WHO Urges Stronger AMR Surveillance as Hospital Data Leave Major Gaps
Antimicrobial resistance is already a top-tier public health threat, but the measurement problem is nearly as urgent as the biology. Reliable surveillance still does not exist at the scale needed to define where resistant infections are occurring, how often they strike, and whether health systems are catching them early enough to change outcomes. In a commentary published in PLoS Medicine, WHO-affiliated authors argue that the field cannot wait for perfect data before acting; it needs better, standardized data now, because fragmented reporting is blunting public health response.
Balasubramanian and colleagues are cited as providing one of the first global incidence estimates of hospital-associated resistant infections, with roughly 136 million hospital-acquired resistant infections annually worldwide between 2010 and 2020. The scale alone is sobering. Yet the estimate also exposes the limits of the current evidence base, because much of what is known about resistant infection burden still comes from sparse hospital point prevalence surveys, uneven microbiology access, and inconsistent reporting practices across countries and health systems.
A single-country estimate can be persuasive. A global estimate built on thin surveillance is harder to trust.
The Data Problem Behind the Resistance Crisis
The commentary makes a simple clinical point: if the denominator is unstable, the burden estimate is unstable. Hospital point prevalence surveys can identify resistant pathogens in inpatient settings, but they are not designed to fully capture national incidence over time. Their usefulness depends on who was tested, which pathogens were included, which antibiotic classes were measured, and whether the participating hospitals resemble the broader country-level case mix. In low- and middle-income countries, only 11 nations reportedly contribute this kind of information, often from small samples that leave wide room for bias.
That matters because AMR is not evenly distributed. It tracks with hospital access, diagnostic capacity, antibiotic exposure, infection prevention quality, and the quality of laboratory systems. When a surveillance network is built mostly from hospitals that already have microbiology services, it may overrepresent settings that are better equipped to detect resistance and underrepresent the patients who never reach testing at all. The result is a system that can confuse visibility with prevalence.
Hospital-based data also struggle with clinical granularity. The estimates discussed here aggregate infections rather than separating them into syndromes, so bloodstream infection, urinary infection, and other clinical presentations are folded into a broad count. That reduces the ability to target syndrome-specific interventions, such as line care in critical care units, urinary catheter stewardship, or pathogen-specific containment strategies.
Why Standardization Matters More Than Ever
The commentary places the new estimates in the context of other burden efforts, including WHO’s GLASS system and the Global Burden of Disease analyses. Each framework uses different bacteria, antibiotic classes, and infectious syndromes. Balasubramanian and colleagues included six bacteria and 13 antibiotic classes across all infectious syndromes, while GLASS currently recommends 11 antibiotic classes, eight bacteria, and specific specimen types for routine reporting. The Global Burden of Disease estimates go even broader, with 12 infectious syndromes, 18 antibiotic classes, and 23 bacteria.
This is not a technical quibble. Different surveillance definitions produce different national and global pictures of the same biological threat. Without shared core definitions, comparisons between countries can become little more than comparisons of methods.
The authors argue that standardization should focus on a common core of infectious syndromes, pathogens, and antibiotic classes, while still allowing local adaptation where needed. That approach would make routine surveillance more comparable across regions and more useful for policy. It would also help interpret trends over time, which is essential if health systems are trying to judge whether stewardship programs, infection control measures, or diagnostic improvements are altering the burden of resistance.
Passive Reporting Alone Will Not Close the Gap
WHO’s proposed approach is two-pronged: strengthen and standardize routine passive surveillance, then supplement it with periodic nationally representative surveys. Passive surveillance can deliver continuous reporting from clinical samples and has obvious value for real-time detection of outbreaks, resistance trends, and treatment guidance. But the commentary is blunt about its limits when coverage is weak. In many settings, laboratory access is incomplete, patients cannot afford testing, and routine quality assurance systems are inconsistent or absent.
The laboratory gap is especially visible in the figure discussed in the article, which shows bacteriologically confirmed bloodstream infections reported to GLASS in 2020. Even for one of the most dangerous infectious syndromes, reporting coverage varied substantially between countries. Some nations could not report AMR data at all during the COVID-19 pandemic, which is a reminder that surveillance systems built on fragile infrastructure can fail precisely when health systems are under the most strain.
Bloodstream infection surveillance is clinically important because it captures some of the sickest patients and some of the most consequential resistance events. But if a country can only measure a fraction of those cases, treatment policy becomes reactive rather than evidence-based.
Periodic National Surveys Offer a More Reliable Anchor
The strongest alternative in the commentary is the periodic nationally representative AMR prevalence survey. Unlike passive reporting, these surveys use probability sampling to select sites and apply standardized methods to estimate prevalence at the population level. The design is meant to create data that are both representative and comparable, not merely convenient. Surveys also build national technical capacity, which can improve patient management and support the longer-term development of stronger continuous surveillance systems.
WHO guidance initially focuses on bacterial bloodstream infections among people needing acute hospital inpatient care, with the intention of expanding to other syndromes. That choice is clinically logical. Bloodstream infection is a high-stakes endpoint, and it can be linked to mortality, treatment failure, and severe illness with relatively direct clinical significance. The survey platform also allows investigators to follow the patient pathway more explicitly, from incidence and diagnosis to treatment and cure.
A few surveillance systems measure the whole cascade. Most do not.
That cascade matters because resistance is not only about which organism is present in the lab. It is also about whether the patient was diagnosed, whether treatment was started, whether effective therapy reached the bedside in time, and whether the infection was ultimately cured. By pairing AMR prevalence surveys with WHO methods for attributable mortality estimation, countries can begin to understand not just how much resistance exists, but how much harm it causes.
What the Field Still Needs Before Policy Can Keep Up
The article closes on a practical warning: existing AMR estimates are useful, but they are built on fragmented data systems that still lean too heavily on high-income-country studies and incomparable methods. Mathematical modelling can speed the production of global estimates, but it cannot rescue poor inputs. Sentinel systems and lot quality assurance sampling may help in some settings, yet they do not replace nationally representative data when the goal is to estimate population burden and track trends over time.
For clinicians, the immediate implication is straightforward. Treatment guidance, stewardship policies, and infection control programs are only as strong as the surveillance architecture beneath them. For public health agencies, the challenge is larger: build systems that are continuous where possible, periodically representative where necessary, and standardized enough to make cross-country comparison meaningful. Without that combination, the world will continue to count resistant infections imprecisely while they continue to rise.