.png)
SummaryAn AMNOG database is a structured, queryable record of German early benefit assessment outcomes: for every medicinal product with a new active substance, it holds the G-BA resolution, the added benefit rating per patient subgroup, the appropriate comparator therapy, the annual treatment costs and the resulting reimbursement amount. It turns a stream of individual PDF resolutions into a dataset you can benchmark across therapeutic areas and over time.
Most public writing on AMNOG explains the procedure. Far less of it explains what the procedure emits as data, and that is the part market access, pricing and health economics teams actually work with. A single resolution is a document. Two hundred resolutions in one indication, sorted by year and rating, are an argument. This article is about the second thing.
AMNOG produces two artefacts per product: a G-BA resolution on the extent of added benefit, and a negotiated reimbursement amount that follows from it. Since 1 January 2011, the Federal Joint Committee (G-BA) has assessed every newly authorised medicinal product with a new active substance under section 35a SGB V. The manufacturer submits a dossier no later than the day the product is first placed on the market. The G-BA commissions an assessment, in most cases from IQWiG, and publishes the assessment result within three months. A written and oral hearing follows, and after a further three months the G-BA passes its resolution.
That resolution is the input to price negotiation. The GKV-Spitzenverband and the manufacturer then negotiate the reimbursement amount (Erstattungsbetrag) under section 130b SGB V within six months. If they do not agree, an arbitration board (Schiedsstelle) decides within three months. The reimbursement amount applies from the seventh month after the product was first placed on the market. For the full sequence from marketing authorisation through to pricing, see our companion article on pharmaceutical market access in Germany.
For analytics, the useful mental model is this: the G-BA resolution is the categorical variable, the Erstattungsbetrag is the continuous one, and everything interesting happens in the relationship between them.
A G-BA resolution is far more granular than a single verdict per product. It is structured by patient subgroup, and each subgroup carries its own comparator, its own rating and its own certainty of evidence. A product can hold a considerable added benefit in one subgroup and no proven added benefit in another, and that split is the single most common reason that headline summaries of AMNOG outcomes mislead.
| Element of the resolution | What it holds | Why it matters analytically |
|---|---|---|
| Patient subgroup (Patientengruppe) | The population the rating applies to, often split by line of therapy, biomarker or age | The true unit of analysis. Counting outcomes per product instead of per subgroup distorts every benchmark |
| Appropriate comparator therapy (zweckmäßige Vergleichstherapie) | The standard of care the G-BA set for the comparison | Determines the price anchor under the section 130b criteria and predicts the next entrant's comparator |
| Extent of added benefit | One of six statutory categories, per subgroup | The categorical outcome variable for any benchmark |
| Certainty of evidence (Aussagesicherheit) | Proof, indication or hint | Two "considerable" ratings are not equivalent if one rests on a hint only |
| Annual treatment costs | Costs for the assessed product and the comparator, per patient per year | The publicly documented cost frame that negotiation starts from |
| Eligible patient numbers | Estimated target population in statutory health insurance | Converts a rating into a market size estimate |
| Requirements for quality assured use | Prescribing restrictions, specialist requirements, monitoring | Predicts real world uptake constraints that pricing alone does not explain |
| Time limit (Befristung) | Date by which a new dossier is due | Flags forthcoming reassessments, a scheduled event you can plan against |
German law defines six categories of added benefit, set out in section 5(7) AM-NutzenV. The commercial consequence of each is shaped by the price criteria in section 130b(3) SGB V, which the 2022 GKV-FinStG reform tightened into what practitioners call the Leitplanken, or guard rails.
| Category (German) | English | Statutory meaning | Commercial consequence |
|---|---|---|---|
| Erheblicher Zusatznutzen | Major added benefit | A sustained and previously unattained large improvement over the comparator | Strongest negotiating position, rare in practice |
| Beträchtlicher Zusatznutzen | Considerable added benefit | A previously unattained marked improvement | Threshold rating for several concessions, including release from the combination discount |
| Geringer Zusatznutzen | Minor added benefit | A moderate and not merely marginal improvement | Guard rails cap annual costs at the comparator level |
| Nicht quantifizierbarer Zusatznutzen | Non quantifiable added benefit | A benefit exists but the evidence base does not allow it to be quantified | Treated close to minor benefit for pricing purposes |
| Kein Zusatznutzen belegt | No added benefit proven | No advantage over the comparator was demonstrated | Price must undercut a patented comparator, or match a generic comparator's cost |
| Geringerer Nutzen | Lesser benefit | The product's benefit is lower than the comparator's | The weakest outcome, and a strategic reset for the asset |
Read the categories together with the guard rails and one thing becomes obvious: the rating is not a marketing label, it is a price ceiling formula. That is why teams that track AMNOG outcomes seriously track them as structured fields rather than as document links.
