Germany G-BA Drug Benefit Rating Lookup

Look up a drug substance's G-BA (Gemeinsamer Bundesausschuss, Germany's statutory health-technology-assessment body) early-benefit assessment rating under Section 35a SGB V ("fruehe Nutzenbewertung") before a market-access or pricing decision: added- benefit rating, evidence category, comparator therapy, indication, and decision date -- straight from G-BA's own official machine-readable data export, not scraped off a web page. Can also run on a schedule to monitor new and changed ratings.

What you get

One record per new, changed, or withdrawn decision/patient-group combination. Every record carries an explicit matchState:

• FIRST_SEEN -- your own baseline run.
• NEW -- a decision/patient-group not seen on a previous run.
• RATING_CHANGED -- the added-benefit rating itself changed (extent, evidence level, comparator, or the headline rating sentence). The record includes both previousBenefitRating (before) and benefitRating (after).
• AMENDED -- some other tracked field changed (validity date, trade name, indication text, ICD code) without the rating itself changing.
• REMOVED -- a decision that was tracked on a previous run has dropped out of G-BA's currently-implementing export (repealed or superseded).

Unchanged decisions are never pushed or charged -- only what actually moved.

Each record includes: decision ID and case number, substance and trade name(s), indication (patient subgroup), ICD-10 code(s), decision date, benefit-rating extent and evidence category, the headline rating sentence, per-endpoint (mortality/morbidity/ quality of life/adverse events) advantage-or-disadvantage symbols where published, and a link back to G-BA's own procedure page.

No personal data

This Actor reads G-BA's own structured XML export (see "How it works" below), which carries no contact/person-shaped field at all -- every one of its ~99 distinct data fields is clinical or administrative (substance, ICD code, trade name, dates, rating category). A full sweep of the live export for title+name patterns found none. As a second-layer guard, every free-text field this Actor does output (indication text, rating text, per-endpoint descriptions) is still checked for a German title (Herr/Frau/Dr./Prof./PD) followed by a capitalised name, and withheld entirely (never partially redacted) if it ever matches -- only a count of withheld fields is reported, never the text itself. The one field in the source closest to a hearing-minutes risk (G-BA's own long-form "Tragende Gruende" reasoning summary) is never read by this Actor at all, by design, regardless of its content.

How it works

G-BA publishes a machine-readable XML export ("AIS -- maschinenlesbare Fassung der Beschluesse nach Section 35a SGB V"), updated twice a month, covering every currently- implementing early-benefit decision with the added-benefit rating as structured data (not locked inside a PDF). Getting it requires accepting a one-checkbox terms-of-use form rather than a plain static download link -- this Actor does that automatically on every run; no account or manual step is needed on your side.

Why you might need this

Pharma market-access and HEOR teams, pricing/reimbursement consultancies, and medical- communications agencies currently track these decisions by paying for a commercial HTA-intelligence subscription or by manually re-checking G-BA's site. This Actor surfaces the same decisions and rating changes directly from G-BA's own official feed. The rating itself is not just informational: under AMNOG (Arzneimittelmarktneuordnungs- gesetz), G-BA's added-benefit category is the direct input to the price negotiation between the manufacturer and the GKV-Spitzenverband, so a RATING_CHANGED record (a downgrade or upgrade in extent) is a signal with a real negotiating-position consequence, not a line item to notice weeks later on a routine manual check.

Scope: Germany (G-BA) only. France (HAS) and the UK (NICE) are separate national HTA bodies with their own sites and schemas -- not covered by this Actor.

Sample output

A RATING_CHANGED record, where the headline added-benefit rating moved between runs:

