In February 2026, researchers at McGill University published a nanoplasmonic device that identifies a pathogen and determines its antibiotic susceptibility in 36 minutes from a urine sample. No overnight culture. No waiting. A chip, a camera, and a machine learning algorithm.
At roughly the same time, the WHO published its 2025 GLASS report confirming that across 14 African countries surveyed, only 1.3% of 50,000 medical laboratories can perform basic bacteriology testing — the foundational step before any susceptibility testing can even begin.
These two facts, held together, define the real crisis in antimicrobial resistance. It is not primarily a crisis of invention. It is a crisis of sight.
The Numbers We Don't Have
The WHO's Global Antimicrobial Resistance and Use Surveillance System — GLASS — is the most ambitious attempt ever made to measure resistance worldwide. By 2024, 127 countries had enrolled and 104 provided data covering over 23 million laboratory-confirmed infections. The headline numbers are grim enough: roughly 1 in 6 bacterial infections globally involves a resistant pathogen, rising to 1 in 3 for urinary tract infections.
But the real story is in what GLASS doesn't capture. Overall national data completeness scored just 53.8% among reporting countries. Large parts of sub-Saharan Africa, Central Asia, and Latin America report limited or no data. The countries with the weakest surveillance and lowest universal health coverage consistently report the highest apparent resistance levels — but even those numbers are likely underestimates, because the sickest patients in the most resource-limited settings never get cultured at all.
Sub-Saharan Africa bears the highest AMR mortality rate globally — 27.3 deaths per 100,000 — yet produces the least surveillance data. We are blindest where the killing is worst.
The MAAP project, mapping laboratory capacity across 14 African member states, found that only 5 of the 15 antibiotic-pathogen combinations prioritized by WHO are being consistently tested. Not because resistance doesn't exist — but because nobody is looking.
Europe's Split Screen
Where surveillance does exist, it tells a story of divergence. The European Centre for Disease Prevention and Control published its EARS-Net Annual Epidemiological Report covering 2024 data from all EU/EEA countries. The results read like a split screen: one success surrounded by accelerating failures.
Source: ECDC EARS-Net AER 2024. EU 2030 targets set against 2019 baselines.
Four of five EU antimicrobial resistance targets are off track. Only MRSA — the pathogen that received the most sustained public health attention over the past two decades — is declining. The rest are worsening. EMA and ECDC jointly called the situation urgent, estimating more than 35,000 deaths per year in the EU/EEA from AMR. The geographic gradient persists: southern, central, and eastern Europe bear the heaviest burden.
This matters beyond Europe because EARS-Net is the gold standard for AMR surveillance. If the region with the best data infrastructure in the world is failing on 80% of its own targets, the picture in regions with no data at all is almost certainly worse.
What 48 Hours Costs
Standard antimicrobial susceptibility testing — the process that tells a clinician which drug will work against a patient's infection — takes 48 to 72 hours. Three sequential steps: grow the bacteria from a clinical sample (up to 24 hours in blood culture, sometimes 5 days), identify the species (another 24 hours), then test susceptibility on pure colonies (4–24 more hours). During this entire wait, patients receive empirical broad-spectrum antibiotics. Sometimes the empirical choice works. Sometimes it doesn't. When it doesn't, people die. When it does work but was unnecessarily broad, it drives resistance.
This wait is the single most consequential bottleneck in infectious disease medicine. Everything downstream — treatment, stewardship, surveillance — depends on closing it.
