PhlexTMF

Vendor: Cencora (PharmaLex / Phlexglobal) · Enterprise

PhlexTMF is a dedicated electronic Trial Master File system: it holds the regulator-mandated evidence file for a clinical trial and manages each document from the moment it is expected to the moment the file is declared inspection ready. Unlike broader platforms it does one job, and its design reflects that - documents are indexed against the TMF Reference Model with AI suggestions at upload, expected-but-missing records are held as intelligent placeholders that drive completeness, and issues are chased through a built-in eQuery mechanism. Ownership has moved twice: Phlexglobal merged into PharmaLex in 2022, and PharmaLex was acquired by AmerisourceBergen, now Cencora, in January 2023.

Trial Master File (TMF)

The lifecycle of a trial master file document, from the moment the trial expects it through capture, classification against the TMF Reference Model, quality review and query remediation, to filing in a steady state and the completion of its milestone. The regulator-mandated evidence file for a clinical trial, and what an inspection actually examines.

Activities

Create Expected Document Placeholder - An intelligent placeholder is raised for a document the trial expects, tied to a collection event or milestone. The placeholder's creation is the moment the document becomes due.
Upload Document - A document is uploaded into the TMF. The upload timestamp starts the document's measurable life and is the anchor for timeliness.
Apply AI Indexing Suggestions - The product's AI proposes Zone, Section, Artifact, Sub-artifact, Country and File Type at the point of upload, trained on millions of documents, to get filing right first time.
Confirm Document Indexing - A user accepts or corrects the suggested indexing. Corrections are the signal worth mining - they show where the trial's filing is genuinely ambiguous.
Fulfil Placeholder - The indexed document is matched to its expected-document placeholder, moving the trial's received-versus-expected position.
Perform Quality Review - A TMF quality review assesses the document against completeness and quality criteria, on a regular, risk-based or milestone-driven cycle.
Raise eQuery - A quality issue is raised against the document or the gap and tracked centrally for remediation.
Respond to eQuery - The owner answers the query, typically by correcting metadata or supplying a better copy.
Close eQuery - The query is closed once the remediation is accepted. Query open-to-close duration is the direct measure of remediation drag.
Reject Document for Reupload - The document is rejected outright and must be uploaded again, rather than corrected in place. This restarts the document's path.
Approve Document - The document passes review and is accepted into the trial master file.
File Document as Final - The document is filed in its final state in the correct place in the TMF structure. The end of the document's own case.
Run TMF Completeness Review - A completeness review compares documents received against documents expected across trial, country, site, zone and vendor.
Declare Inspection Readiness - The trial master file, or a defined slice of it, is declared ready for inspection.

Case attributes

TMF Artifact (string) - TMF Reference Model artifact - the document type.
TMF Section (string) - TMF Reference Model section within the zone.
TMF Sub-artifact (string) - The most granular classification level, where the model defines one.
TMF Zone (string) - TMF Reference Model zone, one of the 11 top-level zones.
File Type (string) - File type of the uploaded document.
Document ID (string) - Identifier of the TMF document. The case id when the case is a single document.
Country (string) - Country the document is filed against.
Site (string) - Site the document is filed against, for site-level records.
Study (string) - Trial the document belongs to.
Vendor (string) - Vendor or organisation that supplied the document, used in received-versus-expected oversight.
Owning Milestone (string) - Collection event or milestone the document is due against.
Indexing Corrected (boolean) - Whether a user overrode the AI's suggested indexing at upload.
eQuery Count (integer) - How many eQueries were raised against the document.
Document Status (string) - Where the document stands in the configured workflow.
Upload Date (datetime) - When the document was uploaded.

Process-mining benefits (historical)

The true time from upload to filed as final [Compliance]: Measure how long documents actually take to move from upload to final filing, and look at the distribution rather than the average. Inspection risk sits in the long tail, and it is usually concentrated in particular artifacts and particular sites rather than spread evenly across the file.
Where the AI's filing suggestions get overridden [Efficiency]: Every time a user corrects the suggested Zone, Section or Artifact at upload, that is a signal. Concentrations of corrections show which artifacts are genuinely ambiguous for this trial, which is both a training opportunity and a candidate for a configuration change.
eQuery loops that quietly consume the team [Cost]: Count documents that went round the query loop more than once and measure open-to-close duration by reason. Remediation drag is invisible in a completeness percentage but it is where a large part of TMF effort actually goes.
Placeholders that were never fulfilled [Compliance]: Every placeholder still unfulfilled is a specific, nameable hole in the trial master file. Ranking them by artifact, country, site and vendor turns a percentage into a worklist and shows whether the gaps follow a particular supplier.
Documents that arrive after the milestone they belong to [Compliance]: Compare when a document was filed as final against the planned and actual dates of its owning milestone. Documents that habitually land after their milestone are exactly what an inspector samples for, and they repeat across trials.
How much rework each vendor and CRO generates [Cost]: Rejections and queries are not evenly distributed across the organisations supplying documents. Attributing them by vendor turns TMF quality from a general complaint into a specific, evidenced oversight conversation.

