Clinical Trial Patient Matching: Why Do Sponsors Need More Than a Public Registry Listing?

Every applicable clinical trial conducted in the United States must be registered in the federal public trial registry maintained by the National Library of Medicine at the National Institutes of Health. An applicable clinical trial is, broadly, a controlled interventional study of a drug, biologic, or device that meets the criteria set out in federal law. For many study teams, that registry record is also the first point at which a trial becomes visible to anyone outside the sponsor and its sites.

That visibility is frequently mistaken for a recruitment channel. A registry record is a disclosure document. It exists so that regulators, researchers, clinicians, and the public can see which studies are planned, underway, and completed, and what those studies found. It was never built to assess whether one particular person is likely to qualify for one particular protocol. Sponsors and contract research organizations that treat a listing as an enrollment engine are asking a compliance instrument to do work it was not designed to do.

What are government trial registries actually built to do?

The federal registry was created by Section 113 of the Food and Drug Administration Modernization Act of 1997 and launched in 2000. Section 801 of the Food and Drug Administration Amendments Act of 2007 substantially expanded it, adding results reporting to registration. The Final Rule at 42 CFR Part 11, which took effect in January 2017, set the operating requirements sponsors work under today: register within 21 days of enrolling the first participant, update the record at least once every 12 months and within 30 days of any change in recruitment status, and submit results generally within one year of the primary completion date.

The purpose stated by the agencies that run the system is transparency. National Institutes of Health policy describes these requirements as measures intended to increase the availability of information about clinical trials to the public, and states plainly that they do not affect how trials are designed or conducted. The registry also carries an explicit caution that listing a study does not mean the federal government has evaluated it. Registration is disclosure, not endorsement and not matching.

The publication side of research reinforces the same purpose. Journal editors have required prospective registration as a condition of publication for two decades, on the grounds that a public record prevents selective reporting of outcomes and unnecessary duplication of studies. Europe follows a parallel logic through the Clinical Trials Regulation, whose single submission portal became mandatory in January 2022 and completed its transition of legacy studies in January 2025. Across these systems the design goal is public access to trial information, not individualized eligibility assessment.

Why does a registry listing alone underperform as a recruitment channel?

The first reason is language. Registry records are written for regulatory precision and for a professional readership. Eligibility criteria, meaning the list of characteristics a person must have or must not have in order to join a study, have been measured in peer-reviewed readability research as requiring a college reading level. Federal health communication guidance recommends a seventh to eighth grade reading level for materials intended for the public, and institutional review boards commonly hold recruitment advertising to a similar standard. A criteria list that clears a regulatory bar often will not clear a comprehension bar.

The second reason is complexity. Industry research tracking protocol design over the past two decades shows sustained growth in the number of eligibility criteria, procedures, and sites per study. More criteria means more opportunities for a candidate to be excluded by something they could not reasonably self-assess from a public summary.

Those factors combine into a self-selection problem. A person reads a listing, recognizes their condition, contacts a site, and is then found ineligible on criteria that were never legible to them. That outcome is a screen failure, meaning a candidate who is formally assessed but does not qualify to enroll. Ineligibility is the leading documented cause, and rates vary widely by therapeutic area, running higher in central nervous system and biomarker-driven studies. The mechanics of how sites improve eligibility matching sit largely outside what a registry record can influence.

The third reason is record currency. Peer-reviewed audits have found that a substantial minority of registry records carry a recruitment status that is either incorrect or was updated long after the fact. In one study, investigators contacted the listed coordinators for trials showing an open recruiting status and were unable to confirm active recruitment for a large share of them. The registry acknowledges the problem structurally: a record not verified within two years is automatically reclassified to an unknown status.

None of this criticizes the registry. Accurate disclosure and real-time recruitment routing are simply different jobs.

What does unqualified referral volume cost a sponsor or CRO?

Recruitment is the dominant operational risk in trial execution. A large analysis of randomized trials published in a major medical journal found that among studies discontinued before completion, poor recruitment was the single most frequent reason. Site-level industry data consistently shows a portion of activated sites enrolling no participants at all, and a larger portion enrolling below target.

The cost is not only in unfilled slots. Every candidate who reaches a site consumes coordinator time whether or not they qualify. Hours spent working through candidates who were never plausible are hours unavailable for consenting, scheduling, and retaining those who are. This is why the cost of delayed enrollment is measured through cycle-time metrics such as time to first participant enrolled and the interval from first participant in to last participant out, rather than through raw inquiry counts.

Delay then compounds. Studies that fall behind on enrollment tend to accumulate protocol amendments, add sites, and extend timelines, each carrying its own cost and monitoring overhead. Referral volume that does not convert is not a neutral input. It is an active operational load.

What does a matching and pre-screening layer actually add?

A matching layer performs the assessment a listing cannot. In plain terms, it takes eligibility criteria written as free text for human readers and represents them in a structured form a computer can evaluate, uses natural language processing to interpret that text, compares the criteria against information a candidate has provided, then ranks the results so the most plausible matches surface first.

