CTIS Trial Results Disclosure 2026: Summary Results, Layperson Summaries and Clinical Study Reports — What Sponsors Must Prepare and When

TL;DR

Takeaway: Public CTIS dates show that 1,100 ended trials with a visible summary-of-results submission date cluster near one year (median 341 days; p75 364). That supports earlier internal targets, but it is not a compliance-rate estimate. Category-specific publication timing, the paediatric six-month deadline, secure-domain activity and other legal facts cannot be resolved from the public extract alone.

Everything in the European disclosure conversation tends to collapse into "post your results within a year." The legal and operational picture is more specific: determine the trial category and applicable clock, separate secure submission from public publication, own the technical and layperson artifacts, and track any CSR obligation tied to a marketing-authorisation procedure [1][2][3].

Visible submission dates cluster near one year. Among 1,100 ended trials with a summary-of-results submission date, the gap from recorded end to public submission date runs median 341 days, p75 364 and p90 399. Layperson-summary dates are similar (n = 1,074 after excluding one pre-end date; median 343, p75 364) [3]. Treat this as a planning distribution, not an adjudication of on-time filing.

A conservative public-visibility screen still finds follow-up work. Among 184 trials ended at least 30 months before the document snapshot, 40 show neither a public summary-submission date nor a public result document: 31/138 Phase I, 6/33 Phase II and 3/13 Phase III [3]. These are records to investigate—not violations—because the public layer cannot establish all category, deferral, waiver, secure-submission or authority facts.

The layperson summary is a governed multi-artifact workstream. 1,081 trials carry 4,052 layperson-summary entries; 627 have multiple entries. Because language metadata is missing on 3,203 entries, document multiplicity cannot be converted into language or translation counts [2]. The sponsor must build its language requirements from the actual concerned Member States and participant populations.

This paper is a disclosure-planning workpaper: the three obligations as the law writes them, the 2024 transparency-rule reset, the measured timing and gap record, and a T-minus calendar with owners per artifact. It is not a compliance audit of any sponsor: a missing public document is not proof of a breach, and the reasons a trial can lawfully show nothing are part of the story.

Three obligations, one article

Takeaway: Article 37(4) of Regulation (EU) No 536/2014 stacks three distinct obligations in one place: the summary of results (Annex IV), the layperson summary (Annex V), and — for trials used in a marketing authorisation — the clinical study report, each with its own trigger and audience.

The Clinical Trials Regulation puts all three duties in the fourth subparagraph of Article 37. First: "Irrespective of the outcome of a clinical trial, within one year from the end of a clinical trial in all Member States concerned, the sponsor shall submit to the EU database a summary of the results of the clinical trial. The content of that summary is set out in Annex IV." Second, in the same subparagraph: "It shall be accompanied by a summary written in a manner that is understandable to laypersons," its content set out in Annex V [1]. Third, the CSR rule: where the trial was intended to support a marketing authorisation, the applicant must submit the clinical study report "within 30 days after the day the marketing authorisation has been granted, the procedure for granting the marketing authorisation has been completed, or the applicant for marketing authorisation has withdrawn the application" [1].

Three audiences, three formats, three clocks. The summary of results is the technical record — the structured scientific summary that a regulator or researcher reads. The layperson summary is a communication artifact written for the public, in the languages of the trial's participants. The CSR is a regulatory dossier document that exists for a different procedure entirely and arrives on a different trigger: not trial end, but the marketing-authorisation decision [4].

The portal's own definitions make the boundaries concrete: the trial's results section "includes the trial's summary of results and layperson summary of results, as submitted by the sponsor on CTIS, and could include the Clinical Study Report, if submitted by the relevant applicant of the Marketing Authorisation" — three document types, three submitting roles, one public surface [4]. Note the last clause carefully: the CSR arrives from the MA applicant, which may be a different legal entity and a different department from the sponsor filing the summaries. In merged and licensed programs it often is. That is why disclosure planning has to name owners across the sponsor–applicant boundary before the clocks start, not at filing time.

Each obligation also has a different control path. The summary and layperson material are linked to trial-results disclosure, while a CSR may be triggered by a later marketing-authorisation procedure and submitted by a different legal function. Public visibility alone cannot establish whether a deadline was missed, so the sponsor's obligation record must retain the category, trigger date, applicable rule, internal target, submission evidence and any accepted exception.

The law also carries its own escape valve, in the third subparagraph: "where, for scientific reasons detailed in the protocol, it is not possible to submit a summary of the results within one year, the summary of results shall be submitted as soon as it is available" [1]. That clause matters for reading the measured record later: some share of late or absent results is legally explained — but it must be detailed in the protocol in advance, which makes it a planning decision, not an after-the-fact excuse.

