Two patients on the same ward test positive for the same organism in the same week. Are those infections from the same source in the hospital or a coincidence that got tracked in from outside? Most hospitals can’t definitively answer that question, even though the technology that’s able to has existed for over a decade.
Hospitals and the people who work in them are at capacity, and the clinical laboratory workflow reflects what a clinician needs—identifying what this organism is and what will kill it. It was built to treat the immediate needs of a patient, not to work out where their infection came from. An outbreak only gets noticed when it rises above the background noise—an exotic organism spreading or a pattern of infections that has already caused real harm.
To make matters worse, antimicrobial resistance is increasing, making many of these outbreaks harder to treat. Between 2019 and 2023, carbapenemase-producing Enterobacterales—a type of bacteria that’s resistant to certain last-line antibiotics—rose 69 percent across 29 states.1 Candida auris, a fungal infection with relatively more intrinsic resistance, increased from 766 US cases in 2020 to more than 6,304 in 2025. However, as resistance rates increase over time, the cost of DNA sequencing decreases, providing a tool to fight back.
Identity is Not Linkage
The resolution of standard diagnostics is capped at the species level. An identity and antibiotic susceptibility profile can determine what an organism is and what should work to treat it, but it can’t say where it originated or how it’s related to other infections. Without that relatedness, scientists can only count similar infections above an expected baseline and risk missing true links or overpredicting false ones.
Relatedness requires the genome: If two pathogen genomes differ at only a handful of sites, there must not have been much time for those differences to accrue. In other words, the ancestor of those genomes must be very recent, meaning those two infections were likely seeded by the same source. Alternatively, thousands of differences point to a more distant ancestor and independent colonizations.
Without the genome, counting alone can fail in both directions. Two patients are colonized in the emergency department and then move to different units, where their infections surface weeks apart. Ward counts increment independently, and transmission gets missed.
The reverse also happens. Several cases of the same species with the same susceptibility profile cluster on one unit, an investigation opens, and contact precautions go up. But the genomes are unrelated, the sources are separate, and the transmission was assumed rather than real.
Genomic linkage corrects both. It draws connections nobody could have seen, and it exonerates coincidences that otherwise may look like clear transmission.
The Forensic Era
In 2011, whole-genome sequencing by Snitkin et al. reconstructed the transmission map of a carbapenem-resistant Klebsiella pneumoniae outbreak at the NIH Clinical Center that affected 18 patients, 11 of whom died.2 The responsible organism was a type of carbapenemase-producing Enterobacterales mentioned earlier.
That retrospective analysis established both the method's power and how it would initially be used. Sequencing became epidemiology's forensics lab, reached for only after an outbreak was already obvious. It stayed there because sequencing was expensive, and expensive tests get rationed to the cases that clearly need them most.
The Economics of Continuous Surveillance
When a measurement gets cheap enough, it changes category from a test one orders to a signal that’s monitored. A continuous glucose monitor is no more accurate than a blood draw. It just tells you that something is wrong sooner, before the symptoms hit.
Since that initial investigation by Snitkin et al., sequencing costs have fallen enormously. For routine isolate sequencing, though, the constraint has been per-sample economics rather than the headline cost per base. Those economics turn out to be a question of volume. In other words, the more isolates get sequenced at a time, the faster the costs fall on a per-unit basis.
Sequencing is also only part of the cost. Linkage requires someone to judge what a given number of differences means in that ward, on that timeline, or for that organism, and that expertise is a fixed cost that a thin isolate stream doesn’t justify.
This work has been operationalized, however. Foodborne pathogen surveillance has already made this exact transition. FDA and CDC genome-based tracking of foodborne pathogens yielded an estimated ~$500 million in annual returns by 2019, against ~$22 million in annual investment.3 Part of that investment was on sequencing over 100,000 pathogen genomes. The thing keeping this same pivot from being widely adopted in hospitals is the siloed nature and insufficient volume at individual institutions.
What Continuous Surveillance Looks Like in Practice
A small number of centers clear that volume bar on their own, and they are where the first evidence comes from. Researchers at UPMC Presbyterian sequenced clinical isolates as routine practice rather than as investigation.4 Their efforts surfaced 172 outbreaks spanning 476 isolated pathogens, a large proportion of which would have been missed entirely by standard surveillance. They caught clusters at two patients that might otherwise not have surfaced until after one year and many more infections. Over the two years, they estimated 62 infections and roughly five deaths avoided, with a 3.2 times return on investment. UPMC's isolate stream is unusual, but nothing about the result depends on it being theirs. Their ability to conduct this continuous surveillance was bought by volume, which aggregation can buy for everyone else.
Importantly, whole-genome sequencing doesn’t replace traditional epidemiology or infection prevention. Rather, it supports them with the highest possible quality of data and strongest confidence in the links it surfaces. A genome can determine if two infections share a source; it cannot walk the ward, find the sink, or change the practice that put it there. That work remains where it has always been.
From Counting to Knowing
Antibiotics helped add decades to modern life expectancy, and infection control helps keep that gain intact.5 Hospitals are where the most drug-resistant organisms are under the highest active selection and closest to the patients least able to survive them. Containing them there is one of the strongest levers we have, and we now have the tools to do it better. The source of the mystery infection of two patients in one ward in one week is now an answerable question.
- Rankin DA, et al. Changes in carbapenemase-producing carbapenem-resistant Enterobacterales, 2019 to 2023. Ann Intern Med. 2025;178(12):1818-1821.
- Snitkin ES, et al. Tracking a hospital outbreak of carbapenem-resistant Klebsiella pneumoniae with whole-genome sequencing. Sci Transl Med. 2012;4(148):148ra116.
- Brown B, et al. An economic evaluation of the whole genome sequencing source tracking program in the U.S. PLoS One. 2021;16(10):e0258262.
- Sundermann AJ, et al. Real-time genomic surveillance for enhanced healthcare outbreak detection and control: Clinical and economic impact. Clin Infect Dis. 2026;82(1):134-141.
- Hutchings MI, et al. Antibiotics: Past, present and future. Curr Opin Microbiol. 2019;51:72-80.
