The real-time trial as default

The conventional clinical trial collects data on a defined schedule, locks the database at fixed timepoints, and analyzes the results after enrollment is complete. The pre-specified endpoint is measured at the pre-specified timepoint. The protocol is rigid because the analytic methods used to interpret the result depend on the rigidity. The price of rigidity is time.

A pivotal trial in a rare disease typically enrolls for two to three years, follows participants for one to three years, and produces a regulatory submission a year or more after follow-up ends. Six to seven years from first patient enrolled to regulatory decision is a normal timeline. For a rare disease where the global eligible population is two hundred patients, that timeline costs the field the data that would have come from treating the next two hundred patients during the same window.

The Real-Time Clinical Trials initiative at the FDA, launched in 2023 and expanded under subsequent guidance, contemplates a different model. Trial data flow continuously from the patient to the trial sponsor to the regulatory reviewer. Endpoints are pre-specified but assessed as the data accumulates rather than at fixed timepoints. The statistical methods that support the approach exist; they are the same Bayesian and adaptive-design methods that the FDA has approved for specific trials over the past decade. The Real-Time initiative formalizes the methods as a category rather than a one-off accommodation.

What real-time trials require

Three infrastructure pieces have to be in place for the model to work.

The first is continuous, structured data flow from the patient to the trial. Quarterly case report form submissions cannot support real-time analysis. Wearable device data, electronic health record integration, patient-reported outcome capture through validated apps, and laboratory data flowing electronically rather than as PDF reports are the substrate. The substrate has matured substantially over the past decade. Most pivotal trials still do not use it.

The second is statistical methodology that updates inferences as data accrues without inflating false-positive rates. Bayesian sequential analysis, alpha-spending functions, and group sequential designs all support the methodology. None requires invention. The approval of specific designs for specific trials happens on a case-by-case basis with the FDA. The Real-Time initiative is normalizing the patterns the agency has already approved into a more predictable framework.

The third is regulatory infrastructure that can review continuously. The traditional model has the agency receiving a complete data package after enrollment ends. The real-time model has the agency receiving data as it flows. The agency's internal review processes have to support continuous review without overburdening reviewers. The FDA's investments in real-time data review under several initiatives suggest the agency is building this capacity, although the rollout is uneven across review divisions.

Why rare disease fits the model

Three structural features of rare disease align with the real-time model better than common disease does.

The patient population is small enough that every data point matters. Losing a quarterly visit's data because the patient was hospitalized is statistically meaningful in a 30-patient trial in a way that would not register in a 3,000-patient trial. Continuous data capture preserves more information per patient.

The endpoints are often objective and continuous. A blood phenylalanine level in PKU. A motor function score in SMA. A creatine kinase value in Duchenne. The measurements are quantitative, repeated frequently, and analyzable with continuous methods. The categorical responder-versus-non-responder analyses that dominate common-disease trials are less central in rare disease, where the trajectory of the continuous biomarker is the primary signal.

The clinical relationship is more longitudinal. A rare disease patient is followed by a metabolic specialist, a pediatric cardiologist, a neuromuscular team, or another subspecialty group across years. The data flow from the clinic is already continuous in the medical record sense. Capturing it for the trial requires structured extraction, not new data generation.

What changes when the trials get faster

The acceleration math compounds across the development cycle. A trial that reads out in two years instead of seven means the regulatory decision arrives five years earlier. The next trial in the same condition or in a related condition starts five years earlier. The accumulating evidence base for the platform technology grows five years' worth of data faster than the conventional model produces.

For the patient currently receiving the standard of care, faster trials change the temporal calculation about whether to enter a trial or wait for approval. If approval is two years away, more patients can wait. If approval is seven years away, the patients who would benefit are aging out of eligibility, dying before they can access the therapy, or accumulating disease that the trial therapy could have prevented.

For the data infrastructure, the real-time trial is the application that justifies the investment. A data trust that delivers structured continuous data to a trial in real time is the infrastructure the real-time model depends on. The trust's value proposition extends across all subsequent trials in the same population. The cost of the infrastructure, paid once, returns across every trial that uses it.

The Nurses' Health Study captured biennial questionnaires for fifty years and produced findings of historic importance. A modern data trust can capture continuous data through wearable devices, structured patient-reported outcomes, and electronic health record integration. The temporal resolution is orders of magnitude higher than the NHS achieved. The findings that emerge from continuous data are qualitatively different from those available at biennial snapshots. The technology to support the difference exists. The infrastructure decisions that turn the technology into routine practice are still being made.