Clinical Supply Label Data: The Fields That Cause Mislabeled Kits

Where Clinical Supply Label Data Comes From

A commercial drug label is written once, approved, and printed many times. A clinical trial label is different. The same study may need a dozen country versions, several kit configurations, blinded and open-label variants, and a new expiry date part way through. Each printed label is the output of a data assembly step, and that step draws on systems owned by different teams who rarely look at the same screen.

Six Sources Feed One Small Piece of Paper

Trace any single label back to its inputs and you will usually find the following sources:

• The protocol. It gives the protocol number, the trial title, the visit schedule, the dosing scheme, and the blinding design. It also tells you whether a visit number belongs on the label and which particulars, if any, are being provided by other means.
• Regulatory label text. In the EU, the sponsor submits the label text as part of the clinical trial application. The European Commission’s Clinical Trials Regulation Q&A says this text should be submitted in a table that follows the structure of Annex VI, and that any particular left off the label must point to the protocol section that explains where it is held or why it is omitted.⁴ That table is, in effect, the approved specification for every printed label in that country.
• Translations. Each country’s text must be translated, reviewed, and approved, and booklet labels add page order and layout on top of the words.
• Packaging and label printing. The packaging site holds the batch record, the label print files, the kit number ranges, and the label reconciliation counts.
• IRT. The IRT system holds the kit list, the randomization or treatment assignment, the country and depot setup, the expiry dates it uses to decide what can be sent and dispensed, and the rules that stop expired or ineligible kits from reaching a participant.
• Stability and release records. The shelf life approved in the investigational medicinal product dossier (IMPD), the Qualified Person (QP) certification, and any later stability data that supports an extension all decide the date that should appear on the kit.

Every one of these sources is owned by someone different. Clinical operations owns the protocol. Regulatory owns the label text. A translation vendor owns the languages. A packaging contractor owns the batch record. An IRT vendor configures the system. Quality and CMC own the stability program. The label is the one place where all of their data meets, and it is printed by the party with the least visibility into the others.

Why Errors Are Harder to Catch Than on Commercial Packs

The European Commission’s detailed guideline on good manufacturing practice (GMP) for investigational medicinal products is direct about this. It says packaging and labeling of investigational products are likely to be more complex, more liable to errors, and harder to check than for authorized products, particularly when blinded products look alike, and that reconciliation, line clearance, and in-process checks should be intensified as a result.⁵

There are three structural reasons. First, blinding means an operator cannot tell active from placebo by looking, so a label applied to the wrong container cannot be spotted at the line. Second, volumes are small and configurations change often, so there is less repetition to build familiarity. Third, much of the label content is variable data (kit numbers, dates, country text) rather than fixed artwork, and variable data errors look correct to anyone who does not hold the source list.

That last point is why this is a data quality problem. A mislabeled kit is rarely the result of a careless operator. It is usually the result of two systems holding different values for the same field, and nobody comparing them before the labels were printed.

What the Record Shows About Label Errors

Publicly documented clinical supply labeling failures are less common than commercial labeling recalls, because most are handled as deviations inside the sponsor and never reach a public database. The record that does exist is consistent, though, and it points to the same fields again and again.

Missing Fields and Inconsistent Formats

In March 2021, FDA announced a joint public meeting with the Reagan-Udall Foundation on potential medication error risks with investigational drug container labels.¹⁴ The Federal Register notice laid out what FDA already knew. It noted that, apart from the required caution statement, much of what appears on a US investigational label (protocol number, strength, dosage form, storage conditions, lot number) is not a regulatory requirement. It also summarized a Canadian study of labels from blinded protocols run by European and American sponsors, which found almost half of the labels missing important information, most often the expiration date, sponsor address, or storage conditions. The same study pointed to small fonts, variable formats for expiration dates and lot numbers, error-prone abbreviations, and highly similar product or protocol identification numbers.¹⁴

That Canadian study, published in Clinical Trials in 2019, reviewed the investigational drug labels for all protocols active from February 14 to 22, 2018 at a 500-bed mother and child hospital, leaving out drugs sourced locally. The authors measured high variability in content and layout, and found that the proportion of free space on a label was correlated with how easily pharmacists could find the kit number.¹⁸ That finding matters for data design: the kit number is the field that links a physical box to its IRT record, and a crowded label makes it harder to read correctly.

