QPPV.COM

GVP Module VIII: Non-Interventional PASS Requirements

A practical guide to non-interventional PASS requirements, including usual care, treatment decisions, study inclusion, additional procedures, epidemiological methods and regulatory classification.

Audio Lesson 23 min
Knowledge Assessment Test your understanding of this article. Take the assessment →

GVP Module VIII: Non-Interventional PASS Requirements

Introduction

A non-interventional post-authorisation safety study is not simply a PASS that happens to use observational data. The classification depends on how treatment is prescribed and managed, how patients enter the study, whether the study introduces additional diagnostic or monitoring procedures, and how the resulting information is analysed.

This distinction matters because imposed non-interventional PASS are subject to specific EU procedures, protocol requirements and regulatory oversight.

EMA's current PASS questions and answers describe a non-interventional imposed PASS as a study in which the medicine is prescribed in the usual way within the marketing authorisation, treatment decisions are based on current practice rather than a study protocol, treatment prescribing is separate from the decision to include the patient, patients do not undergo additional diagnostic or monitoring procedures, and data are analysed using epidemiological methods. Systematic reviews and meta-analyses of safety data can also be considered non-interventional PASS. citeturn0search0

1. The Five-Part Classification Test

The practical classification can be considered as five linked questions:

  1. Is treatment prescribed in the usual way?
  2. Is the treatment decision independent of the study protocol?
  3. Is the decision to prescribe separate from the decision to enrol or include the patient in the study?
  4. Does participation avoid additional diagnostic or monitoring procedures?
  5. Are epidemiological methods used to analyse the data?

These questions should be considered together. A study should not be classified solely because it is called an "observational study" in a protocol.

2. Usual Clinical Practice

The medicinal product should be used according to the terms of its marketing authorisation and in the usual manner of clinical practice.

The study should observe treatment as it occurs rather than direct the prescriber to use the medicine according to an experimental allocation or study-specific treatment algorithm.

This does not mean that every patient must receive treatment in exactly the same way. Real-world clinical practice naturally contains variation in dose, duration, switching, discontinuation and concomitant therapy.

The relevant question is whether that variation arises from ordinary clinical practice rather than being imposed by the research protocol.

3. Treatment Decisions Must Remain Clinical Decisions

A key characteristic of a non-interventional study is separation between treatment decisions and study participation.

The physician or other authorised prescriber should decide whether and how to treat the patient on clinical grounds.

The study should not require the prescriber to choose the medicine, dose or treatment duration in order to satisfy the research protocol.

A protocol may subsequently record those treatment decisions in a structured way. Recording a clinical decision does not itself make the study interventional.

4. Prescription Before Study Inclusion

The decision to prescribe should be clearly separated from the decision to include a patient in the study.

For example, a patient may already have received a medicine as part of ordinary care and subsequently become eligible for inclusion in an observational database study.

The study should not create a situation in which treatment is prescribed primarily because the patient is needed for the research.

This distinction is particularly important when investigators recruit patients prospectively into an observational cohort.

5. Prospective Does Not Automatically Mean Interventional

A prospective observational study can still be non-interventional.

Patients can be enrolled before outcomes occur, followed over time and assessed using prospectively defined methods without the study controlling their treatment.

The defining issue is not whether data are collected prospectively. It is whether the research process changes treatment or introduces study-driven medical decisions.

6. Retrospective Does Not Automatically Mean Non-Interventional

Similarly, calling a study retrospective does not by itself establish that it meets every requirement for non-interventional classification.

The classification should be based on the actual study conduct and data-generating circumstances.

A retrospective analysis may still involve issues requiring careful assessment of its regulatory and methodological classification.

7. No Additional Diagnostic or Monitoring Procedures

Patients in a non-interventional PASS should not undergo additional diagnostic or monitoring procedures solely because they are participating in the study.

This requirement protects the distinction between observing clinical practice and introducing study-related clinical activity.

Routine information already generated during ordinary care can generally be observed and analysed. The protocol should make clear which information is routinely available and which, if any, procedures are introduced specifically for the study.

8. Routine Tests Versus Study-Specific Tests

A common misunderstanding is that any prospective data collection makes a study interventional.

