Estimating Patient Exposure in Pharmacovigilance

Introduction

Estimating patient exposure is one of the most important activities performed during aggregate safety evaluation. Almost every quantitative assessment within pharmacovigilance depends upon understanding how many patients have received a medicinal product, how long they were exposed and the characteristics of the exposed population. Without this information, interpretation of adverse event data becomes extremely difficult because the number of reported events cannot be placed into an appropriate clinical context.

Exposure estimation underpins numerous pharmacovigilance activities throughout the lifecycle of a medicinal product. It is used during preparation of Periodic Benefit-Risk Evaluation Reports (PBRERs), Periodic Adverse Drug Experience Reports (PADERs), Development Safety Update Reports (DSURs), signal detection, benefit-risk evaluations, post-authorisation safety studies and many other aggregate analyses. Although the objectives of these activities differ, they all require an understanding of the denominator against which safety observations should be interpreted.

Despite its importance, patient exposure cannot usually be measured directly once a medicinal product enters routine clinical practice. Unlike controlled clinical trials, where every patient's treatment history is documented prospectively, post-marketing pharmacovigilance relies upon indirect estimates derived from multiple data sources. Sales data, dispensing records, prescription databases, electronic healthcare records, patient registries and observational studies each provide partial information regarding product utilisation. The aggregate report author must understand the strengths and limitations of these sources before selecting an appropriate exposure estimate.

Exposure estimation is therefore an exercise in scientific approximation rather than mathematical certainty. The objective is not to determine the exact number of patients treated worldwide but to produce estimates that are sufficiently robust to support meaningful interpretation of cumulative safety data. Every exposure estimate is based upon assumptions, and those assumptions should be scientifically justified, transparent and appropriate for the intended analysis.

This article explains the principles of exposure estimation in pharmacovigilance, the major sources of exposure data, commonly used calculation methods, situations in which different approaches are appropriate and the limitations that should be recognised during aggregate safety evaluation.

Why Exposure Matters

Pharmacovigilance focuses primarily on identifying, evaluating and managing risks associated with medicinal products. Reports describing suspected adverse reactions provide the numerator for many aggregate safety assessments. However, the numerator alone rarely provides sufficient information to determine whether an observed safety concern represents a meaningful change in the safety profile of the product.

Consider two medicinal products. Each has accumulated one hundred reports of a particular adverse reaction. At first glance, both appear to have similar safety profiles. However, if the first product has been used by one thousand patients while the second has been used by ten million patients, the clinical interpretation is entirely different. The absolute number of reports is identical, yet the extent of patient exposure differs by several orders of magnitude.

Exposure therefore provides the denominator that allows reported safety events to be interpreted within an appropriate clinical context.

Illustrative Example

Product A

Reported Cases: 100

Estimated Patients Exposed: 1,000

Approximate Reporting Rate: 100 reports per 1,000 exposed patients

Product B

Reported Cases: 100

Estimated Patients Exposed: 10,000,000

Approximate Reporting Rate: 0.01 reports per 1,000 exposed patients ```

Although spontaneous reporting rates should never be interpreted as true incidence rates, this simple example illustrates why exposure estimation remains fundamental to aggregate safety evaluation. Interpretation based solely upon case counts may be misleading if the extent of product utilisation differs substantially between products or changes considerably over time.

Exposure also influences interpretation of temporal trends. An apparent increase in adverse event reporting may reflect increased product utilisation rather than deterioration of the underlying safety profile. Conversely, declining case counts may simply reflect reduced market utilisation following commercial decisions or changes in clinical practice. Accurate exposure estimation therefore supports more informed interpretation of cumulative pharmacovigilance data.

Writing Tip

Exposure estimates should always be interpreted together with the characteristics of the exposed population. Changes in indication, dose, treatment duration or patient demographics may alter interpretation even when the overall number of exposed patients remains relatively stable.

The Role of Exposure Throughout the Product Lifecycle

Methods used to estimate patient exposure evolve throughout the lifecycle of a medicinal product because the available data sources change considerably following marketing authorisation.

During clinical development, exposure is measured directly. Every subject enrolled within a clinical trial has documented treatment dates, treatment duration, cumulative dose, protocol deviations and follow-up information. Investigators therefore know precisely how many subjects received the investigational product, how long treatment continued and the extent of follow-up available for safety evaluation.

Following marketing authorisation, this level of precision is rarely available. Patients obtain medicines through diverse healthcare systems, prescribing practices vary between countries, adherence differs considerably between individuals and many products are available through multiple commercial channels. Exposure estimation therefore becomes progressively dependent upon indirect sources of information and scientifically justified assumptions.

