Human Cases of Carbapenemase-Producing Escherichia coli Linked to Spread Between Animals and the Environment in a Veterinary Facility—Massachusetts, USA, 2023

Abstract

Background

Carbapenemase-producing Enterobacterales (CPE), traditionally thought to cause hospital-acquired infection in humans, are now emerging as agents of both community-acquired infections in humans as well as hospital-acquired infections in pets. It is currently unknown what impact this spread among animals may have on human health.

Methods

The Massachusetts State Public Health Laboratory investigated links between 3 genetically related human cases of bla_NDM-5-harboring Escherichia coli. Simultaneously, a separate, collaborative investigation into bla_NDM-5-harboring E. coli animal infections at a Massachusetts veterinary hospital was conducted including animal and environmental screening. Whole-genome sequencing was performed on human, animal, and veterinary hospital environmental isolates. Once a genetic link was established between animal and human isolates, follow-up interviews including questions related to pet exposure were performed with affected humans.

Results

Whole-genome sequencing demonstrated that human, animal, and veterinary environmental isolates clustered within 0–10 single nucleotide polymorphisms. Although initial case interviews failed to identify any epidemiologic linkages between the humans, follow-up questioning revealed that all 3 human cases had pets treated at the same veterinary hospital prior to their diagnosis.

Conclusions

This report suggests the potential for transmission of bla_NDM-5-harboring E. coli between animals and people. There is an urgent need to invest resources and develop a One Health approach to support veterinary facilities in recognizing and preventing transmission of CPE to limit spread within communities. Physicians, veterinarians, and public health professionals conducting outbreak investigations should be aware of this potential route of CPE spread.

We report human infection or colonization with bla_NDM-5-harboring Escherichia coli following hospitalization of their pets at a veterinary facility with transmission of the same strain in Massachusetts. Human health surveys with pet health history and whole-genome sequencing established a link.

Graphical Abstract

Graphical Abstract.

This graphical abstract is also available at Tidbit: https://tidbitapp.io/institutional-portal/clinical-infectious-diseases/tidbits/human-cases-of-carbapenemase-producing-escherichia-coli-linked-to-spread-between-animals-and-the-environment-in-a-veterinary-facility-7ea1f374-3e96-428c-b6a1-1288833c7dd7/update

(See the Editorial Commentary by Perreten and Endimiani on pages e1145–7.)

Carbapenem-resistant Enterobacterales (CRE) are some of the most critical pathogens facing human health today according to the World Health Organization (WHO) and the US Centers for Disease Control and Prevention (CDC) [1, 2]. Carbapenem-resistant Enterobacterales can cause a variety of infections, including urinary tract infections, blood stream infections, and pneumonia, with many being life-threatening given their frequent resistance to initial empirical antimicrobial choices and limited therapeutic options [3–5].

Of particular public health concern are carbapenemase-producing Enterobacterales (CPE), as the genes encoding these carbapenem-inactivating enzymes are often located on mobile genetic elements that can spread between bacterial strains and species [3]. Historically, CPE have primarily been recognized as agents of hospital-associated infections in the United States; however, recent data suggest that community spread of CPE is an emerging problem [6, 7]. There also has been an increasing number of reports of pet animals with CPE colonization and infection, including outbreaks at veterinary hospitals [8–10]. Whole-genome sequencing (WGS) has advanced our ability to track the spread of CPE among people [11, 12]. Whole-genome sequencing presents an opportunity to investigate the role that animals may play in human community spread as preliminary studies have suggested that animal and human isolates are closely related [13]. Here we describe an investigation that links human cases of New Delhi metallo-β-lactamase (bla_NDM-5)–harboring Escherichia coli with spread of the organism at a veterinary hospital in Massachusetts.

METHODS

Initial Human Case Investigations

At the time of the investigation and currently it is a statewide policy in Massachusetts that clinical Enterobacterales isolates from people who are carbapenem-resistant, as well as those demonstrating carbapenemase production, must be reported to the Massachusetts Department of Public Health (MDPH) by the microbiology laboratory. When a CPE is reported, MDPH epidemiologists investigate these cases to identify potential exposures as well as to implement infection-control measures in the healthcare setting (when indicated). Additionally, any CRE or CPE isolate must be submitted to the Massachusetts State Public Health Laboratory (MA SPHL) for confirmation [14]. Whole-genome sequencing is performed on all of the aforementioned isolates and compared with other isolates using the NCBI Pathogen Detection browser to identify clusters [15]. When a cluster of 2 or more matching isolates is identified, MDPH epidemiologists are notified and investigate any epidemiologic link(s) and common exposures to implement control measures.

