M12 · REGULATION & REIMBURSEMENT
The analysis doesn't decide — an institution does.
Modules 5 through 11 built the analyst's toolkit: perspective, costing, discounting, the QALY, the ICER, thresholds, models, uncertainty, real-world evidence. All of it produces a dossier — a case, built with care. But a dossier doesn't decide anything by itself.
It lands on a desk inside a specific HTA agency — a body built with its own decision rule, its own mandate, its own history. That institution's design shapes the verdict as much as the number in the dossier does. This module turns from "how do you build the evidence and the number" to "what happens to that number once it arrives" — and the biggest source of variation, by far, is which agency, built how.
Same drug, opposite verdicts.
Here's the puzzle this lesson exists to solve. Take one drug. One clinical trial. One dossier, submitted on the same day to two different countries: an incremental cost of £16,000, an added 0.4 QALYs, an ICER of £40,000 per QALY gained, and a clinical picture everyone agrees is a minor-to-moderate improvement over the standard comparator.
In England, that dossier is turned down for routine funding. In Germany, the same evidence gets the drug reimbursed from launch. Nobody made an error. No analyst mis-modelled anything. The evidence didn't change between the two submissions — the machine it was fed into did.
That's the real subject of this module: not "is this analysis correct," but "what does this particular institution do with a correct analysis." Two structural differences — two axes — explain almost all of the variation you'll meet across agencies. This lesson builds both.
Axis 1: threshold-and-QALY, or not.
It's tempting to assume "cost-effectiveness analysis" is simply what HTA is — that every agency in the world runs the incremental cost per QALY against some threshold and reads off a verdict. It isn't. That's one design choice among several, not the definition of the field.
England's NICE is the clearest example of the threshold-and-QALY design: an explicit cost-effectiveness threshold range (£20,000–£30,000 per QALY, moving to £25,000–£35,000 from 2026), and the ICER's position relative to it is the test. Above it, the drug isn't recommended at that price — unless the price is quietly reduced, often through a confidential discount known as a managed entry agreement, to bring the ICER back under the line.
Germany's AMNOG system runs on a completely different mechanism: an added-benefit assessment. There's no QALY threshold anywhere in the process. Instead, the question is how much better the drug is than the appropriate comparator — none, minor, considerable, or major — and that rating becomes the basis for a price negotiation. A drug can have an eye-watering ICER by NICE's yardstick and sail through AMNOG, because AMNOG never asks that question in the first place.
Axis 2: who assesses, who decides, who prices.
The second axis is orthogonal to the first, and just as consequential: institutional separation. It helps to name two questions that are easy to blur together as one: assessment vs appraisal. Assessment is the technical evaluation — what does the evidence show? Appraisal is the value judgement — given that, what do we do? Pricing is a third question again: what gets paid?
Some systems keep all three in one body. NICE both assesses the evidence and appraises it against the threshold within a single committee process, with price entering mainly through the discount negotiated on top.
Other systems deliberately split the three into separate institutions, each with one job: in Germany, IQWiG assesses the evidence, the G-BA appraises it and decides the added-benefit rating, and the insurance funds (GKV-SV) negotiate the price — three organisations, three mandates, no overlap.
Neither design is simply "better." Bundling is faster and puts one committee's judgement fully in view. Separating adds friction and time, but keeps the technical question ("what does the evidence show?") insulated from the negotiating table — nobody grading the evidence has a stake in the price it leads to.
Run the drug through different systems.
Below is the drug from Screen 2 — fixed, identical, never changing. Tap through the four decision engines and watch the question each one asks, the verdict it reaches, and what that verdict actually decides. Same evidence in, every time. Watch what comes out.
The drug — identical in every tab
England (NICE-style)
QuestionDoes the incremental cost per QALY gained fall inside our funding threshold?
VerdictICER £40,000/QALY sits above the £20,000–£30,000 threshold.
DecidesNot recommended for routine funding at this price — unless a confidential discount brings the ICER inside the threshold.
Four taps, four fates: not recommended, reimbursed, reimbursed, and a non-binding opinion — from the exact same 0.4 QALYs and £16,000. None of the four engines is lying about the evidence. Each is applying a different test to it, built for a different health system with different priorities.
Now you.
Each description below matches exactly one of the five labels. Pick it.
1. One body assesses the clinical and economic evidence and tests whether the incremental cost per QALY gained falls inside an explicit funding threshold. That single test decides the outcome.
2. There's no cost-per-QALY threshold in sight. The question is how much better the drug is than the appropriate comparator, rated on a fixed scale — and the answer sets the price through negotiation.
3. Whether to fund the drug at all turns on whether its clinical benefit is judged sufficient; a separate rating of how much of an improvement it represents over existing care then sets the price tier.
4. The body that publishes the verdict has no authority to reimburse or refuse anything. It's an opinion that individual payers are free to use, adapt, or ignore.
5. The institute that evaluates the clinical evidence, the committee that decides whether to reimburse, and the body that negotiates what gets paid are three different organisations, each with one job.
