Global value dossiers, explained.

A global value dossier explained: one value story, written once, adapted by every market

One dossier's path from trial results to payers in seven countries

1. The evidence is planned and gatheredReviews, burden studies and models are commissioned, starting about two years before launch in one orphan-drug timeline.3 The Talvoren library holds 142 sources.
2. Value messages are agreedA map of 15 messages under five pillars: burden, efficacy, safety and convenience, comparison, economics.
3. The GVD is written, checked and signedEight sections and 390 claims. This is where "offsets part of its cost" gets traced, or does not.
4. Affiliates adapt itUK appraisal, German benefit dossier, US payer dossier and four more, across 2027 and into 2028.
5. Payers ask, and new data arrivesEach challenge and each new data cut (a study's next report) sends the team back to the claim. One consultancy calls these living documents, updated as data and needs change.1

Step 3 is the subject of this guide. The people, products and figures in this example are fictional.

Why no law governs a GVD, and three more surprises about value dossiers

GVD, value message, HTA and PICO in one line each

Fourteen terms are defined in the one-page summary.

What goes in a global value dossier template: a value story, six evidence sections and a message map

In shortThere is no official template, but most GVDs follow the same order: summary, value story, disease and burden, current treatment and unmet need, product, clinical value, comparisons and economic value. A value story and its message map set a GVD apart from a reimbursement dossier, which gives more detail and argues less.

Trinity Life Sciences lists the usual structure: an executive summary; a value story of messages, sometimes hyperlinked to their support; background and disease burden; current treatment and unmet needs; product information; clinical value, including the trial program, efficacy and safety, and direct, indirect or subgroup analyses; and economic value, meaning cost-effectiveness and budget impact (what the medicine adds to a payer's spending).1 Company templates vary. The fictional Talvoren dossier in this guide uses eight sections, with the comparison broken out on its own because that is where payers press hardest.

Three documents people confuse: GVD, reimbursement (HTA) dossier and AMCP dossier

The names overlap, and many companies write something between a value dossier and a reimbursement dossier.1 A team comparing GVD and AMCP work found both need disease burden, clinical evidence, economics, guidelines and HTA decisions. The AMCP version wants US data and full evidence tables in a set format with limited interpretation, while the GVD takes a global view around a value story and a flexible layout.5

An AMCP dossier is the payer value dossier US health plans ask for, in the Format set by the Academy of Managed Care Pharmacy. Version 5.0, in 2024, pushed toward brevity, visuals, health disparities and early exchange before approval.6 Version 6.0 followed in October 2026, adding guidance for cell and gene therapies and biosimilars.7

The value message map behind a dossier's arguments

Before anyone drafts a section, the team agrees on the messages. A typical map groups them under pillars, and each message lists the sources meant to support it. A well-kept map answers the first question any reviewer asks: which study says that? The Talvoren map has 15 messages under five pillars, including these three. Each one looks routine, and each one raises a different evidence question.

How each market's reimbursement test changes what a value dossier must prove

In shortEvery market asks a different question. Germany asks whether a medicine adds benefit over a named comparator. The UK asks for cost per quality-adjusted life year. France rates the benefit and uses the rating in pricing. US payers weigh clinical evidence and budget impact. The GVD has to hold evidence for all of them, or the gap appears later, in one country.

This is the reason a GVD works best as a menu. A UK appraisal needs utilities and a cost-effectiveness model. Germany needs the comparison the G-BA names, in the population it names, and does not take a cost argument at that step. A US plan wants its budget per member. The dossier lead rarely knows at drafting time which question will be asked first, so each value claim needs evidence that can survive whichever reader meets it.

The rules around a value dossier in 2026: the EU joint assessment, FDA's payer draft and NICE's new range

In shortA GVD has no rulebook of its own, so its writers answer to the rules of whatever it feeds. Four texts changed the job between 2025 and 2026: the EU joint clinical assessment took effect, FDA issued a new draft on payer communications, NICE raised its range, and the EU's HTA group wrote principles for AI in joint dossiers.

