Sample: 6,000 withdrawn studies (never started), ClinicalTrials.gov, 30 October 2026.
| Reason | n | share |
|---|---|---|
| other/unspecified | 2954 | 49.2% |
| enrollment | 1250 | 20.8% |
| funding | 551 | 9.2% |
| efficacy | 393 | 6.5% |
| business | 344 | 5.7% |
| protocol | 322 | 5.4% |
| covid | 126 | 2.1% |
| safety | 60 | 1.0% |
Reading
Withdrawal is a pre-start event: these trials were registered and then cancelled. Comparing this distribution with terminated trials shows whether the failure modes differ before and after launch.
Limits
- Free-text reasons; keyword categories are coarse and first-match.
- 6-page cap, registry order rather than random.
- Registration is voluntary at the planning stage in many jurisdictions.