The two annexes define the artifacts' shape. Annex IV fixes the summary of results as a structured document — trial identification, sponsor, endpoint and outcome data, adverse-event summaries — populated from the trial's final analysis, which is why the disclosure calendar is downstream of the statistical freeze. Annex V fixes the layperson summary's scope and requires a description of the trial in terms a non-specialist can follow [1]. Neither annex is optional and neither is a narrative paper: both are structured deliverables with defined fields, which is why production planning (not writing talent) determines whether the deadlines are met. The summary of results is assembled; the layperson summary is written and then translated; the CSR is compiled. Three different production verbs, three different owner profiles, one calendar.

One more scoping fact, because it drives the Phase I findings later: the obligation binds "irrespective of the outcome" and irrespective of phase. There is no early-phase exemption in Article 37(4). The Commission's Q&A document states it flatly: "According to article 37(4) of the Clinical Trials Regulation a summary of results needs to be submitted to the EU database within 1 year from the end of the clinical trial" — every clinical trial in scope of the Regulation [5].

The deadlines as the law writes them

Takeaway: The one-year clock runs from the end of the trial in all concerned Member States — not the global end — with "end" defined as the last visit of the last subject; paediatric trials get six months; the CSR clock starts at the MA decision. Three definitions decide months of planning.

When the clock starts. "End of a clinical trial" is defined in Article 2(2)(26): "the last visit of the last subject, or at a later point in time as defined in the protocol" [1]. And the one-year period runs "from the end of a clinical trial in all Member States concerned" — not from the global end of a multinational trial that continues outside the EU. The Commission's Q&A spells out the distinction: results are due "within one year from the end of the CT in all MSC in the EU/EEA (and from not the global end of the CT" — and notes that results may legitimately be late "when the clinical trial is still ongoing in third countries and data from that part of the trial are not available" [5]. For a sponsor running a global program, that is a planning fork: the EU clock can expire while the trial is still enrolling elsewhere.

The paediatric short clock. For paediatric trials, the summary of results and layperson summary are due within six months of trial end, not twelve [5]. A six-month clock inverts the production plan: at end of trial, the statistical outputs, the summary writing, the layperson drafting and the translations all have to be essentially staged before the last visit, or the deadline is missed on arrival.

When the CSR clock starts. Not at trial end at all. The CSR is due within 30 days after the marketing authorisation is granted, the procedure completes, or the application is withdrawn [1]. The 30-day trigger is attached to the MA procedure's outcome — whatever it is. EMA's portal documentation confirms the operational consequence: CSRs are submitted "within 30 days from the relevant marketing authorisation decision, regardless of its outcome. Its publication occurs always upon submission" [4]. A withdrawn application still owes its CSRs, thirty days later.

A worked timeline fixes all three clocks in one picture. Take a trial whose last participant in the EU makes their last visit on 15 March 2027, with sites continuing in Brazil until December 2027, under a paediatric investigation plan. The end of trial in all MSCs is 15 March 2027 — the Brazilian continuation does not move it. If the trial includes paediatric participants, the summary and layperson set are due by 15 September 2027; otherwise by 15 March 2028. The global data lock supporting those summaries may be complicated by the ongoing Brazilian arm — which is exactly the situation the Commission Q&A anticipates when it allows lateness "when the clinical trial is still ongoing in third countries and data from that part of the trial are not available," and exactly why the escape valve must be detailed in the protocol in advance [5]. And if the same trial later supports an MA application decided in June 2030, the CSR lands by July 2030 on its own separate 30-day clock, filed by the applicant. One trial, three deadlines spanning years, three filing acts.

Three edge cases that move the clock. The Regulation itself defines what counts as "end" in the situations sponsors actually hit. Early termination: "the date of the early termination shall be deemed to be the date of the end of the clinical trial" — the one-year clock starts at the termination decision, not at the planned end [1]. A trial halted temporarily and not resumed within two years: the two-year expiry date, or the sponsor's decision not to resume, "whichever is earlier, shall be deemed to be the date of the end of the clinical trial" — a trial that quietly died in 2023 has a disclosure clock that started without anyone filing anything [1]. And the least-known clause of the three: where the protocol provides for an intermediate data analysis before trial end and those results are available, "a summary of those results shall be submitted to the EU database within one year of the intermediate data analysis date" — an earlier, parallel clock that exists whenever the protocol says interim, whether or not anyone planned a second submission [1]. All three edge cases share one property: the trigger is automatic and defined by law, not chosen by the disclosure team. A calendar that only models the last-visit-of-last-subject path will misfire on exactly the trials whose stories changed mid-course — the terminated, the halted, and the interim-analyzed — which in any real portfolio is a material fraction.

There is also a notification layer most plans forget. Article 37(2) separately requires the sponsor to notify the Member States concerned, through the EU portal, of the end of the trial in all Member States concerned within 15 days of that point [1]. The 15-day notification is not the results deadline — it is the administrative event that fixes the date from which the one-year deadline is countable in the portal's own record. A sponsor whose end-of-trial notification is late has not formally missed the results deadline, but has created a record mismatch that surfaces in exactly the public view this paper measures.

