Clinical Trial Protocol Amendments — ClinicalTrials.gov version diff

Find out what actually changed in a clinical trial protocol, amendment by amendment.

ClinicalTrials.gov keeps every historical version of a study record. What it does not give you is the diff. The site shows one version at a time plus a list of which modules were touched — so answering "did they change the primary endpoint?" means opening two versions in two tabs and reading them side by side. For a portfolio of 100 competitor trials that is not a task anyone does.

This Actor does it for you. Give it NCT IDs, a condition, or a sponsor, and it returns one structured record per protocol amendment, with before/after values and a significance rating:

• Primary endpoint re-specified — outcome measures added, removed or redefined
• Enrollment cut or raised — with the delta and percent change
• Eligibility widened or narrowed — criteria text, age bounds, sex, healthy volunteers
• Primary completion date slipped — with the shift in days
• Trial terminated, suspended or withdrawn — with the stated reason
• Arms and interventions added or dropped
• Phase, allocation, masking or primary purpose changed
• Lead sponsor or collaborators changed

Who this is for

• Competitive intelligence — an endpoint change or enrollment cut in a rival's phase 3 is a read on their readout
• Feasibility and study startup — see how often a sponsor amends eligibility, and how far completion dates really slip
• Portfolio and BD teams — a standing watch list of trials, checked on a schedule
• Regulatory and medical writing — reconstruct an amendment history without diffing PDFs by hand
• Research on protocol amendments — outcome switching, enrollment attrition, timeline drift, at scale

Input

{
  "nctIds": ["NCT04368728", { "recordId": "WATCH-17", "nctId": "NCT02576574" }],
  "since": "2024-01-01",
  "minSignificance": "major",
  "maxAmendmentsPerStudy": 20
}

Or search instead of listing IDs:

{
  "searchCondition": "glioblastoma",
  "searchSponsor": "Novartis",
  "searchStatus": ["RECRUITING", "ACTIVE_NOT_RECRUITING"],
  "maxStudies": 50,
  "since": "2025-01-01"
}

Field	Default	What it does
nctIds	—	NCT IDs, or objects with your own recordId to join results back to your system
searchQuery	—	Free-text search (query.term)
searchCondition	—	Condition or disease
searchSponsor	—	Lead sponsor or collaborator
searchStatus	—	Recruitment statuses, e.g. ["RECRUITING"]
maxStudies	25	Cap on studies scanned (max 500)
since	—	Only report amendments on or after this date
minSignificance	major	critical, major or minor
includeMinorChanges	false	Include title, summary, contact and site-count edits
maxAmendmentsPerStudy	20	Most recent N version transitions per study
maxVersionFetches	600	Global budget on historical record fetches

Output

One dataset item per amendment. type is amendment, no-amendments or error.

{
  "type": "amendment",
  "recordId": "WATCH-17",
  "nctId": "NCT02576574",
  "briefTitle": "…",
  "leadSponsor": "…",
  "amendmentDate": "2022-12-09",
  "fromVersion": 42,
  "toVersion": 43,
  "significance": "critical",
  "categories": ["arms", "endpoints"],
  "summary": "Primary outcome measures: 4 added, 2 removed — a primary endpoint was REMOVED or redefined; Study arms: 3 added, 3 removed",
  "changes": [
    {
      "field": "primaryOutcomes",
      "label": "Primary outcome measures",
      "category": "endpoints",
      "significance": "critical",
      "added": ["Overall survival [Up to 60 months]"],
      "removed": ["Progression-free survival [24 months]"]
    }
  ],
  "recordUrl": "https://clinicaltrials.gov/study/NCT02576574?tab=history"
}

Date changes carry daysShifted; enrollment changes carry delta and percentChange; list fields carry added and removed.

Significance is critical for endpoint changes, phase changes, terminations, enrollment moves of 30%+ and date shifts of six months or more; major for ordinary eligibility, design, arm, sponsor and timeline edits; minor for titles, summaries, site counts and sub-2% enrollment drift.

Why you can trust a "no amendments" answer

This matters more than any feature. A monitoring tool that quietly returns "nothing changed" because its data source moved is worse than no tool at all — you would conclude a competitor's protocol was stable all year.

So this Actor is built to fail loudly rather than report a clean sheet it cannot stand behind:

• Every history and version payload is shape-checked; a missing version list, a mismatched version number or a record without its core modules is an error, not an empty result.
• A study that does not exist returns study_not_found. A record that cannot be parsed returns history_unreadable. Neither is ever reported as "no amendments."
• If more than 25% of scanned studies fail to parse, the entire run fails — because at that point every clean answer in it is suspect.
• An unrecognised module label is treated as material, so a new ClinicalTrials.gov module cannot silently hide amendments.

A no-amendments record means exactly what it says: the version history was read successfully and nothing at or above your significance threshold happened in the window.

Data source and licence

ClinicalTrials.gov, a service of the US National Library of Medicine. Records are US Government works in the public domain and NLM provides them for programmatic reuse. Study search uses the documented API v2; version history uses the JSON history endpoints behind the site's own "Study Record Versions" tab. No API key, no scraping, no rate-limit circumvention — requests are throttled and identify themselves.

Amendment records reflect what sponsors submitted to ClinicalTrials.gov. They are not a substitute for the sponsor's own protocol amendment documents, and this Actor is not medical, legal or regulatory advice.

Pricing

Pay per event, platform costs included:

Event	Price
Study scanned	$0.08
Amendment reported	$0.04

Monitoring 100 trials monthly typically runs about $18/month — against enterprise trial-intelligence subscriptions that start in the five figures.

Tips

• Set since to your last run date and schedule this daily or weekly; you get only the new amendments.
• Use recordId to join output straight back to your watch list.
• Start with minSignificance: "critical" on a large portfolio, then widen.
• Deep histories are the expensive part — maxAmendmentsPerStudy bounds it.