Decentralized Endpoint Collection: FDA, EMA, and ICH E6(R3) Annex 2 Operating Rules

When clinical development teams evaluate decentralized clinical trials (DCTs), operational discussions often conflate three distinct methodological decisions into a single tactical question: where an endpoint is measured, what evidentiary validity the measurement possesses, and who supplies the digital or physical collection hardware. Moving a trial procedure from an investigator site into a participant's home or a community clinic resolves geographic accessibility, but it does not evaluate whether the underlying instrument measures what it purports to measure, nor does it establish whether Bring Your Own Device (BYOD) software behaves identically to provisioned equipment.

Operationalizing trial endpoints in remote settings means isolating the collection-location decision as an independent protocol governance problem. Demonstrating that a digital health technology (DHT) or clinical outcome assessment (COA) is fit for purpose is a separate verification, validation, and usability job—evaluated in our dedicated analysis of fit-for-purpose evidence for digital health technologies. Similarly, determining whether participants capture patient-reported outcomes on personal smartphones or sponsor-provisioned tablets is an engineering comparability problem governed by psychometric equivalence standards, as detailed in our guide on BYOD versus provisioned eCOA device strategies. Furthermore, while computerized systems capturing endpoints require comprehensive validation, as set forth in our ICH E6(R3) endpoint and eClinical implementation checklist, generic system validation does not determine who has the legal authority to administer an assessment away from an investigator's physical site.

This guide addresses the specific regulatory and operational rules that govern endpoint capture outside traditional clinical trial sites. It answers the central question confronting clinical trialists, biostatisticians, and clinical operations leaders: What operating rules do the U.S. Food and Drug Administration (FDA), the European medicines regulatory network (EMRN), and ICH E6(R3) Annex 2 set for collecting endpoint observations away from traditional trial sites—who may collect which observations, what the protocol and data-management documents should lock about visit location, and what records should reconstruct originator, date, and setting?

In September 2024, the FDA issued its final guidance for industry, investigators, and other interested parties titled Conducting Clinical Trials With Decentralized Elements. Jointly developed by the Center for Drug Evaluation and Research (CDER), the Center for Biologics Evaluation and Research (CBER), the Center for Devices and Radiological Health (CDRH), and the Oncology Center of Excellence (OCE), the document was formally announced in the Federal Register on 18 September 2024 (89 FR 76481; FR Doc. 2024-21078; Docket No. FDA-2022-D-2870).

The 2024 final guidance fulfills section 3606(a)(2) of the Consolidated Appropriations Act, 2023, which directed FDA to issue final guidance that includes recommendations to clarify and advance the use of DCTs. The Federal Register notice of availability states that section 3606(b) content is addressed by reference to FDA's December 2023 guidance, Digital Health Technologies for Remote Data Acquisition in Clinical Investigations, establishing a clear regulatory separation between the operational architecture of remote trial visits and the technical validation of digital sensor measures.

The final guidance replaces the draft version published on 3 May 2023 (88 FR 27900) and incorporates several substantive modifications in response to public comments:

• Elimination of the Local HCP Task Log: The Federal Register notice of the final guidance records removal of draft language on maintaining a task log of local HCPs. The final guidance states that investigators do not need to maintain a log of local HCPs. Case-history reports from local HCPs should include the HCP's name and the date activities were performed.

• Clarification of Data Variability and Confounding: FDA added explicit cautions regarding data variability introduced by mixed-location capture, highlighting how unsupervised participant self-assessments or non-standardized community measurements can alter endpoint distribution and inflate error variance.

• Refined Local HCP Oversight Standards: The final text separates usual-practice activities that do not require detailed protocol, brochure, or investigational-product knowledge—performed by local HCPs who are not trial personnel or subinvestigators—from research-unique activities that should be performed by qualified, protocol-trained trial personnel. Those trial personnel should be listed on Form FDA 1572 when they contribute directly and significantly to the data (for example, assessing adverse events, performing a clinical outcome assessment, or applying a protocol-defined scoring system).

• Physical Location Requirement for Inspections: FDA clarified that a physical location is needed for inspections. For inspections of clinical investigators, the investigator should identify a physical location where a responsible person is available to facilitate inspector access to trial-related records and interviews with trial personnel. For INDs, that location is generally the address on Form FDA 1572; other physical locations may be identified. The guidance does not require the protocol to name a single brick-and-mortar inspection address as a decentralized-design rule.