What "submitted" and "published" mean. This is where sponsors get surprised. Submission is the sponsor's act; publication is the portal's. Under the revised transparency rules (next section), most results documents publish when submitted — which makes the submission date the public date, and makes any redaction or confidentiality review a pre-submission task, not a post-publication repair.

What 18 June 2024 changed

Takeaway: The revised CTIS transparency rules — adopted 5 October 2023, applicable 18 June 2024 — removed the deferral mechanism that let sponsors delay publication up to seven years, moved most results to publish-on-submission, and set the Category 1 adult-only 30-month rule. Post-2024, disclosure planning means planning publication, not planning deferral.

The original CTIS regime included a deferral mechanism that allowed sponsors to delay publication of certain results-related information for up to seven years after trial end — a competitive-protection tool designed for the pre-approval window [6][7]. The revised rules, adopted by the EMA Management Board on 5 October 2023 following a public consultation from May to June 2023, and applicable from 18 June 2024, removed it [6][8].

What replaced deferral is a category-based publication table [6]:

Revised CTIS transparency rules (EMA/263067/2023), publication-timing table — EClinCloud summary of the fetched document [6].

Two consequences deserve a planner's attention. First, the 30-month publication rule for adult-only Category 1 trials is the last remaining time buffer between submission and public visibility — and it is category-specific, not a right the sponsor invokes. Second, everything else publishes on submission. The deferral era's habit — submit, then negotiate visibility — is structurally over; the revised rules move the confidentiality conversation before the submission button [6][7].

It is worth being precise about what died with the deferral mechanism, because the seven-year figure sounds historical and is not. A trial that ended in, say, 2019 under the old rules could have had certain results information withheld from public view until 2026 — overlapping an entire development cycle. The revised rules' rationale, as EMA summarized it, was that "one of the key changes of the revised rules is the removal of the deferral mechanism," replacing a sponsor-controlled visibility timeline with a category-controlled one [7]. For planning purposes the category assignment therefore matters more than any negotiating posture: whether a trial is Category 1 adult-only determines whether submission and publication are separated by 18 months or by zero. Category is driven by the trial's population and its role in pharmaceutical development — decided by the protocol, not by the disclosure team — so the disclosure plan should inherit the category determination as an input, early.

The publication table also quietly answers a question sponsors ask constantly: can we still keep results confidential until the MA? For the summary of results and layperson summary, the answer is essentially no — publication follows the table above, and the table's maximum cushion for adult-only Category 1 is the 30-month point. The CSRs are the partial exception, and only because their trigger is the MA procedure itself: they publish when submitted, which is after the decision. Confidentiality strategy in 2026 is therefore mostly a redaction discipline (commercially confidential information, personal data) applied per document before submission — not a timing game played after it [6].

The CTR applied from 31 January 2022 with a three-year transition, and ongoing EU trials were required to transition by 31 January 2025 [8][9]. The public corpus can still contain trials with different histories and publication states, so a mid-2026 snapshot should not be treated as a clean, homogeneous compliance cohort.

The measured record: visible submissions cluster near one year

Takeaway: Among ended trials with a public submission date, the median recorded gap is 341 days and p75 is 364. The distribution supports setting an earlier internal target, but it mixes legal categories and cannot classify filings as compliant or late.

Joining the CTIS public details file (30 July 2026 snapshot; 3,580 trials with an end date) to submission dates: 1,100 trials carry a summary-of-results submission date. The layperson calculation uses 1,074 non-negative post-end gaps after excluding one record dated before the reported end [3].

CTIS public details, ended trials with submission dates — EClinCloud analysis, accessed August 2026 [3].

The distribution's shape is the planning finding. Median 341 and p75 364 put much of the visible activity near the common adult one-year boundary. P90 399 and p99 645 show a later tail. Without trial-level category, applicable deadline, secure-domain history and regulator determinations, those percentiles cannot be relabelled as on-time or late.

Descriptive cuts show Phase III median 325 days (n = 350) versus 342 for other phases (n = 750), with sponsor medians ranging across the observed groups [3]. These unadjusted comparisons mix trial categories, ending years and portfolio composition; they should not be interpreted as sponsor discipline or causal performance differences.

For planning, the similar layperson and summary dates show that the two public submission records often move together [3]. They do not reveal when drafting or translation began. The control lesson is to work backward from the applicable deadline and start required language work early enough for readability, reconciliation, medical review and submission QA.

Read the percentiles as an operations curve. An adult trial governed by a one-year submission deadline would have little margin if it followed the observed median, while a paediatric six-month clock requires a different workback. The sponsor should calculate margin from the trial's confirmed legal deadline rather than applying 365 days to every record.

The distribution cannot diagnose whether timing reflects capacity, policy, portal operations or the mix of legal categories. It does justify a sponsor-defined internal target earlier than the applicable deadline, with contingency sized to statistical readiness, document complexity, required languages and review cycles. Month ten may be reasonable for some adult trials; it is not a universal legal or operational rule.

