The figure circulates widely in clinical research circles: the large majority of clinical trials run behind their original enrollment timeline. Various sources put it in the range of 70 to 85 percent of trials missing their initial enrollment projections. Whether the precise number is 72 or 84 percent matters less than what the number implies: enrollment delay is not an occasional problem that affects a subset of poorly planned trials. It is the default condition of clinical research.
Understanding why requires getting past the surface explanation, which is usually "not enough patients," and into the structural reasons that enrollment underperforms even when eligible patients exist. There are three persistent structural causes, and each operates at a different point in the trial lifecycle.
Cause One: Protocol Design That Does Not Account for Operational Reality
A protocol is written to answer a scientific question under controlled conditions. The eligibility criteria exist to define a population in which the treatment's effect can be meaningfully measured, with confounders excluded. Those are legitimate goals, but the process of writing eligibility criteria often does not include a systematic assessment of how operable those criteria will be at the sites that will actually run the trial.
The result is protocols that are scientifically sound but operationally difficult in ways that only become apparent after activation. Consider a criterion that reads: "No prior treatment with a CDK4/6 inhibitor within 12 months of study enrollment." That seems straightforward. But at the site level, verifying this requires knowing the patient's complete prior therapy history, including treatments received at other institutions, with enough date precision to confirm the 12-month window. If that history is not fully documented in the site's EHR, which is common for patients who transferred care, the coordinator has to request outside records, wait for them, and abstract the relevant information. For a single criterion on a single patient, that process might take several days. Across dozens of potential candidates, it becomes the primary bottleneck.
Protocols with many criteria of this type, requiring chart abstraction from external sources, narrow temporal windows, or judgment calls about drug class membership, create enrollment timelines that cannot be met by any site operating under normal staffing conditions. The enrollment plan assumes a throughput that the protocol's own complexity makes impossible.
We are not arguing that complex eligibility criteria are always wrong. Sometimes they are scientifically necessary. But there is a category of protocol complexity that is accidental rather than intentional, the result of conservative criteria writing that did not weigh the operational cost. That accidental complexity is addressable at the protocol design stage in ways it cannot be addressed later.
Cause Two: Site Selection That Optimizes for Prestige Over Fit
Site selection is one of the most consequential decisions in a trial's operational design, and it is also one of the decisions most influenced by factors that do not predict enrollment performance.
Sponsors and CROs often prioritize established academic medical centers with strong reputations in the relevant indication. That prioritization is understandable: large academic centers have credibility, experienced investigators, and existing patient populations. But large academic centers also have high trial competition. A patient who qualifies for your Phase 2 oncology trial may also qualify for two others running concurrently at the same institution. The site's patient pool is divided across multiple active trials, and site staff are managing multiple competing protocols simultaneously.
Feasibility questionnaires, the standard tool for estimating site-level enrollment potential, consistently overestimate what sites will actually deliver. The gap between feasibility estimates and actual enrollment rates is a well-documented phenomenon in the clinical research community. Sites tend to estimate based on their total disease population, without accounting for the fraction that will fail eligibility criteria, the fraction that will decline participation, or the fraction that will be claimed by competing trials.
Sites that perform above expectation on enrollment tend to have a different profile: they are often community practices or specialty clinics where the PI's patient panel strongly overlaps with the trial's target population, competition from other trials is lower, and the staff-to-trial ratio allows more focused attention on pre-screening and patient engagement. A community oncology practice enrolling 8 patients per month on a single trial often outperforms an academic center enrolling across four concurrent protocols at 3 patients per month each.
This is not a universal argument against academic sites. Some therapeutic areas and trial phases genuinely require the procedural and subspecialty capabilities that only academic centers provide. But enrollment planning that starts with patient population geography and match quality, rather than institutional prestige, tends to produce more accurate projections and better outcomes.
Cause Three: Pre-Screening Gaps at the Site Level
Even when a trial has a well-designed protocol and appropriately selected sites, enrollment underperforms if sites cannot systematically identify and work up eligible patients from their own patient population. This is the problem we focus on at Enrollvue, and it is worth examining carefully because it is the least visible of the three structural causes.
The conventional view is that sites know their patient populations and can identify candidates for a trial by running a query or reviewing their disease registry. In practice, this works reasonably well for criteria that map to structured EHR fields: ICD codes, lab values, age, and similar parameters can be searched systematically. It works poorly for criteria that depend on information documented in clinical notes: prior therapy history, performance status assessments, comorbidities that appear in narrative documentation but not in coded problem lists.
At a site running multiple trials simultaneously, the coordinator's pre-screening capacity is a finite resource. Each new trial requires the coordinator to understand new criteria, develop an approach for candidate identification, and work through that candidate list while managing their ongoing responsibilities. Sites that have good tools for this work can move through a candidate list in days. Sites relying on manual chart review can spend weeks on the same work, and the throughput limit means that some potentially eligible patients are never formally evaluated before the enrollment window closes or the trial is filled at other sites.
The practical consequence of this gap is a category of missed enrollment that is entirely invisible in the final data. Screen failure rates capture patients who were evaluated but failed criteria. The patients who were never reached by the pre-screening process do not appear in any metric. They are eligible patients who existed in the site's record system and were never found.
Why the Problem Persists Despite Decades of Attention
Clinical trial enrollment delay has been a documented problem for as long as industry analysts have been studying it. Why has it not improved?
Part of the answer is that each of the three structural causes requires a different intervention at a different point in the trial lifecycle, and the parties responsible for each decision often do not bear the cost of getting it wrong. Protocol complexity is determined in the development phase, often far upstream from the operations team that will run the trial. Site selection decisions are influenced by medical affairs and KOL relationships alongside operational considerations. Pre-screening capability at individual sites is largely invisible to sponsors until enrollment numbers start coming in.
The cost of these failures concentrates at the end of the process, in timeline extensions, additional site activations, and per-patient costs that escalate as trials drag. By that point, the original decisions that contributed to the delay are history and often not revisited analytically.
What changes the picture is feedback loops that connect enrollment performance back to the decisions that shaped it, and tools that make the invisible parts of the process, particularly the pre-screening gap, measurable and therefore improvable. That is a long-term project, and it is not one any single company solves alone. But it is worth being clear about what the problem actually is before assuming that the existing approaches to it are exhausting the available options.