The meeting itself, held virtually on May 18 and 19, 2021, drew more than 700 registrants. Site pharmacists and researchers described labels with missing drug names, strength information that was missing or hard to find, small fonts, and unsafe abbreviations.¹⁵

Labels That Slow Down or Mislead the Person Dispensing

A French study known as SIMME-CT ran a simulated dispensing exercise with 63 pharmacy staff at a university hospital investigational drug unit. Labels were scored against ten risk factors related to label information or confusion with another trial. High-risk labels produced more errors and slower responses, with an error rate 5.5 times higher than the low-risk series. Occupation and clinical trial experience did not significantly change the error rate.¹⁹ FDA’s notice summarized the overall error rate in that simulation as roughly 12 percent, with most errors related to dosage unit, trial code, drug confusion, or expiration date.¹⁴ Experience does not protect against a poorly built label. The fix has to be in the data and the design.

The Institute for Safe Medication Practices (ISMP) published a two-part series in 2018 on investigational drug errors. Part I described products identified only by a protocol or study number, strengths printed only beneath a peel-off label, sponsor codes that differ by a single final digit, labels that mix a code number, protocol number, lot number, and kit number with no clear hierarchy, and one label carrying three dates (manufacture, original expiration, and updated expiration). It also described peel-off labels hiding protocol information in up to 15 languages, and bottles that arrived at sites with no label at all even though the outer carton was labeled.¹⁶ Every one of those problems is a decision about which data goes where on the label.

An Expiry Date the Regulator Never Approved

The clearest public enforcement example involving a label field is a UK infringement notice. In July 2024, an infringement notice issued through the MHRA, on behalf of the Secretary of State for Health and Social Care, was served on the UK sponsor Cell Therapy Ltd (trading as Celixir) after a 2020 GCP inspection that found 9 critical and 3 major findings. One breach concerned the expiry date. The QP-signed batch certificates supplied with the product showed a 12-month shelf life, and that shelf life was used to label the product. But the shelf life approved by the MHRA in the dossier was 3 months at the time of labeling and 9 months at the time the product was sent to the site. A pending amendment to extend the stability period to 12 months had not been finalized. The notice concluded that “the expiry dates were beyond those approved by the MHRA.”⁹ The notice also records that the product had technically expired, based on the approved dossier, when a participant was dosed (the participant was later confirmed to have received placebo).⁹

From a data point of view, this is a source-of-truth failure. The date on the label came from a release document, not from the approved regulatory record. Both documents were real and signed. They simply did not agree, and nothing in the process forced a comparison before labels were printed.

When IRT Allocation Goes Wrong

The MHRA’s review of serious breach reports for 2018 included a case where an IRT system did not allocate product as the protocol required. Thirty-two participants received an incorrect dose, most of them overdosed because the system did not down-titrate. The vendor reported the issue as fixed when it was not, more participants were dosed incorrectly, and the sponsor had to unblind the trial as an urgent safety measure.¹⁰ This was a dose assignment error rather than a printing error, but it shows the same pattern: the physical kit was correct, and the data that matched it to a participant was not.

What EU CTR Annex VI Requires

For trials run under Regulation (EU) No 536/2014, the Clinical Trials Regulation (CTR), the label content rules are set out in Chapter X (Articles 66 to 70) and in Annex VI. The European Commission’s GMP guideline for investigational products confirms that labeling must comply with Articles 66 to 69 and that the list of information to appear on the label is in Annex VI.⁵ These requirements replaced the labeling section of the old GMP Annex 13 used under the Clinical Trials Directive.