A conservative public-visibility screen

Takeaway: To reduce false positives from the adult-only Category 1 publication rule, use a 30-month screen rather than a one-year public-visibility screen. Among 184 trials ended by 23 February 2024, 40 have neither a public summary-submission date nor a public result document. This identifies records for obligation-level review; it does not estimate breaches.

The earlier one-year screen counted 1,864 trials and produced 903 records without a public summary date or result document. That was not a valid headline: adult-only Category 1 results can be published 30 months after trial end, and public fields do not expose every secure-domain action. The corrected screen uses an end date at least 30 months before the 23 August 2026 document snapshot. Of 184 records, 144 have at least one public result document and 40 have neither a public summary-submission date nor a document [3][2].

By phase, the pattern is stark [3]:

CTIS public details and public results-document corpus, ended by 23 February 2024 — EClinCloud analysis, accessed August 2026 [2][3].

The phase-specific denominators are small, particularly Phase III. Their descriptive shares are similar enough that this screen does not support a Phase I-specific failure claim. Phase I contributes 31 of 40 follow-up records mainly because it contributes 138 of the 184 screened trials.

Sponsor-type comparisons from the one-year cohort are removed because they inherit the same publication-timing and secure-domain ambiguity and are unadjusted for phase, category and ending year. A due-diligence workflow should review individual obligation records rather than rank sponsor types from these public absences.

How to use the screen. For each of the 40 records, retrieve the trial category, participant population, legally relevant end date, applicable submission and publication deadlines, secure-domain submission evidence, any protocol-based scientific reason, correction status and authority correspondence. Classify only after that review. The public portal is useful for reconciliation and diligence, but absence is not proof of noncompliance or lack of a disclosure function.

Where the control lands organizationally. Assign a disclosure owner at authorization, maintain an obligation record for every trial, connect end/termination/interim/MA events to the calendar, and reconcile submission evidence to the public record after the applicable publication point. That architecture is justified by the complexity of the rules and the observed follow-up queue; no speculative claim about Phase I teams, CRO ownership or sponsor intent is needed.

The layperson summary is a governed workstream

Takeaway: The public record holds 4,052 layperson-summary entries and 1,289 summary-of-results entries. Multiplicity signals document-management work, but missing language metadata prevents a translation count. Budget from confirmed language requirements, not the 3:1 corpus ratio.

The document corpus tells the volume story [2]. By document type: 4,052 layperson summaries (type 46) across 1,081 trials; 1,289 summaries of results (type 103) across 1,107 trials; 1,062 clinical study reports (type 45) across 165 trials. The combination pattern: 1,027 trials carry both a summary of results and a layperson summary; 125 carry only a CSR; 40 carry all three.

The combination counts describe the public corpus: 1,027 trials have both summary and layperson entries, 40 have all three document types, and 125 have only CSR entries [2]. They do not prove that every applicable artifact is complete or timely. Most importantly, 4,052 layperson entries across 1,081 trials is not a count of languages or translations; entries may reflect languages, versions, corrections or other portal structure.

Documents per trial: median 3, p75 6, p90 11, p99 23, max 126 [2]. 627 of the 1,081 trials with layperson entries have two or more. The maximum and multi-entry counts identify records that may require version-level reconciliation; the metadata does not establish that the multiplicity came from languages.

The layperson summary's content requirements come from Annex V and the Commission expert group's 2018 recommendations, including structure, readability and content for non-specialists [1][10]. The sponsor must determine applicable language requirements from the concerned Member States and participant populations; country count alone is not a language count.

One data caveat becomes a planning instruction: most public layperson documents carry no recorded language code in the portal metadata (3,203 of 4,052) [2]. The register does not carry the language layer, so the sponsor cannot manage it from the register—the language plan has to come from the trial's own concerned-Member-State list and confirmed requirements, with any language vendors and QA rounds scheduled against the applicable submission clock. QA matters twice over: a layperson summary is both a compliance artifact and a future public document; the applicable category determines whether publication occurs on submission or later.

A workable sizing method starts with a hypothetical six-country trial but does not assume six or seven languages. For every concerned Member State and enrolled population, record the accepted language requirement, source version, translation owner, independent review method, readability approach, reconciliation step, approval and target date. Vendors should estimate lead time from the confirmed matrix and revision cycles; generic two-to-four-week or four-to-six-month bands are not evidence. The budget unit is each approved artifact and review cycle actually required, with a portfolio contingency for amendment or correction.

The Commission expert-group recommendations structure the document itself: a short, plain-language account of the trial's aim, who participated, what was done, what results were found, and any identified risks—with advice to avoid technical vocabulary, explain unavoidable terms and keep sentence structure simple [10]. Readability is a content requirement, not decorative editing. Readability work with intended users or a justified alternative can reveal two distinct risks before submission: a version that is technically accurate but hard to understand, and language versions that communicate results inconsistently. Both require controlled correction; publication timing follows the trial category.

