latest update:
[TrialBase DB]

Data dictionary · provenance + exported fields

Every value in the dataset is traceable to a source record. This page covers the provenance model, null policy, and the dataset field catalog shipped through the REST API and snapshot exports.

Provenance model

Provenance is a side table, not a footnote. For each exported field in a record there is a corresponding provenance row recording the source identifier, source URL, source class, and effective value state. Values without commercial-use cleared provenance are excluded from the paid export path.

_provenance entry shape (inline on API responses)json
{
  "field":         "field_name",
  "value":         "...",
  "source":        "clinicaltrials-gov",
  "source_class":  "government | registry | manufacturer | aggregator | derived",
  "source_url":    "https://clinicaltrials.gov",
  "effective":     true,
  "superseded_at": null
}

Export shape

The primary entity is trials. The export combines the canonical entity table with active provenance from 7 sources. JSON responses and the full JSON download include nested provenance. CSV, Parquet, and SQLite are explicitly curated flat extracts and do not carry nested or separate provenance tables.

Key fields

These fields are selected from the field catalog as the main dimensions for filtering, grouping, matching, and API usage:

  • idTrialBase DB canonical trial identifier. Multiple registry records can resolve to the same canonical id.Identity and registry
  • primary_registrySource registry selected as the canonical record, using a sources.code value such as clinicaltrials-gov, eu-ctis, eudract, isrctn, or drks.Identity and registry
  • registry_idIdentifier for the canonical source registry record, for example an NCT, EudraCT, CTIS, or ISRCTN id.Identity and registry
  • registry_idsCross-registry identifiers attached to the canonical trial, shaped as objects with registry, id, and hard or soft match method.Identity and registry
  • source_record_urlCanonical public registry page for the source record used in the exported row.Identity and registry
  • sponsor_nameLead sponsor name drawn from the public trial registry record.Sponsor
  • conditionsCondition terms associated with the trial.Conditions and interventions
  • countriesCountries associated with trial recruitment or study locations, used by country filters.Conditions and interventions
  • intervention_typesIntervention categories reported by the source registry.Conditions and interventions
  • interventionsIntervention names associated with the trial.Conditions and interventions
  • outcomesSource-owned primary and secondary outcome definitions from the licensed ISRCTN registry contribution.Outcomes and publications
  • outputsISRCTN output/publication metadata and exact PMID or DOI assertions; external links are relationships, not licenses to copy their targets.Outcomes and publications
  • publicationsCitation-only PubMed metadata for exact accepted ISRCTN PMID assertions, including identifiers, title, journal, dates, authors, publication types and retraction/comment links.Outcomes and publications
  • regulatory_recordsHealth Canada CTD authorization observations linked through exact protocol-number evidence plus at least two independent exact agreements.Regulatory
  • lead_conditionFirst listed condition, materialized as a cheap facet and index for common searches.Conditions and interventions
  • completion_dateOverall study completion date where reported.Enrollment and dates

Field catalog

The catalog below contains 30 documented fields from the same field model rendered on this page. Field types and descriptions come from the export contract and source registry. Nullable means unknown or not applicable; it does not mean zero, false, or inferred.

Machine-readable catalogue: /docs/data-dictionary/fields.json.

Identity and registry5 fields

nametypescopedescriptionsources
idbigintall exported records
TrialBase DB canonical trial identifier. Multiple registry records can resolve to the same canonical id.
example: 982431
Provenance: Generated by the registry matcher after hard and soft identifier resolution.
Caveat: Canonical ids are TrialBase DB-assigned and are not registry-issued trial identifiers.
varies
primary_registrytextsearch
Source registry selected as the canonical record, using a sources.code value such as clinicaltrials-gov, eu-ctis, eudract, isrctn, or drks.
example: clinicaltrials-gov
Provenance: Selected from the registry record chosen as canonical for the resolved trial.
Caveat: Primary registry selection is deterministic but secondary registries can carry fresher or richer fields.
varies
registry_idtextsearch
Identifier for the canonical source registry record, for example an NCT, EudraCT, CTIS, or ISRCTN id.
example: NCT04280705
Provenance: Copied from the canonical registry record and retained in cross-registry identifier arrays.
varies
source_record_urltextsearch
Canonical public registry page for the source record used in the exported row.
example: https://clinicaltrials.gov/study/NCT04280705
Provenance: Built from the registry identifier or copied from the source export when a canonical URL is supplied.
varies
registry_idsjsonbdetail
Cross-registry identifiers attached to the canonical trial, shaped as objects with registry, id, and hard or soft match method.
example: [{"registry":"clinicaltrials-gov","id":"NCT04280705","match":"hard"}]
Provenance: Aggregated from registry records linked by identifier, title, sponsor, and date matching.
varies

