What Is a Global Value Dossier and How AI Is Transforming It

The Global Value Dossier is the most important strategic document in the pharmaceutical market access process. In this guide, we explain what it is, how it is built, why it takes so long — and how artificial intelligence is compressing one of the industry's most critical processes from months to weeks.

Building a complete Global Value Dossier (GVD) takes between four and eight months of work from a specialized team. It is the document that determines whether a drug gains market access, at what price, and under what conditions. And yet, most market access teams in Latin America build it in practically the same way they did ten years ago. Generative AI is changing that — with documented results of up to a 60% reduction in development time.

This article is a comprehensive guide for market access directors, HEOR managers, and medical affairs teams who want to understand what a GVD is, how it works in the Latin American context, and what the transformation that generative AI is driving in this process really means.

1. What Is a Global Value Dossier?

A Global Value Dossier (GVD) is a structured document that compiles, synthesizes, and communicates all clinical, epidemiological, economic, and quality-of-life evidence relevant to justifying the value of a drug to decision-makers: HTA agencies, payers, policymakers, and therapeutic committees.

It is the source document from which all local market access materials are derived: submission dossiers for CONITEC, IETS, or CENETEC, cost-effectiveness models adapted to each market, value arguments for physicians and payers, and medical affairs materials.

Standard sections of a GVD

Although content varies by indication and company, GVDs follow a standardized methodological structure based on ISPOR guidelines and the requirements of major global HTA agencies:

Section	Main content	Typical length
1. Executive summary	Synthesis of clinical and economic value for decision-makers	5–10 pages
2. Disease landscape	Epidemiology, burden of disease, quality-of-life impact, unmet need	20–40 pages
3. Treatment description	Mechanism of action, dosing, target population, comparators	15–25 pages
4. Clinical evidence	Efficacy and safety from clinical trials, systematic review, meta-analysis	40–80 pages
5. Patient-reported outcomes	Health-related quality of life, PROs, utilities for economic models	20–35 pages
6. Economic evidence	Cost-effectiveness model, budget impact analysis, sensitivity analysis	30–60 pages
7. Access and system value	Comparators, formulary context, value arguments differentiated by audience	15–25 pages
Total GVD		145–275 pages

2. Why a GVD takes four to eight months — and why that is a problem

The duration of the GVD development process is not arbitrary. It has well-identified structural causes:

2.1 The volume of literature is massive and growing exponentially

A systematic review for a moderately complex indication requires screening between 2,000 and 15,000 references to identify relevant studies. With traditional methods, this screening process takes between three and six weeks of work from an expert reviewer. For drugs in active therapeutic areas — oncology, immunology, rare diseases — the volume can be considerably larger.

2.2 Local adaptation multiplies the work by the number of markets

A global GVD serves as the foundation, but each Latin American market requires specific adaptations: local epidemiology, health system unit costs, cost-effectiveness thresholds, local comparators, and regulatory context. For a company operating in seven LATAM markets, that is seven times the adaptation work.

2.3 Building the value narrative requires multiple iterations

Translating clinical evidence into value arguments that are comprehensible and convincing for a payer — who is not a physician, who has budget constraints, and who evaluates dozens of technologies simultaneously — is a process that requires multiple revisions. The first draft is rarely the final version.

2.4 The decision window does not wait

In many LATAM markets, evaluation windows for new technologies are specific and do not recur frequently. A two-month delay in submission can mean waiting an entire cycle — sometimes a full year — for the next opportunity. In rare diseases or gene therapies, that delay has direct consequences for patients.

3. How generative AI is transforming GVD development

Generative AI does not replace the HEOR expert or the market access team. What it does is eliminate the volume bottleneck — the part of the process that consumes time without adding specialized judgment.

3.1 Evidence synthesis: from weeks to days

The use case with the strongest evidence. LLMs can screen thousands of bibliographic references, extract efficacy and safety data, and identify relevant studies in a fraction of the traditional time. A study published in the Journal of Medical Internet Research documented that AI methods reduce screening time in systematic reviews by up to 80%.