The clinical study report in practice

Takeaway: CSRs are rare in the public record relative to summaries—165 trials carry CSR entries, 1,062 documents—because their trigger is the MA procedure, not trial end. The 125 trials with a CSR but no visible summary/layperson entry show why the two workstreams must be reconciled; public metadata cannot determine the legal status of the other artifacts.

The CSR population is structurally different. It exists where results feed a marketing authorisation, and it is filed by the MA applicant within 30 days of the relevant decision [1]. In the details file, 165 trials carry CSR entries; in the document corpus, those trials account for 1,062 entries [2][3]. The public metadata does not establish why each trial has multiple entries.

The combination finding is useful for reconciliation: 125 trials carry a CSR entry and no summary-of-results or layperson-summary entry in the public corpus [2]. The pattern does not diagnose an absent disclosure function; category, publication timing, trial history and portal state must be checked. It does show why the MA/CSR workstream and trial-results workstream need a shared trial-level obligation record.

Publication timing closes the loop: CSRs publish upon submission, always, and their redaction (commercially confidential information, personal data) is negotiated in the MA procedure context, not at trial end [4][6]. The Regulation's own recitals frame how narrow that redaction space is: "in general, the data included in a clinical study report should not be considered commercially confidential once a marketing authorisation has been granted," and clinical trial results "including reasons for temporary halt and early termination, in general, should not be considered confidential" [1]. The default is disclosure; protection is the argued exception.

The planning consequence: the CSR trigger belongs to the MA workstream, but its inputs and evidence span clinical writing, the trial master file and disclosure governance. The 30-day window argues for preparation before the decision: maintain the CSR as a controlled living document, prepare cross-references and proposed redactions, and define the sign-off and CTIS access path while the MA procedure is active. Withdrawal is also a trigger, so the operating model must not assume that work continues only after a favorable decision.

Enforcement: what actually backs the deadlines

Takeaway: The CTR's teeth are Member-State penalties under Article 94 — required to be "effective, proportionate and dissuasive" — plus corrective measures up to revocation under Article 77. Enforcement is national, not EMA-central, and non-submission of public information is named expressly.

Article 94 requires each Member State to "lay down rules on penalties applicable to infringements of this Regulation and shall take all measures necessary to ensure that they are implemented. The penalties provided for shall be effective, proportionate and dissuasive" — and point 2(a) covers expressly "non-compliance with the provisions laid down in this Regulation on submission of information intended to be made publicly available to the EU database" [1]. That is the results-disclosure obligation, named.

Article 77 empowers concerned Member States to require modifications, suspend a trial or revoke authorization in specified circumstances [1]. It should not be presented as the expected sanction for a post-trial disclosure issue; Article 94 and the applicable national penalty framework are the more direct legal references for infringement claims.

Two caveats belong next to those articles. Enforcement is national, so the applicable Member-State rules and authority practice must be checked before making a penalty claim. The public portal is a transparency and reconciliation surface, not an enforcement decision: it displays public records but does not by itself establish breach, intent, sanction or sponsor-level performance.

And the transition removes the last excuse: since 31 January 2025, every ongoing EU trial is under the CTR and in CTIS [8][9]. A trial that predates the portal does not predate the obligation.

Do not rank enforcement layers without Member-State evidence. The defensible operational point is that sponsors need both legal compliance evidence and public-record reconciliation. The portal may inform partners and the public, but a missing public item is not itself an authority finding, and this analysis does not measure reputational or enforcement outcomes [3].

The T-minus disclosure calendar

Takeaway: The calendar is an illustrative sponsor workback, not law. First determine the category, trigger, legal deadline, publication timing and required languages; then set internal dates and owners with enough margin for the study's actual document and review load.

EClinCloud synthesis of Regulation (EU) No 536/2014 Article 37(4), the revised CTIS transparency rules and the measured timing record [1][6][3].

Two design rules survive the data correction. Start required layperson and language work before the final review window, because the public dates often move with summary submissions. Set an internal target earlier than the applicable legal deadline, with margin based on the actual language matrix, statistical readiness, review rounds and correction risk. The corpus ratio cannot size translations and month ten is not universal.

Running the calendar is a governance act, not a printout. Three practices make it real. Assign a named owner per row so that each trigger and artifact has an accountable handoff. Review the calendar against the trial register at a risk-based cadence so early termination, a two-year halt, an interim analysis or an MA outcome opens the applicable assessment promptly. And connect each milestone to evidence: the legal/category determination, disclosure plan, confirmed language matrix, readability evidence, redaction log, submitted package and acknowledgement. The public data do not reveal why any record is absent, so this is a control design derived from the obligation—not a claim that ownership caused the measured follow-up queue.

The calendar also scales down. A single-site trial without a later MA/CSR workstream may have fewer artifacts and owners, but it still needs a documented category, legal deadline, required layperson material, language decision and submission evidence. The public screen cannot tell us whether an absent item reflects a process failure, so use it for reconciliation rather than accusation [3].