Methodologically, the foundational principle of FDA's DCT guidance is legal equivalence: FDA's statutory and regulatory requirements for clinical investigations under 21 CFR part 312 (Investigational New Drug applications) and 21 CFR part 812 (Investigational Device Exemptions) apply equally to trials with decentralized elements and traditional site-based trials. The guidance contains nonbinding recommendations representing FDA's current thinking. Unless a specific regulation or statute is explicitly cited, the use of the term 'should' denotes suggested or recommended best practice rather than a legally enforceable requirement. As of 13 September 2026, FDA's HTML landing page still lists the September 2024 DCT guidance (docket FDA-2022-D-2870), with content current as of 16 October 2025. That document is FDA's current thinking on decentralized trial activities, not a substitute for cited statutes or regulations.

The operational control of remote endpoint collection begins in the trial protocol itself. In traditional site-based clinical trials, environmental factors, diagnostic instruments, and assessor administration are tightly constrained by institutional standard operating procedures. When assessments disperse into participant homes, mobile units, or community medical offices, the protocol should include, as applicable, specific instructions that limit variability in the collected data.

FDA's September 2024 guidance states that the protocol should specify which visits will be conducted at traditional clinical trial sites, which will be conducted remotely, and which can be left to participants' choice. Allowing participants to choose whether an assessment is performed at a traditional site or remotely may result in bias; the guidance's example is that sicker participants may prefer remote visits. If place of assessment affects the observed value, mixed visit assignment can confound treatment-arm comparisons. That is a design caution, not a finding that participant choice is prohibited.

Tests performed independently by participants at home may introduce more variability than tests performed under supervision at traditional sites. FDA's example is spirometry. Training or video supervision during a telehealth visit may reduce that variability. The guidance does not publish a numeric comparability threshold, a required coefficient of variation, or a certified-manual specification for home tests.

Section III.A of FDA's guidance flags a related statistical challenge for non-inferiority trials. When the effect size of an active control was determined from data collected at traditional sites, the same effect size may not be seen when that control is assessed remotely, which can challenge calculation of a non-inferiority margin. Sponsors should consult the relevant FDA review division when planning a non-inferiority trial in a DCT setting. The guidance does not require a named historical-comparability package or treat consultation as a binding pre-condition of protocol freeze.

These design safeguards directly align with the critical-to-quality framework articulated in our guide on ICH E8(R1) critical-to-quality endpoints. Visit-location specification belongs in that quality-by-design conversation before protocol freeze. Intercurrent events caused by missed remote visits or device failure remain a separate estimand question, as detailed in our guide to ICH E9(R1) estimands and intercurrent events.

The most consequential collection-location determination is who may capture a given observation. FDA's September 2024 guidance uses a usual-clinical-practice test, not a license-class catalog and not a claim that every community clinician is barred from every trial task.

Local healthcare providers—FDA's examples are doctors or nurses located close to participants—may perform trial-related procedures only if those activities do not differ from what they are qualified to perform in clinical practice and do not require detailed knowledge of the protocol, investigator's brochure, or investigational product. FDA's examples are physical examinations or obtaining vital signs.

Because local HCPs delivering routine care do not make specialized research judgments, they are not considered trial personnel and would not be considered subinvestigators in a drug trial. Investigators do not need to maintain a log of local HCPs. Case-history reports from local HCPs should include the HCP's name and the date activities were performed. Investigators should communicate protocol-specific instructions to local HCPs—FDA's example is having participants rest for five minutes before measuring blood pressure—and should review local-HCP data regularly, following up on values that are missing, concerning, or appear to be in error.

Trial-related activities that are unique to the research study and/or require detailed knowledge of the protocol, investigator's brochure, or investigational product should be performed by qualified trial personnel who have been trained on the protocol. Specifically, clinical outcome assessments (COAs)when trial personnel contribute directly and significantly to the trial data—FDA's examples are assessing adverse events, performing a clinical outcome assessment, or applying a protocol-defined scoring system—or require that detailed knowledge, they should be included on Form FDA 1572 as subinvestigators. That listing rule applies to the people who perform those assessments. It does not convert a participant-completed electronic PRO into a local-HCP task, and it does not by itself qualify the instrument.

Obtaining informed consent is an investigator responsibility. FDA does not consider obtaining informed consent an appropriate activity for a local HCP, because the person obtaining consent should, among other things, have detailed knowledge of the protocol, investigational product, and potential risks. Remote consent may still be considered as part of a DCT under 21 CFR part 50; the local-HCP limit is about who may conduct that process, not a ban on telehealth consent by investigators.

