ISO 14155:2026: Good Clinical Practice for Device Trials

Introduction

ISO 14155 is the operating manual that the EU MDR assumes but never writes. The Regulation tells a sponsor what a clinical investigation must achieve, which documents must exist and which clocks start when something goes wrong. It says almost nothing about how to design an endpoint that will survive a Notified Body review, how to monitor fifty sites without verifying every data point, or how to tell a protocol deviation from a waiver. That is the territory of the standard, and it is where most investigations fail. In our experience the findings that sink device trials are rarely ethical scandals. They are design and documentation failures that were visible on paper before the first subject was enrolled.

This guide is written for manufacturers and sponsors who generate clinical data under the MDR, whether the investigation feeds a first clinical evaluation under the EU MDR or a post-market follow-up study that keeps the evidence current. The investigation report ends up in the clinical evaluation report and from there in the EU MDR technical documentation, so every weakness in the investigation travels with it. The fourth edition, ISO 14155:2026, was published in March 2026 and changes more than its predecessor's reputation for stability would suggest.

We cover scope, the 2026 changes and harmonisation, consent, the three risk streams, the core documents, the mapping to Annex XV and 21 CFR 812, safety reporting, responsibilities, transition and the six recurring findings.

Table of Contents

When ISO 14155 applies, and when the MDR decides that it does

The scope of the standard is broad and deliberately simple. It sets good clinical practice for the design, conduct, recording and reporting of clinical investigations carried out in human subjects to assess the clinical performance or effectiveness and the safety of medical devices. Its stated purposes are to protect the rights, safety and well-being of subjects, to make the conduct scientifically sound and the results credible, to define the responsibilities of the sponsor and the principal investigator, and to give ethics committees, regulators and conformity assessment bodies a common reference. For software as a medical device it applies as far as relevant, and the 2026 scope note points to analytical validity, scientific validity and clinical performance. It does not apply to in vitro diagnostic devices, whose performance studies have their own counterpart in ISO 20916 good study practice.

The standard itself is voluntary. What makes it binding in practice is the legal framework that refers to it, and in the EU that framework distinguishes very clearly between the kinds of investigation a manufacturer can run.

Investigations for conformity assessment under Article 62

Article 62(1) of the EU MDR, Regulation (EU) 2017/745, covers investigations performed as part of the clinical evaluation for conformity assessment purposes: to establish that the device performs as intended, to establish its clinical benefits, and to establish its clinical safety and the acceptability of its risks against those benefits. These are the investigations that Article 61(4) makes the default route for class III and implantable devices, subject to the exemptions in Article 61(4) to (6). Article 62(4) then lists thirteen cumulative conditions, and point (m) requires that the investigation fulfil Annex XV. That is the legal hook: Annex XV is mandatory, and ISO 14155 is the most complete description of how to satisfy it.

The authorisation route depends on the device. Under Article 70, the Member State has ten days to confirm that the application is complete. For class I devices and non-invasive class IIa and IIb devices, the sponsor may start immediately after the validation date unless national law provides otherwise, and provided that no negative ethics committee opinion valid for the whole Member State has been issued. For every other device the sponsor waits for an authorisation, which must be notified within 45 days of validation, extendable by 20 days for expert consultation. Planning a first-in-human study of an implant without those 45 to 65 days in the critical path is a common and entirely avoidable delay.

PMCF, Article 82 and sites outside the EU

Article 74(1) covers post-market clinical follow-up investigations of CE-marked devices used within their intended purpose. Where the study adds procedures that are invasive or burdensome compared with normal use, the sponsor notifies the Member States at least 30 days before the start, and a selected set of provisions applies, including Articles 75, 76, 77 and 80(5). Where the device is studied outside its intended purpose, Article 74(2) applies Articles 62 to 81 in full: the study is treated as if the device were not CE marked. The standard follows the same logic from its side, saying that its principles apply to post-market investigations as far as relevant, with Annex I describing the clinical development stages.

Article 82 is the gap most sponsors underestimate. Investigations not run for an Article 62(1) purpose must meet only Article 62(2) and (3), points (b), (c), (d), (f), (h) and (l) of Article 62(4), and Article 62(6). Point (m), the Annex XV requirement, is not on the list, and Member States fill the space with national rules. The Commission's MDR reform proposal of December 2025, COM(2025) 1023, would delete Article 82 altogether and leave these studies to national law; it remains a proposal. Either way, an investigator-initiated or academic study that may later be used as clinical evidence should be run to ISO 14155 from day one, because nobody can retrofit good clinical practice into data that has already been collected.

Sites outside the EU raise a related question. Article 80(3) requires reportable events from third-country sites running under the same plan to be reported to the Member States concerned, so a global study is one study for safety purposes. A single plan run to one standard across every site is the only sustainable design.