That is not correct.

A study may collect laboratory results, imaging findings, clinical observations or other information when those data arise as part of routine care.

The important question is whether the study requires an additional procedure that would not otherwise have occurred as part of normal patient management.

9. Patient Questionnaires and Patient-Reported Information

Collection of patient-reported information requires careful consideration of the study design.

A questionnaire that records information ordinarily available from the patient's experience does not automatically transform an observational study into an interventional study.

However, if study participation requires additional clinical assessments or procedures that would not otherwise occur, the classification should be reassessed.

The protocol should therefore describe what participants are asked to do and distinguish ordinary information collection from additional clinical intervention.

10. Epidemiological Methods

The analysis of a non-interventional PASS should use appropriate epidemiological methods.

Depending on the question, this can include:

The method should be selected because it answers the pharmacovigilance question, not because a particular method is fashionable or familiar.

11. A Database Does Not Make a Study Non-Interventional

Use of an electronic health-record database, claims database or registry is not sufficient to establish classification.

The underlying clinical conduct remains relevant.

A database may contain information generated under interventional research, routine clinical practice, or a mixture of sources. The study protocol should explain the origin and nature of the data being analysed.

12. Registries and Non-Interventional PASS

A registry can support a non-interventional PASS when treatment decisions remain independent of the registry and the study observes relevant clinical practice.

The existence of a registry does not eliminate the need to define:

The registry should be evaluated for fitness for the particular pharmacovigilance question.

13. Systematic Reviews and Meta-Analyses

EMA's current PASS Q&A explicitly notes that systematic reviews and meta-analyses of safety data should be considered non-interventional PASS. citeturn0search0

Their non-interventional character does not make them methodologically simple. The protocol should still define the research question, eligibility criteria, information sources, search strategy, methods for assessing relevant evidence and analytical approach.

14. Mixed Data Sources

A PASS may combine information from multiple sources.

For example, a study may combine:

The protocol should distinguish the origin of each source and explain how the sources are integrated.

A study should not imply that all observations have the same evidentiary characteristics merely because they are included in one analytical dataset.

15. Study Participation and Clinical Management

The study team should not become an alternative treatment-management function.

If study investigators begin recommending treatment changes, ordering study-specific clinical assessments or otherwise directing patient care because of study participation, the classification and governance implications require reassessment.

The boundary should be understood by investigators, vendors and participating healthcare professionals before study initiation.

16. Safety Follow-Up

Safety follow-up can be part of an observational study, but the nature of the follow-up matters.

The protocol should explain whether information is obtained from routine records, ordinary clinical contacts, structured questionnaires or additional procedures.

Where follow-up introduces clinical procedures beyond usual practice, the study classification should be reconsidered.

17. Study Classification Should Be Documented

The organisation should not rely solely on an informal statement that a study is "observational".

The protocol should contain enough information for a reviewer to understand why the study is considered non-interventional.

A useful internal assessment can map each classification criterion to concrete study procedures and responsibilities.

18. Practical Classification Example: Prospective Cohort

A physician decides independently to prescribe a medicine for a patient. The patient is subsequently enrolled in a prospective safety cohort. The study records routine clinical information and follows the patient for a defined period. No additional diagnostic or monitoring procedure is required by the protocol.

This structure is consistent with the core characteristics of a non-interventional study, subject to the full assessment of the actual protocol and conduct.

19. Practical Classification Example: Protocol-Directed Treatment

A protocol instructs investigators to prescribe a particular medicine and dose to eligible participants and requires treatment decisions to follow study-defined criteria.

Calling the resulting activity an "observational PASS" would not resolve the underlying issue. The treatment decision has become linked to the research protocol and therefore requires a different regulatory assessment.

20. Practical Classification Example: Extra Diagnostic Monitoring

Patients receive treatment according to ordinary clinical practice, but study participation requires additional imaging at defined intervals solely to detect a safety outcome.

The study should not automatically be treated as non-interventional merely because treatment itself was not assigned by the protocol. The additional diagnostic procedure requires careful classification and regulatory assessment.

Key Takeaways

The classification of a non-interventional PASS depends on how the study is conducted, not simply on its title, database, prospective or retrospective nature.