Understanding these rater boundaries is essential to maintaining endpoint integrity, as explored in our related analysis of 21 CFR Part 11 and investigator oversight for eCOA. Incorrect delegation is an inspection and data-reliability risk; FDA's DCT guidance does not state that mis-delegated observations are automatically excluded.

graph TD
    Start["Trial-related assessment identified"] --> Q1{"Does the activity match usual clinical practice AND not require detailed protocol, brochure, or investigational-product knowledge?"}
    Q1 -- "Yes" --> Q2{"Is the activity obtaining informed consent?"}
    Q2 -- "Yes" --> Inv["Investigator responsibility. FDA does not consider informed consent an appropriate local-HCP activity."]
    Q2 -- "No" --> LocalHCP["Local HCP such as a doctor or nurse. Not trial personnel. Not a subinvestigator. Record name and date. No task log. Investigator reviews the data."]
    Q1 -- "No" --> TrialStaff["Qualified trial personnel trained on the protocol. List on Form FDA 1572 when they contribute directly and significantly to the data, for example performing a COA, protocol-defined scoring, or adverse-event assessment."]

Managing endpoint observations across geographically dispersed clinical environments needs explicit origin, flow, and originator documentation. In traditional trials, electronic data capture (EDC) systems ingest data from a finite set of institutional clinical sites. In decentralized operations, data streams converge from patient smartphones, wearable sensor hubs, mobile phlebotomy databases, community clinics, and central reading laboratories.

To account for multiple sources of endpoint data, FDA recommends that the sponsor include at least the following in a data management plan or other trial-related documents:

• End-to-End Data Origin and Flow Diagrams: Data origin and data flow from all sources to the sponsor, for example a diagram from creation to final storage.

• Remote Data Acquisition Specifications: Methods and technologies used for remote data acquisition from trial participants, trial personnel, and contracted service providers, including local laboratories and local HCPs.

• Service Provider Master Directory: A list identifying service providers for data collection, handling, and management.

Study records should capture four visit attributes: (1) visit type (telehealth encounter versus in-person consultation); (2) visit location (participant home, community healthcare facility, or traditional trial site); (3) date of visit; and (4) the authorized data originator. For telehealth visits, study records should also include the name of the person who conducted the visit.

The concept of the authorized data originator is central to FDA's eSource framework. Citing its September 2013 guidance, Electronic Source Data in Clinical Investigations, the DCT guidance incorporates the 2013 eSource recommendation that a list of all authorized data originators (persons, systems, devices, and instruments) should be developed and maintained by the sponsor and made available at each clinical site. Remote trial personnel or local HCPs submitting data directly into the eCRF should be included on that list. That originator-listing rule is a collection-record control; it is not a full ALCOA how-to and is not the unpublished dedicated eSource-originator article.

Section III.J of the DCT guidance distinguishes live telehealth from electronic records. FDA considers real-time video and audio interactions a live exchange of information between trial personnel and trial participants. These live interactions are not electronic records and therefore are not subject to 21 CFR part 11, although other telehealth laws may apply. not an electronic record subject to Part 11 controlsThe visits must be documented (for example, visit notes) under the case-history regulations FDA cites. If those documents are electronic, they must be captured in systems subject to 21 CFR part 11. That split does not replace the published Part 11 eCOA evidence article.

These computerized system requirements link directly to the risk-based assurance strategies detailed in our analysis of validating computerized systems for endpoint capture. When a trial permits participants to use their own DHTs, a sponsor-provided DHT should remain an option so participants without the protocol-specified device are not excluded, and sponsor-provided telecommunication services should be available as needed. That is an access rule. It does not choose BYOD versus provisioned supply as the measurement-comparability strategy, as analyzed in our review of eCOA mixed-mode measurement comparability.

In the European Union, the governance of decentralized clinical trial elements is defined by the European medicines regulatory network (EMRN), comprising the European Commission, the Heads of Medicines Agencies (HMA), and the European Medicines Agency (EMA). On 1 October 2025, the EMRN issued Recommendation paper on decentralised elements in clinical trials Version 02. Developed under the priority Accelerating Clinical Trials in the EU (ACT EU) initiative and endorsed by the Clinical Trials Advisory Group (CTAG) on 15 October 2025, the paper was published on 29 October 2025 in EudraLex Volume 10, officially superseding the network's initial December 2022 recommendation paper.