The Particulars for Unauthorized Products

Article 66 requires the outer and immediate packaging of unauthorized investigational and auxiliary medicinal products to identify the trial contacts, the trial, and the product, and to explain how to use it, in a way that protects participant safety and the reliability of trial data.² Section A.1 of Annex VI lists the particulars for unauthorized investigational products. In plain terms they are:¹

• (a) the name, address, and telephone number of the main contact for information on the product, the trial, and emergency unblinding
• (b) the name and strength or potency of the substance, including comparator or placebo names in blinded trials
• (c) the pharmaceutical form, route of administration, and quantity of dosage units
• (d) the batch or code number identifying the contents and packaging operation
• (e) a clinical trial reference code allowing identification of the trial, site, investigator, and sponsor if not given elsewhere
• (f) the subject identification number and/or treatment number and, where relevant, the visit number
• (g) the name of the investigator, if not included elsewhere
• (h) directions for use, which may refer to a leaflet or other document
• (i) “For clinical trial use only” or similar wording
• (j) the storage conditions
• (k) the “period of use (expiry date or re-test date as applicable)”
• (l) a statement to keep out of reach of children, unless the product is not taken home

Read that list as a data model and you can see why labels go wrong. Items (a), (e), (g), and (i) are trial-level data. Items (b), (c), (d), (j), and (k) are product and batch data. Item (f) is participant-level or kit-level data that may not exist until IRT assigns it. A single label combines three levels of data that live in three different systems.

Immediate Packaging, Outer Packaging, and the 2022 Expiry Change

Annex VI allows a reduced set of particulars on the immediate packaging in two situations: when the immediate and outer packaging are meant to stay together and the outer packaging carries the full set, and when the immediate packaging is a blister pack or small unit such as an ampoule that cannot carry everything.¹

As originally adopted, Annex VI required the period of use on the immediate packaging in every case, which created real difficulty for sponsors who needed to extend shelf life after packaging. Updating an expiry date on a blister card could mean opening the outer pack to reach it. The Commission adopted Delegated Regulation (EU) 2022/2239 on 6 September 2022 to amend Annex VI for unauthorized investigational and auxiliary products.³ Industry commentary at the time reported that the change removes the obligation to put an expiry date on the primary packaging in specific circumstances, and that the adopted text was not changed after the consultation.⁶ A legal note on the proposal described those circumstances as immediate packaging in the form of blister packs or small units such as ampoules, or immediate packaging intended to remain with the outer packaging, with the outer packaging still required to show the date.⁷

EMA’s own statement of the current position is narrow, and it is the one to plan around. In the 2023 disclaimer added to its reflection paper on IRT and expiry dates, EMA states that under the CTR and Delegated Regulation 2022/2239, the expiry date can be omitted only from the inner packaging and only in specific cases set out in Sections A.2.1, A.2.2, B.6.2, B.6.3, and D.9 of Annex VI.³ The outer packaging still carries the date.

What Can Move Off the Label, and What Cannot

Section D of Annex VI allows particulars to be left off the label and made available by other means, such as a centralized electronic randomization system or a centralized information system, provided participant safety and data reliability are not compromised. The omission must be justified in the protocol.¹⁴

Point 9 of Section D then lists the particulars that can never be moved off the label. For the full label set out in A.1, those are the product name and strength (b), the form, route, and quantity (c), the batch or code number (d), the subject or treatment number (f), the storage conditions (j), and the period of use (k).¹ EMA’s reflection paper, written for the old Directive, notes that under the CTR the expiry date cannot be removed from the label regardless of product or packaging type, subject now to the 2022 inner-packaging exceptions.³

This matters for IRT design. A sponsor can put the main contact or investigator details in a system rather than on the label, with justification. A sponsor cannot hold the expiry date only in IRT and print nothing. The Commission’s Q&A goes further on digital approaches, stating that stand-alone digital solutions for primary packaging labels, such as a system relying only on a QR code, are not endorsed.⁴ The same Q&A allows a proportionate exception for deep cryopreserved products, where the trial reference code and subject identification number may be omitted from the immediate packaging if justified and if traceability is assured by other means, such as the secondary packaging or an accompanying letter.⁴

Language Is Set Country by Country

Article 69 of the CTR is short. It says “The language of the information on the label shall be determined by the Member State concerned,” and that the product may be labeled in several languages.² That one sentence is why a multinational study needs a country-to-language map as controlled data.