The essential principles are:

usual treatment → independent treatment decision → independent study inclusion → no additional diagnostic/monitoring procedures → epidemiological analysis.

These principles should be demonstrated in the protocol and understood by all parties involved in study execution.

21. Imposed Non-Interventional PASS

The regulatory consequences become particularly important when a non-interventional PASS is imposed.

EMA's current guidance explains that imposed non-interventional PASS can arise as a condition of a marketing authorisation, as a specific obligation, or through other applicable EU pharmacovigilance procedures. The applicable legal basis determines the submission and assessment route. citeturn0search0

The MAH should therefore establish the legal and procedural basis before treating a study as an ordinary voluntary observational project.

22. Article 107n: Draft Protocol

For an imposed non-interventional PASS subject to the Article 107n procedure, the MAH submits the draft protocol before the study is conducted for PRAC review and endorsement, except where the applicable legal framework provides for national assessment.

EMA's current Q&A states that PRAC assesses these protocols in the applicable EU procedure and describes the procedural timetable for the assessment. citeturn0search0

The practical implication is important: an MAH should not begin a study that requires prior protocol endorsement and then seek regulatory agreement retrospectively.

23. Protocol Endorsement Is Not Merely Administrative

Protocol review addresses whether the proposed study can answer the safety question adequately.

A technically complete submission can still receive an objection if important methodological issues remain unresolved.

The organisation should therefore treat protocol preparation as a scientific and regulatory activity rather than a document-production exercise.

24. Substantial Amendments

After an imposed non-interventional PASS has started, some protocol changes require regulatory review before implementation.

EMA identifies changes that may be substantial when they could affect participant safety or the study results and their interpretation. Examples include changes to:

The organisation should therefore have a controlled protocol-amendment assessment process.

25. Not Every Change Is a Substantial Amendment

A protocol may require administrative or operational changes during its lifecycle.

A change should not be labelled substantial merely because it modifies the protocol document.

The organisation should assess the actual effect of the change on:

The assessment and rationale should be documented.

26. Protocol Deviations

A protocol deviation is different from a planned protocol amendment.

A deviation occurs when the study is conducted differently from the approved or agreed protocol.

Examples can include:

The organisation should determine the impact of significant deviations on study validity and interpretation.

27. Deviations Should Not Be Hidden by Later Amendments

A protocol amendment should not be used to retrospectively legitimise a completed deviation.

The chronology should remain clear:

Approved protocol
      ↓
Deviation identified
      ↓
Impact assessed
      ↓
Corrective action / decision
      ↓
Protocol amendment if appropriate
      ↓
Future conduct

This distinction is important for both scientific integrity and inspection traceability.

28. Joint PASS and Multiple MAHs

A PASS may be conducted jointly by several MAHs.

Joint studies can improve efficiency, particularly when the research question concerns the same active substance or safety issue across multiple products.

However, joint participation does not remove the need to define:

29. Consortium Governance

Where a consortium conducts the study, the organisation should know who is responsible for each regulatory and scientific activity.

A useful responsibility structure can cover:

Responsibility should not become ambiguous merely because work is shared.

30. National Variants of an EU Protocol

A common EU protocol can sometimes require country-specific implementation because of differences in national law, healthcare systems or data availability.

EMA's current procedural advice states that where national variants are necessary for appropriate implementation, they should be submitted as regional appendices to the main protocol. citeturn0search0

The main scientific question should remain coherent while legitimate national implementation differences are controlled.

31. The Protocol Should Explain Data Sources Clearly

For each major data source, the protocol should explain what information it contributes and why it is suitable.

This includes, where applicable:

The purpose is not to create unnecessary technical documentation. It is to establish whether the source can answer the question reliably.

32. Data Quality Is Part of the PASS Requirement

A methodologically appropriate study can still fail if its data are unreliable.

The protocol should therefore identify relevant data-quality controls.

These can include:

Data cleaning should not change the scientific question after results are observed.

33. Analytical Dataset

The analytical dataset should be defined clearly enough to understand when it is complete and suitable for the planned analyses.