The legal nature of the EMRN recommendation paper is nonbinding: it provides harmonized guidance to trial sponsors and national competent authorities but does not establish binding Community law. Under EU jurisprudence, authoritative interpretation of European Union legislation—such as the Clinical Trials Regulation (EU) No 536/2014 (CTR)—remains the exclusive constitutional prerogative of the Court of Justice of the European Union (CJEU).

Version 02 introduces critical substantive advancements over the superseded 2022 release:

• Explicit Cross-Border Trial Statement: The front page of Version 02 adds a statement on cross-border trials with decentralised elements. For those trials, the sponsor should check national requirements and contact the Member State national contact point if authorisation is needed in the participant's home country. V02 does not itself grant or deny that authorisation.

• Updated National-Provision Overview: Version 02 updates the national-provision overview. The appendix is for guidance only and is not a complete scenario list; it remains at the assessing Member State's discretion whether a given decentralised element is acceptable in a specific trial. This page does not reprint that matrix as national law.

• Rejection of Outdated 2022 Positions: Academic landscape articles and vendor pages that still treat the December 2022 paper as current EU advice are stale on the October 2025 cross-border statement and updated national appendix. Vendor claims that EMA legally requires hybrid models in every Member State overstate V02. For home procedures, participants should be given the opportunity to visit the investigator in person if needed or preferred; that is a recommendation, not a legal ban on fully remote designs in every Member State.

Under EMRN V02, introducing decentralized elements is conceptualized as an extension of the clinical trial site that encompasses the participant's home, without waiving investigator or sponsor legal obligations under the Clinical Trials Regulation and ICH E6. The protocol should show that the sponsor and investigator remain in full control of their respective responsibilities, including data processing, communication flow, participant protection, and data reliability. The investigator retains ultimate responsibility for trial-related medical decisions. Investigators should have continuous and complete access to source data generated on-site or off-site and to source data reported to the sponsor.

For endpoints measured and entered directly by participants—especially primary efficacy, key secondary, or safety endpoints captured via digital apps or home sensors—EMRN V02 imposes heightened operational safeguards:

• Participant-Entered Endpoint Controls: The risk of erroneous data entry for data measured and entered directly by participants, especially on primary, key-secondary, or safety endpoints, should be minimised by appropriate measures. V02 does not name a required edit-check catalog.

• Home Suitability and Burden Assessment: For procedures at home, the investigator should ascertain that the home situation and premises are suitable. Inclusion and exclusion criteria may need home-adequacy provisions. Home procedures should not add risk to the participant or to data reliability, and participants performing tasks, including digital collection, need training and a burden assessment. Refrigeration and restricted-access checks in V02 belong to investigational-product storage at home, not to every endpoint observation.

• Safety Alert Workflows: Review frequency of incoming digital data should be based on relevance to safety and efficacy. The protocol should specify how digital-tool safety alerts will be handled. Participants should be told that the investigator may not review ePRO or similar data in real time and that safety concerns must be reported directly to the investigator. V02 asks for a direct contact line for support during home procedures; it does not require a 24/7 channel.

• In-Person Visit Option: Participants should be given the opportunity to visit the investigator in person if needed or preferred. That is current EMRN advice for home procedures, not an unconditional statutory right and not a complete harmonisation of Member State consent rules.

• Computerized Systems Compliance and Contingency: IT devices and technologies used for decentralised collection should be fit for reliable data collection and handling in accordance with the protocol, and computerised systems and electronic clinical data should comply with the EMA Guideline on computerised systems and electronic data in clinical trials (in effect 9 September 2023). A contingency plan should be in place for malfunction of a digital tool or disruption of a planned decentralised visit for identified critical-to-quality decentralised elements. Completing that plan is not computerized-system validation of a capture platform and is not DHT fit-for-purpose evidence.

The overarching international standard harmonizing decentralized trial operations is ICH E6(R3) Annex 2. Reaching Step 4 of the International Council for Harmonisation process on 3 June 2026, Annex 2 was officially consolidated with the core GCP Principles and Annex 1 on 16 June 2026. Annex 2 establishes dedicated Good Clinical Practice considerations for clinical trials incorporating decentralized elements, pragmatic design elements, and real-world data (RWD) sources.

