Feasibility benchmarks
| Patients enrolled | Duration (months) | Sites | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Phase | Trials | p25 | median | p75 | p25 | median | p75 | median | pts / site |
- Patients enrolled
- Actual enrolled participants at trial completion (not the sponsor's original target).
- Duration
- Months from study start to primary completion: the primary-endpoint readout, not final follow-up or database lock.
- Sites
- Registered investigator sites. Pts/site is patients ÷ sites, the recruitment rate your site plan implies.
- p25 / median / p75
- Quartiles across the matching trials: a quarter came in below p25, half below the median, a quarter above p75.
Sponsor landscape
Lead industry sponsors, trials started 2020 onward
| Sponsor | Trials | Ph3 |
|---|
Design patterns
Phase 2-3 trials since 2018
Design precedents
| NCT ID | Phase | Patients | Sites | Countries | Months | Design | Sponsor | Started |
|---|
Full precedent tables, exports and bespoke cuts
This page shows the top of each result. Client access adds the complete precedent list for any indication, CSV/Excel export, site-and-country breakdowns, and comparator sets built to your protocol question, plus post-approval uptake curves from national prescribing data.
Request access Or send us a feasibility question and we'll answer it from the data.These benchmarks describe the trials that finished
The quartiles on this page are computed over completed trials with an actual enrolment count — 39,347 of the 65,249 in the archive. That restriction is deliberate: a sponsor’s original target is an intention, and a running trial has no final number. It also means the benchmarks describe trials that reached the end.
8,509 trials were terminated or withdrawn. Against the 48,356 with a recorded outcome that is 17.6%, and none of them are in the figures above.
| Phase | Recorded outcome | Stopped early | Share | Median enrolled, completed | Median enrolled, terminated |
|---|---|---|---|---|---|
| Phase 1 | 19,081 | 2,345 | 12.3% | 32 | 21 |
| Phase 1/2 | 2,988 | 999 | 33.4% | 48 | 22 |
| Phase 2 | 11,535 | 2,695 | 23.4% | 91 | 31 |
| Phase 2/3 | 792 | 235 | 29.7% | 193 | 71 |
| Phase 3 | 10,872 | 1,762 | 16.2% | 315 | 134 |
| Phase 4 | 3,088 | 473 | 15.3% | 124 | 27 |
The gap holds in every phase. A terminated phase 3 trial had a median of 134 participants against 315 for one that completed, and had been running 23 months against 19; in phase 2 it is 31 against 91. Trials that stop have recruited a fraction of the cohort and have taken longer to get there. Read the benchmarks as the shape of a trial that worked, and size the recruitment risk as a separate question.
What a term search reaches
The search runs over a condition index of 680 distinct condition strings covering 44,449 of the 65,249 trials, across 116 indication terms. Matching is on the registered condition text, so a broad term collects its subtypes — anemia reaches sickle cell and aplastic anemia alongside it, which is usually what a feasibility question wants and is occasionally not.
Where an indication has fewer than three completed trials in a phase, that row is left out rather than shown. Three trials do not make a quartile, and a benchmark built on them would be worse than no benchmark.
Questions about these benchmarks
Why do the benchmarks leave out terminated trials?
Because a terminated trial’s enrolment is where it stopped, not what it needed, and averaging the two produces a number that describes neither. The cost of that choice is stated above rather than hidden: 8,509 trials stopped early and none of them are in the quartiles.
Is the enrollment figure the sponsor's target or what they actually recruited?
What they actually recruited. Every benchmark here uses the participant count reported at completion, and trials still carrying an estimated figure are excluded. An original target tells you what a sponsor hoped for, which is not a planning input.
What does duration measure?
Study start to primary completion — the primary-endpoint readout, not final follow-up and not database lock. For a programme plan that is the date that matters, and it is shorter than the trial’s full life.
Does this include academic and government-funded trials?
No. The archive is industry-sponsored interventional trials with a registered phase, starting 2010 or later — 65,249 of them. Investigator-initiated and publicly funded studies run to different constraints, so mixing them into a commercial feasibility benchmark makes it less useful, not more.
Can I see the trials behind a benchmark?
Yes, with client access: the full precedent list for any indication rather than the top of it, with site and country breakdowns, CSV and Excel export, and comparator sets built around a specific protocol question.
Citing this
Altima Partners, “Feasibility benchmarks from 65,249 industry trials”, altimapartners.co.uk/trial-intelligence. Derived from ClinicalTrials.gov via the AACT database (CTTI); cite the registry for any individual trial.