A wave of technologies is trying to do exactly that. Over 90 rapid AST platforms are now in development or early commercialization. The most striking:
| Technology | Mechanism | Time to Result | Stage |
|---|---|---|---|
| QolorPhAST | Nanoplasmonic colorimetry in microfluidics | 36 min | Research (McGill, 54 samples validated) |
| Astek JIDDU | UTI identification + AST | ~1 hr | Beta prototype, clinical trials planned |
| SoundCell | Graphene nanodrum + ML nanomotion | 1–2 hr | Prototype (TU Delft, 98% precision) |
| Phenotech MultiStar | Fully automated phenotypic AST | 2 hr | Clinical trials planned 2026 |
| bioMérieux VITEK REVEAL | Automated rapid phenotypic | 3 hr | Commercial, deployed |
| Avails Medical eAST | Electrochemical phenotypic | ~5 hr | UK clinical trial started April 2025 |
| QuantaMatrix dRAST | Digital time-lapse imaging in agarose | ~6 hr | Commercial in 26 countries |
| Nanomotion + SepsiSTAT | Magnetic bead enrichment + nanomotion | 9–11 hr | Research (from spiked blood to AST) |
These are not theoretical. Several are in clinical use today. But the crucial caveat from the most comprehensive recent review (Nature Communications, 2024): rapid AST is unlikely to be feasible for all bacterial taxa or all antimicrobials. No single platform covers everything. Implementation demands laboratory expertise, workflow changes, and regulatory clearance — all of which take years. And phenotypic AST of any speed still has inherent imprecision of ±1 log2 dilution, which can push results across breakpoints.
The gap between invention and implementation is not months. It is decades.
The Resistance That Hides
Even when susceptibility testing is available and fast, it can still be wrong — because of a phenomenon called heteroresistance.
In heteroresistant infections, the majority of bacterial cells in a population are susceptible to an antibiotic, but a small subpopulation — sometimes as rare as 1 in 100,000 cells — carries resistance. Standard AST measures the bulk population and reports "susceptible." The clinician prescribes accordingly. The susceptible majority dies. The resistant minority survives, expands, and the patient relapses.
How common is this? Studies report heteroresistance in 15% to 97% of clinical isolates depending on species and antibiotic. A 2026 study applying the new Dilution-and-Delay (DnD) assay across ~120 clinical isolates of five ESKAPE pathogens found that 18.8% of all test results fell into a heteroresistant zone — neither cleanly susceptible nor fully resistant, but harboring subpopulations that standard methods miss entirely.
Standard AST workflow: isolate bacteria from patient → grow on media without antibiotic → test susceptibility. But when an isolate is grown without antibiotic pressure, the resistant subpopulation contracts back to baseline. The test sees susceptibility. The patient has resistance.
The DnD assay resolves this by detecting rare drug-insensitive cells at frequencies as low as 1 in 100 million. It simultaneously reports the conventional MIC and the hidden resistance frequency — two numbers instead of one. Validated against PAP-AUC gold standard across E. cloacae, E. coli, and K. pneumoniae.
This is not a theoretical concern. In Staphylococcus aureus bloodstream infections, heteroresistant VISA (hVISA) was found in 22% of MRSA BSI in an 842-patient cohort — associated with 2.5× vancomycin mortality and 5× relapse rates. For colistin — the last-resort drug for carbapenem-resistant Gram-negatives — heteroresistance rates of 24.4% have been documented in Enterobacter. These patients are being told their infections are treatable. Some of them are not.
The Sequencing Frontier
Genomic sequencing offers a fundamentally different approach to surveillance: instead of growing bacteria and testing them against antibiotics one at a time, sequence their DNA and read out all resistance genes at once. In theory, this transforms surveillance from slow, pathogen-by-pathogen testing to a comprehensive genetic census.
The technology is ready. Oxford Nanopore's portable sequencers can run in a field laboratory. Methylation signatures now allow researchers to attribute resistance genes to specific pathogen species without culture — a critical advance for settings where culturing is impossible. And the scale of genomic surveillance in Africa is expanding fast:
- SeqAfrica has generated 30,000 pathogen genomes across 21 African countries in four years, supported by the Fleming Fund. Regional sequencing hubs are now operational.
- Africa CDC Pathogen Genomics Initiative Phase 2 targets operational next-generation sequencing capacity at all 55 National Public Health Institutes.
