Building global standards for antimicrobial policy

Antimicrobial resistance (AMR) threatens human, animal, and environmental health globally. An international team from leading institutions, including the University of Edinburgh, London School of Hygiene and Tropical Medicine, North Carolina State University, and the International Centre for Antimicrobial Resistance Solutions, identifies fundamental gaps in current mathematical modelling approaches that prevent translation of science into policy, including data limitations, knowledge gaps about AMU-AMR relationships, and the absence of international coordination mechanisms similar to climate change efforts. They call for transdisciplinary collaboration to build integrated modelling architectures.

AMR as a wicked problem

Despite growing evidence of global health and economic impacts, AMR fails to gain sufficient policy traction. Unlike acute crises, AMR impacts are cumulative and largely invisible. Rather than stemming from a single pathogen, AMR involves diverse microbes with varied implications-"bug-drug-context" combinations. This heterogeneity complicates communication and policy engagement.

An asymmetry in intervention costs and benefits creates inertia. While AMU reduction is policy cornerstone, global AMU estimates rely on multiple inference layers with often unavailable or biased baseline data.

Most critically, the AMU-AMR relationship remains poorly understood across different bug-drug combinations. Resistance is shaped by both AMU and co-selection through biocides, metals, and environmental factors. The degree to which livestock and aquaculture AMU causes human AMR remains contested yet central to policy decisions. Environmental sectors, despite their significance to AMR transmission, are frequently overlooked.

Critical deficiencies in current models

Analysis of 273 population-level models revealed alarming gaps: 89% considered only humans, 7% included animals, 2% included plants, and zero integrated all three sectors. Only 9% included economic cost-benefit analysis. Additionally, 40% of models included no sensitivity or uncertainty analysis, and none met TRACE modelling guidelines established in 2010.

The modelling hierarchy problem

Mathematical models classify hierarchically from theoretical models through fitted models with internal validity, external validation with independent datasets, to multi-model comparisons. Current AMR efforts predominantly remain at lower levels. External validation is a critical barrier due to limited independent data. Multi-model comparisons, successful for COVID-19, are currently unfeasible for AMR due to heterogeneity and lack of comparable models.

Learning from climate change

The research team proposes framing AMR as environmental pollution rather than purely medical, allowing lessons from climate change mitigation. Climate science developed abatement cost curves guiding policy across sectors and the "social cost of carbon" metric for cost-benefit decisions, coordinated through the IPCC.

By contrast, AMR policy is enacted without economic efficiency evidence. Even for certain AMU interventions, relative cost-effectiveness in different settings lacks clarity. Most critically, no coherent integrated modelling architecture exists for national or international cost-benefit framing of AMR. This gap represents an important opportunity for the Independent Panel on Evidence for Action against Antimicrobial Resistance (IPEA), currently under negotiation by the UN Quadripartite Group on AMR.

Path forward

Transdisciplinary and international modelling collaborations are essential. Data harmonization across phenotypic, genetic, whole-genome, and metagenomic methods for measuring AMR remains challenging. Due to surveillance program structures, human infection samples are overrepresented while environmental data is scarce. Digital One Health frameworks represent one approach to maximize surveillance efficiency. Scientific publishing must ensure data and code transparency for reproducibility.

Conclusion

AMR as a "largely invisible pandemic" challenges policy responses. Integrated AMR modelling remains in its infancy with significant knowledge and data gaps. Consequently, evidence is unconvincing to mobilize political attention and resources. The research team concludes that a concerted transdisciplinary modelling effort is required to generate political will to manage this planetary health crisis.