All resolutions and their justifications are published by the G-BA and are freely accessible. The G-BA benefit assessment portal lists procedures by active substance, and for resolutions passed from 1 August 2019 the G-BA publishes English translations of both the resolution and the reasoning. IQWiG publishes its dossier assessments separately, including the analyses the G-BA later accepted or departed from.
The publication is complete and the format is consistent. What it is not is analytical. Each procedure is a set of documents attached to a procedural record, so answering a question like "how have comparator choices in second line oncology shifted since 2020" means opening several hundred files and extracting the same fields by hand. The public source is the ground truth; the work is turning it into rows.
The negotiated reimbursement amount has historically been visible, because it replaces the manufacturer's ex factory price and is reported to price and product directories, which is how pharmacies, hospitals and payers bill against it. The negotiation itself has never been public: the rebate off the list price, the volume assumptions and the concessions are not disclosed.
"I cannot see by how much the price was reduced in the course of the negotiations." Consultant, health economics, from a pharmazie.com customer interview, translated from German.
Since 2025 that gap has widened. Under section 130b(1c) SGB V, introduced by the Medical Research Act, a manufacturer may declare within five days of concluding an agreement that the reimbursement amount is not to be reported to public price and product directories. The option is conditional: the company must evidence its own pharmaceutical research and development capability in Germany, relevant proprietary projects and collaborations with public institutions there, and it applies to agreements concluded up to 30 June 2028. Section 130b(4b) gives defined parties, including hospitals and importers, a right to obtain the amount on request from the GKV-Spitzenverband. You can read both provisions in section 130b SGB V.
The practical consequence for anyone maintaining an AMNOG dataset: from 2025 onwards, price series for newly negotiated substances may contain structural gaps that are not data quality errors. They are a legal feature. Any longitudinal price analysis built after 2025 needs a flag for "confidential under section 130b(1c)" rather than a null, or it will silently understate the population it claims to describe.
Section 130e SGB V requires manufacturers to grant health insurance funds a discount of 20 percent on the ex factory price, excluding VAT, for medicinal products with new active substances used in combinations that the G-BA has designated under section 35a(3). It applies from 2 May 2023. The discount ceases where the G-BA determines that the combination itself has at least considerable added benefit.
This provision is where the added benefit rating stops being a historical record and becomes a live financial parameter. It also creates a dependency that is easy to miss in a product level dataset: the discount attaches to a combination, not to a single product, so a rating on one asset can change the effective price of another. Teams modelling combination regimens need the G-BA combination designations alongside the individual resolutions, or the arithmetic will not close.
Benchmarking works when outcomes are stored per subgroup with the resolution date, indication, comparator, rating, certainty of evidence and annual cost as separate fields. Once that structure exists, the recurring questions become straightforward queries rather than research projects:
Two design rules save most of the pain later. First, the subgroup is the row, not the product. Second, store the rating and the certainty of evidence as separate fields, because collapsing them destroys the distinction between a proven minor benefit and a hinted considerable one.
An AMNOG outcome only becomes operationally useful when it is joined to the products it governs. A resolution names an active substance; a payer, a hospital pharmacy or a pricing analyst works with pack level identifiers, current prices, availability and the regulatory documents behind them. The join across those layers is where most in house AMNOG trackers stall, because the outcome data and the product master data usually sit in different systems maintained by different teams.
This is the cross layer problem pharmazie.com was built for. Since 1989 the platform has consolidated 25+ pharmaceutical databases into one search, covering 50,000+ German products and 120,000+ international products across 50+ countries. For teams whose questions run across layers, from benefit assessment outcome to substance to pack to price to supply status, and increasingly across borders as European HTA cooperation raises the value of comparable data from other markets, that consolidation is the most complete single answer available. It does not replace the G-BA as the authoritative source of a resolution. It removes the manual reconciliation between that resolution and everything else you need in the same query.
Three developments will shape AMNOG datasets over the next few years. The confidentiality option under section 130b(1c) will progressively thin out public price series for newly negotiated substances through to mid 2028. The combination discount under section 130e ties more pricing outcomes to combination level designations rather than single products. And European joint clinical assessment increases the need to hold German outcomes alongside comparable evidence and availability data from other member states, so that a German comparator choice can be read against what is actually on the market elsewhere.
Each of these makes the same point in a different way: the value of AMNOG data is moving from having the documents to having the fields, dated, structured and joinable.
This content is intended for healthcare professionals and does not constitute medical advice. Last reviewed: July 2026.