Initial Veterinary Facility Investigations

An in-depth characterization of the methods used for investigation into and mitigation of this outbreak of animal CPE cases has been described previously [16]. Briefly, the veterinary hospital recognized a cluster of clinical infections with imipenem-resistant E. coli, which was novel for the facility. While reporting of CPE isolates in pets is not mandated in Massachusetts, the hospital infection-control committee reached out to other veterinary hospitals with CPE infections and ultimately the MDPH for support in initiating an investigation and response to the cluster. The investigation included identification of environmental contamination via targeted culture and patient point-prevalence surveys to identify CPE colonization of hospitalized animals and assessed the potential for ongoing transmission. All animal and veterinary hospital environment testing was performed at the Carbapenem Resistant Enterobacterales Animal Testing and Epidemiology (CREATE) reference laboratory at the University of Pennsylvania.

Whole-Genome Sequencing and Molecular Epidemiology

The CREATE laboratory provided the Minnesota Department of Health Public Health Laboratory (MDH-PHL) animal and veterinary environmental isolates as part of an ongoing collaboration to perform WGS and analysis. Briefly, sequencing libraries were prepared using the Illumina DNA Prep Library preparation kit and sequenced using the Illumina MiSeq version 2 (2 × 250 Paired-End Reads) chemistry. Human isolates were sequenced and analyzed at the MA SPHL using methods previously described [17]. Deidentified raw sequence reads were uploaded to NCBI (BioProject PRJNA288601 for Human Isolates; BioProject PRJNA715470 for Animal/Environmental Isolates) and assembled and analyzed via the NCBI Pathogen Detection browser, which includes basic single nucleotide polymorphism (SNP) analysis by maximum compatibility algorithm and identifies closely related isolates (<50 SNPs) within the database [15]. A table of WGS repository accession numbers can be found in the Supplementary Materials. Following submission and processing of WGS data by the NCBI pipeline, SNP trees for each isolate were reviewed by the CREATE laboratory to identify any clusters containing animal and human isolates.

Following the detection of clustering on the NCBI Pathogen Detection pipeline, sequence files from the NCBI sequence read archive (SRA) (3256-23; 2023DK-00101; 2023DK-00272; 2023DK-00238) were downloaded for additional bioinformatic analysis with the isolates sequenced at the MDH-PHL. The Spriggan 1.1.2 bioinformatics workflow (https://github.com/wslh-bio/spriggan) was used for genome assembly and to determine multilocus sequence type (MLST) [18]. Reference-based SNP analysis was performed on isolates with the same MLST using isolate 47217-23 as the reference and CFSAN as part of the Dryad 3.0.0 workflow (https://github.com/wslh-bio/dryad) [19]. A phylogenetic tree was generated using RAxML and visualized using MEGA7 [20, 21]. Plasmid contents were assessed using PlasmidFinder, and BLASTn was performed on the contig of 47217-13 harboring a carbapenemase gene to compare relatedness with other published plasmids [22].

Additional Human Interviews and Animal Record Review

The CREATE/MDH-PHL team contacted the MDPH team upon identification of the clustering of isolates from people and animals in Massachusetts to initiate the collaboration described here. Epidemiologists from MDPH then conducted interviews with 2 of the human cases and a proxy for the third case, as this case was unavailable to be interviewed. The interviews covered a variety of domains, including questions about healthcare, pet ownership, pet exposures to veterinary settings, and non-pet animal exposures. For pets associated with human cases that were recognized to have visited the veterinary facility, electronic medical records were reviewed to collect information regarding the pet's signalment and characteristics, reason for and timing of hospitalization, and exposure to antimicrobials.

RESULTS

Initial Human Case Investigations

A timeline of events is shown in Figure 1. Months are defined chronologically, with month 1 representing initial hospitalization of the first pet identified that was linked to a human patient. In month 9 of this investigation period, the MA SPHL identified a cluster of 2 bla_NDM-5-harboring E. coli cases that were closely related by WGS. Human case A was initially identified in month 5; the bacterium was isolated from the urine of a 79-year-old female with a urinary tract infection (UTI). This patient had a history of recurrent UTI but no recent inpatient medical care, which suggested the infection was likely community-acquired.

Figure 1.