The ISPOR Working Group presented at the Montreal 2025 annual meeting an AI-assisted co-authoring tool for the Disease Overview section of the GVD. Result: 80% of AI-generated interpretations required no editing by the expert. That is not marginal assistance — it is a transformation of the process.

3.2 Economic modeling: local adaptation in days

GPT-4 has demonstrated the ability to replicate published cost-effectiveness models with a margin of error below 1% in under 15 minutes. The most valuable practical application for LATAM teams is not building models from scratch, but adapting existing models to the local parameters of each market: Brazil's epidemiology, Colombia's health system unit costs, Mexico's willingness-to-pay threshold.

What previously took four weeks per country — with the associated consulting costs — can be done in two to five days with an expert overseeing the process and AI handling the volume of adaptation.

3.3 Narrative development: from data to argument at scale

idalab documented a 60% reduction in HTA dossier writing time with its EPRI tool. The most important impact is not the speed — it is the ability to generate differentiated versions of the same value argument for different audiences.

A GVD has a single evidence base, but payers are very different from one another: the criteria of a CONITEC committee in Brazil, those of a private payer in Colombia, and those of an IMSS formulary decision-maker in Mexico are radically different. AI enables generating, from the same evidence base, narratives calibrated for each profile without multiplying the work time.

3.4 Translation and multilingual adaptation: the LATAM multiplier

For teams operating in LATAM, the ability to adapt the GVD into Spanish, Portuguese, and English simultaneously — while maintaining methodological coherence and consistent technical terminology — is one of AI's most practical benefits. What previously required specialized translators and multiple rounds of technical review can be done in hours.

GVD process	Time without AI	With AI	Benchmark
Systematic review (screening)	3–6 weeks	2–4 days	~80% reduction in screening time
Disease Overview section	3–4 weeks	1–2 weeks	80% of drafts required no editing
Cost-effectiveness model (local adaptation)	4–6 weeks/country	2–5 days/country	Error below 1% vs. base model
Complete GVD	4–8 months	6–8 weeks	60% reduction in total time
Adaptation to 7 LATAM markets	14–28 weeks	2–5 weeks	Scalable in parallel — not sequential

4. The AI-powered GVD and methodological standards: what HTA agencies require

The question that most concerns market access teams when considering AI for GVD development is legitimate: will it pass the scrutiny of HTA agencies?

The answer depends on the methodological framework used and the transparency with which the process is documented. The relevant standards in 2026:

• ELEVATE-GenAI (ISPOR Working Group, 2025): the first standardized reporting framework for the use of LLMs in HEOR research. It establishes transparency, traceability, and validation criteria that agencies can verify.
• CHEERS-AI (ISPOR, 2025): an extension of the CHEERS checklist for economic evaluations incorporating AI and machine learning methods. It specifies how to report AI use in cost-effectiveness models.
• PALISADE (ISPOR, 2025): a specific checklist for machine learning methods in HEOR research and real-world evidence.

The fundamental principle is that AI in GVD development does not lower the methodological standard — what it lowers is the time required to meet it.

5. The GVD in the Latin American context: critical adaptations

Building a GVD for use in Latin America has particularities that do not appear in European or North American manuals:

5.1 Local comparators are not the same as in Europe

European HTA agencies evaluate drugs against the treatment standards of their own health systems. In LATAM, the comparators available in formularies vary significantly between countries and, in many cases, differ from the comparator used in pivotal trials. A GVD that does not anticipate this problem will require costly and slow adaptation when it reaches the local submission stage.