Process-mining benefits (action)

A daily queue of documents past their expected filing point [Service]: Rank documents and unfulfilled placeholders that are already late against their owning milestone, so the team works what threatens inspection readiness rather than the whole backlog.
Alert when eQueries age past your threshold [Service]: Trigger on queries that have been open, or answered but not closed, longer than you allow. Queries stall silently because nobody owns the waiting, and the ageing ones are the ones that end up in a finding.
Warn before a completeness review runs on a known gap [Compliance]: Where a scheduled review is approaching and its scope still contains unfulfilled placeholders, flag it in advance. Fixing the gap beforehand is cheaper than recording a finding and remediating afterwards.
Route rejected documents straight back to their uploader [Efficiency]: When a document is rejected for reupload, send it back to whoever uploaded it with the reason attached, instead of returning it to a shared queue where it waits for triage.
Catch protocol amendments that move the goalposts [Compliance]: A tracked eTMF event changes what the trial expects. Trigger on new events so the newly expected documents are chased from the day the amendment lands rather than the next review cycle.
Keep a live inspection-readiness view by zone, country, site and vendor [Compliance]: Maintain the received-versus-expected picture broken down the way oversight and inspectors ask for it, so readiness is a standing fact rather than something assembled under pressure when an inspection is announced.

Data tables

document_audit_trail - Conceptual entity, not a verified vendor schema name - PhlexTMF publishes no data model. The per-document history of who did what and when. The vendor states the Exchange Mechanism export preserves audit trails alongside metadata, which is what makes a PhlexTMF extract minable rather than merely a content dump. Confirm the exact columns against a real extract.
equery - Conceptual entity, not a verified vendor schema name - PhlexTMF publishes no data model. An eQuery: a tracked issue raised against a document or a gap, chased to closure through the product's centralised remediation tracking. The direct measure of TMF quality effort.
quality_review - Conceptual entity, not a verified vendor schema name - PhlexTMF publishes no data model. A TMF quality review: a regular, risk-based or milestone-driven completeness and quality assessment over a defined slice of the file.
study - Conceptual entity, not a verified vendor schema name - PhlexTMF publishes no data model. The trial. The top of the hierarchy every other record hangs from.
tmf_document - Conceptual entity, not a verified vendor schema name - PhlexTMF publishes no data model. One TMF document with the metadata that places it in the trial master file. Delivered inside the eTMF-EMS export bundle, inventoried by exchange.xml. The case record when the case is a single document.
tmf_placeholder - Conceptual entity, not a verified vendor schema name - PhlexTMF publishes no data model. An intelligent placeholder: a document the trial expects but has not yet received. Placeholders are what turn completeness from an opinion into a measurement, and an unfulfilled placeholder is a hole in the file.
ems_export_bundle - Published standard, unlike the entities above. The eTMF Exchange Mechanism Standard export itself: a predefined folder tree plus an exchange.xml inventory validated against exchange.xsd, with checksums verified on receipt. This is the container the other entities arrive in.
milestone - Conceptual entity, not a verified vendor schema name - PhlexTMF publishes no data model. A trial milestone that expected documents are grouped under and measured against. The deadline TMF timeliness is judged by.
site - Conceptual entity, not a verified vendor schema name - PhlexTMF publishes no data model. A participating investigator site. Site-level records are the bulk of a trial master file and the usual source of completeness gaps.
tmf_event - Conceptual entity, not a verified vendor schema name - PhlexTMF publishes no data model. An eTMF event: a protocol amendment or collection event that changes what the trial expects. Events are why the expected-document set is not static, and they explain sudden movements in completeness.
tmf_artifact_taxonomy - The TMF Reference Model taxonomy the file is indexed against: 11 zones, their sections, and roughly 250 artifacts with their sub-artifacts. Reference data maintained by CDISC, not by the sponsor.

Explore this interactively in the mindzie Knowledge Base.