The life of a clinical investigation under ISO 14155Five phases, each with the documents the standard requires and the MDR provision that turnsthem into a legal obligation in the EU.1DESIGNClauses 6.1-6.12, Annexes A, B, KRisk management in three streams, design justification, CIP,IB, CRFs, monitoring plan, DMC and CEC decisionsEU MDRArt. 62(4), Annex XVCh. II2APPROVEClauses 5.6, 5.8, 9.4Ethics committee opinion, competent authority application ornotification, informed consent materials, insuranceEU MDRArt. 70, 74(1), 823CONDUCTClauses 7.1-7.11Site initiation, risk-based monitoring, safety events,amendments, device accountability, source data, auditsEU MDRArt. 72, 75, 804CLOSE OUTClauses 8.1-8.3, 8.6Routine close-out or suspension and termination,reconciliation, document retentionEU MDRArt. 77, Annex XV Ch.III5REPORTClause 8.4-8.5, Annex DClinical investigation report signed by the investigators,with risk conclusions that feed the clinical evaluationEU MDRArt. 77, Annex XV Ch.III 7Clause 9.1 requires the sponsor to run all five phases inside a clinical quality management system. Clause 7.11 andAnnex J make auditing part of conduct, not an afterthought.
Figure 1 — The five phases of a clinical investigation under ISO 14155:2026 and the MDR provisions behind each

What ISO 14155:2026 changed, and what is harmonised today

ISO 14155:2026 is the fourth edition, published in March 2026 by ISO/TC 194. It cancels and replaces ISO 14155:2020 as a technical revision, and the foreword lists the changes made. Most of them clarify rather than overturn, but four are structural: the separation of risk streams, the introduction of estimands, the new clinical events committee, and the prohibition on handling eligibility exceptions as deviations.

The changes that matter in practice

The table groups the foreword's list by the document or process it affects. Clause numbers are those of the 2026 edition.

AreaWhat changed in 2026Where
DefinitionsClinical performance redefined around the claimed intended purpose and clinical benefit, with a note for investigations that offer subjects no direct benefit3.12
Ethics and consentDeviations from eligibility criteria not permitted, handled by CIP amendment; consent by a legally designated representative; subjects able to discuss participation with others5.6.4, 5.8.1, 5.8.2
Risk managementDevice-use risks separated from risks of CIP procedures outside routine practice; residual risk assessment required; management of procedure risks added during conduct6.2.1–6.2.4, 7.4.5, Annex H
Clinical investigation planFormer Annex A content moved into the body; estimands introduced; methods, timing and equipment calibration for variables; non-inferiority margins and missing data clarified; continued care where it differs from normal practice6.4, A.5–A.7, A.6.4, A.16, Annex K
CommitteesDMC to confirm conditions for suspending or stopping; absence of a DMC to be justified; new clinical events committee6.11, 6.12, 3.8, A.14
SafetySituations of reduced adverse event reporting clarified; AE terminology updated; an AE associated with a device deficiency falls under both flowcharts7.4.2, Figures F.1 and F.2
Investigator's brochurePrecautions, training on device use and in-silico tests addedB.2, B.3, B.5
Sponsor dutiesLocal representative aligned with national requirements; implant card prepared where appropriate; suspension and termination aligned with the risk assessment flow9.2.1, 9.2.2, 8.2

What the 2026 edition does not do is equally useful to know. We found no source describing new provisions for decentralised or remote investigations, and the structure of clauses 4 to 10 is recognisably that of 2020. A sponsor with a mature 2020 system is updating, not rebuilding.

Why the 2020 edition still carries the presumption of conformity

Here the dates matter. On 28 January 2026 the Commission adopted Implementing Decision (EU) 2026/193, published in the Official Journal on 30 January 2026, which added EN ISO 14155:2020 together with amendment EN ISO 14155:2020/A11:2024 to the list of harmonised standards under the MDR in Implementing Decision (EU) 2021/1182. The amendment rewrote the European foreword and replaced Annex ZA, the annex that maps the standard to the Regulation. Two months later ISO published the fourth edition.

CEN approved EN ISO 14155:2026 on 13 January 2026, with October 2026 as the latest date for national adoption and for withdrawal of conflicting national standards. Its Annex ZA maps the clauses to MDR Annex XV, and states that MDR definitions prevail where they differ from those of the standard. But the later amending decisions of April and June 2026, (EU) 2026/760 and (EU) 2026/1231, did not cite it. As of November 2026 the presumption of conformity therefore rests on the 2020 edition with A11, while the national standards bodies have already withdrawn the 2020 text and the state of the art is the 2026 one.

Claiming the presumption of conformity today means citing EN ISO 14155:2020+A11:2024. Ignoring the 2026 edition means arguing against the state of the art in front of a Notified Body. The defensible position is to work to the 2026 edition and document a gap assessment showing that nothing required by the harmonised version has been lost.

Ethics and consent: the eligibility rule that ends quiet waivers

Clause 5 opens with the familiar principles: no improper influence or inducement, compensation and additional health care for injured subjects, registration in a publicly accessible database, and specific protection for vulnerable populations. The MDR mirrors these in Article 62(4), which requires that subjects receive no undue financial or other pressure, that foreseeable risks are justified by the expected benefits, and that the investigation minimise pain, discomfort and fear, with risk thresholds defined in the plan and monitored. The two texts point the same way. Where they differ, the Regulation governs.