As product utilisation expands globally, exposure estimation often combines information from several complementary sources. Sales data may provide worldwide utilisation estimates, electronic healthcare databases may support analyses within selected populations, post-authorisation safety studies may provide more detailed exposure information for specific patient groups and patient registries may generate long-term exposure data for rare diseases or specialised therapeutic areas.

The appropriate exposure methodology therefore depends upon the purpose of the analysis, the stage of the product lifecycle, the available data sources and the scientific question being addressed.

Sources of Exposure Data

The accuracy of any exposure estimate depends primarily upon the quality and suitability of the underlying data source. No single source is appropriate for every pharmacovigilance activity. Instead, authors should understand the strengths, limitations and assumptions associated with each source before selecting an exposure estimation methodology.

An important distinction should be made between exposure data collected prospectively and exposure estimates derived indirectly. During clinical development, exposure is measured directly because treatment administration is documented for every participant. Following marketing authorisation, however, exposure usually has to be estimated using surrogate measures of product utilisation. Consequently, post-authorisation exposure estimates inevitably contain varying degrees of uncertainty.

The choice of exposure source should therefore be driven by the scientific objective of the analysis rather than by data availability alone.

Clinical Trial Exposure

Clinical trials provide the most accurate measurement of patient exposure because treatment administration is recorded prospectively under controlled conditions. Information is usually available regarding treatment start and stop dates, cumulative dose, treatment interruptions, dose modifications, duration of follow-up and patient characteristics.

This level of detail allows precise calculation of patients exposed, patient-years of exposure, cumulative dose and treatment duration.

Clinical trial exposure is therefore particularly valuable during clinical development and continues to contribute important information to Development Safety Update Reports (DSURs). It may also contribute to PBRERs where important ongoing extension studies or post-authorisation clinical trials remain in progress.

However, clinical trial populations are usually highly selected. Inclusion and exclusion criteria, protocol-defined monitoring and controlled treatment conditions frequently limit the generalisability of exposure estimates to routine clinical practice. Consequently, clinical trial exposure should not automatically be regarded as representative of worldwide post-marketing utilisation.

Medical Review Consideration

Consider whether the clinical trial population differs substantially from the authorised treatment population with respect to age, disease severity, comorbidities, treatment duration or concomitant therapies.

Sales Data

Worldwide sales data remain one of the most frequently used sources of exposure estimation for authorised medicinal products because they are generally available for all marketed countries and can be updated throughout the reporting period.

Sales data estimate the quantity of medicinal product supplied into the market rather than the number of patients treated. Consequently, additional assumptions are usually required before patient exposure can be estimated. These assumptions may include the average daily dose, average treatment duration, dosage strength and expected utilisation patterns.

Sales data are particularly useful for worldwide aggregate reports because they provide broad estimates across large populations and multiple territories. They are commonly used during preparation of PBRERs where precise patient-level utilisation data may not be available globally.

Nevertheless, important limitations should be recognised. Units sold do not necessarily correspond to units consumed. Products may remain in the distribution chain, expire before use, be discarded, stockpiled or supplied to healthcare institutions before administration to patients. Furthermore, prescribing practices may differ considerably between countries, making application of a single conversion factor inappropriate.

Sales-derived exposure estimates should therefore be regarded as informed approximations rather than direct measurements of patient exposure.

Writing Tip

Whenever sales data are converted into patient exposure estimates, the assumptions used for the conversion should be documented clearly and applied consistently throughout the analysis.

Prescription and Dispensing Databases

Prescription databases and dispensing records provide a closer approximation of patient utilisation than wholesale sales data because they reflect medicines prescribed or dispensed to individual patients.

Depending upon the healthcare system, these databases may capture prescription date, quantity supplied, strength, dosage instructions, prescriber specialty and patient demographics. Some databases also permit longitudinal follow-up of individual patients, allowing estimation of treatment duration and persistence.

These sources are particularly valuable when evaluating exposure within specific countries or healthcare systems. They may also support pharmacoepidemiological investigations, utilisation studies and post-authorisation safety studies.

However, prescriptions do not necessarily indicate treatment initiation, while dispensing records do not confirm that patients actually consumed the medicine. As with all secondary data sources, interpretation should therefore consider the distinction between medicine supplied and medicine administered.

Electronic Health Records and Claims Databases

Electronic health records and administrative claims databases provide increasingly important sources of exposure information for pharmacoepidemiological research.

