Paying for a Cure You Can Only Give Once: How Outcomes-Based and Annuity Agreements Work

When a therapy is given once, priced in the millions, and cannot be returned if its benefit fades, payers reach for three contract designs: outcomes-based agreements that refund part of the price when a promised health result is not reached, annuity or installment deals that spread one payment across years and can stop if the treatment stops working, and warranties that guarantee money back if a defined efficacy standard is missed within a set window. Each is a way of buying something whose value you cannot confirm on the day it is sold.

Key points#

• A durable, single-dose gene therapy front-loads both the full cost and the full clinical gamble, while proof of lasting benefit only arrives years later.
• Outcomes-based agreements make payment conditional on a pre-specified result and claw back money if that result is not met.
• Annuity or installment contracts turn one enormous charge into a scheduled stream, and can pause payments when a patient loses response.
• Warranties borrow consumer logic: fail a stated efficacy threshold in a stated period, and the maker refunds or offsets.
• All three collapse without trustworthy long-term data, which real-world registries have repeatedly struggled to supply.

Why a single infusion breaks the old payment logic#

Most medicines are paid for the way a utility bill is paid, a little at a time, for as long as the treatment keeps being used. Stop a daily tablet that is not working and the spending stops with it, so the payer's risk is naturally capped. A one-time gene therapy inverts that arrangement. The clinical bet and, in many contracts, the money both land on a single infusion, while the evidence of durable benefit accumulates over years no one can inspect at the point of purchase.

Two distinct uncertainties sit underneath every deal. One is durability: will an effect seen in a short trial still be there in a decade, or will it erode. The other is measurement: can real-world data actually confirm that persistence in terms both sides will accept. A systematic review of managed entry agreements for advanced therapy medicinal products, published in Clinical Therapeutics, catalogs exactly this bind, describing high launch prices set against thin evidence from short-term, single-arm studies in small populations as the recurring problem these contracts are built to manage.

Three ways to share the risk#

Outcomes-based agreements: pay for the result, not the promise#

An outcomes-based agreement links payment to whether a patient reaches a health target that both parties agreed to in advance. Miss the benchmark, and the manufacturer returns some negotiated share of what was paid. In effect it rewrites an unverifiable promise as a conditional sale.

The most visible current example is a public one. Through its Cell and Gene Therapy Access Model, the Centers for Medicare and Medicaid Services negotiated the core terms of outcomes-based agreements on behalf of states with the makers of the two approved sickle cell disease gene therapies. As the U.S. Department of Health and Human Services framed it, payment is tied to whether patients achieve improved health outcomes over time, and the manufacturer provides a rebate if the therapy does not deliver as agreed. CMS reported that 33 states, plus the District of Columbia and Puerto Rico, agreed to take part, covering roughly 84 percent of Medicaid beneficiaries with the condition.

The genuinely hard part is not the arithmetic of the rebate. It is defining an outcome that is clinically meaningful, plausibly attributable to the therapy, and trackable across different payers over many years.

Annuity and installment models: stretch the payment across time#

An annuity or installment agreement attacks affordability head-on by unbundling one huge charge into a series of smaller ones. Rather than a single seven-figure payment, the payer pays in scheduled pieces over several years, much as a mortgage replaces an impossible lump sum with predictable installments. Tie those installments to continued benefit and the annuity becomes a durability hedge as well, because payments can shrink or stop if the patient loses response.

The reasoning is straightforward: a budget can absorb a large cost delivered in scheduled portions far more easily than the same cost demanded all at once, and linking each portion to sustained performance keeps the maker accountable long after the infusion. Health-policy analyses have proposed annuity payments, often bundled with an outcomes-based element, as one answer to the budget shock these therapies create, while flagging the conditions that make them practical, among them an outsized budget impact, rough cost parity with paying upfront, and a limited annuity period. Whether pay-over-time terms are actually on offer varies by therapy and by payer.

Warranties: a defined promise against early failure#

A warranty is the most intuitive of the three because it copies the logic of a consumer product guarantee. The manufacturer commits up front that if the therapy fails to meet a stated efficacy standard within a specified period, it will refund or otherwise offset the cost. That shifts the risk of early efficacy failure off the payer and back toward the maker.

Warranty structures have surfaced alongside newly approved single-dose therapies as a way to give payers protection against one specific scenario, a treatment that simply does not hold. A warranty is only ever as good as its trigger, meaning the threshold chosen, the length of the follow-up window, and who decides whether the failure actually occurred. Those three details are what you should read first in any warranty.

The measurement problem that limits all three#

Every one of these instruments rests on evidence that a therapy did or did not work, and here the track record is sobering. A Belgian analysis of outcome-based managed entry agreements for rare disease therapies, presented as a 2025 conference abstract in the International Journal of Technology Assessment in Health Care, examined agreements for 57 orphan drugs reimbursed between 2012 and 2024. It found that real-world evidence was frequently rejected at reassessment over concerns about data validity and incomplete registries. Only about 8 percent of those agreements produced a definitive listing decision, with a median duration near 24 months and some extended or renewed up to five times.

That is the practical caution beneath the enthusiasm. A contract can specify an elegant refund the instant benefit is lost, but it pays out only if a registry actually captured the outcome, only if the data withstand scrutiny, and only if both parties agree on what the numbers mean. Clear data-collection protocols and real transparency are the precondition for every model above, so check them before you credit any of the three. Without dependable measurement, an outcomes-based agreement is a promise no one can enforce, an annuity cannot know when to pause, and a warranty cannot know when it has been triggered.

None of these arrangements lowers a therapy's price or settles whether it works. What they do is redistribute the durability and affordability risk between payer and manufacturer, and their credibility lives entirely in the quality of the evidence infrastructure underneath them.