Genomic surveillance of multidrug-resistant Enterobacterales in a long-term care facility in Northern Italy based on a point prevalence screening

Abstract

Long-term care facilities (LTCFs) are increasingly recognised as settings where multidrug-resistant organisms (MDROs) persist and circulate. A point prevalence screening study conducted in 2022 in an LTCF in the Autonomous Province of Bolzano/Bozen, Northern Italy, provided a collection of MDR Enterobacterales isolates from residents and staff. This study aimed to characterise the clonal structure, transmission dynamics, and resistance mechanisms of these isolates through high-resolution genomic analysis. Whole-genome sequencing was performed on third-generation cephalosporin- and/or carbapenem-resistant isolates of Klebsiella pneumoniae, Escherichia coli, Morganella morganii, and Proteus mirabilis. Isolates underwent core genome multilocus sequence typing, phylogenetic reconstruction, and resistance gene profiling using Kleborate and ResFinder. K. pneumoniae isolates formed five major clusters, with ST39 confined to a single section, suggesting intra-section transmission. ST405 and ST307, both high-risk clones, were also detected, with ST307 harbouring the carbapenemase gene bla_KPC−3. CTX-M-15 was present in 90% of K. pneumoniae isolates. E. coli isolates showed high genomic diversity with limited evidence of clonal transmission. M. morganii and P. mirabilis displayed mixed patterns, with sporadic introductions and one clonal cluster for P. mirabilis. Genomic surveillance revealed complex transmission networks within this LTCF, involving both clonal expansion and heterogeneous strain circulation. Organisational factors such as patient transfers, shared staff, and communal living likely contributed to MDROs dissemination. The predominance of CTX-M-15 and detection of carbapenemase-producing K. pneumoniae underscore the need for targeted infection control strategies. Integration of genomic data with contextual factors supports continuous genomic monitoring, antimicrobial stewardship, and environmental decontamination in LTCFs.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-47704-2.

Introduction

Long-term care facilities (LTCFs) are increasingly recognised as critical reservoirs for multidrug-resistant organisms (MDROs), particularly in ageing populations with complex healthcare needs. In Italy, where over 23% of the population is aged 65 or older, LTCFs provide prolonged hospital-level care to individuals with reduced autonomy, chronic illnesses, and frequent exposure to antibiotics and invasive devices¹. Previous studies have documented high colonisation rates of MDROs in Italian LTCFs, including extended-spectrum β-lactamase (ESBL)-producing Enterobacterales, methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant enterococci (VRE), and carbapenemase-producing organisms².

Despite growing awareness, the transmission dynamics of MDROs within LTCFs and between LTCFs and acute care hospitals remain poorly understood. Barriers to effective antimicrobial stewardship (AMS) in these settings include diagnostic uncertainty, variability in prescribing practices, and limited access to local resistance data. In Italy, recent national surveillance efforts have begun to address these gaps, reporting high antimicrobial use and a predominance of agents from the WHO Watch category³.

In the Autonomous Province of Bolzano/Bozen, Northern Italy, a series of point prevalence screening (PPS) studies conducted between 2008 and 2022 in a single LTCF revealed persistently high colonisation rates of MDROs among residents and staff. The most recent PPS in 2022 identified ESBL-producing Enterobacterales as the dominant MDROs, with CTX-M-type enzymes being the most prevalent resistance mechanism⁴. The study used phenotypic screening complemented by targeted PCR to detect ESBL, AmpC and carbapenemases (e.g. CTX-M, SHV, DHA, CMY). In the present analysis, we apply whole-genome sequencing to those MDRO isolates.

To complement these epidemiological findings, the present study applied whole-genome sequencing (WGS) to the MDR isolates collected during the 2022 PPS, including a subset that were phenotypically resistant but PCR-negative. Through core genome multilocus sequence typing (cgMLST), phylogenetic reconstruction and resistance gene profiling, we aimed to elucidate the clonal structure, transmission dynamics, and molecular mechanisms of resistance within the LTCF, providing actionable insights for infection control and AMS strategies.

Methods

All procedures were performed in accordance with the relevant institutional guidelines and regulations, including standard protocols for sample handling, reagent use, and biosafety requirements. The ethical considerations were formally approved by the Ethics Committee for Clinical Trials of the Autonomous Province of Bolzano/Bozen, as documented in the minutes of the 41st ordinary meeting (item 21, November 16, 2022). No protocol ID or approval number was assigned.

Study design and setting

This genomic epidemiology study was conducted in a 120-bed long-term care facility (LTCF) located in the Autonomous Province of Bolzano/Bozen, Northern Italy. The LTCF consists of five units: residents in units 2 and 5 are non-ambulatory and require extensive assistance with daily living activities, along with nursing and medical care; those in units 3 and 4 have less functional disability or comorbidity and those in unit 1 have dementia but are ambulatory. The 2022 screening cohort comprised 89 LTCF residents (Sects. 1–5), 37 staff members, and 20 patients in the adjacent Geriatrics unit. Per‑isolate metadata (ID, species, source category: resident/staff/geriatrics, and LTCF Section) are summarised in Supplementary Tables 2 and are used only to contextualize the genomic findings. The facility had participated in point prevalence screenings (PPS) for MDROs in 2008, 2012, 2016, and 2022⁴. The current analysis focuses on isolates collected during the 2022 PPS.