Unlike sales data, these systems often contain detailed clinical information, including diagnoses, laboratory results, procedures, treatment duration and longitudinal patient follow-up. This allows exposure estimates to be stratified by indication, age, sex, comorbidity or other clinically important variables.

Claims databases may additionally support estimation of healthcare utilisation, treatment persistence and switching between therapies.

These data sources are particularly valuable when evaluating specific safety questions within defined populations. However, they frequently represent selected healthcare systems rather than worldwide utilisation and may contain incomplete capture of medicines obtained outside the participating system.

Consequently, although these databases provide greater clinical detail than sales data, their generalisability should always be evaluated carefully.

Disease Registries and Post-Authorisation Safety Studies

Disease registries and Post-Authorisation Safety Studies (PASS) provide highly characterised exposure information for selected patient populations.

Registries often include detailed information regarding disease severity, treatment duration, concomitant therapies, pregnancy status, long-term outcomes and other variables that are unavailable from routine commercial datasets.

Similarly, PASS may collect prospectively defined exposure information specifically designed to answer important regulatory safety questions. These studies frequently provide the most reliable estimates of exposure for specialised patient populations, although they usually represent only a subset of worldwide product utilisation.

Data generated from registries and PASS should therefore be interpreted within the context of their study design and intended population. They complement rather than replace broader exposure estimates obtained from commercial utilisation data.

Choosing an Appropriate Exposure Source

Selection of an exposure source should begin with the scientific objective rather than the available dataset.

A worldwide PBRER may require broad estimates of cumulative product utilisation across all marketed countries, making sales-derived exposure the most practical approach. By contrast, a PASS investigating the incidence of a specific adverse event within elderly patients may require patient-level exposure obtained from electronic healthcare databases or disease registries.

Similarly, a DSUR may rely predominantly upon clinical trial exposure because the investigational product has not yet entered routine clinical practice.

The objective is not to identify the perfect exposure source, because such a source rarely exists. Instead, authors should select the source that most appropriately answers the scientific question being addressed while acknowledging the limitations of the available data.

Inspection Insight

Inspectors and regulatory assessors generally recognise that exposure estimates involve assumptions. Greater concern usually arises when those assumptions are undocumented, inconsistent between reporting periods or unsupported by the available evidence.

Methods for Estimating Patient Exposure

Exposure can be expressed in several different ways depending upon the scientific question being addressed. There is no universally correct exposure metric. The choice depends upon the medicinal product, therapeutic area, available data, treatment characteristics and the purpose of the analysis.

Some analyses require only an estimate of the number of patients who have received the medicinal product. Others require adjustment for treatment duration, cumulative exposure or dose. Consequently, authors should understand both the advantages and the limitations of each exposure metric before selecting an approach.

An important principle is that the exposure estimate should be appropriate for the adverse event being evaluated. Acute adverse reactions occurring shortly after treatment initiation may be adequately assessed using the number of patients exposed. Conversely, adverse reactions associated with prolonged treatment, cumulative dose or chronic exposure often require person-time measures such as patient-years.

Number of Patients Exposed

The simplest exposure metric is the estimated number of patients who have received the medicinal product during a defined period.

Conceptually,

Estimated Patients Exposed

=

Total Quantity Distributed or Administered

÷

Average Quantity Used Per Patient

The practical implementation of this calculation depends upon the available data source.

When patient-level information is available, the calculation is straightforward because each treated individual can be counted directly.

When sales data are used, however, the estimate depends upon assumptions regarding average daily dose, duration of treatment and expected utilisation patterns. These assumptions should always be documented because they directly influence the resulting exposure estimate.

The number of patients exposed is particularly useful when evaluating products administered as single courses of treatment, vaccines, acute therapies or medicines associated with short treatment duration.

However, this metric becomes progressively less informative for chronic therapies because patients treated for one week and patients treated for five years are both counted equally.

Patient-Time Measures

Many medicinal products are administered continuously for months or years. For these products, the duration of treatment influences the probability of developing many adverse reactions.

Patient-time metrics account for this by incorporating both the number of exposed patients and the duration of their exposure.

The most commonly used metric is the patient-year.

One patient treated continuously for one year contributes one patient-year.

Two patients each treated for six months also contribute one patient-year.

Four patients treated for three months each likewise contribute one patient-year.

Consequently, patient-years provide a common measure of cumulative exposure regardless of the number of individual patients involved.