The Commission’s Q&A includes a table of national positions, and the footnotes show how much they vary. Belgium accepts labels in one language (a national language or English) if the product is administered by the physician rather than given to the patient. Finland may require Swedish where Swedish-speaking participants are recruited, and accepts English if trial staff administer the product at the site. Germany allows English labeling where a physician (or a dentist, for dental trials) on the investigating team administers the product directly.⁴ Each of those rules depends on a protocol fact (who administers the product, and where) and on a site fact (which languages participants speak). A label language decision is therefore a join between protocol data, site data, and national rules. If any of the three changes, the label may need to change.

How the US, Canada, and the UK Differ

A global study must meet the strictest applicable rule for each field in each country. The table below compares the main points for four regulators. It is a summary for planning, not a substitute for the texts themselves.

Two points stand out. The United States sets very little by regulation, which is why US sites rely on sponsor practice and on professional guidance. Health Canada, by contrast, requires the expiry date, storage conditions, lot number, sponsor name and address, and protocol code, all in both English and French.¹³ A label template built for a US-only study will not carry over to Canada or the EU without new fields and new language text.

The UK position changed recently. The MHRA’s labeling guidance, first published in June 2025 and last updated on April 28, 2026, applies now that the amended UK clinical trials regulations have taken full effect. It says the expiry date should be on the primary packaging and should not be available only through an IRT system, that a month and year format is usually enough (with day, month, and year for very short shelf lives), and that a retest date is not acceptable in place of an expiry date.⁸ That last point is a real difference from Annex VI, which allows either an expiry or a retest date as applicable. A study running in both the EU and the UK needs a single date rule that satisfies both.

The Fields That Cause Mislabeled Kits

Not every field on a label carries the same risk. The ones below are where mislabeled and wrong-language kits come from in practice. For each, the useful questions are the same: which system owns the value, which systems copy it, and where is the comparison done before printing?

Protocol Number and Trial Reference Code

The protocol number looks like the most stable field on the label. In practice it changes more than people expect. Sponsors renumber protocols when a study moves from one program to another, when an extension study is created, or when a regional sub-study is added. The trial reference code on the label may also be a different string from the protocol number used in IRT, the EU trial number, or the code the packaging site uses for its own work order.

Common failures include labels printed with the parent study number for an extension study, labels for one protocol packed with kits allocated in IRT to another protocol that uses the same product, and look-alike codes that differ by a single character. ISMP’s examples of codes differing only in the last digit apply here as much as to product names.¹⁶

The control is simple to state: hold one authoritative trial identifier, map every other code to it in a controlled crosswalk, and print from the crosswalk rather than from a copy typed into a label request form.

Kit Numbers, Treatment Numbers, and Randomization

The kit number is the field that links the physical world to IRT. In a blinded study, the packaging site receives a kit list (or a medication number list) that maps each kit number to a treatment, generated from the randomization or packaging schedule. The labels print the kit numbers. IRT receives the same list and uses it to assign kits to participants without revealing the treatment.

That means the same list exists in at least three places: the system that generated it, the packaging site’s label print run, and IRT. If any one copy differs (a range offset by one, a block printed twice, a set of numbers reused from an earlier campaign, a list regenerated after a protocol change), the label will be correct according to one system and wrong according to another. Because the study is blinded, no one at the site can tell.

The GMP guideline asks that blinded products remain identifiable, including batch numbers before blinding, and that rapid identification be possible in an emergency.⁵ That expectation can only be met if the kit-to-treatment mapping is the same in every system that holds it.

