What a traceable AMR funding chain looks like.

AMR financing analysis often concentrates on broken evidence chains. That is necessary work. In many country and portfolio contexts, the analyst can see a policy priority, find a relevant health investment, and still be unable to defend the connection between the two.

But a second question matters just as much: what does a comparatively traceable AMR funding chain look like?

Switzerland's National Research Programme 72, Antimicrobial Resistance, is a useful positive control. It does not solve every attribution problem. It does not prove national resistance outcomes. What it does show is how much easier reconstruction becomes when capital is designed around a clear public mandate.

The visible chain.

The Swiss National Science Foundation states that on 24 June 2015 the Federal Council approved the National Research Programme "Antimicrobial Resistance" and mandated the SNSF to conduct it. The programme had a holistic One Health orientation and an overall funding amount of CHF 20 million.

The NRP 72 website records that the programme ran from June 2015 to January 2023 and now points visitors to programme findings, publications and thematic syntheses. Switzerland's StAR platform describes the closing phase as a process in which researchers presented findings and recommendations to representatives from practice and policymaking.

That gives a clean public chain:

For AMR capital analysis, that is unusually valuable. The programme label, mandate, funding envelope and research outputs point in the same direction. The analyst does not have to infer AMR relevance from a generic health-system project title or from a broad laboratory investment that may serve multiple objectives.

Why this matters for funders.

Some evidence gaps are not created by bad analysis. They are created at programme design stage.

If an AMR-relevant component is buried inside a larger health-security investment without a budget line, component description, implementation marker or output logic, the future evidence chain will be weak before anyone starts analysing it. A donor may have funded meaningful AMR-enabling work, but the public record may not allow that work to be reconstructed later.

NRP 72 shows the opposite case. The mandate was explicit. The envelope was visible. The programme had a coherent AMR theme. The research projects were identifiable. The programme produced synthesis work that grouped findings into a broader interpretation.

That does not make the programme perfect as an evidence object. It does make it reconstructable.

Traceability is not causality.

The most important analytical boundary is also the easiest to miss.

NRP 72 supports a defensible chain from AMR mandate to programme capital and research outputs. It does not automatically support a chain from CHF 20 million to a measurable change in national AMR resistance rates.

Those are different questions. A research programme can generate findings, tools, methods, recommendations and policy input. Whether those outputs changed prescribing behaviour, surveillance practice, infection outcomes or resistance trends requires additional evidence. It would require implementation pathways, time ordering, adoption evidence, exposure mechanisms and outcome measurement.

StAR's account of the NRP 72 closing conference reflects this distinction. Scientific innovation can provide instruments and a basis for decision-making, but real-world impact depends on commitment and coordination across multiple social sectors.

The useful separation.

AMR funding analysis becomes more reliable when it separates three questions:

NRP 72 performs strongly on the first two questions. The mandate and programme envelope make the AMR attribution of the research programme itself much more defensible than in cases where AMR relevance is only one possible interpretation of a broader health investment.

The third question remains separate. A programme can be traceable without proving population-level impact. That is not a weakness. It is the right boundary.

A positive control for AMR capital intelligence.

In evidence modelling, positive controls are useful because they show what the system should be able to recognize when the signal is present. NRP 72 plays that role for AMR capital intelligence.

If an intelligence layer cannot reconstruct a chain as explicit as NRP 72, the model is not yet good enough. But if it turns NRP 72 into an overclaim about national resistance outcomes, the model is also not good enough.

The test is discipline in both directions: recognize the strong chain, and preserve the boundary where the evidence stops.

What funders can design into future programmes.

NRP 72 suggests several practical design principles for AMR funders and research-programme owners.

First, make the AMR mandate explicit. A future analyst should not have to infer AMR relevance from adjacent terms such as laboratories, surveillance or health security.

Second, expose the funding envelope and the programme logic. If there are multiple components, the record should show what each component is intended to do.

Third, preserve project-level identifiers. Capital becomes much more useful when it can be connected to specific projects, teams, work packages, publications and synthesis outputs.

Fourth, separate outputs from outcomes. A publication, prototype, diagnostic method or surveillance recommendation is an output. A measurable change in resistance, prescribing behaviour or system performance is an outcome. Both matter, but they are not the same evidence object.

Finally, record the pathway from research into practice. If scientific work influences policy, surveillance systems or implementation guidance, the adoption pathway should become visible rather than assumed.

The broader point.

AMR capital does not only need more money. It needs money that can be interpreted later.

When capital is visible but its AMR component is not isolatable, analysts are pushed into weak inference. When implementation is reported without clear links to financing, the chain breaks. When research produces outputs without an adoption pathway, outcome attribution remains unresolved.

NRP 72 does not remove those problems for the whole AMR system. It gives a concrete example of how a cleaner chain can look.

That is the standard AMR financing should move toward: not the largest possible attribution number, but the most defensible one.

Image: AgentBrain-generated editorial infographic for this article. It is an abstract representation of capital traceability, not an official NRP 72 diagram.