Evidence Architecture, AMNOG 2.0, and the case for a more intelligent European approach to evidence, value, and negotiation.

On 14 October 2026, I had the privilege of contributing to a parliamentary evening at the German Bundestag, organised by ATSE, BPI and vfa, on the future of rare disease innovation. Brilliantly moderated by the wonderful Inga Bergen, our discussion explored the increasingly difficult balance between encouraging medical progress, ensuring patient access and maintaining the financial sustainability of healthcare systems.

The conversations afterwards, particularly with patient representatives and industry colleagues, were encouraging. They reinforced a conviction I have developed over many years working at the intersection of clinical medicine, health economics and market access: we are spending an extraordinary amount of effort assessing evidence that we should have designed differently in the first place.

The timing was remarkable. On the very same day, the US FDA announced its Rare Disease Framework Initiative, setting out an ambition to tailor regulatory expectations more closely to the scientific realities of rare diseases. It is an initiative under development, not a completed regulatory overhaul. Meanwhile, Germany is debating AMNOG 2.0, including a more adaptive approach to benefit assessment.

These are distinct regulatory initiatives, addressing different decisions. Yet both point towards a fundamental challenge: how should healthcare systems deal with scientific uncertainty when conventional evidence-generation models reach their limits?

The real scarcity in rare diseases

Rare diseases confront us with an uncomfortable reality. Patient populations are small, clinical courses may be highly heterogeneous, relevant outcomes can take years to emerge, and conventional comparative studies are sometimes exceptionally difficult to conduct. These constraints have long been recognised in the EMA’s guidance on clinical trials in small populations.

Yet the most consequential scarcity may not be financial. It is the opportunity to learn. Every patient enrolled in a poorly designed study, every missed opportunity to establish natural-history data, and every clinical endpoint that ultimately fails to address a relevant therapeutic question represents knowledge that may never be recovered.

Of course, some uncertainty is inherent to the biology and epidemiology of rare diseases. Other uncertainty is reducible through better study designs and systematic data collection. And some uncertainty is created unnecessarily by fragmented regulatory requirements, inconsistent methodological expectations and decisions taken too late in the development process.

These distinctions matter enormously. Treating all uncertainty alike creates the wrong incentives, both for pharmaceutical innovation and for those responsible for allocating limited healthcare resources.

We are pricing uncertainty today that we should have started managing years ago.

From Evidence Assessment to Evidence Architecture

I strongly welcome the fundamental direction of the German Ministry of Health’s AMNOG 2.0 proposals: an initial assessment based on the best available evidence, followed by reassessment as knowledge evolves, represents a welcome departure from the idea that everything worth knowing must already be established at market entry.

But an adaptive assessment system will only be as good as the evidence-development strategy supporting it. A second assessment after three or four years will not magically produce the answers that nobody prepared to investigate. Indeed, the IQWiG’s critique of the proposed approach rightly draws attention to the need for prospectively planned, fit-for-purpose data.

This is why I believe we need to move beyond evidence assessment towards Evidence Architecture. By this I mean the deliberate design of an integrated evidence system, starting early in clinical development and extending through regulatory approval, health technology assessment, reimbursement and subsequent clinical use.

Rather than treating these stages as separate hurdles, Evidence Architecture starts with the decisions that will ultimately need to be made. What constitutes meaningful therapeutic improvement? Which comparisons are clinically and methodologically appropriate? Which uncertainties can realistically be resolved before approval, and which require prospective follow-up? How should patient experience, disease progression and long-term outcomes be incorporated?

The answers must inform clinical study designs, natural-history research, disease registries, endpoint development and plans for real-world evidence generation. Existing instruments, particularly European Joint Scientific Consultations and the EMA’s guidance on registry-based studies, provide important building blocks.

Importantly, this is not an argument for replacing rigorous comparative studies with whatever data happen to be available. When randomisation is feasible and appropriate, it remains a powerful instrument. Where it is not, we need scientifically credible alternatives, transparent treatment of bias and an explicit account of residual uncertainty.

Evidence Architecture is not about lowering scientific standards. It is about making scientific standards achievable, meaningful and decision-relevant under the conditions of rare diseases.

It also changes the role of patients. Rather than contributing their perspectives only when an assessment is already under way, they should help determine which questions are worth asking and which outcomes genuinely matter.

Value is more than what our current methodologies capture

A more sophisticated approach to evidence necessarily leads to another question: are we adequately capturing the value of medical innovation?

Traditional comparative benefit assessment rightly emphasises demonstrable effects on patient-relevant outcomes. But therapeutic value may have dimensions that are not fully reflected in the conventional assessment framework.

The ISPOR Value Flower offers an intellectually useful starting point. Disease severity, equity, the value of hope, insurance value and the broader consequences of scientific progress are among the dimensions that deserve consideration. The effects of disease and treatment on families and caregivers can also be substantial.

This does not mean that every conceivable dimension should automatically be translated into a higher reimbursement price. Nor should we confuse evidence of comparative clinical benefit with a broader judgement about social and economic value. On the contrary, distinguishing these questions more carefully would improve the quality of our decisions.

