Introduction
Enterococcus faecium is a common cause of hospital-acquired infections and is frequently associated with vancomycin resistance, resulting in high mortality, longer hospital stays, and higher healthcare costs. Reference Zhou, Willems, Friedrich, Rossen and Bathoorn1 The hospital environment is a reservoir for E. faecium due to its ability to survive on surfaces for prolonged periods. Reference Zhou, Willems, Friedrich, Rossen and Bathoorn1 Prior studies have analyzed patient and environmental reservoirs of E. faecium but have typically been limited to selected hospital wards in settings where either vanA or vanB vancomycin-resistant E. faecium (VRE) predominate. Reference El Haddad, Hanson and Arias2–Reference Lee, White, Monahan, Jensen, Chan and Hal4 Australia has a specific VRE epidemiology with more recent emergence of vanA VREfm on a background of vanB VRE endemicity, with both continuing to circulate. Reference Coombs, Daley, Shoby and Mowlaboccus5 In 2019, we noted increasing prevalence of vanA VRE bloodstream infections. We aimed to assess the hospital environment as a reservoir of vanA VRE by conducting a point prevalence study and comparing environmental and contemporary clinical isolates using genomic analyses.
Methods
The study was approved by the Alfred Hospital Ethics Committee (Project Number 81/24, low-risk pathway). The Alfred Hospital (Melbourne, Australia) is a 600-bed quaternary hospital with state referral services for burns and trauma, as well as stem cell and solid organ (heart/lung/kidney) transplant services. We conducted a point prevalence survey of the hospital environment. We sampled surfaces using FLOQswabs (Copan) over two months (see Supp. Methods and Supp. Table 1), which included all wards and the intensive care unit (ICU) (n = 12). Each ward was sampled once during that two-month period. To minimize disruption to patient care and staff workflows, within individual units, we focused on equipment and rooms that were not in active use at the time of sampling.
Ten surface types were sampled across the hospital (see Figure 1B, Supp. Table 2) with multiple samples of each surface type collected across different wards. Swabs were premoistened with sterile saline and used to sample a 10 × 10 cm area of each surface. After incubation in enrichment broth, samples were cultured on CHROMagar VRE media (BioMerieux) and species were identified using matrix-assisted laser desorption-ionization time-of-flight (MALDI-TOF) (Bruker Daltonics). All E. faecium isolates were presumed to be VRE but no phenotypic antimicrobial susceptibility testing was performed.
Figure 1. Summary of Enterococcus faecium environmental screening and contemporary clinical isolates. (A) Proportion of environmental screening swabs positive for vancomycin-resistant E. faecium (VRE) by ward and (B) by surface type. ‘Other’ includes bedside monitors, trolleys, and call bells. (C) Summary of E. faecium van operon presence by multi-locus sequence types (MLST) and (D) core genome MLST (cgMLST) in clinical and environmental isolates.
To assess for molecular epidemiological links between environmental and clinical isolates, we selected all E. faecium isolates from our established blood culture biobank (approved by Alfred Hospital Ethics Committee, Project Number 533/16, biobank pathway), which comprehensively stores all blood culture isolates from our hospital. Phenotypic antimicrobial susceptibility testing data were not available. We included all E. faecium isolates collected during the period starting 6 months before the first environmental sampling date and ending 6 months after the final environmental sampling date (Supp. Fig. 1 and Supp. Table 2). Clinical data were extracted from the electronic medical record. Our hospital does not perform asymptomatic screening for VRE. See Supp. Methods for details of routine environmental cleaning and infection prevention practices. In brief, surfaces were cleaned daily with 10% sodium hypochlorite solution with some high-risk high-touch surfaces cleaned more frequently in extreme-risk wards.
One colony was selected from each sample and all environmental and clinical isolates from the study period underwent short-read (Illumina) whole genome sequencing, as described previously. Reference Macesic, Hawkey and Vezina6 We detected van operon presence and assigned multi-locus sequence type (MLST) and core genome MLST (cgMLST). Reference Higgs, Sherry and Seemann7 For each cgMLST cluster, we assembled a long-read reference genome and calculated within-cluster pairwise single nucleotide variant (SNV) distances (see Supp. Methods).