The trial-level obligation register

A calendar is only as reliable as its legal inputs. Maintain one obligation record per trial rather than a generic "results due in 12 months" field. At minimum, capture the EU trial number; legal sponsor and MA applicant where different; applicable regulatory regime and trial category; adult/paediatric/PIP status; whether participants were recruited; protocol-defined end event; end date in all concerned Member States; early termination, temporary halt and interim-analysis events; scientific-reasons provision and its protocol location; concerned Member States; required artifact and language decisions; submission deadline; publication rule; internal target; responsible owner; secure-domain evidence; public URL/date; corrections; and authority correspondence [1][4][5].

The distinction between four dates is essential: factual event date, legal trigger date, secure submission date and public publication date. A dashboard that stores only "results posted" cannot explain a Category 1 adult-only record that was submitted securely but is not yet public, or a correction that changes public visibility after the original filing. The obligation record should calculate nothing until the category and trigger have been approved by regulatory affairs.

Assign data provenance per field. Protocol and end-of-trial notification support the trigger; CTIS acknowledgement supports submission; current transparency rules support publication timing; the public portal supports reconciliation. If a field is unknown, mark it unknown—do not infer category from phase, paediatric status from age text, or compliance from a missing public document. The 40-record conservative screen is a queue for this evidence collection.

Reconciliation and escalation workflow

Run reconciliation at three points. Before the internal target, verify that statistical, medical-writing, layperson, language and regulatory owners agree on the version and numbers. After secure submission, retain the acknowledgement, submitted package, timestamps, user and any validation messages. After the applicable public point, compare the public record with the obligation record: correct trial number, document types, versions, dates and visibility. A mismatch becomes an investigated exception, not an automatic breach.

Classify exceptions so portfolio metrics remain honest:

This taxonomy prevents two equally damaging errors: reporting a false compliance failure from public absence, and dismissing a real overdue obligation as a portal problem without evidence.

CRO transfer and MA-applicant handoff

Disclosure ownership often crosses organizational boundaries. A CRO may hold the trial operational file, the sponsor may control CTIS, a medical-writing vendor may hold source drafts and a later licensee may become the MA applicant. Contracts and transition plans should identify who monitors each trigger, who owns CTIS access, who approves content, who funds translations, who retains acknowledgements, who handles corrections and how CSR obligations transfer.

At asset transfer, include an obligation schedule in the data room: all trials, end/termination/interim events, category determinations, submitted/public artifacts, pending deadlines, protocol-based provisions, CTIS evidence and open authority questions. The 125 CSR-only public patterns show why trial-results and MA workstreams must reconcile at trial level, while not proving that any one record is deficient [2]. A buyer should test the evidence chain, not judge from document presence alone.

Portfolio reporting should separate operational metrics from legal conclusions. Useful measures include obligations with approved inputs, packages ready by internal target, secure acknowledgements reconciled, public records checked after the applicable point, open exceptions by class and aging, and records lacking an accountable owner. Avoid publishing a "compliance rate" unless legal reviewers have classified every denominator record against the applicable rules and evidence.

Build an inspection-ready trial sample

A portfolio dashboard can look complete while the underlying evidence is fragmented. Test the system by selecting a small, risk-based sample and asking an independent reviewer to reconstruct each obligation without help from the usual owner. Include at least one completed adult trial, one paediatric or PIP-linked trial if present, one early termination or temporary halt, one Category 1 trial subject to delayed publication, one record with a correction, and one trial connected to an MA decision. The sample is not a statistical compliance estimate; it is a control test covering different clocks and publication states.

For each sampled trial, the reviewer should be able to move from the protocol-defined end event to the factual event evidence, approved category and legal trigger, applicable deadline, internal target, final artifact versions, language decision, secure-submission acknowledgement, publication rule, public reconciliation and any correction or authority correspondence. Record the source, owner and timestamp for every link. If the reviewer has to infer an input from phase, search an inbox for an acknowledgement or ask which version was filed, the control did not pass even if the public document is visible.

Grade findings by control failure, not by embarrassment. A missing evidence link with a timely submission is a documentation/control deficiency; an unresolved mismatch between the filed and public version is a reconciliation exception; an apparent missed deadline requires legal classification and escalation. Preserve both the finding and closure evidence. Repeating the sample after owner, CRO or system changes tests whether the process—not one experienced employee—carries the obligation.

This exercise also improves acquisition and partnering diligence. A buyer can request the same reconstruction pack for material trials and distinguish three things that public CTIS screening cannot: a legally classified obligation, an operationally complete filing and a publicly visible artifact. That separation prevents a portal observation from becoming an unsupported compliance representation while still forcing missing records into an accountable remediation queue [1][3].

Portfolio metrics should be derived from this classified register, not scraped visibility alone. Define denominators before reporting: obligations due in the period; obligations with approved legal inputs; submissions due versus publications expected; and open exceptions eligible for aging. Freeze an as-of date and retain the trial-level population behind each numerator. Report unknown/unclassified records separately rather than excluding them, because a rising completeness rate can otherwise hide an expanding unreviewed backlog.

