Analysis
No proprietary data, no NDA, nothing borrowed from a client. Every piece in this section is built from ClinicalTrials.gov registrations, FDA records and published literature, with the method stated so anyone can reproduce it.
The question running underneath most of this work is how old the data behind a recruitment campaign actually is — and what that means when an AI matching engine reads it. The argument is set out in why AI EHR parsing won't fix an under-enrolling study.
Every terminated and withdrawn record on ClinicalTrials.gov carries a free-text reason. Coded across 43,563 interventional studies: accrual is named in 35% of terminations and 23% of withdrawals — more than safety, futility and funding combined.
Read the analysisStart to primary completion across 122,063 completed studies started 2015-2023. Median 23 months in Phase 2, 37 in oncology, 14 in infectious disease — with the 75th percentile you should actually be planning against.
Read the analysisTotal pace rises with site count; per-site productivity falls from 3.06 to 0.11 patients per site per month, and large studies take longer, not shorter. What to assume before approving an expansion.
Read the analysisPublished EHR and registry work on record staleness, contactability decay and eligibility drift between capture and outreach — and the one metric no vendor evaluation currently reports.
Read the analysis