Results
Environmental point prevalence survey
We swabbed 357 surfaces (median 31.5 per ward [range 10–60]). A total of 73/357 (20%) surfaces isolated VRE, with median 20% positive per ward (IQR 11.6%–30.6%) (Figure 1A, Supp. Table 2), indicating that VRE was endemic in the hospital environment. Rates of VRE positivity varied widely, with the highest rate in the Burns ward (10/10 swabs positive) and the lowest rate in Surgical Ward 3 (0/34 positive). Patient chairs had the highest rate of VRE positivity (13/25 [52%]) (Figure 1B, Supp. Table 3).
Environmental and clinical genome characteristics
All 72 E. faecium blood culture isolates from the study period were sequenced and compared to the 73 positive environmental swabs (Figure 1C, Supp Fig. 1). Of 72 blood culture isolates, 63 (88%) were hospital–onset (collected >48 hours after hospital admission). vanA and vanB were detected in 12/73 (16%) and 58/73 (79%) environmental genomes, respectively (Figure 1C). In contrast, 32/72 (44%) clinical genomes carried vanA (P <.001, χ2 test), and 31/72 (43%) carried vanB. Environmental genomes belonged to 10 MLSTs and 10 cgMLSTs, with ST796/ST78 accounting for 54/73 (74%) genomes (Figure 1C and 1D). Clinical STs/cgMLSTs overlapped with environmental genomes but were more diverse. Specifically, ST1421/1424 carried vanA and were more frequent in clinical genomes (30/72 v 7/73 genomes, P <.001, χ2 test). A single cgMLST cluster (B) contained the majority of vanA genomes (25/44 genomes), while vanB genomes were noted across two cgMLST clusters (A and C, 84/89 genomes) (Figure 1D). STs/cgMLSTs carried either vanA or vanB operons, except for 3 genomes in ST796/cgMLST cluster A that had concurrent vanA and vanB carriage (Figure 1C/1D).
Geography of E. faecium on hospital wards
We mapped E. faecium ST/cgMLST and presence of van operon across individual wards (Supp. Fig. 2). High-risk wards (ICU, Haematology/Oncology) had the majority of clinical genomes with a resultant high diversity of STs/cgMLSTs (7 STs/10cgMLSTs for Haematology/Oncology and 7 STs/7 cgMLSTs for ICU, respectively). The Haematology/Oncology ward had the highest proportion of vanA genomes (12/25, 48%), with 7/12 of these being ST1424/cgMLST cluster B genomes. Other wards had uniform ST/cgMLST composition with no differences between wards (Supp. Fig. 2).
E. faecium transmission networks
Using a 6 SNV cutoff, Reference Gouliouris, Coll and Ludden3 there were 895 putative genomic transmission links resulting in 13 clusters (Figure 2). For genetically-related isolates, median time between isolate collection dates was 37 days (IQR 14–124) (Supp. Fig. 3). While 292/895 (33%) links were environmental-clinical, of these only 88/292 (30%) were between vanA genomes. Most clusters had a predominance of either environmental-environmental links or clinical-clinical links (Supp. Table 5). A single cluster contained 25/44 (57%) vanA genomes, with the majority being clinical (22/26, 85%). Environmental genomes in that cluster came from three wards, while clinical genomes were collected from patients in 11 different wards. In contrast, vanB genomes belonged to three major (>5 genomes) clusters, containing predominantly environmental genomes (42/56, 75%).
Figure 2. Network analysis of Enterococcus faecium putative genomic transmission links. Clinical and environmental E. faecium genomes are shown as nodes and putative genomic transmission events (defined as pairwise single nucleotide variant distance ≤6) as edges. Edges resulting from clinical-environmental links are shown in black, and clinical-clinical or environmental-environmental in gray. A single major vanA cluster is noted with majority of clinical genomes. In contrast, three major vanB clusters are noted with environmental genomes predominating.
van operon analysis
In completed assemblies (n = 26), the vanA and vanB operons were located on pRE25-like plasmids and the chromosome, respectively. The pRE25-like plasmids were heterogeneous and did not indicate spread of a single plasmid across multiple cgMLST clusters (Supp. Fig. 4).