Use leading and lagging measures together. Leading measures include trigger inputs approved before the planning window, draft packages ready by internal target, language decisions complete and accountable owners assigned. Lagging measures include secure submissions completed by legal deadline, acknowledgements reconciled, public records checked when publication is expected, corrections closed and sampled evidence chains reconstructed. A red leading measure permits intervention; a perfect lagging percentage discovered after an omitted trial offers false comfort.

Changes require denominator control. Trial acquisitions, sponsor transfers, newly discovered end events, category reclassification, corrections and revised publication expectations should create dated adjustments, not overwrite prior portfolio results. That auditability allows governance teams to explain whether a trend reflects process performance or a changed population. Legal/compliance reviewers should approve any metric labeled "on time" or "compliant"; operational teams can safely own readiness, reconciliation and evidence-completeness metrics.

Close each reporting period with an exception review, not just a chart. For every overdue internal target, missing acknowledgement, unexpected publication state or incomplete legal input, record the trial, obligation, aging start, current classification, accountable owner, next action and decision date. Carry unresolved items into the next period with their original age. This prevents a refreshed dashboard from resetting the clock or making a transferred obligation disappear.

Test access and continuity as well as content. Confirm that at least two current sponsor-controlled users can retrieve the obligation record, filed package and acknowledgement; that CRO or vendor credentials are not the only route; and that departure of the named owner triggers reassignment. A technically complete record that cannot be found or submitted by the responsible organization is not operationally ready.

Frequently asked questions

What is the CTIS summary-of-results deadline?

One year from the end of the trial in all concerned Member States — "end" being the last visit of the last subject unless the protocol defines a later point. Paediatric trials: six months. The clock does not wait for the global end of a multinational program [1][5].

Which trials must post summary results?

All clinical trials within the CTR's scope, "irrespective of the outcome" and irrespective of phase—including Phase I [5]. The corrected public-visibility screen does not support claiming that results specifically disappear in Phase I [3].

Is a layperson summary really required for every trial?

Article 37(4) attaches layperson material to the summary of results, with content per Annex V [1]. Applicable language requirements must be confirmed for the trial; 4,052 public entries versus 1,289 summary entries cannot be interpreted as a language count [2].

When is the clinical study report published?

Where Article 37(4) requires the MA applicant to submit a clinical study report, the trigger is the relevant MA outcome—grant, completed procedure or withdrawal—plus 30 days, not trial end. Under the current CTIS transparency rules, that CSR is published when submitted [1][4].

Can a sponsor lawfully post nothing?

In defined circumstances, including the protocol-detailed scientific-reasons provision and cases addressed in the applicable guidance. Determine the legal status from the trial file and authority record. The corrected 30-month public screen does not show a material phase gradient and cannot quantify overlap between public absence and legal noncompliance [1][3].

What changed with the transparency rules in 2024?

Adopted 5 October 2023, applicable 18 June 2024: the deferral mechanism (up to seven years) was removed; results publish on submission, with the 30-month publication rule remaining only for adult-only Category 1 trials; CSRs publish when submitted [6][8].

Who enforces the deadlines?

Member States, through penalties that must be "effective, proportionate and dissuasive" (Article 94, with non-submission of public information expressly covered) and corrective measures (Article 77). Enforcement varies nationally; visibility is universal, because the portal is the record [1].

How many languages does a layperson summary need?

The regulation does not fix one number. Determine requirements from the concerned Member States and participant populations, using the applicable guidance and authority expectations [10]. Public multiplicity—up to 29 layperson entries in the details file and 627 trials with multiple entries—does not identify distinct languages [2][3].

When does the clock start for a trial that ends early?

On the early-termination date itself — the Regulation deems it "the date of the end of the clinical trial," and the one-year (or six-month paediatric) deadline runs from there. A trial temporarily halted and not resumed within two years is deemed ended at the earlier of the two-year expiry or the decision not to resume [1].

Does an interim analysis trigger its own results deadline?

Yes, when the protocol provides for it and the results are available: a summary of those interim results is due within one year of the intermediate data analysis date — a parallel, earlier clock independent of the main trial-end deadline [1].

What is the difference between the summary of results and the clinical study report?

The summary of results is the sponsor-filed, Annex IV structured summary due one year after trial end, published per the transparency table. The CSR is a dossier-grade document filed by the MA applicant within 30 days of a marketing-authorisation decision (grant, completion or withdrawal) and publishing on submission. Different triggers, different filers, different audiences — and, as the record shows, different pipelines that do not automatically know about each other [1][4].

Methodology and limitations

Takeaway: Two joined public-data snapshots support descriptive timing and visibility screens. Primary law determines obligations; the public extract cannot determine compliance by itself.

Data. (1) CTIS public details and summary files, NAS snapshot 30 July 2026, verified byte-identical to the R2 import of 16 August 2026: 12,123 public trials, 3,580 with end dates, submission dates from results fields [3]. (2) R2 public results-document corpus, 23 August 2026: 6,403 entries (types 45/46/103) linked by CT number [2]. The conservative screen uses end date on or before 23 February 2024. (3) Primary law and EMA documents fetched and verified on 30 August 2026.

