Approved but unpaid: when clearance does not convert to revenue

Cause of death: the company optimised for the regulator and then met a second decision maker it had never designed evidence for.

A monoclonal antibody for early Alzheimer's disease received a UK marketing authorisation in August 2024.¹ On the same day, the health technology assessment body opened its first draft guidance consultation.² Two years later, after three rounds of draft guidance, an appeal, and a fourth committee meeting in June 2026, there is still no published appraisal and therefore no NHS funding mandate. A second antibody in the same class, licensed two months after the first, sits in the same position.³

Neither decision was about safety. Neither was about whether the drug works. The committee accepted a treatment effect. It said the effect was small and the cost was not.⁴ For the second product, the plausible incremental cost effectiveness ratio landed somewhere between roughly £113,000 and £126,000 per quality adjusted life year, against a threshold that in April 2026 rose only to £25,000 to £35,000.⁵

Both products are licensed. Both are available privately. Neither is paid for.

What actually killed it

Approval and payment are two decisions, made by two bodies, against two different questions.

The regulator asks whether the product is safe and whether it works, typically against placebo or an existing standard of care. The payer asks something harder: whether it works better than what we already buy, by enough, at this price, for our population.

The clearest statement of that gap comes from a French payer, which in 2016 delisted an entire class of symptomatic Alzheimer's medicines that held valid marketing authorisations throughout. Its stated reason was that the efficacy evidence rested "essentiellement sur les troubles cognitifs, dans des études cliniques versus placebo dont la pertinence clinique et la transposabilité en vie réelle ne sont pas assurées."⁶ A ministerial order struck roughly 250 presentations from the reimbursement list with effect from 1 August 2018.⁷ The products remained legal to sell. Patients simply paid for them.

That is the sentence to internalise. A placebo controlled trial adequate for approval was found inadequate for payment, by a payer, in writing.

The pattern is measurable, not anecdotal

In Germany, 22 medicines were withdrawn from the market between January 2011 and June 2016 after early benefit assessment, 16 percent of the 139 products assessed in that window. Twenty one of the 22, or 95 percent, had been rated as offering no additional benefit.⁸ The payer's own magazine put it more bluntly a few years later: of 445 assessed medicines as at January 2023, 47 were no longer regularly available, "gut jedes zehnte."⁹

Across 216 German dossier assessments between 2011 and 2017, only 54, or 25 percent, showed major or considerable added benefit.¹⁰

The European Commission's own figure is the one to quote if you only quote one. Of 152 centrally authorised medicines between 2016 and 2019, up to 88 percent were accessible to patients in the larger member states. In small or lower GDP member states, fewer than 32 percent.¹¹ Same authorisation. Same continent.

The version that kills a company

Approval without payment is survivable if the product has other markets. It is terminal if the failed negotiation is the launch market.

A one time gene therapy for a transfusion dependent inherited blood disorder held a conditional EU marketing authorisation from May 2019. In April 2021 the company announced that "reimbursement negotiations in Germany did not result in a price for [the product] that reflects the value of this one time gene therapy."¹² In March 2022 the marketing authorisation was withdrawn at the company's own request. The regulator recorded the reason as "commercial reasons."¹³

The product was approved, effective, and legally marketable across the entire European Union. It has no marketing authorisation today.

Devices have more gates

For medicines there is at least one identifiable decision to lose. For devices there are three or four, and a device can clear every regulatory gate and still never get paid because it never got a code, or got a code with no coverage policy, or got coverage with no rate.

The US payer says this itself, in the Federal Register: "FDA approval or clearance alone does not entitle that technology to Medicare coverage, given separate Medicare statutory coverage requirements," and "clinical studies that are conducted to gain FDA market authorization are not necessarily required to include participants with similar demographics and characteristics of the Medicare population."¹⁴ Its own guide for medical technology companies is blunter still: a code "does not automatically result in Medicare coverage for that technology, nor does it result in an established Medicare payment rate," and the three processes "are not necessarily carried out in any particular order."¹⁵

The French position is stated in law rather than inferred. CE marking harmonises market access; "en revanche, chaque État membre dispose d'une indépendance pour la prise en charge ou non par la solidarité nationale," and the assessment mission begins "une fois le marquage CE obtenu."¹⁶ Two separate bodies then apply two separate criteria: one binary test decides whether there is reimbursement at all, a five level scale drives the price.

Germany's digital route is the sharpest illustration of what "paid" really means. Of 74 digital health applications ever listed, 16, or 22 percent, were struck off for failing to demonstrate a care effect. Among those that survived, the average manufacturer set price of €544 became an average negotiated price of €227, a mean reduction of 59 percent, with the largest single cut running 88 percent.¹⁷

Listed is not paid. Paid is not paid at your price.

How to see it coming

Three questions, none of which a regulatory affairs function will answer for you.

Who is the comparator? Not the placebo arm. The thing the payer currently buys, in that country, at the price they currently pay for it. If your trial did not run against it, your evidence package has a hole that no amount of regulatory success will fill.

Which body decides, and is it the same body that decides price? In some markets one body does both. In France the scientific opinion and the price negotiation sit with two different committees and the final call belongs to two ministers.¹⁶ In Germany the benefit rating is not a coverage decision at all, it is an input to a price negotiation you can lose.

What happens if you lose? In some systems a negative decision means no routine funding. In others it means a price you will not accept, which is a different failure and needs a different plan. Know which one you are walking into before you file, because the answer determines whether "no" is a delay or an exit.

The EU's Health Technology Assessment Regulation, applying since January 2025, is sometimes read as fixing this. It does not. It centralises the clinical half of the assessment and expressly leaves the payment half national: the regulation "shall not interfere with the exclusive national competence of Member States, including those for national pricing and reimbursement decisions."¹⁸ It removes duplicated dossier work. It does not remove this failure mode.