5.2 Economic parameters are radically different between countries

Country	Cost-effectiveness threshold (reference)	Unit cost source	Currency
Brazil	3x GDP per capita (~R$30,000–50,000/QALY)	SIGTAP	BRL
Colombia	Up to 3x GDP per capita (~COP$120M–180M/QALY)	ISS / SOAT tariff manual	COP
Mexico	Not formalized — informal reference ~1–3x GDP	Cuadro Basico IMSS / ISSSTE	MXN
Argentina	Not formalized	National Services Nomenclator	ARS
Chile	~1–3x GDP per capita (informal reference)	FONASA tariff schedule	CLP

5.3 Local real-world evidence is increasingly relevant

CONITEC in Brazil has consistently increased its requirement for local real-world data in submissions. The launch of Brazil's National Health Data Network (RNDS) in July 2025 opens unprecedented possibilities for complementing clinical trial evidence with local effectiveness data. GVDs that integrate this dimension will have a significant methodological advantage before CONITEC.

5.4 The EU JCA standard is now relevant for LATAM

The implementation of the European Union's Joint Clinical Assessment (JCA) starting in 2025 establishes a new comparative clinical evaluation standard that LATAM HTA agencies are actively monitoring. Companies that build their GVDs under JCA methodological criteria will be better positioned before agencies such as CONITEC and IETS, which are raising their standards in that direction.

Frequently asked questions

How much does it cost to build a GVD and who does it?

GVDs are typically built by specialized HEOR and market access consultancies, or by in-house pharma teams with experience in pharmacoeconomics and evidence synthesis. The cost of a complete GVD ranges between USD $150,000 and $500,000 depending on the complexity of the indication, the number of comparators, and the level of economic modeling required. With AI, the most time-intensive components can be completed in a fraction of the time, which directly impacts cost.

Does an AI-built GVD carry the same weight with HTA agencies as one built with traditional methods?

Yes, as long as the relevant methodological frameworks are followed: ELEVATE-GenAI for the use of LLMs, CHEERS-AI for economic modeling. NICE has established that AI use must be declared and documented, but that the quality standard for the output remains the same. In LATAM, where no agency has its own guidelines, this principle applies by extension.

Which sections of the GVD can be built with AI and which necessarily require human work?

The sections that benefit most from AI are: (1) the systematic literature review — especially the screening process; (2) the Disease Overview section, where AI can synthesize epidemiology and burden of disease with high accuracy; and (3) the adaptation of the economic model to local parameters. The sections that require more human judgment are: designing the economic model from scratch, interpreting efficacy results in clinical context, and crafting the strategic value narrative for specific audiences.

How does a GVD differ from an AMCP Dossier or an HTA submission?

They are related but distinct documents: the GVD is the global base document, built in English, containing all available evidence. The AMCP Dossier is the format specific to the U.S. market. Local HTA submissions (for CONITEC, IETS, NICE, G-BA, etc.) are adaptations of the GVD to the format and specific requirements of each agency. A strong GVD is the base investment that makes producing all the others efficient.

What is the first step to modernizing the GVD development process on my team?

The first step is identifying the current bottleneck in your process: is it the literature review time? The local adaptation of the economic model? Building the value narrative? Each of these bottlenecks has specific AI solutions with documented benchmarks. The second step is ensuring your team understands the validation frameworks (ELEVATE-GenAI, CHEERS-AI) so that AI-generated output meets HTA agency standards from the first draft.

Conclusion: the GVD is too important to build slowly

The Global Value Dossier is the foundation of pharmaceutical market access. Everything else — the price, the coverage, the negotiation with the payer — depends on how well that foundation is built. And yet, most teams continue to spend four to eight months on a process where AI has already demonstrated 60% reductions in development time.

In Latin America, where HTA evaluation windows are narrower and market access team resources are more limited than in Europe or North America, efficiency in GVD development is not a competitive advantage — it is a strategic necessity.

The methodological frameworks already exist. The benchmarks are already documented. The question is not whether AI will transform GVD development in LATAM — it already is. The question is whether your team will lead that transformation or adapt once competitors have already pulled ahead.