How to Write Aggregate Safety Reports

Aggregate safety reporting is the point at which pharmacovigilance moves from individual observations to an organised assessment of what is known about a medicinal product over time. Individual case safety reports, studies, literature, exposure estimates, signals, regulatory actions and product information all become inputs into a higher-level question: has the accumulated evidence changed the current understanding of the product, and does that change require action?

Different aggregate reports ask that question from different regulatory perspectives. A PBRER evaluates the post-authorisation benefit-risk balance. A DSUR focuses on the evolving safety of an investigational drug and protection of clinical-trial participants. A U.S. PADER or PAER fulfils a different periodic post-marketing reporting framework unless an alternative PBRER format has been accepted. These reports share scientific disciplines, but they are not interchangeable merely because they all contain cumulative safety data.

The first skill in aggregate writing is therefore not prose. It is scope control: understanding which regulatory question the report is required to answer, which population and time interval it covers, which reference information applies and which evidence belongs inside the assessment.

Purpose and Regulatory Context

An aggregate report should transform heterogeneous evidence into a reproducible scientific conclusion. The regulatory framework determines what type of conclusion is required.

PBRER and EU PSUR

ICH E2C(R2) defines the Periodic Benefit-Risk Evaluation Report. It focuses on cumulative knowledge while retaining emphasis on information that became available during the reporting interval. In the EU, the PSUR follows the PBRER architecture and is governed by EU pharmacovigilance legislation, Commission Implementing Regulation (EU) No 520/2012 and GVP Module VII.

Its central question is whether the benefit-risk balance remains favourable for approved use and whether new information requires changes to product information, pharmacovigilance or risk minimisation.

DSUR

ICH E2F defines the Development Safety Update Report. The DSUR is an annual review of the safety of an investigational drug during clinical development. Its central purpose is ongoing protection of trial participants and evaluation of whether the evolving safety profile changes the conduct of the development programme.

The DSUR therefore places different emphasis on cumulative subject exposure, serious adverse events, important identified and potential risks, clinical-trial findings, changes to the investigator's brochure and actions affecting ongoing studies.

U.S. periodic post-marketing reports

U.S. periodic adverse drug experience reports and periodic adverse experience reports arise from U.S. post-marketing requirements. FDA allows applicants, under the conditions described in its guidance, to use the ICH E2C(R2) PBRER format as an alternative to PADER/PAER or older PSUR formats for applicable products.

The important writing principle is that an international company should not assume one aggregate format automatically satisfies every jurisdiction. Regulatory applicability must be confirmed before scientific content is reused.

Common Scientific Architecture

Despite their procedural differences, robust aggregate reports share a common reasoning chain:

source evidence → data quality and context → pattern recognition → clinical interpretation → cumulative knowledge → regulatory significance → action.

The chain should remain visible even when the report template differs.

Observation is not interpretation

An aggregate report may state that 42 serious cases were received, that a study showed a hazard ratio above 1, or that an eRMR produced a disproportionate reporting pattern. Those are observations. Scientific interpretation asks what the observations mean, how reliable they are, whether they are consistent with other sources and whether they alter previous knowledge.

A large volume of correctly tabulated data cannot compensate for the absence of interpretation.

Interpretation is not action

A medically plausible association does not automatically determine regulatory action. The reviewer must consider strength of evidence, seriousness, preventability, uncertainty, exposure, therapeutic alternatives, benefit and the consequences of intervention.

The report should make that reasoning explicit rather than jumping from data to recommendation.

Define the Reporting Frame Before Collecting Data

Every aggregate project should begin by fixing the analytical boundary.

The team should confirm:

• report type and legal or regulatory basis;
• data lock point and reporting interval;
• active substance, formulation and indication scope;
• countries or regions covered;
• reference safety or product information;
• cumulative period relevant to the report;
• required regional appendices or modules;
• partner-data responsibilities; and
• submission timeline.

A report can be scientifically sophisticated yet non-compliant if it uses the wrong DLP, omits a partner dataset, applies the wrong reference information or answers the wrong regulatory question.

Build the Evidence Map

Before drafting, map the sources that can change the report's scientific conclusions. Depending on report type, these may include:

• ICSRs and cumulative safety tabulations;
• clinical-trial data;
• non-interventional studies and PASS;
• scientific literature;
• non-clinical findings;
• epidemiology and real-world evidence;
• exposure and utilisation estimates;
• signal assessments;
• product-information changes;
• RMP or development-risk information;
• regulatory correspondence and actions;
• benefit or efficacy/effectiveness information; and
• quality information with safety implications.

The evidence map should identify source ownership and cut-off dates. Where datasets come from different systems, reconciliation may be necessary before the report can be trusted.

