The most consistent finding in clinical trial recruitment research is that enrollment projections are wrong. Not slightly optimistic — substantially wrong. Eighty percent of clinical trials fail to meet their initial enrollment timelines, and the primary driver is not unexpected patient scarcity but predictable overestimation at the planning stage.
The mechanics of overestimation are well understood. A PI is asked how many patients per month they can enroll. The PI thinks about their own patient panel, recalls how many patients in the relevant indication they see per week, applies a rough conversion estimate, and returns a number. That number represents approximately 15 to 30 percent of the actual eligible patient population within the site's geographic area. The remaining 70 to 85 percent are patients being seen by other physicians in the region — physicians the PI does not know personally and has no systematic way to reach.
The Geographic Population Gap
Enrollment projection accuracy is fundamentally a data problem. A site estimating enrollment from PI recall is using a sample — the PI's own patient experience — to estimate a population. The estimate is systematically biased because the PI only knows about patients they personally treat. A site using NPI-based geographic population data — counting the actual number of licensed specialists of the relevant type within the trial's enrollment radius and applying regional prevalence statistics — produces a more accurate estimate of how many eligible patients exist, and therefore how many physician relationships need to be activated to hit the enrollment target.
The Screen Failure Multiplier
Enrollment projections are further distorted by underestimating screen failure. A site that projects enrolling 50 patients often does so by calculating how many patients need to express interest to generate 50 enrollments. What they underestimate is the proportion of interested patients who fail eligibility screening. Screen failure rates of 40 to 60 percent are common for many indications. A site projecting 50 enrollments that needs 100 screen-eligible patients actually needs 150 to 250 interested patients to generate 100 who qualify — a recruitment load three to five times larger than the enrollment number suggests.
What Accurate Projection Requires
Accurate enrollment projections require two data points that most sites do not have at feasibility: the actual eligible patient population in the geographic area, derived from prevalence data and NPI physician counts, and the realistic conversion rate from physician contact to enrolled patient, based on the site's historical outreach results. Sites that have run physician outreach programs for previous studies can produce empirically grounded projections. Sites that have not must estimate — and estimation, in clinical trial enrollment, is consistently and significantly optimistic.