Rating value claims against evidence: why the study design and the kind of claim both count

In shortAn evidence matrix lists value claims down the side and sources across the top, and each cell records what that source shows for that claim. Study design sets how much a cell can bear. The kind of claim sets how much it needs. Those two together show which claims are thin, with less evidence than their kind needs, and those are the ones a payer will find.

When a message map and its references sit in separate files, nobody sees claim strength at a glance. An evidence matrix puts them on one page. The columns group sources by kind: clinical trials, reviews, real-world studies, guidance and models. The rows are the value messages. A filled cell means that source says something about that claim, and a shade says how much weight it can carry.

How much weight each study design can carry in a value claim

This grid is a working convention, drawn from common appraisal practice. No regulator publishes it, and your own company's grid should be written down, agreed and versioned.

The cost-offset claim as a chain of three links

Take the message that Talvoren offsets part of its cost through fewer flare-related emergency visits. It needs three things to be true, and each comes from a different kind of source. Two are measured. The third is assumed.

A team can write a rule for this: a cost-offset claim needs at least two real-world studies, or the claim is thin and goes to the dossier lead. That rule is the fictional team's own, and a sensible one. No regulator requires it. Write the rule down before drafting, so that nobody has to argue about it the week a payer asks.

Checking a value claim's numbers: testing hierarchies, credible intervals, data cuts and model outputs

In shortFour checks cover the errors this guide describes. Was the timepoint part of the trial's testing plan? Does the interval on a comparison cross 1? Is this the latest data cut? Is the number a result or a model's assumption? Each one is worked through below on the fictional example.

The testing hierarchy: why week 2 and week 4 are not equal

A trial that tests many endpoints (the outcomes it measures) at many timepoints will produce some positive results by chance. So the statistical plan fixes an order in advance, and only results tested within that order count as confirmed. FDA's guidance on multiple endpoints and ICH E9, the international statistical guideline for trials, both deal with this risk.25,26

A result outside the hierarchy carries a "nominal" p value. The p value is the usual measure of statistical significance, and a nominal one describes the data and falls short of proof. FDA's guidance applies the same idea to labeling: a descriptive analysis outside the prespecified testing strategy should not be presented in ways that imply a statistically rigorous conclusion, and an analysis meant to support a definitive conclusion belongs in that strategy.25

The Talvoren draft said patients felt meaningful skin-pain relief "within 2 weeks of the first dose". Skin-pain relief here means NRS30: at least a 30% drop on a zero-to-ten pain scale. The week-2 result was 24.1% against 17.9% with a nominal p value of 0.08, outside the hierarchy. The first tested timepoint was week 4: 38.6% against 21.0%, p below 0.001.

The honest sentence says "from week 4" and cites the table. If the team wants to mention week 2, it says the result was exploratory.

Reading an indirect comparison: what an interval crossing 1 does and does not show

With no head-to-head trial, a network meta-analysis compares treatments through a shared comparator, usually placebo. The result is often an odds ratio with a 95% credible interval, the range that holds the true value with 95% probability given the model. An odds ratio of 1 means no difference. When the interval spans 1, the data do not settle which treatment is better. The example compares Talvoren with anti-TNF and anti-IL-17A therapy, two classes of biologic already used in HS.

Against anti-TNF therapy the odds ratio is 1.21, with an interval of 0.88 to 1.67. In plain words, the odds of response could be 12% lower or 67% higher. "More effective" overstates that.

The reverse also needs care: an interval that crosses 1 does not prove two treatments are the same. Altman and Bland made the point in 1995: absence of evidence is not evidence of absence.27 "No conclusive difference" says exactly what the data show. A word like "comparable" is a stronger statement, and it needs either an agreed margin or a judgment your medical reviewer is ready to defend.