Regional implementation timing still differs, and those dates are not endpoint acceptance:

• European Union Implementation: The EMA Committee for Medicinal Products for Human Use (CHMP) adopted Annex 2 on 25 June 2026, publishing the Step 5 text (EMA/CHMP/ICH/495903/2024). Annex 2 will officially come into effect across the EU on 15 January 2027. (Principles and Annex 1 became effective in the EU on 23 July 2025).

• United States Implementation: As of 13 September 2026, FDA's public guidance portal still lists E6(R3) Annex 2 as December 2024 draft guidance, not for implementation (docket FDA-2024-D-5601). Until FDA finalizes that annex, the September 2024 DCT guidance remains the current U.S. collection-location document. It is still nonbinding except where a cited statute or regulation applies.

Methodologically, ICH E6(R3) Annex 2 defines decentralized elements as trial-related activities conducted outside the physical location of the investigator, encompassing visits at participant homes, community medical centers, mobile trial units, or virtual interactions conducted via DHTs. In the Annex 2 glossary, a DHT used to capture participant data is classified as a 'data acquisition tool'—a definition establishing operational governance that must not be confused with regulatory device clearance or clinical outcome assessment qualification.

Annex 2's collection-location operating rules are protocol specification, investigator oversight, and remote-data governance. They use should-language and are not an endorsement of any specific decentralised methodology:

• Section 3.2.1 (Methodological Justification): Section 3.2.1 states that specific design elements, data sources, and, where appropriate, operational approaches should be adequately described in the protocol and that their use should be justified. The rationale, fitness for purpose, and feasibility of the methodologies should be briefly explained and may be supplemented in protocol-related documents.

• Section 3.2.2 (Usual Clinical Practice Integration): Section 3.2.2 states that when usual-clinical-practice activities are incorporated, the protocol should clearly describe how, by whom, and under what circumstances they will be performed.

• Section 3.2.3 (Data Source Variability): Section 3.2.3 states that the impact of data variability within or between sources or practice settings should be considered in design and discussed in the protocol or protocol-related documents, for example the SAP. Annex 2 does not prescribe a numeric site-versus-home comparability threshold.

• Section 3.2.5 & Section 3.9 (Safety Information Routing): Sections 3.2.5 and 3.9 state that the protocol or related documents should describe how safety information from DHTs, in-person visits, and remote visits is collected and made available to the investigator in a timely, relevant, meaningful, and manageable way, and what actions the investigator should take.

• Section 3.5.2 (Data Security and Privacy): Section 3.5.2 requires special attention to data-security vulnerabilities, including cybersecurity and privacy, for remote data collection such as remote visits, DHTs, and EHR extraction. Annex 2 does not name HIPAA or GDPR in that subsection; those statutes remain separate regional duties.

To guide trial development teams in navigating these overlapping regulatory frameworks, the table below synthesizes the core operational requirements across the FDA, EMRN, and ICH E6(R3) Annex 2.

Four operating rules follow from the current FDA, EMRN, and Annex 2 texts. They do not constitute FDA or EMA acceptance of an endpoint:

1. Lock Visit Setting in the Protocol and SAP: Specify which visits are on site, which are remote, and which may be participant choice, and include performance instructions that limit variability. If location-driven variability is a design issue, discuss it in the protocol or SAP. Estimand-aligned sensitivity analysis for missing data remains a separate job, as detailed in our guide on missing data and estimand-aligned sensitivity analysis.

2. Enforce the Local HCP Boundary: Do not treat local HCPs as subinvestigators when they perform usual-practice activities that do not require detailed protocol knowledge. Research-unique assessments, clinical outcome assessments performed by trial personnel, protocol-defined scoring, and informed consent belong with qualified trial personnel. Participant-entered ePRO data are still participant-entered data, with the extra EMRN controls for primary and key-secondary endpoints.

3. Reconstruct Every Encounter in the Audit Trail: Capture visit type (telehealth or in person), visit location, visit date, and data originator. Include telehealth visit personnel by name. Put remote and local-HCP direct eCRF entry on the authorized originator list.

4. Respect Regional Legal and Implementation Dates: Do not treat FDA's December 2024 Annex 2 draft as in-force U.S. GCP. Use the September 2024 DCT guidance as current U.S. thinking on collection location. In Europe, use EMRN V02 as current advice, check Member State requirements for cross-border trials, and date EU Annex 2 as coming into effect on 15 January 2027.

Treat location of collection as its own protocol decision. Naming a home visit, a telehealth visit, a local healthcare provider, or a DHT in the protocol is not regulatory acceptance of the endpoint, COA qualification, DHT qualification, or device clearance.