EMA's current PASS Q&A links the concept of the end of data collection to the point at which the analytical dataset is completely available and defines the analytical dataset in relation to the minimum data needed for the statistical analyses supporting the primary objectives. citeturn0search0

This distinction matters because data collection may continue in some operational sense while the dataset required for the primary analysis has already become complete.

34. Interim Results Are Not the Final Study Report

Interim results and feasibility information should not automatically be treated as final study results.

EMA's current Q&A states that interim results and feasibility studies of imposed non-interventional PASS do not fall under the Article 107n-q provisions as final study reports; where submission is requested, the appropriate procedural route should be followed. citeturn0search0

The organisation should therefore label and govern interim outputs correctly.

35. Final Study Report

Only a study report considered final by the MAH should be submitted as the final study report under the applicable imposed PASS procedure.

The final report should allow the reviewer to understand:

EMA provides dedicated guidance for the format and content of final reports for non-interventional PASS. citeturn0search20

36. HMA-EMA Catalogue

For imposed non-interventional PASS, the study protocol, final-study-report abstract and final study report are subject to the transparency requirements associated with the HMA-EMA Catalogue of real-world data studies.

EMA's current Q&A states that the MAH should enter the protocol before data collection starts and the final-report abstract within one month after finalisation of the final study report. citeturn0search0

The catalogue requirement should therefore be included in the study lifecycle plan rather than handled as an afterthought.

37. IRIS and Submission Management

EMA states that, from January 2025, MAHs should use the IRIS platform when managing PASS after the original submission, while IRIS and the existing submission gateway continue to serve different functions. citeturn0search0

The operational workflow should therefore distinguish the platform used for the particular activity and should follow current EMA submission instructions rather than relying on historical practice.

38. Voluntary PASS

Not every non-interventional PASS is imposed.

Voluntary PASS can be sponsored or conducted by an MAH on its own initiative. EMA states that non-imposed studies requested in RMPs can fall within this category. citeturn0search0

EMA recommends that companies submit protocols and reports for voluntary PASS in the same manner as imposed PASS, although this is not mandatory in the same way.

39. Category III PASS

For a non-imposed PASS that is an additional pharmacovigilance activity in the RMP, the regulatory status should be understood separately from an imposed PASS.

EMA's current post-authorisation Q&A states that where a category III protocol is requested by the Committee and reflected as a milestone in the RMP, it can be submitted as a post-authorisation measure. It also states that there is no general legal obligation for category III studies to submit the protocol to PRAC unless such submission is requested. citeturn0search0turn0search2

This distinction prevents voluntary RMP activity from being incorrectly treated as though every requirement applicable to an imposed PASS automatically applies.

40. Scientific Advice

Scientific advice can be useful for complex or controversial PASS methodologies.

EMA states that scientific advice is voluntary and that PRAC is systematically involved in PASS-related scientific advice procedures. citeturn0search0

Scientific advice does not replace a mandatory Article 107n protocol procedure where one applies.

Key Takeaways

The non-interventional PASS requirements extend beyond the definition of observational research.

For imposed studies, the organisation must control the regulatory lifecycle from protocol submission through endorsement, amendments, conduct, finalisation and regulatory follow-up.

The distinction between imposed and voluntary PASS is equally important. The organisation should establish the legal basis and procedural route before relying on an assumption that a study is merely a routine observational activity.

41. Quality Management for Non-Interventional PASS

A non-interventional PASS should operate within the pharmacovigilance quality system appropriate to its importance and regulatory status.

The organisation should define controls for:

The existence of a research department or external study provider does not remove the MAH's responsibility for appropriate pharmacovigilance oversight.

42. Vendor Oversight

Many PASS activities are outsourced to contract research organisations, academic groups, database providers or other specialist organisations.

The MAH should retain sufficient oversight to understand:

Outsourcing execution is not outsourcing accountability.

43. Data Governance

Data governance should address the complete chain from source information to the final analytical dataset.

A useful model is:

Source data
    ↓
Extraction
    ↓
Transformation / coding
    ↓
Quality checks
    ↓
Linkage / derivation
    ↓
Analysis dataset
    ↓
Statistical analysis
    ↓
Results
    ↓
Scientific interpretation

Each transformation that could affect interpretation should be controlled and traceable.