Expiry and Retest Dates

The date field causes the most trouble, for four reasons:

• Multiple sources. The approved shelf life lives in the IMPD. The calculated date lives in the batch record and the QP certificate. The operating date lives in IRT. The printed date lives on the label. The Celixir case shows what happens when the certificate and the dossier disagree.⁹
• Formats. Labels for global studies must be read by people used to different date conventions. The Canadian label study and FDA’s summary both flag variable date formats as an error risk.¹⁴ The MHRA guidance accepts month and year for most products.⁸ A label that prints 03/04/2027 means two different things in two countries.
• Expiry versus retest. Annex VI accepts either an expiry date or a retest date as applicable, while the MHRA does not accept a retest date in place of an expiry date.¹⁸ A label template must know which type of date it is printing, and for which country.
• Change over time. Dates are extended as stability data accumulates, which means the printed value goes out of date while the kit is still in use. The relabeling section below covers that in detail.

Storage Conditions

Storage conditions are a non-omittable particular on the full EU label and a required element in Canada.¹¹³ They are also one of the fields the Canadian study found missing.¹⁴ Storage text causes errors in two ways. The first is simple mismatch: the label says 2 to 8 degrees Celsius while the IRT temperature excursion rules or the pharmacy manual use a different range, or the comparator was repackaged and the original storage statement no longer fits the new container. The second is translation: a storage statement that means “do not freeze” in one language can be rendered loosely in another, and a booklet page may be updated in one language but not the others.

Storage statements belong in a controlled phrase library with an approved translation for each language, keyed to a storage condition code that IRT and the depot system also use. When the code changes, every language and every system changes together.

Country-Specific Statements and Language Text

Country-specific text includes national contact details, any statements a Member State requires, the investigational use statement in the local language, and in some countries the name of the investigator or the site. This is where wrong-language kits come from.

A wrong-language kit is almost never a printing error. It is a supply allocation error: a kit labeled for one country is sent to a site in another. That can happen when depots serve several countries, when a country is added after packaging, when a site moves to a different depot, or when an IRT country setting does not match the label variant that was printed. The national rules in the Commission’s Q&A make it more complex, since whether English is acceptable can depend on who administers the product at that site.⁴

Product Name and Strength in Blinded Trials

In a blinded trial, Annex VI requires the label to name the comparator or placebo as well as the investigational product.¹ The GMP guideline adds that, where products are blinded, the expiry date stated should be that of the shortest-dated product so that the blinding is maintained.⁵ Both rules mean that a blinded label cannot be built from a single product master record. It needs data from every arm.

For open-label studies with several strengths, ISMP’s findings about strengths printed only under a peel-off label, or not printed prominently, are the main hazard.¹⁶ ISMP’s companion article recommends making the drug name and strength the most prominent elements on the container and using a font size of at least 8 points.¹⁷

The table below pulls these fields together into a single map that a clinical supply lead can use to assign owners.

Booklet Labels: Where Multilingual Data Breaks

Booklet labels are how most sponsors fit many languages onto a small container. An ISPE task team white paper from 2012 described them as current practice for multinational trials and noted that “Booklet labels have been established and used in clinical trials for more than 20 years.”²⁰ The format solves a real problem, since it lets one packaged kit serve several countries and gives a supply team flexibility to move kits where they are needed. It also concentrates a great deal of data into one physical item.

How a Booklet Is Built

The ISPE paper describes the parts of a booklet label. The base label is the part applied to the primary or secondary packaging. The cover page is the front page of the booklet. The inner pages hold a table of contents and the country-specific translated pages in the official local languages. Some designs add a removable panel, sometimes used for site documentation.²⁰

The base label typically carries the variable data: kit number, batch number, use-by date, and the fields filled in at the site such as subject number and visit. The inner pages carry the fixed translated text. The cover page tells the reader where to find their language. Each part comes from a different source, and they are joined at printing.

