Healthcare has long operated under a clear and sensible principle: any new diagnostic test must demonstrate performance that is equivalent to or better than the existing gold standard. In many contexts, this is not only appropriate but essential. We would not, for example, accept a new HIV test that sacrificed accuracy for speed or convenience. Precision matters when the consequences of error are severe.

Yet the application of this paradigm becomes problematic in areas where clinicians have few useful diagnostics to use at the point of care.

This is precisely the situation faced by primary care clinicians managing suspected infections, particularly upper respiratory tract infections (URTI) and urinary tract infections (UTI) – the two most common reasons for antibiotic prescribing. At present, clinicians are largely unable to rapidly differentiate between bacterial and non-bacterial infections, or to determine which antibiotics are likely to be effective. As a result, prescribing decisions for URTIs and UTIs are often made blind, though the medical community prefers the term ‘empiric’ even though it means the same thing. There’s no avoiding the fact most antibiotics prescribed in primary care are selected on the basis of clinical experience, population-level guidance and risk aversion rather than patient-specific data.

The gold standard for infection diagnosis – laboratory-based bacterial identification followed by minimum inhibitory concentration (MIC) determination – requires time, infrastructure and expertise that are simply not available in primary care. A full microbiology work-up typically takes 48 hours or more – well beyond the timeframe in which prescribing decisions must be made. Yet because this gold standard is so deeply embedded in regulatory frameworks and clinical thinking, any new diagnostic that cannot replicate this full dataset is deemed inadequate.

This creates a paradox.

Rapid diagnostics that could provide partial but clinically meaningful information – such as indicating whether an infection is likely bacterial, or suggesting whether antibiotics are likely to be beneficial at all – are held to a standard that is effectively unattainable in a point-of-care setting. The result is that potentially useful tools are excluded, not because they fail to improve decision-making, but because they fail to mirror a laboratory process designed for a fundamentally different context.

From a practical perspective, this rigid adherence to equivalence may be counterproductive. Some information, delivered at the right time, may be far better than no information at all. A test that offers a strong steer, even if not a definitive answer, could meaningfully reduce unnecessary antibiotic prescribing, improve patient care and support stewardship efforts. Yet current evaluation frameworks struggle to accommodate this kind of value proposition.

This tension presents challenges across the ecosystem. Developers of new diagnostics are constrained in what they can realistically build. Regulators face difficulty approving products that fall short of entrenched definitions of diagnostic completeness. Health systems and researchers struggle to design studies that capture the real-world impact of partial information on prescribing behaviour and outcomes.

If we are serious about tackling AMR, this is a moment that calls for clear thinking, creativity and openness to new approaches. The fight against resistance will not be won solely through perfect information delivered too late, but through better decisions made earlier, under real-world constraints. That will require regulators, clinicians, developers and policymakers to reconsider how diagnostic value is defined, and to recognise that in primary care, progress may depend not on replacing the gold standard, but on pragmatically complementing it.

In short, antimicrobial stewardship cannot advance if innovation is judged only by its ability to replicate laboratory certainty. Sometimes, the most responsible step forward is enabling clinicians to see a little more clearly, rather than insisting they see everything at once.

Expert authors