Evidence hierarchy depends on the question

There is no single fixed hierarchy for every safety issue. A rare anaphylactic reaction may be established largely through characteristic case evidence. A suspected increase in myocardial infarction may require comparative epidemiology. A developmental toxicity concern may depend heavily on non-clinical findings until sufficient human exposure accumulates.

The aggregate writer's task is to understand what each source can and cannot establish.

From Evidence to Cumulative Interpretation

Aggregate writing requires the reviewer to move repeatedly between interval information and cumulative knowledge. New information matters because it is compared with what was known before the reporting period.

A useful sequence is:

1. What new information appeared?
2. Is the information reliable and clinically meaningful?
3. Does it confirm, weaken or change previous understanding?
4. Does it alter the characterisation of a known risk or create a new safety question?
5. Does the updated knowledge require action?

This sequence should be visible throughout the report rather than reserved for the conclusion.

Working With ICSR Data

ICSRs are often numerically prominent in aggregate reports, but raw case counts can be misleading. Spontaneous reporting is affected by under-reporting, publicity, geography, product age, indication, reporting source and many other factors.

When discussing an ICSR pattern, consider:

• case validity and duplicate management;
• clinical phenotype and diagnostic certainty;
• chronology and time to onset;
• dose and duration where relevant;
• dechallenge or rechallenge information;
• concomitant medicines and alternative explanations;
• seriousness and outcome;
• exposure context; and
• consistency with other evidence sources.

A spontaneous reporting rate calculated using estimated exposure is still a reporting rate, not necessarily an incidence rate. True incidence generally requires systematic case ascertainment and a valid exposed denominator.

Working With Clinical-Trial Data

Clinical trials provide controlled exposure and often comparator information, but their interpretive value depends on study design, population, duration and event frequency.

Aggregate reviewers should distinguish:

• numerical imbalance from statistically or clinically meaningful difference;
• pre-specified analyses from post hoc observations;
• common events from rare events for which the trial is underpowered;
• trial-specific findings from reproducible programme-level patterns; and
• protocol-selected populations from routine clinical use.

For a DSUR, these distinctions are central to participant protection and development decisions. For a PBRER, relevant post-authorisation or continuing trial findings contribute to cumulative benefit-risk assessment.

Working With Observational and Real-World Evidence

Observational studies can provide exposure denominators, comparative rates and evidence from broader populations, but they also introduce confounding, selection bias and misclassification.

Do not reduce a study to its effect estimate. The report should consider:

• study design;
• source population;
• exposure definition;
• outcome definition and validation;
• comparator selection;
• confounding control;
• missingness and data completeness;
• sensitivity analyses; and
• generalisability to the population relevant to the report.

A large database does not remove systematic bias.

Working With Literature

Literature can contribute individual cases, mechanistic evidence, clinical studies, epidemiology and external expert interpretation. Search and screening processes should be reproducible where they support regulated pharmacovigilance obligations.

For aggregate assessment, however, the key question is not how many papers were found. It is which publications materially change the evidence base and why.

A well-written report distinguishes a single hypothesis-generating case report from a well-designed comparative study and from a replicated body of evidence.

Exposure as the Denominator Context

Exposure helps explain whether changes in case counts may reflect changes in use and supports interpretation of rates where the data permit them.

Common metrics include:

• estimated patients exposed;
• treatment courses;
• patient-days or patient-years;
• prescriptions or dispensings;
• sales-derived treatment units; and
• directly measured clinical-trial exposure.

The method should fit the treatment pattern. Patient-years may be informative for chronic therapy but misleading for a one-time treatment; treatment courses may be more appropriate for short-course medicines.

Every estimate depends on assumptions. Changes in dose, duration, adherence, indication or market penetration can alter the conversion from product volume to exposed patients.

Signals and Risks Are Not the Same Thing

A signal is information suggesting a new potentially causal association or new aspect of a known association that merits further evaluation. A risk is a more established safety concept that can be characterised and managed.

Aggregate reports should preserve that distinction. A signal can be:

• newly identified;
• ongoing; or
• closed after evaluation.

A closed signal may result in a newly identified risk, refinement of an existing risk, or a conclusion that the evidence does not support a causal association.

The report should therefore show the path from signal evidence to cumulative risk knowledge rather than simply listing both categories independently.

Characterising Risk

Risk characterisation should convert a diagnostic label into clinically useful knowledge. Depending on evidence availability, consider:

• frequency or occurrence pattern;
• seriousness and severity;
• time to onset;
• duration and reversibility;
• dose or exposure relationship;
• susceptible populations;
• risk factors;
• preventability and detectability;
• outcome;
• mechanism or biological plausibility; and
• remaining uncertainty.

This level of characterisation supports regulatory decisions far better than a list of adverse-event terms.