Patient-time measures are particularly valuable when evaluating adverse reactions that may be related to cumulative exposure or prolonged treatment.

Writing Tip

Whenever patient-years are used, clearly describe the assumptions used to estimate treatment duration, particularly when calculations are based upon sales rather than individual patient records.

Patient-Years of Exposure

Patient-years are widely used within pharmacovigilance because they standardise exposure across different treatment durations.

The general principle is straightforward.

The duration of treatment contributed by every patient is summed and expressed in years.

Examples include:

One patient treated for twelve months

=

1 patient-year

Two patients treated for six months each

=

1 patient-year

Ten patients treated for thirty-six days each

≈

1 patient-year

Patient-years facilitate comparison between reporting periods, studies and products where treatment duration differs substantially.

However, patient-years remain estimates unless calculated from individual treatment records. When derived from commercial utilisation data they depend upon assumptions regarding average treatment duration and adherence.

Authors should therefore distinguish clearly between measured patient-years and estimated patient-years.

Exposure Derived from Sales Data

Sales-derived exposure estimation remains one of the most common approaches used for worldwide aggregate reports because comprehensive patient-level utilisation data are rarely available.

The calculation generally involves several stages.

First, the total quantity of medicinal product distributed during the reporting period is determined.

Second, this quantity is converted into total treatment days using assumptions regarding average daily dose.

Finally, total treatment days are converted into patient-years.

Each stage introduces additional assumptions.

These assumptions may include:

• average prescribed dose;
• average treatment duration;
• treatment adherence;
• wastage;
• stock remaining within the supply chain;
• regional prescribing differences.

The resulting estimate should therefore be regarded as an approximation intended to support aggregate interpretation rather than an exact measurement of patient utilisation.

Medical Review Consideration

Review whether assumptions remain consistent with previous reporting periods. Changes in methodology may produce apparent changes in exposure that reflect calculation differences rather than genuine changes in product utilisation.

Defined Daily Dose

For certain medicinal products, exposure may be estimated using the World Health Organization Defined Daily Dose (DDD).

The Defined Daily Dose represents the assumed average maintenance dose per day for the principal indication in adults.

DDD methodology allows standardisation of utilisation across healthcare systems and is widely used in pharmacoepidemiology.

However, DDD has important limitations.

The WHO specifically states that the Defined Daily Dose should not be interpreted as the recommended prescribed dose or the actual dose received by individual patients.

Medicinal products with weight-based dosing, highly individualised treatment regimens, oncology protocols or multiple approved indications often show substantial differences between prescribed dose and the Defined Daily Dose.

Consequently, DDD should only be used where appropriate for the scientific objective and should not automatically replace product-specific exposure estimates.

Exposure Estimates Based on Treatment Courses

Some medicinal products are administered as complete treatment courses rather than continuous therapy.

Examples include:

• antibiotics;
• antiviral therapies;
• fertility treatments;
• selected oncology regimens;
• vaccines administered according to predefined schedules.

For these products, estimating completed treatment courses may provide a more meaningful denominator than patient-years.

The exposure metric should therefore reflect the biology of treatment rather than following a predetermined calculation method.

The objective is always to produce an exposure estimate that supports appropriate interpretation of the available safety data.

Selecting the Appropriate Exposure Metric

There is no universally correct method for estimating patient exposure. The most appropriate exposure metric depends upon the medicinal product, its approved indications, treatment regimen, duration of therapy, available data sources and the scientific question being addressed. Choosing the wrong denominator may result in misleading conclusions, even when the calculations themselves are mathematically correct.

The first step is therefore not to perform a calculation but to understand how the medicinal product is used in routine clinical practice. Products intended for lifelong treatment require different exposure measures from medicines administered as a single dose, while medicines prescribed intermittently require different assumptions from products given according to fixed treatment schedules.

The exposure metric should reflect the clinical reality of treatment rather than forcing every product into the same mathematical model.

Continuous Long-Term Therapy

Medicinal products intended for long-term or lifelong treatment are usually best described using patient-time measures such as patient-years of exposure.

Examples include antihypertensive medicines, oral antidiabetic therapies, lipid-lowering agents, antiepileptic medicines, disease-modifying antirheumatic drugs and many maintenance biologics.

For these products, treatment duration is clinically important because the probability of many adverse reactions increases with cumulative exposure. Simply counting the number of patients exposed may underestimate important differences between patients treated for several weeks and those treated continuously for many years.

Patient-years therefore provide a more meaningful denominator for many chronic therapies because they incorporate both the number of exposed patients and the duration of treatment.