A phrase such as "numerically higher, a difference that was not conclusive" gives the direction and its limit in one line, and it too needs your reviewer's agreement.

Data cuts: the same study can say two things

Long studies report in stages, and each report takes its data up to a cut-off date. The Talvoren draft said 74.0% of week-16 responders kept their response, citing the week-36 cut of the extension study, which keeps following trial patients, dated December 15, 2025. The week-52 report, with data to June 30, 2026, arrived on October 7 and says 71.3% of 512 patients.

The older figure is true of its report. It is wrong for a dossier dated after the new one arrives. The FDA 2026 draft treats outdated information as something firms must correct when it was shared before approval under 502(gg).19

Economic results: arithmetic a reader can do, and what the words must admit

Six ways a value claim drifts from its evidence, and why careful writers still let them through

In shortWording errors in value dossiers are ordinary, because they are ordinary in all medical writing. In one meta-analysis of medical journals, about one quotation in four carried an error. No one publishes the rate for GVDs, so the way to learn yours is to re-check a sample.

Spin: favorable wording around a neutral result

The published record shows how natural this is. Boutron and colleagues read 72 randomized trials whose primary outcome was not significant, and found wording that suggested benefit anyway in the conclusions of 58.3% of abstracts.30 Authors of those papers had peer review behind them. A value dossier is written by a company for a purpose, so a reviewer should expect at least the same pull.

What a thin value claim risks: no added benefit in Germany, rejected comparisons and lost price

In shortA weak GVD claim rarely causes a rejection on its own, but the evidence gap behind it does. Germany gives the clearest public data: no added benefit in 64% of assessments since 2020, indirect comparisons accepted in 22.5% of cases, and a deeper rebate when no added benefit is found. These are abstracts from consultancies and authors in the field, so read them as direction.

The authors of the 15-year German review add a point worth a dossier lead's attention. Where the comparator was a therapy individualized to each patient, they suggest that most "no added benefit" results reflected evidence that could not be generated or accepted, rather than a medicine shown to have no effect: accepted evidence failed to show sufficient effects in only 14% of those cases.31 That is a planning problem, and a value dossier is where it first becomes visible.

A separate analysis found launch-price rebates averaged 24% for drugs with an added benefit and 33% for those without.34

The joint European dossier raises the cost of a late evidence gap

Under the EU rules, the dossier is due 100 days from the first request, or 60 in accelerated and new-indication cases.22 Those are too short to generate evidence, only to present it. When 24 scopes were simulated, the worst case reached 88 PICOs, and the minimum analyses ran from 175 to 720 depending on what was feasible.33

Trinity's advice is to start global dossiers as early as possible, and to expect the clinical sections to need substantial reworking.1 A GVD whose comparative claims already point to their analyses shortens that work. One that has only prose has to rebuild it.

Credibility, the risk with no number

When a UK lead asks what a line is based on and the answer takes a week, or when a payer finds a superseded number, the next message gets read with less trust. Your own log of market questions is the best measure you will get: how many, on which claims, and how long each answer took.

Why GVD volume and pressure are rising: more launches, slower access and joint European assessment

In shortNobody publishes a count of GVDs written a year or the size of the writing market. Three public trends set the pressure: dozens of new medicines a year on each side of the Atlantic, access in Europe that is slowing, and a joint assessment that grows in 2028 and 2030.

The same survey shows the average rate of availability falling from 54% in 2019 to 44%, and the authors point to differences in national pricing, reimbursement and HTA processes as a reason for the spread between countries.37

Delay has a price, and the Tufts Center for the Study of Drug Development has published two figures for it from one dataset of 645 drugs: about $500,000 a day in lost sales in its 2024 journal article and commentary,38 and about $800,000 in its August 2024 white paper.39 Both give $40,000 a day in direct Phase II and III trial cost.

These are averages for delays in drug development across marketed drugs. They are a rough guide to what a day means, and they need care before being applied to a day lost in one country's review.