Timeline of human, animal, and veterinary environmental samples containing a highly related ST162 bla_NDM-5-harboring Escherichia coli. Month 1 was defined as the first visit of a pet associated with a human case to the affected veterinary facility. Pets and owners are labeled with the same letter (A–C). Animals on the bottom row are labeled with isolate identification numbers found as part of point-prevalence surveillance. The hospital icon represents time points with positive veterinary environmental samples. Abbreviations: F, female; M, male. Created in BioRender. Cole, S. (2025) https://BioRender.com/pxp7ft1.

Human case B was identified in month 9 when a 63-year-old male was screened via rectal swab for CPE colonization following potential in-hospital exposure to a different strain of CPE. At the time of screening, the patient was admitted to an acute care hospital for lymphoma treatment. Screening revealed bla_NDM-5-harboring E. coli and not the CPE strain that the patient was potentially exposed to in the hospital. Other patients in the facility were also screened after the same exposure, but none tested positive for bla_NDM-5-harboring E. coli. A public health investigation to identify common healthcare exposures for human cases A and B was conducted but failed to reveal epidemiologic links across healthcare facilities.

In month 10, a third, highly related bla_NDM-5-harboring E. coli isolate was linked to the WGS cluster (human case C). This isolate was identified in a 23-year-old female's urine and found to be 5 and 8 SNPs different from the isolates for case B and case A, respectively. The patient was being treated for recurrent UTI. The third patient had no history of inpatient medical care, although she had previously undergone urinary tract surgery more than 12 months previously. No epidemiologic links, including common healthcare exposures, could be made to either of the first 2 cases. All 3 cases lived in the same county with no known recent foreign travel or medical tourism.

Initial Veterinary Facility Investigations

For context, review of pharmacy prescriptions identified only 16 carbapenem prescriptions, all in inpatient dogs, in the year prior to the investigation, representing a relative frequency of 0.01% of inpatient antimicrobial prescriptions. As previously described, animals associated with this outbreak included a total of 19 dogs with various disease syndromes, including pneumonia, UTI, surgical site infections, and bloodstream infections; unfortunately, clinical isolates were not available for additional analysis [16]. As part of the response to these cases, environmental sampling and testing at a veterinary reference laboratory were performed beginning in month 2 of this investigation, and a total of 22 environmental surfaces were found to culture positive for bla_NDM-5-harboring E. coli intermittently through month 8. Common sites found to be contaminated included animal housing units and sinks/drains. Additionally, rectal screening of dogs and cats hospitalized between months 5 and 13 was performed as part of point-prevalence surveys recommended by the MDPH. A total of 17 carbapenemase-producing E. coli isolates from environmental surfaces and 9 (8 dogs and 1 cat) rectal swab isolates were available for WGS as part of the investigation described here.

Whole-Genome Sequencing and Molecular Epidemiology

Initial NCBI Pathogen Detection pipeline analysis generated a cluster of animal, veterinary environment, and human isolates within 14 SNPs. No isolates other than those from the humans in this study and from the veterinary hospital outbreak were found to cluster within 50 SNPs. Further in-house analysis using the bioinformatics pipeline Dryad 3.0.0 revealed that all isolates of E. coli from animal, veterinary environment, and human sources were ST162 and clustered within 10 SNPs based on a core genome of 4 821 710 bp (98.47% of mapping reference genome). Human isolates were highly related to animal and veterinary environment isolates (Figure 2). The isolate from human case A (2023DK-00101) was 0 SNPs from an isolate collected from an intensive care unit table monitor (3256-23) at the veterinary facility as well as a canine rectal swab isolate from a dog tested in month 13 (68821-23). The isolate from human case B (2023DK-00238), which was collected from the patient in month 9, was 0 SNPs from 5 environmental isolates (43091-23, 43108-23, 44597-23, 43090-23, 44600-23) and 1 animal isolate (23595-23) obtained in month 5 via surveillance. The human case C (2023DK-00272) isolate was the most divergent but was still only 5 SNPs from a veterinary isolate (23595-23). All isolates in this investigation were confirmed to harbor the bla_NDM-5 carbapenemase gene. Other antimicrobial resistance genes identified included acrF, bla_CTX-M-14, bla_EC, bla_TEM-1, ble, mdtM, and tetB. Plasmid analysis revealed this outbreak strain carried IncFIB, IncFIC, and IncX3 type plasmids; and BLAST of the 47217-23 contig (DARQMO010000042.1; 9082 bp) harboring the bla_NDM-5 gene identified several highly related IncX3 plasmids (100% identity) also harboring bla_NDM-5 (GenBank Accessions MK986791.1, PV171503.1, CP060901.1). This suggests that the IncX plasmid was not novel to this outbreak and the outbreak described here is connected to the global increase in bla_NDM-harboring plasmids [23].