However, patient-years should still be regarded as estimates unless derived from individual patient-level treatment records. When calculated from sales or utilisation data, the assumptions regarding average treatment duration and adherence should always be documented and periodically reviewed.

Short-Course Therapy

Medicinal products administered for defined treatment courses often require different exposure measures.

Examples include antibacterial agents, antiviral medicines, Helicobacter pylori eradication regimens, oral corticosteroid courses and many peri-operative treatments.

For these medicines, the number of completed treatment courses or the estimated number of treated patients frequently provides a more meaningful denominator than patient-years. A patient receiving a seven-day course of antibiotics contributes very little person-time, making patient-years difficult to interpret and potentially misleading.

Treatment-course estimates may therefore better reflect actual clinical practice, particularly when evaluating adverse reactions occurring during or shortly after treatment.

Where repeated treatment courses are common, authors should explain whether exposure represents treatment courses, individual patients or both.

Intermittent or "As Required" Therapy

Medicinal products prescribed on an intermittent basis present additional challenges for exposure estimation.

Examples include antihistamines, non-steroidal anti-inflammatory drugs, migraine therapies, rescue inhalers, anxiolytics used intermittently and medicines prescribed on a "when required" basis.

For these products, sales volume may correlate poorly with treatment duration because utilisation varies considerably between patients. Some individuals may use the medicine daily for prolonged periods, whereas others may use only a few doses each year.

Patient-years derived from average daily dose assumptions should therefore be interpreted cautiously. Depending upon the scientific question, the estimated number of exposed patients or treatment episodes may provide a more appropriate measure than cumulative treatment duration.

Authors should clearly describe any assumptions used to estimate utilisation patterns and acknowledge the uncertainty associated with intermittent therapy.

Fixed-Interval Therapy

Some medicinal products are administered according to predetermined intervals rather than daily dosing schedules.

Examples include annual bisphosphonate infusions, six-monthly monoclonal antibodies, depot antipsychotics, long-acting hormonal preparations and other prolonged-release injectable therapies.

For these medicines, conversion of units sold into patient-years using daily dose assumptions is inappropriate because treatment frequency is defined by the dosing schedule rather than continuous administration.

Exposure estimation should instead consider the recommended dosing interval, expected treatment persistence and the clinical objectives of therapy. In some situations, the number of administered doses or the number of patients receiving scheduled treatment may provide the most informative denominator.

Medical reviewers should confirm that exposure assumptions reflect the approved dosing schedule rather than applying methodologies developed for continuously administered medicines.

Cyclical Therapy

Some medicinal products are administered in repeated treatment cycles separated by defined treatment-free intervals.

Common examples include cytotoxic chemotherapy, immunotherapy, assisted reproduction protocols and certain immunosuppressive regimens.

For these products, exposure estimation requires consideration of both the number of treated patients and the cumulative intensity of treatment. Important measures may include completed treatment cycles, cumulative dose, average number of cycles received or patient-years, depending upon the adverse reaction being evaluated.

For example, acute infusion reactions may be interpreted using the number of treatment cycles, whereas cumulative toxicities such as cardiotoxicity or peripheral neuropathy may be more closely associated with cumulative dose or prolonged exposure.

Consequently, no single exposure metric adequately describes every aspect of cyclical therapy. Authors should select the denominator that best reflects the biological mechanism underlying the safety concern under evaluation.

Medical Review Consideration

Review whether the chosen exposure metric reflects the pharmacology of the medicinal product and the mechanism of the adverse reaction under evaluation rather than simply using the same denominator throughout the report.

Exposure Estimation for Special Treatment Models and Patient Populations

Most medicinal products can be described adequately using the exposure metrics discussed previously. However, certain treatment models and patient populations require additional consideration because standard assumptions regarding dose, treatment duration or utilisation may no longer apply.

The objective is not to create unique exposure calculations for every product. Instead, the author should understand which characteristics of the medicinal product influence exposure estimation and select the methodology that best reflects actual clinical practice.

Weight-Based and Body Surface Area Dosing

Many medicines are prescribed according to body weight (mg/kg) or body surface area (mg/m²) rather than as fixed doses.

Common examples include monoclonal antibodies, chemotherapy, paediatric medicines, enzyme replacement therapies and several biological products.

Exposure estimation for these medicines is considerably more complex than for fixed-dose products because the amount of medicine administered varies between patients. Consequently, simple conversion of units sold into patients treated may produce misleading results.