The consent provisions in 5.8 were tightened in two respects. Consent may now explicitly come from a legally designated representative where the subject cannot give it, and the process must give the subject the opportunity to discuss participation with others, such as family, before deciding. Neither is a revolution, but both translate directly into the consent form and into the site training records, and both will be checked.

The change with real operational consequences is in 5.6.4. Deviations from the eligibility criteria are not permitted. A subject who does not meet the inclusion and exclusion criteria cannot be enrolled by logging a deviation; if the criteria are wrong, the CIP is amended and the amendment goes through the ethics committee and, where required, the competent authority. MDR Annex XV Chapter II Section 3.10 already required a policy on deviations with an explicit prohibition of waivers, so this was never permitted in the EU. The standard now says so in terms that leave no room for interpretation.

This matters because enrolment pressure is constant, and the temptation to admit a subject who is “almost” eligible is strongest exactly where the investigation is most fragile: rare indications, slow sites, first-in-human studies. An eligibility waiver contaminates the analysis population, and a pattern of them is treated by inspectors as a sign that the sponsor does not control the study. The fix is upstream. Criteria that sites cannot apply are a design defect, and the CIP should be stress-tested against real screening logs before submission.

Three risk streams: separating the device from the procedure

The most consequential change in ISO 14155:2026 is the restructuring of clause 6.2. The 2020 edition already required risk management in accordance with ISO 14971, with Annex H explaining how. The 2026 edition separates three sources of risk, each with its own treatment, and anyone who has run an investigation will recognise why.

The first stream, in 6.2.2, covers risks related to the use of the investigational device and their disclosure. These remain under ISO 14971, applied through Annex H, and the new requirement is an explicit residual risk assessment. The residual risks that subjects accept must be identified, evaluated and disclosed consistently in the investigator's brochure, the CIP and the informed consent materials. During conduct, clause 7.4.4 defines the process for assessing potentially unacceptable risks of device use, and clause 8.2 aligns suspension and termination with it. If your ISO 14971 risk management file and your consent form describe the same hazard with different frequencies, an ethics committee will notice before you do.

The second stream, in 6.2.3 and 7.4.5, covers risks of clinical procedures required by the CIP outside routine clinical practice: an extra biopsy, additional imaging, a study-only visit, a washout. These are not device risks and they do not belong in the device risk management file, yet they are real risks to subjects that the investigation itself creates. Commentators who have reviewed the new text read it as asking for a proportionate, descriptive assessment of these procedures rather than a full ISO 14971 analysis. For multinational studies the practical consequence is that “routine practice” differs between countries, and what counts as an additional procedure in one may be standard of care in another.

The third stream, in 6.2.4, covers risks related to the clinical investigation process: consent failures, data integrity, monitoring gaps, device accountability errors. This is the risk assessment that should drive the monitoring plan, and its existence as a named clause is the closest the standard comes to codifying risk-based monitoring.

The separation pays off in safety reporting and in the final benefit-risk conclusion. An adverse event caused by a study-only procedure is reportable under Article 80(2) of the MDR when it is serious and causally related, but it says nothing about the device. Conflating the two inflates the apparent device risk and distorts the benefit-risk analysis that the clinical evaluation builds on the investigation.

Write one residual-risk table and cite it verbatim in the IB, the CIP and the consent form. Three documents drafted by three people will describe the same risk three ways, and the 2026 edition now treats that inconsistency as a defect rather than a style issue.

The CIP, the investigator's brochure and the CRFs decide the outcome

Every investigation is won or lost on paper before the first subject is enrolled. The MDR fixes the minimum content of the dossier in Annex XV Chapter II: the application form in Section 1, the investigator's brochure in Section 2, the clinical investigation plan in Section 3 and the other information in Section 4, including insurance, consent documents and data protection measures. The MDCG has added two guidance documents that follow the numbering of the Annex, MDCG 2024-3 on CIP content, with a synopsis template, in March 2024 and MDCG 2024-5 on investigator's brochure content, with a checklist, in April 2024. ISO 14155 supplies the method behind each heading.

The clinical investigation plan and the estimand

In the 2026 edition the core CIP requirements moved from Annex A into clause 6.4, which signals what the committee thought of protocols treating the annex as optional reading. Annex XV Chapter I Section 2.6 already required endpoints that address the intended purpose, clinical benefits, performance and safety, determined with scientifically valid methods, and a clinically relevant primary endpoint. The standard now adds the vocabulary to do this properly.

The estimand framework, introduced through 6.4, clauses A.5 to A.7 and the new informative Annex K, comes from ICH E9(R1). An estimand states precisely what treatment effect the investigation is meant to estimate: the treatment, the population, the variable, the population-level summary and the handling of intercurrent events such as device removal, crossover or death from an unrelated cause. Device trials have historically been vague about intercurrent events, and the vagueness surfaces at the analysis stage, when the sponsor discovers that the primary endpoint can be computed three defensible ways with three different answers. Clause A.7 adds that non-inferiority margins must be justified and that missing data handling and sensitivity analyses are specified in advance.