Discussion
We noted a 20% prevalence of VRE environmental colonization across a diverse range of hospital surfaces, but only 12/73 (17%) environmental genomes carried vanA. This indicated that while VRE was endemic in the hospital environment, it likely did not make a major contribution to the increased vanA VRE prevalence in bloodstream infection isolates. Indeed, we noted distinct environmental and patient reservoirs: an environment dominated by vanB ST796/78 genomes, and a clinical reservoir comprising vanA ST1421/1424 genomes. There was evidence of spill-over with putative clinical-environmental links, but these were a minority (33%) and vanA putative genomic links only contributed 88/895 (10%) of these in turn.
Our findings stand in contrast to prior work showing significant links between VRE in the environment and clinical colonization and infection. Reference El Haddad, Hanson and Arias2–Reference Lee, White, Monahan, Jensen, Chan and Hal4 This difference may be due to these studies describing non-outbreak settings and focusing on limited wards (ICU and Haematology/Oncology). Potential contributors to our findings include our institution’s differing infection prevention approaches to vanA and vanB VREfm-colonized patients: Reference Howard-Anderson, Gottlieb, Beekmann, Polgreen, Jacob and Uslan8,Reference Morgan, Murthy and Munoz-Price9 vanA patients are routinely placed in contact precautions with dedicated equipment, while for vanB patients this occurs only on high-risk wards, and if they have diarrhea or non-contained wounds. The environmental isolates in our study may represent persistent strains of VRE. E. faecium ST796, a key vanB lineage in our study, has been shown to develop biocide tolerance that may provide it with a fitness advantage in hospital environments. Reference Pidot, Gao and Buultjens10 Of note, the vanA clinical reservoir that prompted our investigation was subsequently resolved during the COVID-19 pandemic (Supp. Fig. 1), perhaps because of these factors.
Our study had several limitations. To assess the clinical reservoir, we focused on bloodstream infection isolates. However, these likely represent the ‘tip of the iceberg’ of VRE colonization with only 8% of VRE-colonized patients developing infection. Reference Willems, van Dijk and Vehreschild11 Similarly, our institution does not conduct asymptomatic VRE screening, limiting our ability to quantify rates of VRE colonization in patients. We also included all E. faecium bloodstream infection culture isolates (both VRE and non-VRE), potentially impacting comparison with the environmental VRE isolates. However, 68/72 (94.4%) clinical isolates carried the van operon, thus limiting this impact. Sampling equipment and rooms not in active use were a potential source of bias, however, this pragmatic approach allowed us to conduct sampling efficiently while avoiding interference with clinical activities. We also selected single colonies for whole genome sequencing, which may not be representative of the entire enterococcal population, particularly for environmental samples. Finally, our analysis of transmission networks was limited by the absence of patient movement data.
Although the environment plays a major role in healthcare-associated VRE spread, our study shows that distinct clinical and environmental reservoirs of VRE may exist, particularly in outbreak settings where new clones are emerging and have not yet colonized the environment. This provides support for use of molecular techniques to focus infection prevention strategies according to the reservoir. While cleaning strategies aiming to reduce the environmental burden form an essential part of any multifaceted approach to VRE control, Reference De Angelis, Cataldo and De Waure12 the clinical reservoir may require more active attention through measures such as introduction of active surveillance, redoubled hand hygiene efforts, and more stringent application of contact precautions in order to prevent both between-patient spread and future environmental colonization.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/ice.2025.27
Financial support
This work was supported by the National Health and Medical Research Council of Australia (Emerging Leader 1 Fellowship APP1176324 to N.M., Practitioner Fellowship APP1117940 to A.Y.P.).
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Competing interests
Nil relevant.
Conference presentation
Portions of this work were presented at European Society of Clinical Microbiology and Infectious Diseases Global 2024, held in Barcelona, Spain.
Open access
