Request a demo specialized to your need.
3 Ways a 98% Complete eTMF Can Still Fail an Inspection
A high eTMF completeness score has always felt like a reasonable proxy for inspection readiness. Under ICH E6(R3)'s Appendix C, it isn't anymore and the gap between the two shows up in specific, predictable ways once essential records start spanning more than one system.
Appendix C defines essential records as whatever documents, data, and metadata "facilitate the ongoing management of the trial" and allow trial conduct to be reconstructed not a static list of PDFs sitting in an eTMF. That definition, combined with how EU Clinical Trials Regulation and the Clinical Trials Information System already spread key documentation across shared regulatory infrastructure, means essential records now routinely live in CTIS, CTMS, quality systems, safety and pharmacovigilance platforms, and partner repositories not just the eTMF application.
A completeness percentage measured against only one of those systems was never designed to answer the question EMA's 23 July 2025 E6(R3) implementation date is actually asking.
Here are three specific ways that gap plays out.
1. A Country Can Look "Green" While Key Documents Sit Unfinished Elsewhere
A CTMS country view can show every expected milestone hit, while the actual Part II documents supporting one of those milestones are still being finalized in a separate system. The CTMS status reflects operational progress. It doesn't reflect whether the specific document an inspector would ask for is actually ready and those are two different questions with two different answers.
2. An eTMF Zone Can Look Complete While the Real Version Lives in a Quality System
Under Section C.2, some essential records SOPs, validation evidence, master service agreements — can legitimately sit outside the formal TMF, as long as they remain accessible and governed. That's a reasonable provision on its own. But it means an eTMF zone can show 100% of its expected placeholder documents filed, while the most recently approved version of a monitoring plan actually lives in a quality repository the eTMF completeness score never checks.
3. A Safety Signal Can Be Thoroughly Documented While the TMF View Looks Thin
Pharmacovigilance workflows can capture a safety escalation completely and appropriately, entirely within a dedicated safety system. If the TMF's own view of that same trial doesn't reflect a link to that documentation, the TMF looks incomplete on a topic that was, in reality, handled correctly — just somewhere the completeness metric doesn't reach.
What These Three Have in Common
None of these are failures of an individual system. Each system, on its own, may be doing exactly what it's supposed to do. The gap is that no single completeness number reflects whether someone can move from a specific inspection question show me the essential records behind this protocol amendment, this monitoring strategy change, this safety escalation to the full set of records it touches, across every system that holds a piece of the answer.
Appendix C's real shift is from treating the TMF as a box to be filled to treating it as a navigational layer across systems that were never going to be consolidated into one.
Single-System Score vs. Manual Cross-Reference vs. a Navigational Layer
| Single-System Completeness % | Manual Cross-System Check | Unified CTMS↔eTMF Navigational Layer | |
|---|---|---|---|
| What it actually measures | Filing rate in one system | Whatever someone manually reconciles that day | Records across every connected system |
| Catches a document finalized elsewhere | No | Only if someone happens to check | Yes linked via governed pointers and metadata |
| Speed of answering a specific inspection question | N/A — the score doesn't answer questions | Hours to days of manual tracing | Minutes, via a direct navigation path |
| Version control across systems | Only within the one system it measures | Manually verified, inconsistently | Tracked as records are classified and linked |
| Reflects real inspection readiness | Partial, sometimes misleadingly so | Depends entirely on the person doing the check | Closer to what Appendix C is actually asking for |
What a Navigational Layer Actually Looks Like
Cloudbyz runs CTMS and eTMF on a shared Salesforce data, security, and audit spine, which gives sponsors a natural place to build the kind of cross-repository navigation Appendix C assumes. CTMS holds the canonical view of studies, countries, sites, milestones, and oversight activity.
eTMF holds the sponsor's TMF artifacts, structured against the TMF Reference Model taxonomy. Inside that shared spine, the AI eTMF Agent classifies incoming documents, tags them with study, country, and site context, captures author and approver metadata, and links each one back to the CTMS milestone or activity it supports.
When a document legitimately lives elsewhere a CTIS record, an SOP in a quality repository, an agreement in an MSA library the agent stores a governed pointer and its metadata rather than a duplicate file, so navigating from CTMS into "essential records" surfaces both what's held in eTMF and what's held externally but still in scope, with an audited link connecting them.
When it detects something ambiguous conflicting versions, inconsistent country assignments, unclear ownership it escalates that to a person with enough context to resolve it, rather than resolving it silently or leaving it buried in an aggregate score.
How completely this closes the gap for a given organization depends on how many external systems are actually connected into the spine and how consistently new documents get classified but the underlying shift, from a single number to a navigable path across systems, is what Appendix C is actually asking sponsors to demonstrate.
See What Navigating Essential Records Looks Like
Subscribe to our Newsletter