Three shifts a value team should plan around

How affiliates adapt a global value dossier, and why one weak claim reaches every market at once

In shortAffiliates reuse sections and rewrite the words. Which sections each one needs depends on its submission. A weak claim therefore arrives in several documents at different times, under different wording, so it is hard to pull back. A good global team knows which markets depend on each claim.

The Talvoren dossier feeds seven market documents, from the UK appraisal in April 2027 to the Japanese price listing in January 2028. None of them copies the GVD as it stands. The study comparing GVD and AMCP work is direct: content must be tailored to each audience and purpose, and planning before work begins is where the efficiency lies.5 What the markets share is the claim and the source. That is what a global team has to keep straight.

Economics arrives last, and the joint dossier skips it

The sections also run on different clocks. Early clinical and background sections can start before the study reports are final. The full clinical value section often has to wait for them. Because a joint European dossier leaves economics to each country, the economic section is often written after that dossier goes in.1 So a GVD is usually several versions at once, and each market may hold a different one. When a section changes, the affiliates working from the older one have to hear about it.

What AI changes in GVD development, what HTA bodies expect of it and where the evidence stops

In shortPublished figures are promising and thin: conference abstracts, often from firms selling the work. HTA bodies now say how they want AI used: a human accountable, each AI step disclosed, and in the EU the prompts kept. We found no regulator that bars AI from drafting a value dossier. Value messaging itself still needs experts.

What HTA bodies and regulators say about AI in evidence work

The work AI handles well in a GVD, and the work that stays with experts

Finding the passage, carrying a number across exactly, spotting a citation to an older report: these are matching tasks, and they suit software. Deciding what a dossier should claim, how to phrase a comparison, and whether a thin claim stays or goes are judgments that carry the company's name. The two dossier abstracts above keep experts on those steps, and so do the European principles. A useful test of any tool is whether each output points to a passage a person can open in one click.

How market access teams write a GVD today: who does what, and where the section days go

In shortA global value and access lead owns the dossier, a HEOR scientist owns the evidence and models, medical writers draft, and medical and legal reviewers sign. Agencies may do much of the writing. In the fictional team's estimate most of the days go to finding passages and writing cited sentences, and no public benchmark says otherwise.

From the value and access lead to the affiliate: who handles a dossier

The steps and hand-offs from study reports to a signed GVD section

Where nine working days on one GVD section go

Four ways companies get value dossier writing done

None of these is free of review. Whichever way you staff it, a named person still signs each section. The next part shows one product doing this work on a fictional dossier.

The Talvoren evidence matrix: a week-52 report arrives, one claim turns thin, and the UK gets a cited answer

In shortThis is the example from the top of the page, inside the product. On October 7, 2026 the LUMEN-X week-52 report arrives. Dana Okafor, the global value and access lead, re-maps the evidence. Agents read the report, retire the older data cut, rate 31 cells and apply the claim rules. One claim fails, Dana opens what it rests on, and she answers the UK's question that morning.

1

Every value claim against every source, on Dana's home screen

Dana's first screen is the evidence matrix. 15 value claims under five pillars run down the side. Across the top are the 15 sources cited in version 0.6: clinical trials, reviews, real-world studies, guidance and models. There are 31 supported cells, and a number in a cell counts the passages behind it.

The colors show design: green for a controlled trial or the label, blue for an extension, review, indirect comparison or real-world study, amber for a model. The Strength column gives each claim's overall rating, and Traced counts the sentences, tables and figures built on it. Burden claims earn "Strong" on epidemiology and reviews, where no trial applies.

2

"Re-map the evidence" after the week-52 report

The report reached the library this morning. Dana clicks once. The Evidence Librarian reads it and marks the week-36 interim report of the same study superseded. The Evidence Matrix Mapper rates the 31 claim-source cells by study design, column by column. The Claim & Label Checker applies three claim rules to the 15 claims.