A world tour: agencies as points on the axes.
With both axes in hand, the world's major agencies stop looking like an arbitrary list and start looking like points on a map.
- NICE (England) — the archetype of threshold-and-QALY, bundled: one body assesses, appraises against the threshold, and its recommendation is implemented directly by the NHS.
- IQWiG + G-BA (Germany) — no threshold, added-benefit assessment, and deliberately separated across three institutions, ending in a negotiated price with the insurance funds.
- HAS (France) — clinical benefit first: sufficiency for reimbursement (SMR) and degree of improvement (ASMR) are rated before any price conversation starts, with the ASMR level driving that negotiation.
- CDA-AMC (Canada) — threshold-and-QALY like NICE in method, but the recommendation feeds into a separate, joint price-negotiation body rather than being implemented directly.
- PBAC (Australia) — QALY-centric and, historically, the pioneer: Australia was the first country to make economic evaluation a formal requirement for reimbursement, back in 1993.
- The Institute for Clinical and Economic Review (US) — an independent, non-profit body with no decision-making power at all. It publishes value-based benchmark prices; individual US payers decide for themselves whether to use them.
- HITAP (Thailand) — a leading agency from a middle-income country, with a threshold and methods adapted to its own context, and wide influence on HTA across lower- and middle-income countries.
Notice how little the list has to do with wealth or clinical sophistication, and how much it has to do with each system asking a different question of the same evidence.
Why the differences exist (and why they matter).
These aren't random design quirks. They trace back to how each health system is financed and governed. A tax-funded national service with one payer, like the NHS, can run a single explicit threshold because there's one budget to protect and one body accountable for it. A social-insurance system with many competing sickness funds, like Germany's, has no single "the budget" to test a price against — so it rates benefit first and negotiates price as a separate, ongoing commercial process.
Political appetite for an explicit price on a life-year also differs sharply. Some systems are comfortable stating a number in public (NICE's threshold is published). Others are not, and route the same underlying trade-off through a rating scale and a private negotiation instead — the trade-off doesn't disappear, it just isn't stated as a number anyone can quote.
For anyone working across borders, the consequence is direct: the same global value dossier cannot simply be translated and resubmitted. A submission built to defend an ICER against a threshold is answering a question AMNOG never asks; a submission built around added-benefit categories is answering a question NICE never asks. Reading — and building — a dossier means first identifying which machine it's about to enter.
The other chair
If you sit on the submission side, this cuts both ways: the strongest honest case for a threshold-based agency foregrounds the ICER and the discount that closes any gap; the strongest honest case for AMNOG foregrounds comparator selection and the added-benefit evidence. Same drug, same trial data — a genuinely different argument, because the two engines are asking genuinely different questions.
What's the right explanation?
A drug has an ICER of £45,000 per QALY, based on the same clinical trial evidence submitted to both agencies. NICE (England) does not recommend it for routine funding. Germany's AMNOG system reimburses it from launch, at a negotiated price, based on that same evidence. What best explains the difference?
Why this matters for HTA
Here's where this lands on your desk: any time you read, compare, or write a verdict that crosses more than one country, you're implicitly comparing decision machines, not just evidence. A few disciplines follow.
- Before judging a verdict, name the axis. Ask whether the agency runs on a QALY threshold, an added-benefit rating, or a clinical-sufficiency test — and whether assessment, appraisal, and pricing sit in one body or three. The verdict only makes sense once you know which machine produced it.
- Don't read cross-country disagreement as a quality signal. A "no" in one country and a "yes" in another, on the same dossier, is not automatically evidence of a weak submission, inconsistent regulators, or manipulated analysis — it can simply be two different tests applied to the same input.
- Build the dossier for the machine, not just the disease. A value story optimised for a threshold-and-QALY agency will not automatically persuade an added-benefit agency, and vice versa — the emphasis, the comparator argument, and even which numbers are headline all need to shift.
The number an ICER produces is portable. The rule that turns that number into a verdict is not.
HTA agencies around the world, in one breath.
- An HTA agency's verdict depends on its institutional design as much as on the evidence — the same dossier can get opposite outcomes in different countries.
- Axis 1: does the agency test the ICER against an explicit QALY threshold (NICE), or does it skip the threshold entirely and rate added clinical benefit instead, negotiating price against that rating (AMNOG)?
- Axis 2: are assessment, appraisal, and pricing bundled into one body (NICE), or deliberately separated across independent institutions, each with one job (Germany's IQWiG / G-BA / GKV-SV)?
- The same £16,000-for-0.4-QALYs drug produced four different fates across four real-world-style engines — not recommended, reimbursed, reimbursed, and a non-binding opinion — from identical evidence.
The number an ICER produces is portable. The rule that turns that number into a verdict is not.
So far, every one of these engines has run entirely within its own national border — each country assessing the same dossier separately, from scratch. That's now changing: the EU's HTA Regulation introduces a Joint Clinical Assessment, a single shared clinical evaluation done once at EU level, while pricing and reimbursement stay firmly national. That reshaping of the map is next.