect with FDA decision-making, who has priority under what circumstances, and whether the independence that makes DSMB governance structurally meaningful would be preserved.

Thoelke draws a clear operational boundary: “The RTCT model does not replace the role of the DSMB or transfer DSMB responsibilities to the FDA. The FDA has stated that DSMB workflows should generally continue business as usual unless a specific study design requires otherwise.” The distinction, as he frames it, is between a DSMB evaluating safety and efficacy data within a predefined governance framework designed to protect trial integrity, and an RTCT model focused on enabling earlier access to predefined, decision-relevant metadata tied to agreed regulatory questions. “Those functions can intersect operationally, but they are not the same thing.”

Fakhouri sees the DSMB’s role not shrinking under RTCT but becoming more critical. “The role of monitoring committees would become even more important because you’ll have that independence from what the reviewer is doing with the data. The monitoring role would become even more critical and faster,” she said. That model only works if DSMB processes can be accelerated to keep pace with real-time operations, and if governance explicitly delineates the relationship between DSMB authority and FDA review. Neither of those conditions is yet formally defined.

The Honest Case For, And Against

The case for RTCT is real and deserves direct acknowledgment. Faster regulatory decisions mean faster access to therapies. Better safety visibility means adverse events can be identified and acted on before they compound. Compressed timelines between trial phases mean less dead time, more experiments, and more opportunities for patients to access trials that might benefit them.

Fakhouri adds an upstream dimension that is underappreciated in the policy debate: AI is already accelerating the discovery of new compounds eligible for human trials, but the number of available patients and sites has not kept pace with that growth. The pipeline is expanding; the infrastructure is not. “One of the most frustrating things for patients is how slow things are and how hard it is to find trials,” she said. “Anything that we could do to increase this flow of data and information is actually a good thing.”

The risks are equally concrete. Real-time data creates conditions for overreaction to partial information. A signal that would have been contextualized during analysis before submission may prompt regulatory questions before the sponsor has the operational capacity to respond. There is ambiguity around detection timing and response obligations. There is workflow overload at sites already stretched thin. And there is a lack of definitive guidance on the questions listed throughout this article. These are foreseeable operational realities, not hypothetical concerns.

“The pilot is not the end state. It is the process by which the industry and the FDA jointly work out answers to the questions that remain open. Participation means shaping that process. Waiting means implementing its results.” – Tala Fakhouri, Parexel

What Sponsors, CROs, And Sites Should Do Now

Whether or not an organization is in the pilot, the operational discipline RTCT demands is good practice in any trial environment. That means defined and enforced data entry standards with hard 24-hour observation-to-EDC windows; live audit trail accountability with documented reasons for every edit and correction; written and practiced cross-party escalation protocols that exist before a signal appears at 11 p.m. on a Friday; a unified data dictionary across sites, sponsors, and CROs; and staff training that goes beyond standard GCP basics to include real-time dashboard interpretation and signal response.

Thoelke identifies operational readiness, not technology acquisition, as the primary gap for most organizations. “Sponsors should identify studies that could realistically benefit from near-term regulatory decision-making use cases, particularly in early-phase development where faster regulatory interaction could create meaningful value,” he said. For CROs and sites, the opportunity is to move toward workflows that are more integrated with real clinical operations, not simply faster versions of existing batch-oriented processes.

For sponsors specifically, Thoelke points to a structural restructuring requirement: “The RTCT model will push trial sponsors to restructure their teams to accommodate the new processes.” That is not a technology deployment; it is an organizational change, and it requires cross-functional alignment between clinical operations, regulatory affairs, and pharmacovigilance that does not exist by default in most organizations today.

Fakhouri’s guidance for organizations that want to shape what comes next is direct: “Respond to the RFI. Participate in the pilot if you can. Make sure that your data and technology are in a place where you would be able to participate. And for those who are lucky enough to be selected to participate, I would encourage sharing the findings, because scaling RTCTs across the ecosystem will require shared operational knowledge.”

This concludes the three-part series on the FDA’s real-time clinical trials initiative. Return to Article 1 for the full announcement context and Article 2 for the technology and infrastructure implications.