A source is only marked superseded when a newer report of the same study exists. Each step stays in view while it runs.

3

The cost-offset claim turns amber

One claim fails. M5.3, "Offsets part of its cost through fewer flare-related emergency visits", has one clinical trial, one real-world study and one model behind it. The team's rule asks for at least two real-world studies, so the row turns amber, the claim goes to Dana and gap G-1 is updated.

The panel beside it spells out the reason: one real-world study found; the trial shows fewer flares; the model assumes fewer visits. It also lists what rests on the claim in the dossier: two sentences and one table.

4

What the one real-world study measures

Dana opens the US claims cell. The passage is Table 3 of the burden study: 29% of patients had at least one HS-related emergency department visit a year. The panel says it plainly: this measures how often patients need emergency care, and shows no fall in visits with Talvoren.

The cell also lists what is built on this source: a sentence in section 3 and Table 3-1. One more click opens the study itself at that passage.

5

A cited answer to the UK, drafted from the row

Hannah Price, the UK lead, had asked on October 6: "Section 8 says Talvoren offsets part of its cost through fewer emergency visits. What is that based on?" Section 8 has not been drafted at this point, so her question is about the message planned for it on the value message map.

Dana drafts the answer from the row. It cites the US claims study, the LUMEN-1 flare result (24.3% to 9.6%) and the cost-effectiveness model, says surgery is not modeled, and presents the offset as a model assumption until a second real-world study is available. She sends it, and the record keeps it.

Inside the dossier: the section board, cited drafts, findings, the evidence library, gaps and claim rules

In shortThe matrix sits on top of a full dossier workspace. The section board shows where each section stands. Each section opens as a cited draft with findings beside the source. The library holds every source, the map holds the messages, the gap list holds what is missing, and the settings hold your rules.

The section board: 61 pages, 386 of 390 claims clear, four findings to decide

Section 6: three findings, each with the passage beside it

The Clinical Value Writer's draft of section 6 runs to 6,920 words and 112 claims, and every sentence carries a numbered citation. Three claims need a person. Lena Ortiz, the medical writer, sees the source on the right with the passage highlighted, so she never leaves the sentence to check it.

Section 7: "more effective than anti-TNF" against the indirect comparison

The draft says Talvoren was more effective than anti-TNF therapy. The network meta-analysis gives an odds ratio of 1.21 with a 95% credible interval of 0.88–1.67, which crosses 1. The checker cannot change text: it raises a finding with the evidence and a proposed wording, and links the sentence to matrix row M4.1.

On this team's rules, the suggestion is "comparable" to anti-IL-17A and "numerically higher" than anti-TNF, a difference that was not conclusive. That wording rule is a sample setting. As Part A explains, an interval crossing 1 does not by itself show two treatments are comparable, so set the wording your own medical reviewers will defend.

Section 8: the Economic Value Writer's draft and the sentence the checker stops

Section 8 was ready to draft when the cost-effectiveness report arrived. Dana starts it. The Economic Value Writer works from the model reports, and the checker catches the kind of sentence this guide opened with: "By controlling flares, Talvoren is expected to reduce HS surgery and hospital admissions." No source shows that, and the model does not include surgery. The suggested wording says the model assumes fewer flare-related emergency visits.

The evidence library and one claims study, opened at its passages

The library holds 142 sources, 141 of them current, and shows each one's design and how many sentences cite it. The matrix columns are the 15 current sources the dossier cites so far; with the superseded week-36 report, that makes the "16 cited" the library shows.

The value message map and the evidence gaps

Progress: drafting against plan and what the checker found

Your claim rules, and a section signed with its meaning

How agents built each dossier section, and the record it leaves

Six dossier agents, from Evidence Librarian to Evidence Matrix Mapper, and the five people who decide

In shortSix specialist agents index the evidence, draft sections, check claims and rate them against sources. None of them can change a source, rewrite a value claim, clear outside-label content or sign. The medical writer decides wording, the dossier lead decides thin claims, the legal and regulatory reviewer decides outside-label content, and reviewers sign.

