France's PMSI hospital database counts 897 PNH patients (2018-2022), putting 2022 prevalence near 1 in 94,000, below the 1-in-70,000-to-80,000 range Orphanet and France's national rare disease plan have cited for years.
Three independent French data sources size the PNH population, and none of them was built to be compared against the others. The most systematic is administrative: a 2025 real-world study built on France's PMSI national hospitalization database, accessed via the CASD secure data platform, identified 897 PNH patients between 2018 and 2022, with the annual count rising from 581 to 725 and roughly 100 newly diagnosed patients added each year. That puts 2022 prevalence near 1 in 94,000. The second source is literature-derived: Orphanet and France's national rare disease plan have long cited a higher prevalence range of 1 in 70,000 to 80,000, implying roughly 850 to 1,000 patients nationally, an estimate not verified against hospital records.
The third source, the French National Observatory of PNH Clones, is not a national census at all: its first interim analysis validated 126 clone-positive cases from only 24 of more than 50 participating French-speaking centres. Triangulating the three means treating PMSI as the primary, most current and complete count, cross-checking it against the Orphanet range rather than averaging the two, and treating the clone registry as a partial, flow-cytometry-confirmed subset that will always undercount by design until its full centre network reports in. A sizing model anchored only on the Orphanet literature figure, or on Germany's DGHO Onkopedia guideline, which itself borrows a 16-per-million rate from French and UK registries, would overstate the addressable French population relative to what the country's own administrative data now shows: roughly 10.6 per million in 2022.
France PNH sizing — three sources, three population counts
| Data Source | Population Figure | What It Captures |
|---|---|---|
| PMSI national hospitalization database (2018-2022) | 897 patients; ~1/94,000 prevalence in 2022 (725 patients) | Hospitalized/treated patients captured in French Social Security hospital claims data |
| Orphanet / France's national rare disease plan | 1/70,000 to 1/80,000 (approx. 850-1,000 patients nationally) | Pre-PMSI literature-derived estimate, not verified against hospital records |
| French National Observatory of PNH Clones | 126 validated clone-positive cases from 24 of 50+ network centres | Flow-cytometry-confirmed clone diagnoses, not a national census |
Sources: Epidemiology and care management of Paroxysmal Nocturnal Hemoglobinuria (PNH) in a real-world setting in France: Description from the French National Hospitalization Database, PLOS One, 2025 (PMID 41990028); Orphanet PNH disease page; French National Rare Disease Plan, 2023; First Interim Analysis of the French National Observatory of Paroxysmal Nocturnal Hemoglobinuria Clones, Blood 130(Suppl 1):4960, 2017; MaRIH (Filière de Santé Maladies Rares Immuno-Hématologiques).
What this model answers
Every section answers a named commercial question your team is asking, scoped to your asset.
Delivers
- The PMSI methodology and its 2018-2022 cohort build
- the Orphanet/national rare disease plan literature estimate
- why PMSI, not the older literature figure, is the primary anchor
Delivers
- The clone registry's partial network coverage (24 of 50+ centres)
- why it structurally undercounts
- how it functions as a flow-cytometry-confirmed cross-check rather than a population count
Delivers
- The 16-per-million figure's French/UK registry origin
- France's own roughly 10.6-per-million 2022 PMSI-derived rate
- the sizing implication for any model still anchored on the borrowed figure
Custom model delivered in 72 hours.
Commission This ModelWhat's inside
- Why PMSI's 897-patient administrative count, not the older Orphanet literature estimate, anchors French PNH sizing
- Pressure-tested against the clone registry's partial coverage before the rest of the model is built out
- The PMSI cohort build via the CASD secure data platform and its 897-patient, five-year count
- 2022 prevalence of roughly 1 in 94,000 from the 725-patient annual count
- The 1-in-70,000-to-80,000 estimate and its 850-1,000-patient implied population
- Why this figure predates and was not verified against hospital administrative records
- 126 validated clone-positive cases from 24 of more than 50 participating centres
- Why the registry is a partial flow-cytometry cross-check, not a national census
- Where PMSI, Orphanet, and the clone registry agree and diverge
- The sizing implication versus Germany's borrowed 16-per-million DGHO Onkopedia figure
- The full triangulated model, re-runnable with your own assumptions
- The open sizing questions your team must close before the number is used in planning
Included with every brief
How AXLRx builds this model
Prepared by MoatRx analysts.
Every AXLRx market sizing model triangulates at least two independent methods before accepting a patient count. This is explicitly a sizing model (static patient count), distinct from a Patient Flow or forecasting model (dynamic revenue/uptake).
PNH France sizing sources: the PLOS One 2025 PMSI real-world study (PMID 41990028), Orphanet's PNH disease page, France's National Rare Disease Plan (2023), and the First Interim Analysis of the French National Observatory of PNH Clones (Blood, 2017).
- PMSI-based prevalence and patient counts verified against the PLOS One 2025 publication and its PubMed record (PMID 41990028)
- Orphanet and national rare disease plan literature estimate verified against Orphanet's PNH disease page and France's National Rare Disease Plan (2023)
- Clone registry validated-case count and centre coverage verified against the First Interim Analysis abstract published in Blood (2017)
Frequently asked questions
Commission this model
AXLRx delivers rare disease market sizing models built for forecasting and strategy teams sizing the France PNH opportunity. Custom model in 72 hours.
Specify your indication, market, and cohort definition.
AXLRx analyst confirms triangulation methods and comparator set before building.
Research-verified sizing model in 72 hours with optional analyst readout.