Computations. Gap = first public summary-submission date minus recorded end date, in days, over trials with both. One negative layperson gap was excluded rather than clamped to zero. Public-visibility follow-up = trials ended at least 30 months before the document snapshot with neither a public summary date nor any result-document entry. Phase bands use the CTIS phase field.

Limitations. The corpus is public: secure submissions, documents under correction or other non-public states may be invisible, so "no public item" never means breach. The snapshots differ by under a month. Trial category and every exception/authority decision are not available for compliance classification. Language metadata is absent on 3,203 of 4,052 layperson entries, so language-level counts are not computable. Document entries may represent languages, versions, corrections or other portal structure.

What this paper is not. Not a legal opinion, not a sponsor audit, and not a compliance-rate claim about any named organization. It is a measured map of the public disclosure record with the law that shapes it.

Conclusion

Takeaway: CTIS public data can improve planning and reconciliation, but it cannot adjudicate compliance. Determine each trial's obligation first, start required layperson/language work early, set a risk-based internal target and reconcile evidence after the applicable publication point.

Three takeaways survive the join.

Visible dates cluster near one year. Median 341 days and p75 364 leave little margin for an adult one-year workback, but the corpus mixes categories. Use the distribution to justify contingency, not to label 17.6% late or mandate month ten [3].

The corrected visibility screen is a reconciliation queue. Of 184 trials beyond 30 months, 40 lack both a public summary date and result document. The similar descriptive shares across phases do not support a Phase I-failure narrative. Close each record with trial-specific legal and submission evidence [2][3].

The layperson summary needs its own owner and evidence trail. Public entry counts show material document-management volume but cannot reveal language count. Size the budget from the confirmed artifact/language matrix and review cycles [2][3].

For different readers the same record yields different tools. Regulatory affairs gets an obligation register and workback. Clinical operations gets event triggers for end, termination and interim analysis. Due diligence gets a public-record reconciliation screen that must be followed by document review. Medical writing gets an artifact/language matrix based on real requirements rather than corpus ratios. The value lies in joining those controls without pretending the public layer contains facts it does not.

EClinCloud's eTMF is one relevant controlled-document surface for the evidence this calendar names; product scope does not calculate Article 37 obligations or establish compliance [11]. The measured record is ours; the obligation register remains sponsor-owned.

Sources

1. European Union, Regulation (EU) No 536/2014 on clinical trials on medicinal products for human use — Article 37(4) (summary of results within one year; layperson summary; CSR within 30 days of MA decision; scientific-reasons escape valve), Article 2(2)(26) end-of-trial definition, Articles 94 and 77.

2. European Medicines Agency, CTIS public results documents — EClinCloud analysis of the public results-document corpus, 23 August 2026 snapshot: 6,403 documents (1,289 summaries of results, 4,052 layperson summaries, 1,062 CSRs) linked to 1,246 trials by CT number, accessed August 2026.

3. European Medicines Agency, CTIS public trial details — EClinCloud analysis of 12,123 public trial detail and summary records (30 July 2026 snapshot): end dates, public submission dates and phases; conservative public-visibility screen end ≤ 23 February 2024, n = 184, accessed August 2026.

4. European Medicines Agency, CTIS public portal: Trial Results (EMA/441151/2024), 21 August 2026 — document-type definitions; CSRs submitted within 30 days from the MA decision regardless of outcome and "publication occurs always upon submission."

5. European Commission, Clinical Trials Regulation (EU) No 536/2014 Questions & Answers, Version 7.2, EudraLex Volume 10, 27 March 2026 — Q&A 6.1 (one year; paediatric six months), point 480 (clock runs from end in all MSCs, not the global end; third-country continuation).

6. European Medicines Agency, Revised CTIS Transparency Rules (EMA/263067/2023) — adopted by the EMA Management Board 5 October 2023; deferral mechanism removal (previously up to seven years); Category 1 adult-only publication at 30 months after end of trial; publication upon submission otherwise.

7. European Medicines Agency, Revised transparency rules for the EU Clinical Trials Information System, 6 October 2023 — announcement of the revised rules and the removal of the deferral mechanism.

8. European Medicines Agency, Guidance and Q&As — CTIS — revised CTIS transparency rules applicable from 18 June 2024; CTR transition from 31 January 2022 to 30 January 2025.

9. European Medicines Agency, Clinical Trials Regulation — from 31 January 2025 all ongoing EU trials must comply with the CTR and be recorded in CTIS.

10. European Commission, Summaries of Clinical Trial Results for Laypersons — Recommendations of the expert group on clinical trials for the implementation of Regulation (EU) No 536/2014, Version 2, 22 February 2018, EudraLex Volume 10 — content and structure of layperson summaries.

11. EClinCloud, eTMF — Electronic Trial Master File — the document-workflow surface for disclosure artifacts; product scope, not a compliance-rate claim.