The Failure Modes

The ISPE paper ran a survey of investigational sites, with responses from more than 250 sites around the world. It reported that “78% of respondents indicated that the information contained within a booklet label is hard to read (print/font too small).”²⁰ The paper’s failure scenarios include a participant who cannot find their language because nothing shows where the label text is, booklets too bulky to handle, labels torn off when opened, and labels damaged in transport so that product information can no longer be read.²⁰

From a data point of view, booklets fail in these ways:

• A language page is out of step. One country’s approved text changes after an amendment, and its page is updated while the other pages keep the old version. The booklet as a whole now carries two versions of the same particulars.
• A language is missing. A country is added after packaging, and the kits already in the depot do not carry that country’s page. If IRT does not know which booklet version each kit carries, those kits can be allocated to the new country.
• The base label and the inner pages disagree. The base label prints the date in one format while the translated pages describe it in another, or the base label carries a storage condition that differs from the translated text.
• Page order changes. A new language is inserted in the middle of the booklet, which shifts the table of contents and the page each site has been trained to use.
• Too many pages. Duhamel and colleagues concluded that the labels in their study often lacked key information even when they ran to many pages.¹⁸ More pages do not guarantee more complete content, and every added page is another component to keep in step.

Managing Booklet Text as Versioned Data

The fix is to stop treating a booklet as a print file and start treating it as an assembly of versioned components. Each language page has its own version and approval status. Each booklet version is a defined list of page versions in a defined order. Each packaging campaign records which booklet version it used. IRT holds the booklet version for each kit, or at least for each packaging lot, so that it can refuse to allocate a kit to a country whose page is missing or out of date.

The ISPE paper recommended that sponsors instruct site staff on completing the empty fields on the label, and that evidence of label review and training be kept.²⁰ The same applies to the data behind the booklet. If no one can say which version of the Polish page is on the kits sitting in a given depot today, the booklet is not under control.

Expiry Date Extensions and Relabeling

Most long-running studies extend shelf life at least once. An extension is good news for supply, since it means less waste and fewer emergency packaging runs. It is also the single most error-prone data change in clinical supply, because the new date has to reach the dossier, the release records, IRT, and the physical kits, in that order, with evidence at each step. SD’s earlier article on supply forecasting discussed expiry extension as a way to reduce overage. This section is about the data mechanics of doing it safely.

The Chain of Approvals Before a New Date Exists

EMA’s reflection paper on IRT and expiry dates sets out the sponsor-side process. The update should only be made after review of stability data and extension of the expiry date by the QP. The change should be communicated through a revised certificate of analysis or certificate of conformance that includes the expiry date, and the extension must have been approved through an amendment to the clinical trial application unless the extension plan was already authorized in the IMPD.³

The paper carries a disclaimer that it was written for trials under the Clinical Trials Directive and has not been revised for the CTR.³ Its description of the data flow is still the clearest regulator account available, and the sequence it describes (stability data, then QP decision, then regulatory approval, then system update) is the one the Celixir notice shows being skipped.⁹

Updating IRT

This is where the reflection paper is most specific, and most useful for system design:³

• There should be a defined process between the sponsor and the IRT provider so that the new date is communicated with enough time to be implemented and verified. The paper says that “A regular email is generally not sufficient for this purpose,” and notes that inspectors have seen expiry updates sent by clinical project managers. It expects staff involved in manufacturing, for example the QP, to be part of the process.
• The provider should know who is authorized to request an expiry update and how.
• Where the sponsor updates dates through a web interface, access should be limited to people with authority to change expiry dates.
• Where a drug supply system feeds expiry data to IRT, that interface should be validated so that any change in the supply system is mirrored in IRT.
• Changes should have an audit trail. For critical updates such as expiry, a second person should verify that the correct data was entered and released to the live environment, and kit-level changes should also be second-person verified.
• The provider should confirm completion, and the sponsor should have documented confirmation in the trial master file that the update was done in an appropriate timeframe.