Health economics is fundamentally concerned with allocating scarce resources in ways that generate the greatest possible benefit, while recognising that efficiency is not society’s only legitimate objective. Severity, fairness, unmet medical need and future innovation also matter. And our methods inevitably influence the innovation that follows.

What we choose to measure determines what we reward. What we reward influences which innovations patients will see tomorrow. This is why methodological modernisation is not an academic exercise. It is an essential component of innovation policy.

Why I trust negotiation more than algorithms

Once we acknowledge that therapeutic value has multiple dimensions, and that evidence evolves over time, the economic question becomes unavoidable: how do we translate these considerations into a fair and sustainable reimbursement price?

My position is straightforward. I believe in negotiation.

Payers understandably value predictability, consistency and cost control. Explicit rules and quantitative methods have an important role in achieving these objectives. But increasingly prescriptive pricing algorithms risk creating an illusion of objectivity. An algorithm cannot eliminate difficult judgements about value, uncertainty, affordability and fairness. It merely incorporates them into parameters that may become less transparent than the decisions they are intended to replace.

One of AMNOG’s institutional strengths has been its combination of evidence-based benefit assessment and negotiated reimbursement. Rather than continuously narrowing this negotiating space, we should strengthen the conditions under which meaningful agreements can be reached.

Germany’s framework agreement under Section 130b SGB V provides an institutional basis that could be developed further: with clearer provisions for evidence uncertainty, differentiated contractual arrangements, subsequent reassessment, price adaptation and, where appropriate, risk-sharing. Such flexibility must be accompanied by transparent principles, credible evidence and enforceable commitments. Negotiation should never become an excuse for arbitrary prices or for avoiding accountability.

But different therapeutic situations legitimately require different solutions. An ultra-rare genetic disease with no treatment alternatives presents different challenges from a condition with several established therapies and substantial comparative evidence. The alternative to a pricing algorithm is not arbitrariness. It is intelligent, well-governed negotiation. And the ability to negotiate responsibly becomes even more valuable when the evidence itself is expected to evolve.

Europe cannot afford fragmented thinking

The international context adds considerable urgency to this discussion: the Trump administration presents Europe with a striking combination of signals. On the regulatory side, the FDA is exploring more tailored approaches to rare-disease development. On the economic side, the administration’s Most-Favoured-Nation pricing agenda has introduced substantial uncertainty into international pharmaceutical pricing and launch strategies.

The connection matters. When a reimbursement decision in one European country can influence pricing expectations in the much larger US market, national decisions acquire international consequences. Reuters has reported examples of manufacturers delaying European launches amid uncertainty about US pricing policies, although the longer-term and Europe-wide effects remain to be established.

This does not mean that European healthcare systems should abandon affordability objectives or simply accept higher prices. It means that national reimbursement policies can no longer be understood solely through their domestic budgetary effects.

Europe needs greater strategic coherence.

The extension of EU Joint Clinical Assessments to orphan medicines in January 2028 offers an important opportunity. But European HTA must become more than a mechanism for producing common clinical assessment reports. We should be building a genuinely European infrastructure for learning: interoperable disease registries, stronger cross-border natural-history cohorts, coordinated scientific advice, more consistent evidence expectations and meaningful involvement of European patient communities.

These are particularly important for rare diseases. Europe collectively offers possibilities for generating clinical knowledge that individual national populations often cannot provide. National healthcare systems will retain responsibility for assessing their specific needs, negotiating reimbursement and managing their budgets. That distinction is important. European cooperation in evidence generation does not require uniform European prices.

But 27 national reimbursement systems should not mean 27 disconnected approaches to scientific learning. And in an international environment in which regulatory and pricing strategies are increasingly interdependent, Europe needs to recognise the strategic significance of its collective market.

A different ambition for AMNOG 2.0

At the end of our parliamentary discussion, Inga Bergen challenged me to summarise the essential priorities in a single sentence.

My answer was to modernise the methodology of benefit assessment, broaden our understanding of patient-relevant value, strengthen negotiation as a mechanism for reconciling competing interests, and think European rather than becoming trapped in national regulatory detail. I remain convinced that these priorities belong together.

Evidence Architecture can help us generate more meaningful knowledge. Better methodologies can help us understand therapeutic value more comprehensively. Well-governed negotiations can translate that knowledge into fairer economic arrangements. And European cooperation can create the scale, consistency and scientific infrastructure that rare-disease innovation desperately needs.

The proposed direction of AMNOG 2.0 is encouraging. The FDA’s latest initiative suggests that the wider international debate is moving as well. Europe should seize this opportunity, not simply to reform another assessment procedure, but to reconsider how we generate knowledge, recognise value and reward innovation. The future of rare-disease innovation will not be determined solely by how rigorously we assess yesterday’s evidence. It will depend on how intelligently we architect tomorrow’s.

My sincere thanks to ATSE, BPI and vfa for the invitation, to Inga Bergen for her excellent moderation, and particularly to the patient representatives and industry colleagues for the stimulating exchange of perspectives.