The other half of the CIP problem is strategic rather than statistical. Annex XIV Part A Section 1(a) requires the clinical evaluation plan to include a clinical development plan showing the progression from exploratory investigations, such as first-in-human, feasibility and pilot studies, to confirmatory pivotal investigations and PMCF, with milestones and acceptance criteria. An endpoint that does not answer a question in that plan is an endpoint that will not support a claim in the clinical evaluation report, however well it is measured.

The investigator's brochure, CRFs and monitoring plan

The investigator's brochure in clause 6.5 and Annex B is the document that tells investigators what they are implanting or applying and what can go wrong. The 2026 edition added precautions, training on the use of the device and in-silico tests within the preclinical evidence. MDR Annex XV Chapter II Section 2 asks for the same core: identification and description, instructions for use and training, preclinical evaluation, existing clinical data, a benefit-risk and risk management summary, information on medicinal substances or tissues where applicable, the GSPR compliance status including standards applied, and the clinical procedures with any deviation from normal practice. MDCG 2024-5 adds that the anticipated frequencies of serious adverse events and serious adverse device effects should be stated, which is the reference safety information against which unanticipated events will later be judged.

Case report forms, in clause 6.6 and informative Annex C, are where the endpoint definition becomes data. A CRF that collects the variable in a different unit, at a different time point or with a different definition from the CIP will produce a dataset that cannot answer the protocol question. Clause A.6.4 now requires the methods and timing for assessing, recording and analysing each variable, together with calibration arrangements for study equipment, which closes a gap that every monitor has met: a blood pressure endpoint measured on uncalibrated devices at different sites.

The monitoring plan in clause 6.7 completes the set. Article 72(2) of the MDR requires the sponsor to determine the extent and nature of monitoring on the basis of the objectives and methodology of the investigation and the degree of deviation from normal clinical practice. That is a risk-based mandate written into law. A plan that applies the same visit frequency and full source data verification to every site and every variable has not made that determination; it has copied a template.

✦ Clinical Documentation Kit · EU MDR

Plan the investigation where the clinical evidence strategy lives.

A CIP is only as good as the clinical development plan behind it. The kit gives you the clinical evaluation plan and report, the clinical development plan, the PMCF plan and evaluation report, the PMS plan, the PSUR and the SSCP, cross-referenced so that every investigation answers a gap the CER has already named.

✓ 8 Word templates plus an Excel index

✓ Aligned with MDCG guidance

✓ CDP references CER gaps, PMCF plan references CDP strategy

Get the Clinical Documentation Kit → €399

ISO 14155 against MDR Annex XV and 21 CFR 812

The standard, the EU Regulation and the US regulation cover the same ground with different instruments. The standard tells you how to do it well. The two regulations decide what is legally required and by when, and they do not always ask for the same things.

Annex XV, clause by clause

The table maps the main sections of MDR Annex XV to the clauses of ISO 14155:2026 that implement them. The Annex ZA of EN ISO 14155:2026 contains the formal mapping, with its indication of full or partial coverage; this one is a working map for authors.

MDR Annex XVRequirementISO 14155:2026
Chapter I — General requirements
Section 1Ethical principles at every step, from justification to publicationClauses 4 and 5; Annex G
Sections 2.1–2.6Appropriate plan, procedures and methods; representative users and setting; endpoints with a clinically relevant primary endpoint6.3, 6.4, Annex A, Annex K
Section 2.7Investigator access to device data; training of staff, verified and documented by the sponsor7.2, 9.2.1, 10.2
Section 2.8Report signed by the investigator, with critical evaluation of all data including negative findings8.4, Annex D
Chapter II — Documentation for the application
Section 2Investigator's brochure, 2.1 to 2.86.5, Annex B
Section 3.1–3.7CIP general content, risk-benefit, objectives, design, monitoring plan, statistical design6.2, 6.4, 6.7, Annex A
Sections 3.8–3.11Data management, amendments, deviations policy with prohibition of waivers, device accountability7.5.1, 7.8, 5.6.4, 7.9
Sections 3.13–3.16Informed consent, AE and device deficiency reporting, follow-up after the end, continued care5.8, 7.4, 10.8, A.16
Chapter III — Other obligations of the sponsor
Sections 2–6SAE reporting agreement, retention of 10 or 15 years, independent monitor, follow-up, evidence of GCP6.9, 8.6, 9.2.4, 7.10, 7.11
Section 7Content of the clinical investigation report8.4, Annex D

Two differences between the instruments recur in Notified Body discussions. The MDR's definitions prevail where they differ from the standard's, which Annex ZA confirms, so an investigation run to ISO 14155 must still report under the MDR's categories. And the MDR imposes deadlines the standard does not, from the 30-day notification of Article 74(1) to the one-week notification and 38-day wait for substantial modifications under Article 75, on which MDCG 2021-28 gives guidance.