What each dossier agent does, and what it never does

Who decides findings, and who signs each section

Sections are signed only after their findings are decided. A signature binds the signer's name, the date and time, and its meaning ("scientifically accurate and balanced" or "consistent with the expected label") to the section version. Those are the same elements FDA's electronic records rule, 21 CFR Part 11, asks a signed record to show, for records under its rules.53 These are controls that support your compliance. Your own quality system decides what they need to do.

The people in the Talvoren screens

Study reports and models in, a cited dossier and an evidence matrix out, in the systems you already use

In shortThe solution works from the evidence you already keep, and it keeps the sources as they are. Your template, styles and rules become settings. What comes out is a cited draft, an evidence matrix, a gap list, answers to markets and a signed review record. Scale and access control are built for a global team.

Systems a dossier rollout connects to

Where seven markets pick up the Talvoren dossier

Two design targets for dossier work, one published figure, and how each is labeled

Roles, record and security around a global dossier team

What thin claims and slow value dossier drafting cost you, and how to find your own numbers

In shortStart from your own costs, because the numbers that matter are yours. Published figures only frame the stakes: how payers treat manufacturer models, how often Germany accepts an indirect comparison, and what a day of delay is worth.

Six GVD costs, where each one hides and how to size it in a week

Costs a value dossier calculator leaves out

What faster value dossier drafting returns when you count sections and market questions against cost

In shortPut in your own numbers. The calculator counts two things: the working days released per section, and the hours released on market questions. Every default is an assumption to replace, because no public source supplies one. Dossier count matters most: at the default inputs, one dossier a year barely pays on hours alone.

Six places a GVD team saves time, each with its formula

The calculator below counts the first and the fourth. Estimate the others from your own records.

Dossiers a year, section days and your hourly cost

A company launching one product every few years may be better served by a pilot on one section and a smaller scope. A large portfolio, or an agency writing many dossiers, has a different sum.

How ready your value dossier team is: template, library, message map, baseline and AI approval

In shortPick the answer closest to your team today. Eight questions score 0 to 2; the last is for reflection. The meter adds up your score as you go, and each question says why it matters and what a strong answer sounds like.

A low score means building your evidence library and message map first

What to ask a GVD writing vendor about citations, thin claims, label checks and AI disclosure

In shortAsk every vendor the same questions, us included, and any agency offering an AI accelerator. The answers show whether a sentence traces to a passage, how comparative wording is tested, who decides, and what you can hand an HTA body if asked.

We found no published independent test of any GVD writing tool's accuracy, as of November 2026. The studies cited here are self-reported. A planted-error test on your own section is the fairest comparison, and the same test measures your current process.

Test GVD authoring software on one section with planted errors, then decide on three numbers

In shortMeasure first, then test on your own material with errors you planted, then decide with three numbers. None of these steps commits you to a vendor, and the first two pay off whatever you decide.

1. Re-check 100 claims from a signed section against their passages, and log the days that section took. That is your baseline.
2. Write down your rules for evidence strength, comparative wording, cost-offset claims and outside-label content. Have medical and legal reviewers approve them.
3. Gather one section's sources with your template and message map, and an approval path to share them. Copy one draft and plant known errors in it.
4. Run in parallel on a live section, with your reviewers signing everything, and answer a few real market questions through it.
5. Decide with three numbers: catch rate by error type, false alarms on clean text, and working days to a reviewed section.

Six pilot measures, from catch rate by error type to days to a cited answer

Catch rate, false alarms and days to a reviewed section are the three numbers you decide on. The other three explain them.

Questions market access and HEOR teams ask about global value dossiers

In shortDirect answers to the questions people search, each one complete on its own.