{
  "recordKey": "1334:7",
  "decisionId": "1334",
  "caseNumber": "2026-04-01-D-1287",
  "patientGroupId": "7",
  "substance": "Retifanlimab",
  "tradeNames": ["Zynyz"],
  "indication": "Erwachsene mit metastasiertem Merkelzellkarzinom",
  "decisionDate": "2026-10-01",
  "benefitRating": {
    "extent": "betraechtlich",
    "evidence": "Hinweis",
    "comparator": "Watch and wait / bestmoegliche unterstuetzende Behandlung",
    "text": "Ein Hinweis auf einen betraechtlichen Zusatznutzen liegt vor."
  },
  "previousBenefitRating": {
    "extent": "gering",
    "evidence": "Anhaltspunkt",
    "comparator": "Watch and wait / bestmoegliche unterstuetzende Behandlung",
    "text": "Ein Anhaltspunkt fuer einen geringen Zusatznutzen liegt vor."
  },
  "matchState": "RATING_CHANGED",
  "changedFields": ["benefitRating.extent", "benefitRating.evidence"],
  "matchesWatchlist": true,
  "sourceUrl": "https://www.g-ba.de/beschluesse/1334/",
  "checkedAt": "2026-10-10T06:00:00.000Z"
}

Input

{
  "watchlist": ["Sugemalimab", "C61"],
  "maxResults": 300
}

• watchlist (optional, default empty): substance names, trade names, or ICD-10 codes. Every record gets a matchesWatchlist flag (case-insensitive substring match) so you can filter to drugs/indications you actually track. Leave empty to track everything.
• maxResults (optional, default 300): caps how many changed records are pushed and charged in a single run. Important for your first run: G-BA's export currently covers roughly 1,700 decision/patient-group records; a first-ever run with no prior state treats every one of them as a new baseline record, so with the default cap your first run reports the newest ~300 and the rest follow automatically on your next scheduled run (nothing is lost or double-charged either way). If you want the full history backfilled in one run, raise maxResults and raise your own run cost cap in Console accordingly (see Pricing below).

Change tracking across runs

Run this Actor on a schedule -- weekly is a reasonable match for G-BA's own twice- monthly update cadence. Each decision/patient-group's rating and key fields are stored per Actor run, so every run after the first reports only what actually changed.

Pricing

Pay per run started, plus per new/changed/removed decision record reported (unchanged records are never charged). See the Pricing tab for exact rates. A first run with a high maxResults will cost proportionally more (see the Input section above) -- raise your own maximum-cost-per-run setting if you deliberately want a one-run full backfill.

Worked example. A weekly scheduled run with no watchlist (full feed, after the one-time backfill) typically sees 5-20 real new/changed decisions a week given G-BA's twice-monthly update cadence, so roughly USD 1.40-2.60/month. A watchlist-scoped run tracking a handful of named substances pays close to zero most weeks (just the run fee) and only incurs a per-record charge on an actual rating change for one of your own tracked drugs -- the scenario this Actor is built for.

---

Deutschland G-BA Fruehe-Nutzenbewertung-Monitor

Verfolgt den Gemeinsamen Bundesausschuss (G-BA) auf neue und geaenderte Beschluesse zur fruehen Nutzenbewertung nach Section 35a SGB V: Ausmass des Zusatznutzens, Wahrscheinlichkeit/Evidenzkategorie, zweckmaessige Vergleichstherapie, Anwendungsgebiet und Beschlussdatum -- direkt aus G-BAs eigenem maschinenlesbarem Datenexport, nicht von einer Webseite gescrapt.

Keine personenbezogenen Daten: Der genutzte XML-Export enthaelt keinerlei Kontakt-/Personenfeld; alle enthaltenen Freitextfelder werden zusaetzlich auf ein Titel-plus-Name-Muster (Herr/Frau/Dr./Prof./PD + Name) geprueft und bei einem Treffer vollstaendig verworfen, nie teilweise geschwaerzt. Das Feld mit dem hoechsten theoretischen Risiko (die "Tragenden Gruende" als Volltext) wird von diesem Actor grundsaetzlich nicht gelesen.

Umfang: ausschliesslich Deutschland (G-BA). Frankreich (HAS) und Grossbritannien (NICE) sind eigene nationale HTA-Behoerden mit eigenen Datenquellen und werden von diesem Actor nicht abgedeckt.

Eingabe: watchlist (Wirkstoff-/Handelsnamen oder ICD-10-Codes, optional) und maxResults (Obergrenze geaenderter Datensaetze pro Lauf, Standard 300 -- beim ersten Lauf ohne vorherigen Zustand werden alle ca. 1.700 Datensaetze als neu gewertet; mit der Standardgrenze werden die juengsten 300 gemeldet, der Rest folgt automatisch bei den naechsten planmaessigen Laeufen).