What the FDA accepts

In the United States, investigations conducted in the country fall under 21 CFR 812, the investigational device exemptions regulation, with its own vocabulary: an investigational plan under 812.25, a report of prior investigations under 812.27, and unanticipated adverse device effects that the investigator reports to the sponsor and the IRB within ten working days and that the sponsor, after evaluation, reports to FDA, IRBs and investigators within ten working days under 812.150. For data generated outside the US, 21 CFR 812.28, in force since 21 February 2019 for investigations enrolling their first subject from that date, accepts clinical data in support of an IDE or marketing submission if the investigation was conducted under good clinical practice, including independent ethics committee review and informed consent, and if FDA can validate the data through an onsite inspection when it considers it necessary.

In the preamble to the final rule, FDA said that sponsors following ISO 14155 should be able to meet the good clinical practice requirement of 812.28, and that it would accept a clinical investigation report prepared under Annex D as the required protocol and results summary. FDA's database of recognised consensus standards lists ISO 14155:2020 under recognition number 2-282, entered on 21 December 2020 with complete recognition. When we checked the entry, last updated in September 2026, no recognition of the 2026 edition was listed. A sponsor submitting to FDA should therefore declare conformity to the 2020 edition and treat the 2026 changes as voluntary improvements until the database says otherwise.

ISO 14155:2026, MDR Annex XV and 21 CFR 812 side by sideThe standard describes how to do it; the two regulations decide what is legally required, andthey do not ask for the same things.TOPICISO 14155:2026EU MDR21 CFR 812Legal statusVoluntary. EN ISO14155:2020+A11:2024 is thecited harmonised version; the2026 edition is not yet citedBinding regulation; Annex XVapplies through Art. 62(4)(m)Binding regulation for USinvestigations; ISO14155:2020 recognised as2-282ProtocolCIP per 6.4 and Annex A, withestimands and Annex K designconsiderationsCIP content in Annex XV Ch.II 3.1-3.19; MDCG 2024-3Investigational plan per812.25BrochureIB per 6.5 and Annex B, nowincl. precautions, training,in-silico testsAnnex XV Ch. II 2.1-2.8; MDCG2024-5Report of priorinvestigations per 812.27ApprovalEC opinion before enrolment;regulatory approval whererequiredArt. 70: validation, thenstart or 45-day authorisationby classIRB approval; FDA-approvedIDE for significant riskdevicesSafety reportingAE, ADE, SAE, SADE, USADE anddevice deficiencies; Annex FflowchartsArt. 80(2) reportable events;2 or 7 calendar days per MDCG2020-10/1Unanticipated adversedevice effects: 10 workingdays, 812.150Final reportCIR per 8.4 and Annex DArt. 77: 1 year after end, 3months after earlyterminationFinal report per812.150(b)(7)For data generated outside the US, 21 CFR 812.28 accepts studies run under GCP. Following ISO 14155 is the practicalroute to meet it.
Figure 2 — ISO 14155:2026, EU MDR and 21 CFR 812 compared on six topics

Safety reporting: from adverse event to USADE

Safety reporting is where ISO 14155 and the MDR interlock most tightly, and where the terminology causes most errors. The MDR defines three terms in Article 2: adverse event in point 57, serious adverse event in point 58 and device deficiency in point 59. ISO 14155 defines a richer set, and where the two differ the MDR definitions prevail. MDCG 2020-10/1 rev.1 of October 2022 uses the standard's categories to make Article 80 operational.

CategoryWhat it meansConsequence in the EU
Adverse event (AE)Any untoward medical occurrence, unintended disease or injury, or untoward clinical sign in subjects, users or other persons during the investigation, whether or not related to the investigational deviceRecorded if the CIP identifies it as critical to the evaluation of results (Art. 80(1))
Serious adverse event (SAE)An AE leading to death, serious deterioration in health, or foetal distress, foetal death or congenital anomalyAlways recorded (Art. 80(1)); reportable when causally related or reasonably possibly related to the device, comparator or procedure
Adverse device effect (ADE)An AE related to the investigational device or comparatorAssessed for relationship; recorded per the CIP
Serious adverse device effect (SADE)An ADE with any of the consequences of an SAEReportable under Art. 80(2)
Unanticipated SADE (USADE)A SADE whose nature, severity or outcome is not consistent with the current risk assessment and reference safety informationReportable; feeds the risk assessment process of clause 7.4.4, which can lead to suspension
Device deficiencyInadequacy in identity, quality, durability, reliability, usability, safety or performance, including malfunction, use error and inadequate labellingReportable under Art. 80(2) if it might have led to an SAE without suitable action or intervention

Article 80(2) requires the sponsor to report without delay, through the electronic system of Article 73, every serious adverse event with a causal relationship, or a reasonably possible one, to the investigational device, the comparator or the investigation procedure; every device deficiency that might have led to a serious adverse event; and new findings on either. The period must take account of the severity of the event. MDCG 2020-10/1 translates that into two clocks, which the national authorities apply: immediately and no later than two calendar days after the sponsor becomes aware, for events indicating an imminent risk of death, serious injury or serious illness that requires prompt remedial action; and no later than seven calendar days for all other reportable events.