- India wastewater genomics revealed a homogeneous resistome across the subcontinent — resistance genes in rivers matching those in hospital patients, confirming the environmental reservoir at national scale.
But genomic surveillance has a fundamental limitation that recent reviews emphasize: a resistance gene detected is not necessarily a resistance gene expressed. Molecular panels can only search for known targets, overestimate resistance when genes are silent, and miss novel mechanisms entirely. Phenotypic confirmation — actually testing whether the drug kills the bacteria — remains irreplaceable. The future is not genomics instead of phenotypic AST. It is genomics plus phenotypic AST, each covering the other's blind spots.
A Proof That Diagnostics Change Prescribing
If the argument for rapid diagnostics sounds abstract, Rwanda provides the concrete evidence.
The ePOCT+ digital diagnostic and clinical decision support tool was deployed across primary care facilities and subjected to a rigorous evaluation published in PLOS Medicine (March 2026). Before ePOCT+, 71% of consultations resulted in an antibiotic prescription. After deployment: 25%.
A two-thirds reduction in unnecessary antibiotic prescribing — in a low-resource setting — through a digital tool that guides clinical decision-making based on patient presentation and point-of-care tests. No expensive laboratory required. No sequencer. Just structured clinical reasoning supported by technology.
This is what diagnostics can do when deployed. Not in the future. Now.
The Stewardship Arithmetic
The WHO AWaRe classification divides antibiotics into three categories: Access (first-line, narrow-spectrum, low resistance risk), Watch (broader-spectrum, higher resistance risk), and Reserve (last-resort). The goal, raised by the UN General Assembly in September 2024, is for at least 70% of all human antibiotic consumption to be Access antibiotics by 2030 — up from the previous 60% target.
The current global average: ~52.7%. Only 14.1% of studies from low- and middle-income countries meet the 70% target. The UK committed to reaching 70% by 2029 in its national AMR action plan.
Stewardship — using the right antibiotic, at the right dose, for the right duration — is impossible without knowing what you're treating. Every hour a clinician waits for susceptibility results is an hour of empirical prescribing, which in practice means broad-spectrum Watch or Reserve drugs. The diagnostic gap and the stewardship gap are the same gap. Close one and you close the other.
What Is Actually Missing
Ninety-plus rapid AST technologies. Portable sequencers. AI-powered clinical decision support. Digital prescribing tools proven to cut unnecessary antibiotic use by two-thirds. The technology exists. What doesn't exist is the infrastructure, the funding, and the political will to deploy it where it matters most.
1.3% of African labs perform bacteriology
Only 5 of 15 WHO-priority pathogen-drug combos tested
53.8% global data completeness
Zero pediatric formulations in 17 sub-Saharan African countries
18.8% of AST results miss heteroresistance
90+ rapid AST platforms in development
36-minute specimen-to-result nanoplasmonic AST
30,000 African pathogen genomes sequenced
ePOCT+ cuts prescribing 71% → 25%
DnD assay detects resistance at 1-in-100-million frequency
The barrier, as LSHTM researchers wrote in March 2026, is no longer primarily technological. It is structural, political, and financial. WHO has set the target: 80% of countries should be able to test all GLASS priority pathogens by 2030. At current rates of laboratory capacity building, that target is fantasy.
What the AMR field needs now is not another diagnostic invention. It needs the political commitment to deploy the diagnostics that already exist — to close not the technology gap but the implementation gap. Every undiagnosed infection is a missed opportunity to use the right drug, to spare the broad-spectrum reserve, to contribute a data point to the surveillance systems that guide global policy. Every blind spot is a breeding ground.
We have built extraordinary eyes. Now we need the will to open them.
Previously: After Antibiotics: Engineering Biology to Fight Itself (Post #23). For the NDM surveillance crisis: The NDM Emergency. For the diagnostic landscape in depth: Seeing the Enemy: The Diagnostic Revolution. For Europe's hidden heteroresistance: The Immune Hacker (hVISA in MRSA).