When a Payer Says Yes, For Now: Coverage With Evidence Development

Coverage with evidence development is a conditional yes. A payer agrees to fund a promising treatment now, before the evidence is fully mature, on the condition that data keep being collected in routine care to confirm whether the therapy actually works and for whom. It is a way to move faster than a strict evidence standard would allow, without pretending the uncertainty has been resolved. The wager is that real-world data arriving over the next few years will either justify the decision or correct it.

Why a payer would say yes before the evidence is in

For serious conditions with few options, waiting has a cost measured in patients. Regulators increasingly clear such products through accelerated pathways, and the Orphanet Journal of Rare Diseases review by Dayer and colleagues (2024) describes what that leaves on the payer's desk: approvals resting on small single-arm trials rather than randomized comparisons, and endpoints that are often surrogate or intermediate rather than long-term clinical outcomes. The treatment may well help. But durability, comparative effectiveness against existing care, and real budget impact are genuinely unknown at launch.

A payer facing that gap has three broad choices. Refuse coverage until better evidence exists, and accept that some patients go without a therapy that may work. Cover unconditionally, and accept that some spending will go to treatments that turn out to be ineffective or no better than cheaper options. Or cover conditionally, releasing access while requiring the manufacturer to generate the missing evidence. Coverage with evidence development is that third path.

How the mechanism works in practice

The Institute of Health Economics rapid review (Akpinar and Warkentin, 2025) catalogs how different systems structure the conditional yes. In the United States, the Centers for Medicare and Medicaid Services uses coverage with evidence development to limit coverage to patients enrolled in approved studies, with the stated intent to revise the decision based on results. The review cites aducanumab for Alzheimer's disease as an example: CMS granted coverage tied to participation in real-world evidence studies as a condition of continued Medicare coverage.

Elsewhere the same idea travels under the label managed entry agreement, which the IHE report defines as an arrangement allowing the managed introduction of a technology, often involving risk-sharing, performance-based reimbursement, or time-limited recommendations. The report documents CAR-T cell cancer therapies such as tisagenlecleucel and axicabtagene granted conditional reimbursement in France, England, and Scotland, each requiring registry-based evidence to support future reassessment. England has used its Cancer Drugs Fund to approve treatments with real-world follow-up; Scotland's ultra-orphan pathway builds in reassessment after three years. Tumor-agnostic drugs such as entrectinib and larotrectinib, the IHE review notes, first drew negative recommendations on thin trial data and were later reimbursed conditionally once real-world evidence supported resubmission.

The shared structure is a clock and a data plan: coverage begins, evidence accrues, and a scheduled reassessment decides whether the yes holds.

The trade-offs of paying first

The appeal is obvious. Patients get earlier access, and the health system learns how a treatment performs outside the controlled world of a trial, in the messier population that actually receives it. But paying before the evidence matures carries costs that deserve equal billing.

The first is that the required evidence is hard to produce well. Dayer and colleagues lay out why real-world data for these decisions is methodologically demanding: without randomization, patients on different treatments differ systematically, so selection bias and confounding threaten any comparison. Rare diseases sharpen the problem, because small, heterogeneous populations make bias-correction methods like matching or stratification difficult to apply. Historical or external controls are shaky when diagnostic standards and usual care shift over time. Registries and claims data carry missing or incomplete information, and routine data-quality monitoring, standard inside a trial, is not the norm outside one. The IHE review adds that long timelines, high administrative burden, and difficulty aligning incentives across payers and manufacturers all hinder evidence generation, and that uptake of outcome-based agreements remains low in part for these reasons.

The second cost is fragmentation. Because different payers set different requirements, the same product can face annual review in one country, three-year reassessment in another, and five-year reassessment in a third, each with its own registry mandate. A study designed to satisfy one system may not answer another's question, so effort is duplicated and comparability suffers.

The hard part is the exit

The deepest trade-off is that a conditional yes is far easier to grant than to withdraw. Once patients are established on a therapy, reversing coverage is disruptive and unpopular, and the evidence generated is often too weak to force the issue. Dayer and colleagues point to the sobering track record: among United States coverage-with-evidence-development programs, only a small number ever retired the requirement or revoked coverage, and a review of Cancer Drugs Fund reassessments found real-world data played a limited role in resolving the original uncertainty. The uncomfortable implication is that a mechanism justified by its promise to correct mistakes may rarely do so.

That gap between intent and practice is the crux. Coverage with evidence development is defensible precisely because it is meant to be revisited. If the reassessment is toothless, the arrangement quietly becomes permanent coverage dressed as a trial, and the uncertainty it was supposed to manage is simply absorbed by the payer.

What separates a real conditional yes from a hopeful one

Both reports converge on the same design principles. The IHE review emphasizes early alignment among regulators, health technology assessment bodies, and payers about what the evidence must show, backed by transparent protocols, robust data infrastructure, and clear stakeholder roles. Dayer and colleagues offer a parallel checklist: state the intended use up front, prespecify study length against dropout risk, justify outcome measures with validated instruments, choose methods that adjust for confounding, and define who funds and owns the data. Underneath all of it sits one condition that cannot be waived, a credible and enforceable point at which the decision is genuinely reopened, with the possibility of no left on the table.

For patients and clinicians, the practical read is this. A conditional yes can be the fastest ethical route to a treatment that might help, and it is not a verdict that the treatment definitely works. The value of the arrangement rests almost entirely on whether the data collection is fit for purpose and whether the promised reassessment is real. This is educational information, not medical advice.