Patient Recruitment · 2026

Why Is Your Clinical Trial Under-Enrolling?

A single enrolment percentage can tell you that something is wrong. It rarely tells you what, and it almost never tells you why — here's how to find the stage actually driving the shortfall.

Why is your clinical trial under-enrolling? Usually because one of three distinct stages — sourcing, referral handling, or outcome tracking — is quietly failing, not because recruitment overall is broken.

What does "under-enrolling" actually mean for a clinical trial?

Under-enrolling means a study is behind where it expected to be on patient enrolment, but that single fact says nothing about where the shortfall originates. Recruitment shortfalls rarely have one cause: a study running behind isn't necessarily failing on identification, or engagement, or reporting — it may be doing two of those three perfectly well while one quiet gap absorbs all the difference.

Most teams still reach for a single number when something feels off — enrolment rate, screen-fail percentage, a stalled dashboard. A single percentage can tell you that something is wrong. It rarely tells you what, and it almost never tells you why, which is the gap a proper recruitment diagnosis is built to close.

Why does the headline enrolment percentage hide more than it reveals?

The clearest evidence that a topline number can mislead comes from Tufts Center for the Study of Drug Development's analysis of more than 150 studies across roughly 16,000 sites. Overall, 89% of trials met their enrolment goal — a result that reads as reassuring until the site-level breakdown shows 11% of sites enrolled zero patients and 37% under-enrolled, with only 39% meeting target and 13% exceeding it. As Ken Getz, director of sponsored research at Tufts CSDD, put it: "Patient recruitment and retention are among the greatest challenges that the clinical research enterprise faces today, and they are a major cause of drug development delays."

Site-level variance this wide is exactly why a single trial-level percentage isn't diagnostic on its own — and why, per the same research, developers typically need to nearly double their original timelines to reach target even when they get there in the end.

Where does a trial actually break down — Pre-Referral, Referral, or Post-Referral?

It's useful to separate recruitment performance into three distinct stages, because each one fails for different reasons and is fixed by different people.

Pre-Referral Position is the gap between what the study expected and where it currently stands, and which parts of the study that gap actually touches. A study not yet recruiting is assessed on readiness, not performance; a study underperforming against its own stated benchmark is a different conversation entirely from one that never set a realistic benchmark to begin with.

Referral Capability covers everything between a patient being identified and a referral reaching the site in usable form — where patients are sourced from, how quickly they're contacted, how consistently follow-up happens, how accessible the recruitment team actually is, how much clinical detail is captured before referral, and how cleanly that information hands over. A study can have excellent sourcing and still lose patients here if contact is slow or follow-up is inconsistent.

Post-Referral Integration is what happens to a referral after it leaves the recruitment team's hands. Is the outcome tracked back to its original source? Does the site's feedback reach the people running recruitment? Is any of this used to adjust activity mid-study, or does the same spend keep going to the same channels regardless of what's working?

A shortfall can originate in any one of these stages, or in the relationship between two of them. Strong upstream sourcing that's wasted by weak downstream tracking is a genuinely different finding from weak sourcing alone, even if the two produce an identical headline number.

How does Referral Capability quietly cost you enrolments that sourcing never lost?

Referral Capability is where a surprising amount of the loss happens, and it's easy to misdiagnose as a sourcing problem because the symptom — patients not reaching enrolment — looks identical from the outside. In a 2025 oncology trial analysis, roughly 25% of consented patients never reached enrolment, and a separate study of nearly 6,000 patients found only 5.5% agreement between self-reported eligibility and the actual health record (Inovalon, 2026). Both failure modes sit squarely in pre-screening depth, not in how many leads came in the door.

Operational friction compounds it. Site coordinators typically juggle 22+ separate logins per study (Society for Clinical Research Sites), which is exactly the kind of handover friction that slows a referral between clinical pre-screening and a usable referral reaching the site. A limited source mix is a minor observation for a study recruiting an easy-to-reach population, but a material finding for a rare-disease study with a narrow pool of eligible patients — the same gap means something different depending on what kind of study is asking the question.

Why does a strong referral pipeline still fail without Post-Referral Integration?

A dominant referral source that looks strong on paper can still be a visibility gap dressed up as a strength, if nobody is tracking it through to enrolment. Post-Referral Integration is the stage most often skipped entirely, because it requires connecting data that sits with different teams — call and contact tracking, site-level outcome data, and whatever system decides where the next round of spend goes.

The Society for Clinical Research Sites frames the underlying principle well: "Algorithms surface candidates; relationships convert them." A diagnosis has to check whether the loop actually closes — whether site feedback and enrolment outcomes travel back upstream to change what recruitment does next — not just whether a channel produces referrals. Without that loop, the same spend keeps going to the same channels regardless of what's actually working, and a strong-looking source can be quietly underperforming for months before anyone notices.

What does a proper recruitment diagnosis actually check?

A recruitment diagnosis worth doing should be able to answer specific, stage-level questions — not restate a single overall score.

• Is the source mix appropriate for this population, not just broad? Three well-targeted channels can outperform five poorly matched ones — channel count alone is a weak signal.
• Is a dominant referral source being tracked all the way to enrolment, or is it a visibility gap dressed up as a strength?
• Where does the pathway actually slow down — at first contact, at follow-up, or somewhere in pre-screening?
• Is accessibility (contact hours, channels) quietly excluding a working-age or carer population the study needs?
• Does outcome data ever travel back upstream to change what recruitment does next, or does the loop stay open indefinitely?

None of these questions are answered by a single overall score. They're answered by looking at the relationships between specific answers — which is the difference between a report that restates numbers back at you and one that tells you something you didn't already know.

What should you do first if your trial is behind on enrolment?

The most useful first step isn't another dashboard review — it's establishing which of the three stages is actually driving the gap, and how significant that gap is once the study's own population and protocol are taken into account. A thin pre-screen matters more in a complex protocol than a simple one; the same underlying score means something different depending on context, which is exactly why a diagnosis needs to read the answers in context rather than just add them up.

That's what a structured recruitment diagnostic is built to surface: not a score to compare against someone else's study, but a clear read on your own pathway, stage by stage, with the areas most worth examining ranked in order of impact — covering everything from how referrals reach the site to how supporting records get collected along the way.

Frequently asked questions