ClinicalTrials.gov version history, for every trial at once
Every ClinicalTrials.gov record is a living document. Sponsors revise status, enrollment, dates, design and interventions over a trial's life, and each revision becomes a new version of the record. This guide explains how those versions work and how to analyze them across the whole registry.
What a record version is
When a sponsor or investigator submits a change to a registered study, ClinicalTrials.gov posts the updated record as a new, numbered version, starting from the original registration. On the website, each study's History of Changes tab lists its versions and lets you view the record as it stood at each one.
Each version carries two dates that matter for analysis:
- Submission date: when the registrant submitted the change.
- Posting date: when ClinicalTrials.gov made that version public. For anything tied to what the public could know and when (an event study, for instance), this is the date to use.
How much records change
Amendments are the norm, not the exception. Across the 604,583 trials in the registry:
- 78.8% have been amended at least once.
- The median trial has 3 versions; the median interventional trial has 4.
- In total the registry holds 4.44 million trial versions.
- 46.9% of interventional trials have had their primary completion date pushed later at least once (more on slippage).
What you can learn from the history
The current record tells you where a trial ended up. The version history tells you how it got there, and when each step became public:
- Timeline slippage: how often and how far start, primary completion and completion dates move.
- Enrollment revisions: targets raised or cut, and anticipated versus actual enrollment.
- Status trajectories: the path from not yet recruiting to recruiting, suspended, terminated or completed, with the stated reason for stopping.
- Design and intervention changes: arms, masking, allocation and registered interventions added or dropped.
- Event studies: lining up public disclosures of protocol changes with market reactions.
- Predictive features: modeling a trial's outcome from how it has changed so far, without look-ahead bias from the final record.
Ways to get the version history
The History of Changes tab
Free and authoritative, and the right tool for a handful of trials. It shows one trial at a time, though, so it doesn't scale to questions about thousands of trials.
Bulk downloads and database snapshots
Registry-wide downloads and relational mirrors describe each trial as it stood on the day the data was taken. Some projects keep archived snapshots, which you can compare with each other. Snapshot comparisons still miss any amendment that was posted and then superseded between two snapshots, anything from before the first archived snapshot, and the exact date each change was posted.
A compiled version history
Anamnesis Data compiles every posted version of every trial into relational tables with one row per trial version. Each row carries both the submission and the public posting date, so one query covers the whole registry:
-- Every recruitment status change across the registry, with the date it became public
WITH v AS (
SELECT nct_id, nct_version, last_update_post_date,
overall_status,
LAG(overall_status) OVER (
PARTITION BY nct_id ORDER BY nct_version
) AS prev_status
FROM trial_versions
)
SELECT *
FROM v
WHERE overall_status IS DISTINCT FROM prev_status
AND prev_status IS NOT NULL;The data comes as versioned Parquet releases, so an analysis can be pinned to an exact snapshot, or as a hosted relational database you can query with SQL.
Common questions
Does ClinicalTrials.gov keep old versions of a trial record?
Yes. Every posted version is retained and can be viewed on the study's History of Changes tab.
Which date should I use to know when a change became public?
The version's posting date, not its submission date. The two can be days or weeks apart. Older versions sometimes carry an estimated rather than an actual posting date, so check the date type when timing precision matters.
Does the history include results?
Each version records whether results had been posted as of that version. Per-version posted results, outcomes, locations and collaborators are planned additions.
How do I cite the dataset?
See How to cite Anamnesis Data. Please also credit ClinicalTrials.gov as the source of the underlying records.