For example, a patient weighing 40 kg and another weighing 100 kg may receive substantially different quantities of the same medicine while each contributes one treated patient. Likewise, paediatric patients often receive considerably smaller doses than adults.

Where exposure estimates are derived from sales data, authors should consider whether average dose assumptions adequately reflect the treated population. In some situations, indication-specific dosing assumptions or registry-derived utilisation data may provide more reliable estimates.

For adverse reactions related to cumulative exposure, patient-years alone may be insufficient. Cumulative dose or total administered drug quantity may provide additional clinical context.

Vaccines and Other Single-Administration Products

Vaccines illustrate why exposure estimation should follow clinical practice rather than a predetermined mathematical formula.

Many vaccines are administered once or according to predefined schedules consisting of a limited number of doses. Similarly, some medicines, including certain gene therapies and radiopharmaceuticals, may be administered only once during a patient's lifetime.

For these products, the number of vaccinated or treated individuals generally represents the most informative denominator. Patient-years contribute relatively little because treatment duration is negligible despite potentially long-term follow-up.

Some vaccination programmes include primary schedules, booster doses and catch-up programmes. Exposure estimation should therefore distinguish between administered doses and vaccinated individuals where this distinction influences interpretation of safety data.

Authors should also recognise that vaccine distribution data may differ from vaccine administration data. Large vaccination campaigns may involve stockpiling, redistribution between regions or delayed administration, making doses distributed an imperfect surrogate for doses administered.

Paediatric Exposure Estimation

Estimating exposure in paediatric populations presents unique methodological challenges that extend beyond simple dose adjustment.

Children frequently receive weight-based or age-based dosing, undergo rapid changes in body weight during prolonged treatment and may switch between formulations as they grow. Oral suspensions, dispersible tablets, sachets and injectable preparations may all be used for the same medicinal product depending upon age and clinical circumstances.

Consequently, sales data require careful interpretation because identical quantities of active substance may correspond to substantially different numbers of treated patients across different age groups.

Additional complexity arises because many paediatric medicines are used in specialist hospital settings where purchasing patterns may not accurately reflect patient utilisation. Off-label prescribing and extemporaneously prepared formulations may further complicate exposure estimation.

Where paediatric exposure is an important component of the benefit-risk evaluation, patient registries, electronic healthcare records or dedicated utilisation studies frequently provide more reliable estimates than wholesale sales data alone.

Medical Review Consideration

Confirm that exposure estimates adequately reflect the paediatric population under evaluation and are not based solely on adult dosing assumptions.

Medicines with Multiple Indications

Many medicinal products are authorised for several therapeutic indications, each with different dosing schedules, treatment durations and patient populations.

Methotrexate provides a useful illustration. It is used in oncology, rheumatoid arthritis, psoriasis and several other conditions, each with markedly different doses and treatment patterns. Similar complexity exists for corticosteroids, biologics and many targeted therapies.

A single worldwide exposure estimate may therefore conceal important differences between indications.

Where safety concerns relate predominantly to one indication, indication-specific exposure estimates may provide more meaningful interpretation than aggregate worldwide exposure. Authors should consider whether the available data permit stratification by indication and clearly explain any limitations where such stratification is not possible.

Combination Therapy

Some medicines are almost always administered in combination with other treatments.

Examples include antiretroviral therapy, anti-tuberculosis regimens, combination chemotherapy and certain immunosuppressive protocols.

Exposure estimation should focus on the medicinal product being evaluated while recognising that treatment duration, dosing schedules and treatment discontinuation may be influenced by the accompanying regimen.

Where combination therapy substantially influences exposure assumptions or interpretation of safety findings, this relationship should be explained within the aggregate report.

Pregnancy and Rare Diseases

Exposure estimation may be particularly challenging in pregnancy and rare diseases because routine utilisation data are often limited.

Pregnancy exposure may be identified through pregnancy registries, spontaneous reports, observational studies or healthcare databases. Each source captures different aspects of exposure and may substantially underestimate the true number of exposed pregnancies.

Similarly, orphan and ultra-orphan medicines often have relatively small patient populations. Exposure estimates may therefore rely on disease registries, manufacturer patient-support programmes, compassionate use programmes or specialist treatment centres rather than conventional sales data.

For these products, transparency regarding assumptions and data limitations is particularly important because relatively small differences in estimated exposure may significantly influence interpretation of reporting rates.

Writing Tip

Rare diseases do not justify less rigorous exposure estimation. Instead, they require greater transparency regarding the methods used and the uncertainty associated with the resulting estimates.