Two subtleties generate most of the misclassification. The first is that relationship is assessed separately for the device and for the procedure, and the revision of the MDCG guidance made that explicit; an event can be unrelated to the device and still reportable because the study procedure caused it. The second is that a device deficiency is reportable even when nobody was harmed. Sites that think of safety reporting as “reporting injuries” will miss both. The 2026 edition adds a third: when an adverse event is associated with a device deficiency, both flowcharts of Annex F apply, so the same occurrence is assessed as an event and as a deficiency.

Post-market studies follow a different route. For PMCF investigations under Article 74(1), Article 80(5) replaces the clinical investigation reporting with the vigilance provisions of Articles 87 to 90, and Article 80(6) brings Article 80 back where a causal relationship between a serious adverse event and the preceding investigational procedure has been established. Clause 7.4.2 of the 2026 edition also clarifies when adverse event reporting may be reduced, which is mainly relevant to post-market studies of well-characterised devices; the rationale belongs in the CIP, not in a monitor's memory.

Classifying a safety event under ISO 14155 and the MDRAnswer the gates in order. The classification drives what is recorded, what is reported, andhow fast.1. Did an untoward medical occurrence affect asubject, user or other person during theinvestigation?Yes: adverse event (AE). Record it asthe CIP requires, related to thedevice or not.yes, continue2. Did it lead to death, serious deterioration inhealth, or foetal distress, death or anomaly?Yes: serious adverse event (SAE).Always recorded under Art. 80(1).yes, continue3. Is it related, or reasonably possibly related, tothe device, the comparator or an investigationprocedure?Yes: reportable under Art. 80(2). Ifdevice-related it is also a seriousadverse device effect (SADE).yes, continue4. Was the nature, severity or outcome of the SADEidentified in the current risk assessment and IB?No: unanticipated SADE (USADE). Feedsthe 7.4.4 risk assessment, which canlead to suspension.yes, continue5. Separately: was there a device deficiency thatmight have led to an SAE without action orintervention?Yes: reportable under Art. 80(2) eventhough no one was harmed. If an AEoccurred too, both F.1 and F.2 apply.Reporting clock per MDCG 2020-10/1: immediately and no later than 2 calendar days for imminentrisk of death, serious injury or illness needing prompt action; otherwise no later than 7 calendardays.
Figure 3 — Classifying a safety event from adverse event to USADE, and the MDR reporting clocks

✦ Risk + Clinical Bundle · ISO 14971 · EU MDR

One risk file and one clinical file that tell the same story.

The 2026 edition makes residual risk disclosure explicit, and reviewers now compare the risk file with the clinical documents line by line. The bundle combines the Clinical Documentation Kit with the Risk Management Kit: risk management plan and report, hazard analysis, FMEAs and GSPR checklist, built to cross-reference each other.

✓ 14 templates across both kits

✓ ISO 14971 risk files linked to the CEP, CER and PMCF plan

✓ Save €149 against buying the kits separately

Get the Risk + Clinical Bundle → €599

Sponsor, investigator, monitor and committees: who carries what

Clauses 9 and 10 divide the work. The sponsor carries clinical quality management in 9.1, selection and training of clinical personnel, preparation of documents and materials, conduct, monitoring, safety evaluation and reporting, and close-out in 9.2, the conditions for outsourcing in 9.3 and communication with regulatory authorities in 9.4. The principal investigator carries qualification of self and site, communication with the ethics committee, the consent process, compliance with the CIP, medical care of subjects and safety reporting to the sponsor in 10.1 to 10.8. The MDR places the same weight on the sponsor: Annex XV Chapter III requires an agreement under which investigators report serious adverse events promptly, an independent monitor, complete follow-up of subjects, and evidence of good clinical practice.

Outsourcing to a CRO moves the work but not the responsibility. Clause 9.3 lets the sponsor transfer duties and functions, but the sponsor remains accountable for the quality and integrity of the data, and a CRO contract that does not state which party holds which duty is a finding in itself. The same applies to the local representative, now aligned in 9.2.1 with national requirements, and to the EU legal representative that Article 62(2) of the MDR requires of a non-EU sponsor.

Two committees deserve particular attention because the 2026 edition changed both. A data monitoring committee reviews accumulating safety and, where relevant, efficacy data, and recommends whether the investigation should continue. Clause 6.11 now requires its charter to confirm the conditions under which it would recommend suspension or termination, and A.14 requires the CIP to justify the absence of a DMC. A small single-site feasibility study can justify it easily; a pivotal randomised trial in a high-risk indication cannot. The clinical events committee in 6.12 is new: an independent group that adjudicates clinical events against the endpoint definitions and classifies their relationship to the device or to study procedures. Sponsors are expected to consider one, and where it exists its charter must set out its methods, its independence and the criteria it applies. For multicentre trials with adjudicated endpoints, an investigator's own classification of a stroke or a myocardial infarction is exactly the kind of variability a CEC is designed to remove.