Benefit-Risk Reasoning

Not every aggregate report requires the same benefit evaluation. The PBRER explicitly requires integrated benefit-risk evaluation; a DSUR is principally a development safety report, although overall risk considerations remain essential to decisions about ongoing trials.

For reports requiring benefit-risk assessment, the writer should avoid a mechanical “benefits outweigh risks” statement. The analysis should identify the benefits that matter for the indication, the risks that materially affect those benefits, the populations in whom the balance may differ and the uncertainties that could change the conclusion.

A useful conceptual structure is:

important benefit → important risk → magnitude and uncertainty → available risk control → therapeutic alternatives → effect on overall balance.

Formal quantitative models may be useful in some settings, but are not universally required for routine aggregate reports.

Writing Proportionately

Proportionality is a scientific principle, not a shortcut. A mature product with a stable safety profile may need concise discussion of unchanged areas and detailed discussion only of the few developments that matter. A newly authorised medicine with several evolving signals may require much greater depth.

The report should allocate space according to decision relevance, not according to how much source data a function supplied.

Avoid equal-weight writing

A common weakness is giving every event, publication or study the same narrative prominence. This hides the safety story. The report should make important developments easy for the assessor to recognise.

Managing Multiauthor Reports

The lead author should function as the integrator of scientific reasoning. Contributors from safety operations, clinical development, epidemiology, regulatory affairs and other functions may supply accurate text that nevertheless conflicts in terminology, cut-off date or conclusion.

Useful integration checks include:

• one definition of the reporting interval;
• one controlled set of exposure numbers;
• consistent signal status across sections;
• consistent risk terminology;
• consistent reference information;
• reconciliation of regulatory actions with safety discussions; and
• a conclusion that follows from the evidence presented earlier.

Shared content can be reused between reports where guidance allows, but it should be reviewed for the receiving report's purpose and current DLP.

Review and Quality Control

A robust review process separates scientific challenge from technical quality control.

Scientific review

Ask whether conclusions are supported by the evidence and whether important alternative explanations have been considered. Challenge causality, study interpretation, exposure assumptions, risk characterisation and benefit-risk reasoning.

Consistency review

Compare the report with relevant controlled documents: reference safety information, product information, investigator's brochure where applicable, RMP, signal records, previous aggregate reports and important regulatory correspondence.

Data review

Confirm important numbers, dates, DLPs, exposure estimates, case counts and study statuses against controlled sources. Where two sections legitimately use different populations or cut-offs, explain the difference.

Editorial and submission review

Only after the scientific content is stable should the team finalise cross-references, formatting, appendices and technical submission requirements.

This staged approach reduces the risk that a polished document conceals an incoherent scientific assessment.

Special Situations

Products still in development after marketing authorisation

A product may require both a DSUR and a PBRER/PSUR. ICH guidance supports sharing of some modules when DLPs and content permit, but the two reports retain different purposes. Development safety information should not be copied into the PBRER without considering approved-use relevance, and post-marketing information used in the DSUR should be interpreted in relation to trial-participant safety.

Multiple partners or licence holders

Written agreements should define data exchange, preparation and submission responsibilities. Aggregate conclusions can only be reliable if relevant partner information arrives before the report cut-off and is incorporated consistently.

Combination products or multiple indications

Different indications may have different benefits, doses, durations, populations and risk tolerances. An aggregate conclusion that is reasonable overall may conceal an unfavourable balance in a particular indication or subgroup. Structure the analysis accordingly.

Limited exposure

For orphan products, new medicines or specialised therapies, quantitative estimates may be unstable. A small number of high-quality cases can therefore carry substantial weight, while absence of reports provides limited reassurance. The report should state the uncertainty explicitly.

Potential Failure Modes

The following are illustrative failure modes rather than cited inspection findings.

Data collection starts before the regulatory question is defined

Teams request every available dataset without first confirming report scope, creating large volumes of material that may not answer the required question.

The report becomes a case-count catalogue

Numbers are presented accurately but without explaining clinical coherence, exposure context or cumulative meaning.

A reporting rate is presented as incidence

Spontaneous reports divided by estimated exposure do not automatically produce epidemiological incidence. The limitations of both numerator and denominator must be respected.

Signals, risks and adverse reactions are used interchangeably

This blurs the status of evidence. A signal under evaluation is not the same as an established adverse reaction or characterised risk.

Observational associations are treated as causal findings

Large sample size does not eliminate confounding or misclassification. Study design determines evidentiary strength.

The conclusion appears for the first time at the end

A well-written conclusion should be the logical consequence of discussions already developed throughout the report.

Report types are treated as interchangeable

PBRER, DSUR and PADER/PAER may share data, but their legal basis, populations, reference information and regulatory objectives differ.