Figure 2.

A, Pairwise comparison heatmap demonstrating reference-based single nucleotide polymorphism (SNP) differences among isolates included in the investigation using isolate 47217-23 as the reference genome and (B) SNP-based phylogenetic tree generated using RAxML with isolate identification numbers and collection sources as labels. The SNP heatmap utilizes a color gradient to represent the degree of genetic similarity among isolates. Red indicates the highest genetic similarity (smallest SNP differences). Yellow represents intermediate SNP differences, while green denotes the largest SNP differences (lowest genetic similarity). Created in BioRender. Cole, S. (2025) https://BioRender.com/itaq1wm

Additional Human Interviews and Animal Record Review

Interviews of human cases identified an epidemiological link between the human cases and veterinary isolates. All 3 human cases had pets in their home who had received care at the described veterinary facility within 2 months prior to the associated human case testing positive for bla_NDM-harboring E. coli. Human case A (positive for CPE UTI in month 5) owned a dog that was initially hospitalized in month 1 for surgical intervention. Human case A's dog also returned to the veterinary facility for a total of 4 outpatient oncology visits between month 2 and month 4 but had no exposure to systemic antimicrobials other than perioperative cefazolin during the initial hospitalization. Human case B (rectal screen positive in month 9) owned a cat treated for an infection in month 8 with a long-acting third-generation cephalosporin (cefovecin sodium). The outpatient care of the cat belonging to human case B overlapped with hospitalization of human case C's dog. Human case C (diagnosed with a CPE UTI in month 10) had a dog managed for heart disease in month 8, including 2 admissions, an interventional radiology procedure, and 5 total days of hospitalization. The dog's antimicrobial therapy included ampicillin-sulbactam, amoxicillin-clavulanate, enrofloxacin, and doxycycline at various points. None of the veterinary patients owned by human cases were included in screening for CRE colonization during their hospitalizations or developed recognized CRE infections.

DISCUSSION

Our investigation combined both molecular and applied epidemiological investigation methods to characterize community spread of a bla_NDM-5-harboring E. coli between people and companion animals. Specifically, we determined that multiple, seemingly unrelated cases of human CPE were linked to an outbreak among veterinary patients by comparing veterinary and human isolates using WGS. Broadening the scope of standard public health investigation of human CPE to include pet exposures and animal health history aided in the identification of the likely exposure source for these cases. With evidence of spread of enteric bacteria between pets and people, the emergence of CPE among animals represents a significant public health concern [24, 25].

A 2021 meta-analysis revealed that dog owners are at higher risk than non-pet owners for methicillin-resistant Staphylococcus aureus carriage but did not find the same association for third-generation cephalosporin or carbapenem-resistant Enterobacterales. This meta-analysis, however, did not investigate recent pet healthcare exposures [26]. Despite these findings, some studies have demonstrated that people (eg, owners, veterinary staff) and animals in proximity to each other during the same timeframe can be colonized with related strains of CPE [27, 28]. Additionally, several studies have genomically linked human and animal isolates, but these studies often lacked epidemiological context to elucidate transmission [11, 12].

Our investigation adds to the growing body of evidence that CPE can spread between animals and people and, ultimately, lead to human infections. In this investigation, we demonstrated a temporal relationship between prior exposure of pets to a veterinary hospital where ongoing CPE transmission was documented and subsequent CPE infection or colonization in their owners. The timeline suggests that the animals became colonized during veterinary care and spread CPE to their owners following discharge; however, we cannot definitively say that the transmission occurred directly or indirectly between the pet and their owner because the pets owned by human cases A, B, and C were not tested for CPE. For example, we cannot rule out the role of the veterinary hospital environment as another potential source for owners as they could have been exposed during visitation at the facility; however, we believe this to be less likely given that no other human cases linked to the veterinary hospital were identified despite ongoing statewide surveillance activities [15]. Testing of the pets was not elected because of both feasibility challenges and that a significant period of time (>6 months) had passed at the time of the investigation. This lack of testing of the human cases' pets does represent a significant limitation of both this investigation and, more broadly, current One Health surveillance of CPE in animals, people, and their shared environment. This report highlights the critical need for infrastructure to complete WGS analysis and epidemiologic investigation of veterinary CPE isolates to reduce data gaps and allow for more rapid infection prevention and control (IPC) interventions.