During conduct, clause 7 governs the evidence trail. Investigational device accountability in 7.9 requires that every unit shipped, used, returned or destroyed is traceable to a site and, where applicable, a subject, and Annex XV Chapter II Section 3.11 adds access control and returns. Document and data control in 7.8, including electronic clinical data systems in 7.8.3, sets the expectations for traceability and audit trails. Auditing in 7.11 and informative Annex J make independent audits part of conduct. Annex E lists the essential clinical investigation documents that make the investigation reconstructable, and clause 8.6 governs their retention; the MDR sets the floor in Annex XV Chapter III Section 3 at ten years after the investigation ends or after the last device is placed on the market, and fifteen years for implantable devices.

Close-out ends with the clinical investigation report under clause 8.4 and normative Annex D. Article 77 of the MDR fixes the timeline: the end of the investigation is notified within fifteen days; the report and a summary written for its intended user are submitted within one year of the end, or within three months of an early termination or temporary halt. A temporary halt or early termination on safety grounds must be notified to all Member States concerned within 24 hours. The report must include negative findings, which Annex XV Chapter I Section 2.8 makes explicit, and its conclusions feed straight into the clinical evaluation.

Moving an ongoing investigation from the 2020 to the 2026 edition

ISO publishes no transition period, and for the EU none has been set because EN ISO 14155:2026 has not yet been cited in the Official Journal. There is therefore no official date by which an ongoing investigation must comply with the fourth edition. What follows is our practitioner position, not a regulatory requirement.

New investigations should be designed to the 2026 edition from the start. A CIP written in November 2026 to the 2020 edition will be reviewed by people who know the new text, and the gaps will be visible: no estimand, no stated position on a clinical events committee, no justification for the absence of a DMC, one undifferentiated risk section. Each is cheap to add at drafting and expensive to add by substantial modification.

Ongoing investigations need a documented gap assessment, not a rewrite. Most conduct requirements carry over unchanged, and amending a running study to adopt the new structure would create the very deviations and re-consent burden that good clinical practice is meant to avoid. The assessment should check five points: whether the risk documentation separates device and procedure risks and states residual risks consistently; whether the statistical analysis plan addresses missing data and sensitivity analyses; whether the committee governance is justified; whether any eligibility exceptions have been logged as deviations; and whether sites classify device deficiencies associated with adverse events under both Annex F flowcharts. Where a gap affects subject safety or data reliability, it is closed at the next substantial modification. Where it does not, the assessment records why the 2020 approach remains adequate for the study.

The clinical investigation report is where the transition becomes visible. A report issued in 2027 for a study designed in 2024 should state the edition the investigation was conducted to and summarise the gap assessment. Reviewers can accept an investigation run to the edition in force at the time; they cannot accept silence about it.

Six findings that recur, and none of them is about ethics

The common image of a failed clinical investigation is an ethical breach: missing consent, an unapproved site, a concealed death. Those happen, but they are rare. The findings that recur across sponsor audits, Notified Body reviews of clinical data and competent authority inspections are mundane, and they share one property: each was visible in the documents before the first subject was enrolled.

The first is a primary endpoint that cannot carry the claim, because it measures a surrogate the clinical evaluation does not accept or because it was chosen for feasibility rather than for the benefit claimed. The second is waivers disguised as deviations, now expressly ruled out by 5.6.4. The third is late or wrong classification of safety events, typically assessing relationship to the device but not to the procedure, or missing a reportable device deficiency because nobody was harmed. The fourth is a monitoring plan copied from a template, which fails the risk-based determination that Article 72(2) requires. The fifth is device accountability that does not reconcile by lot and site. The sixth is risk documentation that contradicts itself across the IB, the CIP and the consent form.

Six ISO 14155 findings that recur, and the fix for eachNone of them is an ethics failure. All of them are design or documentation failures that werevisible before the first enrolment.HIGHPrimary endpoint thatcannot carry the claimThe endpoint measures somethingother than the benefit the CERneeds. Fix: derive it from theclinical development plan anddefine the estimand in the CIP.HIGHWaivers disguised asdeviationsSubjects enrolled outsideeligibility criteria and loggedas deviations. Fix: 5.6.4 andAnnex XV 3.10 forbid it; amendthe CIP instead.HIGHSafety eventsclassified late orwronglyDevice relationship assessedonly for the device, not theprocedure; clocks missed. Fix:assess both and train sites onAnnex F.MEDIUMMonitoring plan copiedfrom a templateSame visit frequency and 100%SDV for every site and variable.Fix: tie the plan to the 6.2.4process risks and critical data.MEDIUMDevice accountabilitythat does notreconcileShipped, used, returned anddestroyed quantities do not addup by lot. Fix: a running logper site, reconciled at everymonitoring visit.LOWRisk documents thatcontradict each otherRisks in the IB, CIP and consentform differ in wording orfrequency. Fix: oneresidual-risk table, citedverbatim in all three.
Figure 4 — Six recurring ISO 14155 findings, rated by severity, with the fix for each

The first three carry the highest severity because they damage either subject protection or the usability of the data, and none can be fixed after the fact. A miscalculated endpoint cannot be re-measured, an ineligible subject cannot be un-enrolled, and a late safety report is late forever. The fourth and fifth are serious but recoverable with corrective action during conduct. The sixth is the cheapest to prevent and, since the 2026 edition made residual risk disclosure explicit, the easiest for a reviewer to find.