The same paper lists date-related functions an IRT system should support: dates after which product should not be sent from the manufacturer or warehouse to sites, dates after which it should not be dispensed (taking treatment length into account), and offsets for the time it takes to reach countries further away.³ These are all derived from the expiry date. When the expiry date changes, every derived date changes with it. A sponsor that updates the expiry date but not the derived dates has only done half the change.

Putting the New Date on the Kit

Once the new date is approved and loaded, the physical kits must be relabeled, or at least the outer packaging must carry the new date. The Commission’s GMP guideline sets the content and conditions:⁵

• An additional label is applied. It should state the new expiry date and repeat the batch number and clinical trial reference number.
• It may be placed over the old expiry date, but for quality control reasons not over the original batch number.
• The work should be done by trained staff under GMP and specific procedures, checked by a second person, and documented in the batch records.
• It should be done in a partitioned or separated area, with line clearance at the start and end and a label reconciliation. Any discrepancies should be investigated and accounted for before release.
• Relabeling may be done by authorized personnel at a hospital, health center, or clinic that meets Article 61(5)(a) of the CTR, which does not require a manufacturing authorization.

The MHRA’s guidance for the UK is closely aligned. A label that extends or amends the expiry must show the new date and repeat the batch number, and may cover the old expiry date but not the original batch number. Where possible the work should be done at a site holding a manufacturing authorization for investigational products. Where it is not, the sponsor needs a formal procedure, documented training and delegation, and an independent second check. The MHRA does not require its own approval for this post-certification labeling, but it must be documented in trial and batch records.⁸

ISMP’s sponsor recommendations add the US site view: relabeling for extended expiration dates should be the sponsor’s responsibility and should happen before the product is dispensed, and if expiration dates are not on the container label, the packing slip with each delivery must list them.¹⁷ ISMP’s example of a single label showing three different dates is what happens when relabeling is layered without a clear rule for what the reader should trust.¹⁶

Blinded Products and the Shortest-Dated Component

For blinded studies, the GMP guideline’s rule that the stated expiry should be that of the shortest-dated product means an extension is only usable when every arm’s shelf life supports it.⁵ If the active product’s stability data supports a 24-month date and the placebo or comparator only supports 18 months, the label stays at the shorter date. IRT must apply the same rule, or it will treat kits as usable that the label says have expired. That is a direct conflict between the printed record and the electronic one, and in a blinded study the person dispensing cannot resolve it.

Data Controls Between IRT and Labeling

The fields above share a pattern. The label and IRT each hold a copy of the same facts, and the errors come from the copies drifting apart. The controls below are about keeping them together. None requires a new system. Each requires an agreed data owner and a check that is run and recorded before labels are printed or kits are released.

Agree on Shared Identifiers

Write down, at study start, the identifiers that every system will use. That means the trial identifier, the product or treatment codes, the kit number format and range, the packaging lot or label group identifier, and the label variant or booklet version code. Agree who owns each one. Put the list into the IRT user requirements and the packaging specification so that both vendors work from it.

EMA’s reflection paper expects a sponsor-approved user requirements specification for IRT, with validation documents mapped back to it down to individual test scripts.³ The shared identifier list is part of those requirements. If it is not written down, each vendor will make its own choices, and the crosswalk will be built after the first mismatch.

Reconcile the Kit List Before Release

Before kits from a packaging campaign are released for use, compare three lists: the list the randomization or packaging schedule generated, the list the packaging site printed and applied (from the batch record and label reconciliation), and the list loaded into IRT. Compare counts, the first and last kit numbers in each block, any numbers voided or destroyed during packaging, and the treatment mapping for a sample of kits (done by an unblinded person where the study is blinded).

The GMP guideline already requires label reconciliation and investigation of any discrepancy before release.⁵ Extending that reconciliation to include the IRT load is a small step that closes the most common kit number error: a list that is correct in the packaging record and wrong in IRT.

