ProofPrism

The Corpus

What 52,408 stopped trials look like, all at once

Every trial ClinicalTrials.gov marks terminated, withdrawn, or suspended, machine-classified into why it actually stopped. This page is the wide view. The 10 Evidence Briefs are the narrow one — the only cases here that have been checked against real sources.

These are machine-classification counts across the full corpus — not verified facts. Only the 10 Evidence Briefs have been checked against real web sources. See methodology.

Does the stated reason tell us anything about the science?

The first cut every record gets. Only about one in nine says anything about whether the intervention itself worked.

Operational
65.6%
34,374 trials · About trial conduct — funding, staffing, business — not the science.
Unknown
20.8%
10,882 trials · The stated reason does not say enough to classify.
Scientific
11.2%
5,886 trials · Tells us something about whether the intervention worked.
Not a failure
2.4%
1,266 trials · Registry status is TERMINATED/WITHDRAWN, but this was not a failure.

All 19 reasons, ranked

Accrual failure alone accounts for more trials than the five scientific-failure codes combined.

  • Scientific
  • Operational
  • Unknown
  • Not a failure
  1. Accrual / enrollment14,373 · 27.4%
  2. Business / strategic6,987 · 13.3%
  3. Not stated5,790 · 11%
  4. Funding4,575 · 8.7%
  5. Staff / site3,152 · 6%
  6. Efficacy / futility2,487 · 4.7%
  7. External event2,144 · 4.1%
  8. Stated but uninformative1,954 · 3.7%
  9. Infeasible, cause unstated1,760 · 3.4%
  10. Supply / manufacturing1,643 · 3.1%
  11. Regulatory / conduct1,500 · 2.9%
  12. Toxicity / safety1,406 · 2.7%
  13. Never started, no cause given1,378 · 2.6%
  14. Superseded by external evidence1,077 · 2.1%
  15. Goal met / graduated807 · 1.5%
  16. Design / endpoint / power failure773 · 1.5%
  17. Not a trial / registry error313 · 0.6%
  18. Requirement withdrawn146 · 0.3%
  19. No target engagement / PK-PD failure143 · 0.3%

Corpus fingerprint

Who ran these trials, and how far they got.

64.3%

Academic / other

33,680 trials

29.9%

Industry

15,685 trials

2.5%

NIH

1,329 trials

3.3%

All other sponsor types

1,714 trials

44.6%

Not applicable

23,361 trials

19.3%

Phase 2

10,104 trials

10.6%

Phase 1

5,575 trials

9%

Phase 3

4,711 trials

Read the full methodology·Source: Project Lattice classified.jsonl + stopped.jsonl, joined on nct_id