The wait, by cohort
How long results spent between being submitted and being readable, grouped by the year they were submitted.
Grouping by submission year removes one bias and exposes another. A cohort can only show the trials that have emerged; the ones still in review are invisible. So a recent cohort displays only its fast members and its median is biased downward. Those rows are shown, greyed, and marked not quotable — the shape of the censoring is the thing this project is about, and hiding it would be worse than showing it.
| Cohort | n | Median | p90 | > 1 year | Quotable |
|---|---|---|---|---|---|
| 2008 | 204 | 174 d | 490 d | 21.1% | yes |
| 2009 | 1,664 | 101 d | 662 d | 16.7% | yes |
| 2010 | 1,768 | 70 d | 558 d | 13.4% | yes |
| 2011 | 2,530 | 77 d | 530 d | 14.5% | yes |
| 2012 | 3,094 | 76 d | 476 d | 12.5% | yes |
| 2013 | 3,909 | 114 d | 602 d | 16.3% | yes |
| 2014 | 4,160 | 44 d | 343 d | 9.5% | yes |
| 2015 | 4,605 | 66 d | 447 d | 12.3% | yes |
| 2016 | 5,050 | 114 d | 539 d | 15.0% | yes |
| 2017 | 6,053 | 120 d | 633 d | 24.1% | censored |
| 2018 | 5,015 | 120 d | 506 d | 18.3% | censored |
| 2019 | 5,089 | 68 d | 342 d | 9.2% | censored |
| 2020 | 5,871 | 48 d | 213 d | 5.5% | censored |
| 2021 | 5,716 | 66 d | 353 d | 9.8% | censored |
| 2022 | 5,140 | 107 d | 710 d | 24.1% | censored |
| 2023 | 5,818 | 103 d | 516 d | 19.3% | censored |
| 2024 | 5,725 | 90 d | 317 d | 7.4% | censored |
| 2025 | 5,991 | 58 d | 189 d | 1.7% | censored |
| 2026 | 2,473 | 54 d | 112 d | 0.0% | censored |
n is a lower bound in every row, including the quotable ones. It counts the trials from that cohort which have posted. Those still in review cannot be seen or counted, which is the premise of the whole project. A cohort is mature 3,285 days after 31 December of its year — measured from the end, because a trial submitted in December had the least time of anyone in its cohort.
Quotable cohorts: 2008–2016. Snapshot computed 2026-08-30.