Make Country Eligibility a Data Rule

A wrong-language kit should be impossible to allocate, not just unlikely. That means IRT must know, for each kit or packaging lot, which label variant or booklet version it carries, and must hold a table of which variants are approved for which countries. When a site requests resupply, IRT should only offer kits whose variant is approved for that site’s country.

The table is not static. A country can be added, a national text can change after an amendment, or a national rule can depend on who administers the product, as the Commission’s Q&A footnotes show for Belgium, Finland, and Germany.⁴ Each change must update the eligibility table, and the update should go through the same change control as any other IRT configuration change.

Treat Dates as Controlled Master Data

The expiry date for a batch should exist in one authoritative place, with every other copy derived from it. In most organizations that place is the release record, once the QP has certified the batch against the approved shelf life. The control that would have caught the Celixir problem is a check, at release, that the calculated expiry does not go beyond the shelf life approved in the current dossier for that batch.⁹

From there, IRT’s expiry dates and derived dates should be loaded through a defined interface or a controlled, second-person verified manual entry, as EMA’s reflection paper describes.³ The label’s printed date should be checked against the same source. The MHRA’s statement that the expiry date should not be available only through IRT means the label and IRT must both carry it, which in turn means both must be right.⁸

Run Protocol Amendments Through Both Systems

Many label errors start with a protocol amendment. A new visit, a new dose level, a new country, or a change to who administers the product can change label content, kit configuration, and IRT setup at the same time. EMA’s reflection paper says that when the protocol changes, the sponsor should assess the impact on IRT, that changes do not always appear to affect supplies at first, and that it is good practice to include clinical supplies staff in those discussions.³

The practical step is to add two questions to every amendment impact assessment. Does this change any label particular or any country’s label text? Does this change any IRT setting that depends on label content (kit types, country eligibility, dates, dose assignment)? If the answer to either is yes, the label change and the IRT change go through change control together and are released together.

Test the Edges in UAT

EMA’s guideline on computerized systems and electronic data in clinical trials asks sponsors to verify IRT dose look-up tables against the approved protocol, including test data on the borderline between doses, and to test every combination of strata in stratified randomization.²² The reflection paper expects sponsors to write their own UAT scripts covering both sponsor and investigator activities.³ The MHRA’s 2018 down-titration case shows what happens when IRT dose allocation does not match the protocol.¹⁰

For label-related functions, the edges to test are predictable: a kit whose expiry falls inside the dispense window but outside the visit window; a kit whose label variant is not approved for the requesting country; a kit from a packaging lot that has not yet been reconciled; an expiry update applied to one batch but not another in the same lot; and a blinded kit where the shortest-dated arm has not been extended. Each should be a written test case with an expected result.

A Label Data Review for the Next Study

The sequence below is a way to put these controls into a study that is about to start packaging. It can also be run on a study already in progress, starting at step three.

None of these steps is new to a mature clinical supply organization. What is often missing is the connection between them: the label team runs its reconciliation, the IRT vendor runs its UAT, the regulatory team approves the text, and no one compares the three. The review sequence is mostly about making that comparison someone’s job.

Conclusion

Mislabeled and wrong-language kits rarely come from the printing line. They come from clinical supply label data held in several systems that drift apart: an expiry date on a certificate that the dossier never approved, a kit range in IRT that is one block off from the printed range, a booklet page that was updated for one country but not another, or a depot that sends a kit to a country its label was never approved for. The rules in EU CTR Annex VI, the GMP guideline for investigational products, and the MHRA’s current guidance all assume the sponsor can keep these copies in step. The 2022 amendment gave sponsors a little more room on inner packaging, but it did not move the expiry date, the batch number, the storage conditions, or the subject number off the label. The paper label and IRT must both be right, and must agree.

Sakara Digital works with pharma and biotech organizations on the data and systems behind regulated operations, including the hand-offs between IRT, packaging, and regulatory records that this article describes. If you are planning a multinational study, preparing for an expiry extension, or reviewing how your IRT vendor and packaging contractor share data, and you want an independent view on where the gaps are, we are happy to have that conversation.

For Further Reading