The limits of that process are structural rather than a defect of any particular technology. Eligibility criteria are authored for clinicians, not machines, so a proportion of them resist automated evaluation entirely: those turning on clinical judgment, on imaging or laboratory interpretation, or on a subjective assessment of a candidate's status. Performance is also bounded by the completeness and timeliness of the underlying data, and models trained on data that underrepresent particular populations can carry that forward. Published informatics research is consistent: automated matching narrows and prioritizes, and human review remains the determinative step. Understanding where AI creates new failure points is part of evaluating any matching claim responsibly.

A credible matching layer is therefore two components rather than one. Software narrows a large population to a plausible set. A clinically trained human then verifies the match, works through the criteria a model cannot evaluate, and confirms the candidate understands what participation involves and is genuinely interested before anything reaches a research site.

DecenTrialz operates as that layer. It is a clinical trial recruitment and pre-screening platform that combines AI-assisted participant matching with pre-screening conducted by registered nurses, then delivers qualified referrals into a structured workflow with dashboards for sponsors, CROs, and sites. Final eligibility determination, informed consent, the study walk-through, and enrollment always belong to the research site team, never to DecenTrialz. The platform is designed to raise the quality of what arrives at a site, not to substitute for the site's clinical and regulatory authority.

Recruitment of this kind sits inside an established regulatory frame that sponsors should expect any partner to respect. The Food and Drug Administration treats direct recruitment advertising as the beginning of the informed consent and subject selection process, which places recruitment materials within institutional review board oversight. Screening procedures performed to establish research eligibility likewise fall under review board scrutiny. Where pre-screening touches protected health information, the Health Insurance Portability and Accountability Act requires an appropriate legal basis such as authorization or a documented waiver. Regulatory attention to artificial intelligence is relevant as well: draft Food and Drug Administration guidance issued in January 2025 sets out a risk-based framework for assessing an AI model's credibility against its specific context of use, and a European Medicines Agency reflection paper adopted in September 2024 takes a comparably risk-based, human-centered position.

How should sponsors and CROs evaluate the two layers together?

The registry and the matching layer are complementary, not competing. Matching platforms typically consume registry data through its public application programming interface, and the registry remains the authoritative public record of a study, including its identifier, protocol summary, status history, and results. Weak registry hygiene degrades the matching layer above it.

Registry-level work remains firmly the sponsor's job: keeping recruitment status current within the required windows, writing the brief summary in plain language, maintaining monitored site contact information, and giving the study a title a non-specialist can recognize. These are transparency obligations in their own right, and they also determine whether the record is usable downstream.

On the matching layer itself, evaluating a recruitment platform is a due diligence exercise. Questions worth asking include:

• What is the model's defined context of use, and what validation evidence supports it there?
• Who performs human verification after the algorithmic step, and what are their clinical credentials?
• How is referral quality measured? Conversion from referral to randomization is informative; raw referral volume is not.
• What is the legal basis for handling candidate health information, and how is consent to be contacted documented?
• Have all participant-facing recruitment materials been submitted for institutional review board approval?
• What visibility do sponsors and sites get into funnel performance, and at what level of granularity?

A platform that answers these questions clearly is describing an operational process. One that answers them with claims about volume is describing a lead source.

Questions sponsors and CROs ask about trial matching layers

Does registering a trial satisfy any recruitment obligation?

No. Registration and results reporting are transparency requirements under federal law, enforced through notices of noncompliance and civil money penalties. They are unrelated to whether a study meets its enrollment targets, and compliance carries no recruitment benefit beyond public visibility of the record.

Can a matching layer replace the registry?

No, and it should not attempt to. Registration is a legal requirement and the registry is the authoritative public record of the study. A matching layer operates on top of that record.

Is AI matching alone sufficient to produce site-ready referrals?

Published research and current regulatory thinking both point away from that conclusion. Automated matching narrows and ranks candidates effectively, but criteria requiring clinical judgment, and verification of a candidate's actual interest and understanding, call for a trained human step before referral.

Where does the research site's authority begin?

At every clinical and regulatory decision. Determining final eligibility, obtaining informed consent, conducting the study walk-through, and enrolling participants are functions of the research site team. A pre-screening platform prepares candidates for that conversation and takes no part in those determinations.

What this means for recruitment planning

The framing worth carrying into a recruitment plan is additive. Maintain the registry record well, because it is a legal obligation, a transparency commitment, and the foundation other systems read from. Then add the layer that performs individualized assessment, because the registry does not and was never intended to. Treating a disclosure record as an enrollment channel is where recruitment plans tend to quietly fail.

Sponsors and CROs evaluating how a matching and nurse-led pre-screening layer would fit alongside their existing recruitment approach can contact the DecenTrialz team to discuss study-specific requirements.