Rescue · Diagnosis

Your study is behind on enrollment. Diagnose it before you spend more.

Adding sites and adding media budget are the two default responses to a lagging study. Both are expensive, both are slow, and neither works if the constraint is somewhere else in the funnel. Here is how to find the actual bottleneck first.

Rock Enroll Editorial · September 2026

There is a predictable sequence that plays out when a study falls behind.

Enrollment tracks below plan for two months. The clinical operations lead raises it. Someone proposes adding sites. Someone else proposes increasing the recruitment spend. A vendor is invited to present. Within a quarter, the sponsor has committed to both, and six months later the study is still behind — now with more sites to manage and a larger budget consumed.

The problem is not that either intervention is inherently wrong. It is that both are being applied before anyone has established what is actually failing. Adding sites solves a geographic access problem. Adding media budget solves a top-of-funnel volume problem. If the study has neither of those problems, both interventions are money spent to accelerate a process that is leaking somewhere further down.

This matters more than it might sound, because the failure is common and the delays are severe. A widely cited Tufts Center for the Study of Drug Development analysis found that 53% of studies ran past their planned timelines, with one in six taking more than twice as long as originally scheduled. A separate analysis of clinical trial accrual found that roughly 19% of trials either terminated for failed accrual or completed with less than 85% of expected enrollment — enough to seriously compromise statistical power.

And the causes are not few. Beth Harper, who has spent more than two decades working on studies in rescue, has mapped over 150 distinct root causes for why sites fail to enroll. When there are 150 possible explanations and you pick your intervention before diagnosing, the odds are not with you.

The funnel is six stages, and they all look identical from the top

Most sponsor-side enrollment dashboards report two numbers: patients screened and patients randomized. Occasionally referrals. That is enough to tell you that enrollment is behind. It is not remotely enough to tell you why.

The actual path a patient takes has at least six distinct stages, and a failure at any one of them produces exactly the same signal on a sponsor dashboard — a low randomization count.

1. 1 Reach. The right people see something about the study.
2. 2 Response. Someone who saw it takes an action — clicks, calls, fills something out.
3. 3 Pre-screen. They complete an eligibility questionnaire.
4. 4 Handoff. A pre-screened, apparently eligible person is passed to a site, and the site makes contact.
5. 5 Visit. That contact converts into a scheduled and attended screening visit.
6. 6 Randomization. The screened patient passes full eligibility and enters the study.

A study starved at stage 1 and a study hemorrhaging at stage 4 present identically in the weekly report. They require opposite interventions. Spending more at stage 1 when the leak is at stage 4 means paying to push more people into a pipe that is already spilling.

The five ratios that locate the leak

You do not need industry benchmarks to do this. You need your own numbers, stage by stage, and you need to compare them against the assumptions in your own enrollment plan — the plan that projected the timeline you are now missing. That plan contained implicit conversion assumptions. Making them explicit and testing them against reality is the entire diagnostic.

Compute each of these, by site and in aggregate:

Responses ÷ reach

Are the right people seeing it, and does the message land? A low ratio here points to targeting or creative, not to volume. Buying more of the same impressions will reproduce the same ratio at higher cost.

Completed pre-screeners ÷ responses

How many people start the questionnaire and abandon it? Screener abandonment is one of the most fixable problems in recruitment and one of the least examined. A long screener, a screener that asks for a phone number before establishing interest, or one that opens with disqualifying questions will quietly discard a large share of genuinely eligible people.

Site contact made ÷ pre-screen passes

This is the handoff, and it is frequently the largest leak in the entire funnel. It is also the one sponsors have the least visibility into, because it happens between two systems — the recruitment vendor's portal and the site's own workflow. CISCRP's 2025 Perceptions & Insights study found that, aside from simply not qualifying, the single biggest obstacle patients reported to trial participation was never being contacted by study staff at all, or waiting so long to hear back that they moved on.

Read that again in the context of a rescue decision. If pre-screened, interested, potentially eligible patients are being referred and not contacted, then more media spend buys more patients who will also not be contacted. The money converts to nothing. This is the specific failure mode that makes recruitment vendors look ineffective when the vendor is performing fine.

Screening visits ÷ contacts made

Site capacity, scheduling friction, distance, and visit burden. If contact is happening but visits are not, the constraint is operational at the site — staffing, appointment availability, or a screening visit that asks too much of the patient too early.

Randomizations ÷ screening visits

Your real screen failure rate. If this is far below what your protocol assumed, the problem is not recruitment at all. It is eligibility.

What each pattern actually means

Three failures that are not recruitment problems at all

Before commissioning a rescue, rule these out, because a recruitment vendor cannot solve any of them and will not tell you so before signing.

The eligibility criteria are too narrow

If the protocol excludes most of the treated population for the indication, the patients do not exist in the numbers the plan assumed. The remedy is a protocol amendment, and the honest question is whether the amendment is worth its own timeline cost.

The sites were selected wrong

Site selection is often driven by prior relationships, investigator reputation, or geographic coverage rather than by demonstrated access to the specific eligible population. A high-prestige site with no patients matching your criteria will enroll nothing regardless of how much support it receives.

A competing trial is absorbing the population

In crowded indications, several studies are chasing the same finite pool of eligible patients through the same sites. Check what else is recruiting in your indication and geography before concluding that your recruitment is underperforming.

What to do in the first two weeks

1. Pull the raw stage-level counts, by site, for the entire enrollment period to date. Not the summary dashboard — the underlying counts.
2. Identify which stages you cannot measure. In most studies there will be at least one, and it is usually the handoff. That gap is itself a finding, and closing it is often the first intervention.
3. Compute the five ratios and compare each against what your enrollment plan implicitly assumed. The largest gap is your bottleneck.
4. Only then decide what to buy. If a vendor proposes a solution before you have completed this, ask them which stage their intervention addresses and how they will measure it. A vendor who cannot answer that question specifically is selling volume, not enrollment.

The uncomfortable conclusion most sponsors reach partway through this exercise is that they cannot complete it — the data to compute the ratios was never captured, because no one instrumented the path between the advertisement and the randomization. That is a fixable problem, but it is worth being clear that it is the real one. A study that cannot see its own funnel cannot be rescued reliably. It can only be spent on.

Sources

• Applied Clinical Trials, “The Enrollment Rescue Dilemma: How Sponsors and Sites Can Make the Most of a Tough Situation” — Tufts CSDD timeline findings and accrual analysis. appliedclinicaltrialsonline.com
• Clinical Leader, “Study Rescue: How To Get Your Clinical Trial Back On Track” — interview with Beth Harper, Clinical Performance Partners. clinicalleader.com
• Citeline, “How to Avoid a Rescue Trial in the First Place” — citing CISCRP 2025 Perceptions & Insights Study. citeline.com

Editorial note: the diagnostic above depends entirely on having stage-level data from advertisement through randomization. If your study cannot currently produce those counts, DYNO's predictive enrollment engineering documentation covers how that instrumentation is structured.