Researchers at the University of Pittsburgh have been deploying real-time whole-genome sequencing (WGS) to help monitor and curb healthcare-associated infections (HAIs) within the university hospital.

Presenting at this year’s American Society for Microbiology Conference on Rapid Applied Microbial Next-Generation Sequencing and Bioinformatic Pipelines (ASM NGS) on Monday, Alexander Sundermann, an infectious disease professor at the University of Pittsburgh, showed preliminary data that demonstrates the feasibility of real-time WGS surveillance and its potential utility in containing hospital outbreaks.

Healthcare-associated infections are “unfortunately common, costly, and deadly,” said Sundermann, adding that in an inpatient setting, up to 10 percent of patients in US hospitals will develop HAIs.

The traditional approach of identifying hospital-associated outbreaks can be somewhat arduous, often including chart reviews, staff interviews, and a series of other procedures. After that, hospitals frequently deploy “a kitchen sink of interventions” to stop the outbreak, Sundermann said, and WGS might not be used until the end to help confirm or refute the suspected HAI and its transmission — an approach that researchers have termed “reactive WGS.”

Meanwhile, genomic surveillance, which entails routine sequencing of pathogen isolates regardless of the presence of an outbreak, has become an emerging approach for infection prevention, he noted.

Previously, Sundermann and his colleagues developed a WGS- and machine learning-based method for identifying and preventing HAIs dubbed Enhanced Detection System for Healthcare-Associated Transmission (EDS-HAT).

As a proof of concept, the researchers applied EDS-HAT to a retrospective genomic surveillance study and sequenced patient samples from suspected healthcare-related transmissions over a two-year period. They clustered the sample isolates using a threshold of 15 SNPs for all target pathogens except Clostridioides difficile, which was analyzed with a two-SNP threshold.

The results from the pilot showed that out of over 2,750 unique patient isolates, 297 samples were genetically associated with around 100 clusters. In comparison, the traditional infection prevention approach only requested performing sequencing on 15 outbreaks involving roughly 130 patients, Sundermann said. Merely five of these sequenced isolates were genetically related, indicating that the conventional infection prevention scheme had missed and misidentified transmissions.

From a cost-effectiveness standpoint, Sundermann said, using Pseudomonas aeruginosa as an example, the cost for a hospital to treat a healthcare-related infection is almost $25,000, while the cost of sequencing one isolate is about $80. “If you prevent one Pseudomonas aeruginosa infection, you actually save a lot of money in the long term,” he said.

Building on the retrospective study, Sundermann’s team sought to test the utility of EDS-HAT for real-time WGS surveillance. The group applied the method to culture-positive pathogen samples from patients suspected of HAIs at the university hospital. After performing WGS on the isolates using the Illumina NextSeq platform, the researchers analyzed the sequencing data using a custom-developed bioinformatics pipeline to determine the genetic relatedness between the positive sample and isolates in the existing HAI database.

Once a likely HAI is identified, a preliminary report is shared with the hospital infection prevention staff for follow-up investigation and the initiation of any interventions that are deemed necessary.

Download Modern Healthcare’s app to stay informed when industry news breaks.

Since November of last year, the researchers have sequenced samples from over 1,500 patients. Of those, 182, or about 11 percent, clustered into 58 different outbreaks. More than half of the cases in those clusters had some epidemiological evidence of transmission within the hospital, Sundermann said. The results also indicated an average cluster size of three patients, with a median of two. “We want to keep these clusters small,” Sundermann noted, “because that means that the pathogen might not be spreading anymore.”

So far, a few HAIs identified in the study were endoscope-related, Sundermann said, with one outbreak involving four patients and two different pathogens pointing to one endoscope. In another notable outbreak, a handful of patients from a third-party chronic care facility embedded in the hospital were impacted.

To make EDS-HAT an effective tool for real-time WGS surveillance, a rapid turnaround time is “extremely key,” Sundermann emphasized. The current average turnaround for the workflow — from a culture being ordered in the hospital to making actionable plans based on the sequencing data — is 16 days, with a median of about two weeks. “We want to squeeze the time down as much as we can,” he said. “Because within those two weeks, there is transmission happening in the hospital that is not intervened upon.”

In addition, Sundermann said a good outcome measure for the new approach’s effectiveness is to investigate how many downstream infections still occur even after WGS surveillance-informed interventions have been initiated to curb an outbreak. So far, an interim analysis at the six-month time point of the study found zero more infections after interventions were in place, indicating success. Still, more analyses at the study’s one-year mark, which is next month, will show if these preliminary results hold, he noted.

Beyond that, Sundermann said, the team also plans to examine the hospital’s overall HAI rate after implementing the program, comparing it with a peer hospital that has a similar patient capacity but has not been using a genomic surveillance program for infection prevention.

“HAI rate is what the hospital is going to care about,” he said. “If the HAI rate is going down from this program, our steps are more likely to be widely adopted.”