Reconciling and Validating Exposure Estimates

Exposure estimation rarely involves a single complete data source. Most aggregate reports require information from several internal and external systems, each providing different aspects of product utilisation. Sales data may originate from commercial systems, prescription data from healthcare databases, patient-level information from registries, and additional exposure estimates from licensing partners or affiliated companies.

The objective is not simply to combine these datasets but to produce a scientifically defensible estimate of patient exposure while avoiding omissions, inconsistencies and double counting.

Consequently, exposure estimation should be regarded as a controlled scientific process rather than a mathematical exercise.

Reconciling Multiple Data Sources

Many medicinal products are marketed across numerous countries through different commercial arrangements. Exposure information may therefore be received from affiliates, regional companies, licensing partners, distributors, contract sales organisations or third-party vendors.

Before combining these data, authors should determine precisely what each dataset represents.

Important questions include:

• Does the dataset represent units manufactured, units shipped, units sold or patients treated?
• What geographical regions are included?
• What reporting period has been used?
• Which dosage strengths and formulations are included?
• Have returns, expired stock or replacement stock been excluded?
• Does the dataset overlap with another exposure source?

Only after these questions have been answered should datasets be combined.

Failure to understand the origin of exposure data may result in inaccurate estimates despite technically correct calculations.

Avoiding Double Counting

Double counting is one of the most common causes of inaccurate exposure estimation.

This may occur when exposure from the same country is reported by both a global commercial database and a regional affiliate, when licensing partners provide overlapping utilisation data or when different commercial systems measure different stages of the supply chain.

For example, counting product shipments from a manufacturing site together with sales reported by the receiving affiliate may artificially inflate estimated exposure because the same product has been counted twice.

Similarly, combining registry-derived patient counts with sales-derived estimates without understanding the relationship between the datasets may introduce substantial duplication.

Exposure reconciliation should therefore include clear documentation of the origin and scope of every dataset included in the calculation.

Medical Review Consideration

Where multiple exposure sources have been combined, verify that each contributes unique information and that overlap between datasets has been assessed and documented.

Managing Missing Exposure Data

Complete worldwide exposure information is not always available.

Commercial partners may provide sales data after the reporting deadline, utilisation data may be unavailable for newly launched markets or historical datasets may contain missing periods.

The absence of complete data does not necessarily prevent preparation of an aggregate report. However, authors should document the limitations of the available information and explain any assumptions used to estimate missing exposure.

Examples may include extrapolation from previous reporting periods, use of historical utilisation patterns or temporary reliance on alternative commercial datasets.

Such assumptions should always be scientifically justified, clearly documented and reviewed when complete information becomes available.

Whenever exposure estimates have been derived using imputation or extrapolation, this should be explained transparently within the report where appropriate.

Consistency Between Reporting Periods

Exposure methodology should remain as consistent as possible across successive aggregate reports.

Changes to calculation methods, conversion factors or data sources may create apparent changes in product utilisation that reflect methodological differences rather than genuine changes in exposure.

When improvements to exposure methodology become necessary, authors should evaluate whether historical exposure estimates should also be recalculated to preserve comparability across reporting periods.

Where recalculation is not feasible, the report should explain the reason for the methodological change and discuss its potential impact on interpretation of cumulative safety data.

Consistency of methodology is often as important as precision of the estimate itself.

Documenting Exposure Assumptions

Every exposure estimate is based upon assumptions.

These assumptions may relate to average daily dose, treatment duration, adherence, dosing interval, body weight, treatment persistence or utilisation patterns.

Assumptions should never remain implicit.

Instead, they should be documented in sufficient detail to allow another appropriately qualified reviewer to understand how the estimate was derived and, where necessary, reproduce the calculation independently.

Good documentation also facilitates future aggregate reports because subsequent authors can understand the rationale for previous calculations rather than developing new assumptions for each reporting interval.

Quality Control of Exposure Calculations

Exposure calculations should undergo independent quality review before incorporation into aggregate reports.

Quality review should extend beyond mathematical accuracy.

Reviewers should confirm:

• that the correct data source has been used;
• that all relevant markets have been included;
• that units have been converted appropriately;
• that assumptions are scientifically reasonable;
• that calculations can be reproduced independently;
• that exposure estimates remain consistent with previous reports unless justified otherwise.

Unexpected changes should always be investigated before the report is finalised.

Large increases or decreases in estimated exposure may represent genuine changes in product utilisation. Equally, they may indicate changes in commercial reporting systems, calculation methodology or data quality.