44. Missing Data

Missing data should be considered before the analysis begins.

The protocol should describe, where relevant:

A large dataset does not compensate for systematic missingness in variables central to the research question.

45. Confounding Control

Observational studies are particularly vulnerable to confounding because treatment is not randomly assigned.

The protocol should identify clinically important confounders and define how they will be measured and addressed.

Potential approaches include:

The method should be justified in relation to the study question and available data.

Time-related biases can produce misleading results even when the overall study design appears appropriate.

For example, an exposure definition that requires a patient to survive or remain under observation for a period before becoming classified as exposed can introduce immortal time.

The protocol should define time zero, exposure windows and follow-up periods carefully.

This is an example of why a PASS protocol requires epidemiological expertise rather than merely a database query.

47. Outcome Validation

Where the primary outcome is clinically important, the organisation should consider whether the data source identifies it accurately enough for the research question.

Possible approaches include:

The appropriate approach depends on the study question and available evidence.

48. Study Population Generalisability

The population included in the study may differ from the broader population exposed to the medicine.

The final interpretation should therefore consider:

A statistically precise estimate can still have limited relevance if the study population does not represent the population for which the safety question matters.

49. Multinational Studies

A multinational PASS can increase sample size and improve generalisability, but it also introduces heterogeneity.

The protocol should consider differences in:

The analysis should distinguish genuine clinical differences from differences caused by data collection or healthcare systems.

An EU-level PASS framework does not eliminate applicable national requirements.

The study team should identify relevant national requirements concerning matters such as:

The regulatory classification of the PASS and the legal requirements governing its data collection are related but should not be treated as identical questions.

51. Protection of Human Subjects and Data Protection

Non-interventional does not mean that participants or their information have no protections.

The protocol and study governance should address applicable requirements for:

The exact requirements depend on the study design, data source and applicable law.

52. Study Progress Monitoring

The organisation should monitor whether the study remains capable of meeting its objectives and regulatory commitments.

Useful indicators can include:

Monitoring should trigger escalation when the study can no longer reasonably answer the agreed question.

53. What If the Study Becomes Infeasible?

A study can become infeasible because of:

The organisation should not conceal infeasibility by continuing to generate low-quality data.

Instead, it should assess the scientific and regulatory consequences and determine whether a controlled amendment, alternative data source, replacement study or other regulatory action is appropriate.

54. What If the Safety Question Changes?

New evidence can change the question the PASS was intended to answer.

For example, a signal initially focused on incidence may evolve into a question about:

The organisation should determine whether the existing protocol remains fit for purpose. A new question should not simply be inserted into the analysis after the data have been examined.

55. Inspection Perspective: Can the Organisation Explain Why the Study Is Non-Interventional?

An inspector may ask a simple question:

Why is this study classified as non-interventional?

A strong answer should not depend on terminology alone.

The organisation should be able to demonstrate that:

These conclusions should be supported by the protocol, contracts, operational procedures and evidence of actual conduct.

56. Inspection Perspective: Protocol Versus Actual Conduct

An organisation may have an excellent protocol but conduct the study differently.

Inspectors may therefore compare:

The objective is to establish whether the study actually remained within its approved or agreed design.

57. Inspection Perspective: Hidden Intervention

A particularly important concern is when study procedures unintentionally influence clinical care.

For example, study personnel may begin recommending additional tests, altering treatment decisions or applying study-specific clinical criteria.

Such conduct can undermine the original non-interventional classification.

The organisation should therefore train investigators and study staff on the boundary between data collection and clinical intervention.

58. Inspection Perspective: Data Cannot Be Reproduced

An inspector may ask how a key result in the final report was produced.

The organisation should be able to trace the result through:

Final table
   ↓
Statistical program
   ↓
Analysis dataset
   ↓
Derived variables
   ↓
Source data

The exact technical implementation can vary, but the scientific result should be reproducible to an appropriate degree.

59. Inspection Perspective: Amendment Decisions Are Missing

A study may contain many protocol changes but no evidence explaining whether those changes were substantial.

This creates a governance gap.