✦ MDR Complete Bundle · EU MDR

The investigation report has to land somewhere.

The clinical investigation report feeds the clinical evaluation, which sits in the technical documentation, which rests on the risk file. The MDR Complete Bundle covers all three: the Technical Documentation Kit, the Clinical Documentation Kit and the Risk Management Kit, so the evidence chain holds from investigation to certificate.

✓ 22 templates in 3 kits

✓ Technical documentation, clinical and risk management in one set

✓ Save €178 against buying the kits separately

Get the MDR Complete Bundle → €999

Frequently asked questions

What is ISO 14155?

ISO 14155 is the international standard for good clinical practice in clinical investigations of medical devices in human subjects. It covers the design, conduct, recording and reporting of investigations that assess clinical performance or effectiveness and safety, defines the responsibilities of the sponsor and the principal investigator, and protects the rights, safety and well-being of subjects. It does not apply to in vitro diagnostic devices.

What is the latest version of ISO 14155?

The latest version is ISO 14155:2026, the fourth edition, published in March 2026. It cancels and replaces ISO 14155:2020. The European adoption, EN ISO 14155:2026, was approved by CEN in January 2026 and supersedes EN ISO 14155:2020.

Is ISO 14155 mandatory under the EU MDR?

No standard is mandatory under the MDR, but Annex XV is, through Article 62(4)(m), and ISO 14155 is the recognised way to meet it. Applying the harmonised version gives a presumption of conformity with the requirements it covers. A sponsor that does not follow it must show by other means that every Annex XV requirement is met, which in practice is harder and rarely attempted.

Is ISO 14155:2026 harmonised under the MDR?

Not yet. The harmonised reference in Implementing Decision (EU) 2021/1182 is EN ISO 14155:2020 with amendment A11:2024, added by Implementing Decision (EU) 2026/193 of 28 January 2026. EN ISO 14155:2026 has an Annex ZA mapping it to Annex XV but had not been cited in the Official Journal as of November 2026.

Does ISO 14155 apply to PMCF studies?

Yes, as far as relevant. The standard says its principles apply to post-market investigations according to their nature, with Annex I describing the clinical development stages. Under the MDR, PMCF investigations with additional invasive or burdensome procedures are notified under Article 74(1) at least 30 days before the start, and those studying a device outside its intended purpose fall under Article 74(2) with the full clinical investigation regime.

Does the FDA accept ISO 14155?

Yes. FDA recognises ISO 14155:2020 as a consensus standard under recognition number 2-282, and the preamble to 21 CFR 812.28 states that following ISO 14155 should let sponsors meet the good clinical practice requirement for data collected outside the US. As of September 2026 the FDA database listed no recognition of the 2026 edition.

Does ISO 14155 apply to in vitro diagnostic devices?

No. The scope excludes in vitro diagnostic devices, although some sections may be considered depending on the device and the applicable rules. IVD performance studies are covered by ISO 20916, which plays the equivalent role under the IVDR.

Is there a transition period for ISO 14155:2026?

ISO sets no transition period and, because the 2026 edition is not yet harmonised, the EU has set none either. New investigations should be designed to the 2026 edition. Ongoing investigations should document a gap assessment against it and close the gaps that affect subject safety or data reliability at the next substantial modification.

Conclusions

ISO 14155:2026 does not rewrite good clinical practice for devices. It closes the loopholes that experienced sponsors already avoided and that inexperienced ones fell into: device and procedure risks treated as one, eligibility exceptions logged as deviations, primary endpoints with no stated estimand, committees chosen by habit. Each of those is a design decision, and each is taken before the ethics committee submission.

Three actions follow. Design every new investigation to the 2026 edition, and declare conformity to EN ISO 14155:2020+A11:2024 for the presumption while it remains the cited version. Run a written gap assessment on every ongoing investigation and file it with the study documents. And derive every CIP from the clinical development plan, because an investigation that does not answer a question the clinical evaluation is asking produces data no reviewer will use.

The EU MDR Clinical Documentation Kit on MD Regulatory includes the clinical evaluation plan and report, the clinical development plan, the PMCF plan and evaluation report, the PMS plan, the PSUR and the SSCP, with coordinated cross-references and alignment to MDCG guidance, ready to deploy in an existing EU MDR quality management system.

Related articles