Titles and summaries3 fields

nametypescopedescriptionsources
brief_titletextsearch
Short public title from the registry record; included in the flat export columns and search results.
varies
official_titletextdetail
Full official study title where supplied by a registry.
varies
brief_summarytextdetail
Registry-provided study summary or synopsis.
varies

Status and design6 fields

nametypescopedescriptionsources
overall_statustextsearch
Raw recruitment or lifecycle status from the source registry, preserving registry-specific vocabulary.
varies
phasetextsearch
Clinical trial phase as reported by the source registry.
varies
study_typetextsearch
Study type, commonly interventional, observational, or registry-specific uppercase equivalents.
varies
primary_purposetextsearch
Primary purpose of the study, such as treatment, prevention, diagnostic, supportive care, or another registry value.
varies
results_postedtextdetail
Registry value indicating whether study results have been posted or made available.
varies
statustextcomputed
Canonical status derived from overall_status for cross-registry filtering, such as completed, recruiting, terminated, suspended, or unknown.
varies

Enrollment and dates4 fields

nametypescopedescriptionsources
start_datetextsearch
Study start date as provided by the source registry. Registry dates can be partial before ingestion normalization.
varies
completion_datetextsearch
Overall study completion date where reported.
varies
enrollment_countintsearch
Participant enrollment count normalized to an integer from registry-provided enrollment text.
varies
primary_completion_datetextdetail
Primary completion date for the study's primary outcome collection where available.
varies

Provenance1 fields

nametypescopedescriptionsources
_provenancejsonbprovenance payload
Per-field active provenance keyed by field name. Each entry lists source, source URL, confidence, and fetched timestamp.
varies

Conditions and interventions5 fields

nametypescopedescriptionsources
lead_conditiontextsearch
First listed condition, materialized as a cheap facet and index for common searches.
varies
conditionstext[]search
Condition terms associated with the trial.
Caveat: Registry-reported condition terms are trimmed and exact repeats within one source observation are removed; assess field fill and selected provenance in each release's measured coverage.
varies
interventionstext[]search
Intervention names associated with the trial.
Caveat: Registry-reported intervention names are trimmed and exact repeats within one source observation are removed; assess field fill and selected provenance in each release's measured coverage.
varies
intervention_typestext[]detail
Intervention categories reported by the source registry.
Caveat: Registry intervention vocabularies differ; exact repeats within one source observation are removed, while release-level coverage and selected provenance remain measured evidence.
varies
countriestext[]search
Countries associated with trial recruitment or study locations, used by country filters.
Caveat: Countries are registry-reported location values with exact repeats removed; missing values do not prove no locations, so review release-level coverage and provenance.
varies

Outcomes and publications3 fields

nametypescopedescriptionsources
outcomesjsonbdetail
Source-owned primary and secondary outcome definitions from the licensed ISRCTN registry contribution.
Caveat: Definitions are registry-reported metadata, not measured result values or efficacy claims.
isrctn
outputsjsonbdetail
ISRCTN output/publication metadata and exact PMID or DOI assertions; external links are relationships, not licenses to copy their targets.
Caveat: Linked files, abstracts, full text, contacts and artifact bodies are excluded.
isrctn
publicationsjsonbdetail
Citation-only PubMed metadata for exact accepted ISRCTN PMID assertions, including identifiers, title, journal, dates, authors, publication types and retraction/comment links.
Caveat: Abstracts, full text and LinkOut bodies are excluded; each item carries a retrieval-date staleness notice and NLM attribution.
pubmed

Regulatory1 fields

nametypescopedescriptionsources
regulatory_recordsjsonbdetail
Health Canada CTD authorization observations linked through exact protocol-number evidence plus at least two independent exact agreements.
Caveat: Health Canada states CTD is not a comprehensive registry; status is source-reported and is not medical, safety or recruitment advice.
health-canada-ctd

NULL policy

A missing field on a record is a missing field: not a zero, not a negative assertion, and not an imputed value. Fields that do not apply to a record are explicitly NULL. Coverage gaps are visible by design.

Audit trail

Every provenance row carries the observed value, source code, source record URL, effective state, and ingest context. Snapshot manifests and source licence notes document which upstream records were cleared for commercial use.

Field requests are reviewed against the same commercial-use source gate as the rest of the export. A field only ships when its source terms permit commercial use in the buyer-facing dataset.