For significant changes, the organisation should retain the assessment of:

60. QPPV Oversight

The QPPV does not need to personally perform every epidemiological analysis.

The QPPV should, however, have sufficient oversight to understand whether the PASS is:

For significant or imposed PASS, escalation should occur when study progress, methodology or results could affect the product's benefit-risk balance or regulatory obligations.

61. Minimum Evidence Package

For an important PASS, a practical inspection evidence package should allow the organisation to retrieve:

The exact document structure will depend on the quality system.

62. Final Decision Checklist

Before classifying or conducting a non-interventional PASS, ask:

  1. Is the medicine being used according to usual clinical practice?
  2. Are treatment decisions independent of the study?
  3. Is study inclusion independent of prescribing?
  4. Are there any study-specific diagnostic or monitoring procedures?
  5. Is the analytical approach epidemiological?
  6. What is the legal basis for the PASS?
  7. Is the study imposed or voluntary?
  8. Does Article 107n apply?
  9. Could any proposed protocol change be substantial?
  10. Are national legal requirements addressed?
  11. Is the data source fit for purpose?
  12. Can the final results be reproduced and interpreted?
  13. Are regulatory and RMP consequences controlled?
  14. Can the organisation demonstrate effective QPPV oversight?

63. Final Perspective

The strongest non-interventional PASS are not defined merely by what they do not do. They are defined by a deliberate separation between routine clinical decision-making and evidence generation.

That separation must remain intact from protocol development through study conduct and final analysis.

For imposed studies, the regulatory framework adds another layer: the protocol, amendments, results and subsequent actions must follow the applicable EU procedures.

A defensible PASS therefore has two characteristics at the same time:

scientific validity and regulatory traceability.

Neither is sufficient by itself.

Key Takeaways

A non-interventional PASS must be non-interventional in actual conduct, not merely in its title or protocol description.

The organisation should demonstrate the distinction between clinical treatment decisions and research activities, maintain appropriate data and methodological controls, and manage the study within its pharmacovigilance quality system.

For imposed non-interventional PASS, the regulatory lifecycle is especially important: protocol endorsement, substantial amendments, final reporting, transparency requirements and regulatory follow-up must be controlled as part of one integrated process.

References

  1. European Medicines Agency. GVP Module VIII — Post-authorisation safety studies (Rev. 3).
  2. European Medicines Agency. Post-authorisation safety studies (PASS), including current questions and answers.
  3. European Medicines Agency. Guidance for the format and content of the protocol of non-interventional post-authorisation safety studies.
  4. European Medicines Agency. Guidance for the format and content of the final study report of non-interventional post-authorisation safety studies.
  5. European Medicines Agency. GVP Module VIII Addendum I — Requirements and recommendations for submission of information on non-interventional PASS.
  6. Directive 2001/83/EC, as amended, including Articles 107n–107q.
  7. Regulation (EC) No 726/2004, as amended.
  8. Commission Implementing Regulation (EU) No 520/2012, as amended.
  9. ICH M14. General principles on plan, design and analysis of pharmacoepidemiological studies that utilise real-world data for safety assessment of medicines.

Regulatory Note

This article is an educational explanation of EU requirements for non-interventional PASS. It does not replace current EU legislation, GVP Module VIII, EMA procedural guidance, applicable national requirements or an organisation's approved procedures.

PASS procedures and guidance may change. Current regulatory requirements should be verified before a study is classified, submitted, amended or implemented.

Examples and inspection scenarios are illustrative unless an authoritative source is specifically identified.

Revision History

Related Articles

GVP Module VIII Addendum I: Submission of Information on Non-Interventional PASS

A practical guide to the submission architecture for non-interventional PASS, distinguishing legal submission obligations, recommendations, national procedures, EU procedures and study registration.

GVP Module VIII: PASS Protocols, Objectives and Study Design

A practical guide to PASS protocol development, covering the research question, objectives, study design, population, variables, data sources, analysis, bias and regulatory expectations.

GVP Module VIII: Imposed and Non-Imposed PASS

A structured explanation of imposed and non-imposed PASS, showing how the regulatory basis determines the applicable procedure, responsibilities, documentation and oversight.

Last reviewed: 2026-08-24