Scientific review should distinguish between these possibilities before the exposure estimate is used for interpretation of safety findings.

Regulatory Review and Inspection Expectations

Regulatory assessors recognise that post-authorisation exposure estimates are rarely exact. The primary expectation is not mathematical perfection but scientific credibility and methodological transparency.

During assessment of a PBRER, reviewers may examine whether the chosen exposure metric is appropriate for the medicinal product, whether the assumptions have been justified and whether the exposure estimate adequately supports interpretation of the cumulative safety data.

During pharmacovigilance inspections, inspectors may additionally review the systems used to obtain exposure data, responsibilities for calculation, quality control procedures, documentation of assumptions and governance surrounding changes in methodology.

The ability to explain how an exposure estimate was derived is often more important than the precision of the numerical estimate itself.

Inspection Insight

Inspectors commonly request evidence supporting exposure calculations, including the underlying commercial datasets, calculation worksheets, documented assumptions and quality review records. Organisations should therefore retain sufficient documentation to reconstruct exposure estimates used within aggregate reports.

Selecting the Appropriate Denominator

One of the most important decisions in exposure estimation is selecting the denominator against which safety observations will be interpreted. There is no single denominator that is appropriate for every medicinal product or every pharmacovigilance activity. The denominator should instead reflect the scientific question being addressed, the pharmacology of the medicinal product and the mechanism of the adverse event under investigation.

Selecting an inappropriate denominator may lead to misleading conclusions despite mathematically correct calculations. For example, expressing acute infusion reactions per patient-year may underestimate the relationship between the event and individual infusions, while expressing cumulative hepatotoxicity per administered dose may fail to account for prolonged exposure. The denominator should therefore describe the aspect of exposure that is most relevant to the biological mechanism of the adverse reaction.

Experienced aggregate physicians often begin by asking a simple question:

What type of exposure is most likely to influence the occurrence of this adverse event?

Only after answering this question should an exposure metric be selected.

Matching the Denominator to the Safety Question

The denominator should represent the exposure most closely associated with the adverse event being evaluated.

Acute events occurring immediately after administration are frequently related to individual doses, injections or infusions.

Adverse reactions associated with prolonged treatment are usually better interpreted using cumulative treatment duration, such as patient-years.

Events related to repeated treatment cycles may require cycle-based exposure estimates, while pregnancy outcomes should generally be interpreted using the number of exposed pregnancies rather than the number of treated women.

The objective is not to identify a universally superior denominator but to select one that best supports scientific interpretation of the available evidence.

This table illustrates principles rather than mandatory regulatory requirements. Alternative exposure measures may be more appropriate depending upon the medicinal product and the scientific objective.

Scientific Principle

The denominator should describe the opportunity for the adverse event to occur. Whenever possible, it should reflect the biological mechanism of the event rather than simply the data that happen to be available.

Interpreting Exposure Estimates

Exposure estimates should always be interpreted within their clinical and regulatory context.

An increase in the number of reported adverse reactions does not necessarily indicate deterioration of the safety profile. Increased utilisation, expansion into new indications, changes in prescribing practices, enhanced pharmacovigilance activities or stimulated reporting may all increase the number of reported cases without altering the intrinsic safety of the medicinal product.

Similarly, declining numbers of reported cases should not automatically be interpreted as improved safety. Commercial withdrawal from major markets, reduced prescribing, supply interruptions or changes in treatment recommendations may all reduce the number of reported cases by decreasing patient exposure.

Exposure estimates should therefore be interpreted alongside other information including changes in utilisation, patient demographics, treatment duration, indication, geographical distribution and regulatory actions occurring during the reporting interval.

The purpose of exposure estimation is to improve scientific interpretation rather than to generate isolated numerical values.

Key Principles for Aggregate Report Authors

Although exposure estimation methods differ between medicinal products, several principles apply consistently across pharmacovigilance activities.

Exposure estimation should always begin with the scientific question rather than the available data.

The selected denominator should reflect the clinical use of the medicinal product and the mechanism of the adverse event being evaluated.

Assumptions should be transparent, scientifically justified and documented sufficiently to allow independent review.

Methodology should remain as consistent as possible between reporting periods, and any changes should be explained clearly.

Exposure estimates should be interpreted together with the characteristics of the exposed population rather than as isolated numerical values.

Finally, exposure estimation is an aid to scientific judgement rather than a substitute for it. The objective is not mathematical precision but a robust and defensible estimate that supports meaningful interpretation of cumulative pharmacovigilance data.