Human health practitioners should include animals when considering possible sources of multidrug-resistant organism (MDRO) infections. While suspicion for, or diagnosis with, infections due to classically zoonotic organisms (eg, Campylobacter, Pasteurella, Bartonella) often prompts the physician to ask about pet ownership as part of the medical history, it is less intuitive to consider pets as sources of exposure to MDROs such as CPE. Veterinarians also have a role by making recommendations for the prevention of transmission within a household when an animal is diagnosed with any zoonotic organism infection or colonization, including MDROs [29, 30]. These include frequent, proper hand hygiene, routine cleaning and disinfection of the home, and consultation with the owner's healthcare provider [29–31].

Pet ownership in the United States is very common; it is reported that approximately 45.5% (59.8 million) and 32.1% (42.2 million) of households have dogs and cats, respectively [32]. Attempts to estimate the prevalence of CPE or CRE in pets in the United States are fairly limited but have suggested low rates of colonization [11] and infection [33] (0.2%–1.2%). However, given the contact between large numbers of animals at dog parks, veterinary hospitals, and boarding facilities, the potential for amplification of pets and humans exposed is significant and could result in sporadic cases, clusters, or outbreaks. Although there are established guidelines for IPC programs in veterinary medicine [34], changes in veterinary care, such as novel invasive procedures requiring intensive care, use of higher-tiered antimicrobials, broader use of immunosuppressive medications, and the increase in specialty care centers, likely present new challenges for IPC and antimicrobial stewardship programs [35]. There is a need for additional public and animal health agency support for, and outreach to, veterinary clinics to limit the spread of CPE among pets. Efforts must be made to further educate the veterinary community (eg, veterinarians, technicians/nurses, laboratorians) on CPE and related IPC practices [35–37].

There is also a need to support case confirmation and investigations by public health departments or animal health authorities. Currently, required reporting of CPE from companion animals is limited to only a few regions—for example, New York City, which made them reportable in 2024. Clusters of atypical infections or infections of public health concern in animals are often reportable to animal health authorities; however, how frequently this is done is unknown. A study of veterinary laboratory capacity revealed that testing capabilities for carbapenemase detection are currently limited in many states and collaboration across laboratories could be valuable [35]. For example, leveraging existing infrastructure at public health laboratories or funding collaborations with state veterinary diagnostic laboratories to perform phenotypic and genotypic confirmatory testing of CPE would help identify cases of CPE without adding a cost burden directly to pet owners or veterinary hospitals. Support should also be provided for inclusion of isolates from animals into surveillance programs, which may be important to identify ongoing multi-host species outbreaks.

This investigation demonstrates a blueprint of how One Health collaborations across sectors could function. A formal relationship between a veterinary laboratory and a public health laboratory, established as part of the Pathogen Genomics Centers of Excellence Network established in 2022 by the CDC, laid the foundation for our findings. Once a One Health connection was established using WGS analysis, the state public health department could initiate a local investigation, which was possible because of robust surveillance approaches within that region. There is an urgent need for collaboration across the human and veterinary health sectors to address the emergence of CPE among pets and, thus, prevent potential transmission to people.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author contributions . Investigation design (I. M. D., C. L. F., P. M. S. V., M. A. C., C. M. B., A. B., S. D., S. D. C.), data collection (I. M. D., M. A. C., J. L. D., A. R., J. D., E. F., N. N. E., K. J., J. L., M. D.), data analysis (M. A. C., A. R., J. L. D., J. D., S. D. C.), manuscript and figure drafting (J. L. D., J. D., S. D. C., and I. M. D.), manuscript and figure revisions (all authors).

Data availability. Sequence data are available via the NCBI Sequence Read Archive (BioProject PRJNA288601 for Human Isolates; BioProject PRJNA715470 for Animal/Environmental Isolates).

Disclaimer. This study’s contents are solely the responsibility of the authors and do not necessarily represent the official views of the Centers for Disease Control and Prevention.

Financial support . This work was supported by the Office of Advanced Molecular Detection, Centers for Disease Control and Prevention, through Cooperative Agreement Number CK22-2204 to S. D. C. and P. M. S. V.