Uncertainty is edited out

Overconfident language can misrepresent the evidence. Remaining uncertainty should be identified and connected to any proposed monitoring or data-generation plan.

Inspection and Regulatory Review Considerations

An assessor or inspector may test whether the final report is supported by the pharmacovigilance system that produced it.

Illustrative questions include:

• How was the report scope and DLP established?
• Which source systems contributed data?
• How were partner data reconciled?
• Can key exposure estimates be reproduced?
• How were important signals selected for discussion?
• Why did a particular association become a risk, remain a signal or get closed?
• How were conflicting study results interpreted?
• Are risk conclusions consistent with reference information and the RMP or investigator's brochure as applicable?
• How did the final conclusion lead to regulatory or risk-management action?
• Were actions from the previous aggregate report completed?
• Can the organisation retrieve the source evidence supporting a material conclusion?

These are illustrative questions, not a universal regulatory checklist.

Practical Final Checklist

Before approving an aggregate report, confirm that:

• the correct report type and regulatory framework have been applied;
• DLP, interval, product scope and reference information are correct;
• all material source datasets are included or omissions are justified;
• interval evidence is interpreted against cumulative knowledge;
• case counts are not mistaken for incidence;
• exposure estimates are reproducible and limitations stated;
• clinical-trial and observational evidence are interpreted according to study design;
• signals, adverse reactions and risks are distinguished consistently;
• important risk characterisation is clinically meaningful;
• benefit evaluation is adequate where required by the report type;
• important uncertainty is visible rather than suppressed;
• conclusions follow logically from the preceding evidence;
• proposed actions are traceable to conclusions;
• terminology and numbers are consistent across sections;
• shared content from other reports has been reviewed for current scope and DLP; and
• the final document can be reconstructed from controlled source evidence.

Key Takeaways

Aggregate safety reporting is not one document type. PBRER/PSUR, DSUR and U.S. periodic post-marketing reports each have distinct regulatory purposes, even though they share important scientific methods.

The common discipline is evidence integration. The writer defines the analytical frame, establishes cumulative context, evaluates interval evidence, characterises signals and risks, communicates uncertainty and derives actions from the resulting scientific understanding.

High-quality aggregate writing therefore depends less on how much information is included than on whether the reader can follow the reasoning from observation → interpretation → cumulative knowledge → regulatory significance → action.

The governing standard is traceability. Every important conclusion should be supported by identifiable evidence, and every important proposed action should arise from a conclusion that the report has already explained.

References

1. International Council for Harmonisation. ICH E2C(R2): Periodic Benefit-Risk Evaluation Report (PBRER). https://database.ich.org/sites/default/files/E2C_R2_Guideline.pdf
2. International Council for Harmonisation. ICH E2F: Development Safety Update Report. https://database.ich.org/sites/default/files/E2F_Guideline.pdf
3. International Council for Harmonisation. ICH E2C(R2) Questions and Answers. https://database.ich.org/sites/default/files/E2CR2_Q%26As_Q%26As.pdf
4. European Medicines Agency. Periodic safety update reports (PSURs), including GVP Module VII and current explanatory material. https://www.ema.europa.eu/en/human-regulatory-overview/post-authorisation/pharmacovigilance-post-authorisation/periodic-safety-update-reports-psurs
5. European Medicines Agency. Guideline on good pharmacovigilance practices (GVP) Module VII – Periodic safety update report (Rev. 1). EMA/816292/2011 Rev. 1. Available from the EMA PSUR page.
6. European Commission. Commission Implementing Regulation (EU) No 520/2012, consolidated version. https://eur-lex.europa.eu/eli/reg_impl/2012/520/2026-02-12/eng
7. U.S. Food and Drug Administration. Providing Postmarket Periodic Safety Reports in the ICH E2C(R2) Format (Periodic Benefit-Risk Evaluation Report). November 2016. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/providing-postmarket-periodic-safety-reports-ich-e2cr2-format-periodic-benefit-risk-evaluation
8. U.S. Food and Drug Administration. Providing Submissions in Electronic Format — Postmarketing Safety Reports. April 2022. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/providing-submissions-electronic-format-postmarketing-safety-reports

Regulatory Note

ICH E2C(R2) and E2F are internationally harmonised guidelines for PBRER and DSUR respectively. EU PSUR requirements arise from applicable EU legislation and are implemented through Commission Implementing Regulation (EU) No 520/2012, GVP Module VII, the EURD list and current EMA procedural guidance. U.S. PADER/PAER obligations arise from the applicable U.S. regulatory framework; FDA guidance describes conditions under which a PBRER format may be used instead. Scientific workflows, review models, source maps, checklists and inspection questions in this article are recommended or illustrative practices unless an authoritative requirement is specifically identified.