Detailed per‑isolate metadata for the 2022 cohort (source of isolation, LTCF residents, staff, or adjacent Geriatrics unit, and LTCF section information) are available in the previously published study by Nitti et al.⁴, therefore are not reproduced here in full. In the present study, provenance is used only to inform genomic interpretation.

Genomic DNA extraction and sequencing

Genomic DNA was extracted from overnight cultures grown on sheep blood agar (Thermo Fisher Scientific, USA) using the DNeasy Blood and Tissue Kit (Qiagen, Germany), following the manufacturer’s protocol optimised for Gram-negative bacteria. DNA libraries were prepared using the Nextera XT DNA Library Preparation Kit (Illumina, USA) and sequenced on the Illumina MiSeq platform, generating high-quality paired-end reads.

Genome assembly and annotation

Raw reads were assembled and annotated using the PathoSystems Resource Integration Center (PATRIC) platform^5,6. Species identification and multilocus sequence typing (MLST) were performed using MLST 2.0 (Center for Genomic Epidemiology, Denmark), based on conserved housekeeping genes⁷.

Resistance gene detection

Acquired antimicrobial resistance genes were identified using ResFinder 4.0⁸. For K. pneumoniae, Kleborate v3.2.4⁹ was used to determine species identity, MLST, virulence factors, and resistance determinants.

Plasmid reconstruction and mobile‑element analysis were not performed; consequently, gene co‑occurrence and inter‑lineage sharing cannot be attributed to shared plasmids or mobile genetic elements within this study.

High-resolution typing and phylogenetic analysis

Genetic relatedness was evaluated using Core genome multilocus sequence typing (cgMLST) and Ridom SeqSphere+ v10.0 (Ridom GmbH, Germany)¹⁰. For this study, we adopted operational thresholds published in Enterobacterales: a genetic cluster was defined as isolates with pairwise distances ≤ 15 alleles^11,12. This threshold is heuristic and scheme‑dependent and should be interpreted together with epidemiological data.

Minimum spanning trees (MSTs) were constructed to assess genetic relatedness and infer transmission events. MSTs were generated in Ridom SeqSphere+ from cgMLST allelic distances, ignoring missing alleles; for reporting purposes, two‑isolate components meeting the criterion are also considered clusters. Edge labels denote pairwise cgMLST allelic differences. The ≤ 15‑allele cut‑off is applied uniformly to facilitate consistent visualisation and is interpreted alongside epidemiological context. MSTs were shaded to indicate groups meeting the cluster threshold. Unrooted phylogenetic trees were generated using Ridom SeqSphere+, based on allelic profiles derived from cgMLST. The MST is presented for visualisation only. Missing alleles were ignored for MST construction, and cluster assignment relied on the thresholded pairwise graph rather than on MST topology.

Results

Overview of isolates and genomic typing

A total of 67 multidrug-resistant (MDR) isolates were analysed, including K. pneumoniae (n = 29), E. coli (n = 19), M. morganii (n = 8), and P. mirabilis (n = 11), collected in 2022 during a point prevalence screening in a long-term care facility (LTCF) in the Autonomous Province of Bolzano/Bozen, Northern Italy⁴. Provenance (LTCF residents, staff, or Geriatrics unit) and section information for these isolates were reported in detail previously and are referenced here where relevant for interpretation⁴. All isolates underwent cgMLST, phylogenetic reconstruction, and resistance gene profiling. MSTs and unrooted phylogenetic trees were used to assess clonal relationships and transmission dynamics.

Klebsiella pneumoniae

cgMLST analysis revealed five major clusters (Fig. 1), defined as single‑linkage connected components with all edges ≤ 15 alleles (including two‑isolate clusters). MST Cluster 1 corresponds to ST39 and includes 12 isolates; ST39 was predominantly observed in Sect.  5, with two isolates (34b_KP and 35_KP) from Sect.  3, underlining a mainly Section‑specific clustering; in the absence of environmental sampling, environmental persistence remains speculative. ST405 and ST307, both recognised as high-risk clones, were also detected.

Fig. 1.

MST of K. pneumoniae isolates based on cgMLST allelic distances (missing alleles ignored). MST illustrating the genetic relatedness among isolates based on allelic profiles derived from cgMLST. Each node represents a unique isolate. Edge labels indicate pairwise allele differences. Clusters are defined as single‑linkage components with all edges ≤ 15 alleles (two‑isolate components included). Coloured shaded areas (clouds) group closely related isolates into clusters, suggesting potential transmission events or clonal expansion within the facility. The large red shaded cloud in the upper part of the MST corresponds to ST39. Node colour denotes Sequence Type (ST); nodes with a unique ST are shown without fill. Isolate IDs appear inside the circles; cgMLST-identical isolates are collapsed into a single circle. Per-isolate ST and provenance are listed in Supplementary Table 2.

ST307 was represented by a single isolate (48b_KP), which harboured the carbapenemase gene bla_KPC−3. CTX-M-15 was detected in 26 of 29 isolates (90%), frequently co-occurring with bla_OXA−1 and bla_TEM−1D, suggesting plasmid-mediated dissemination. Additional resistance genes included aac(3)-IIa, aac(6’)-Ib-cr, sul2, qnrB1, dfrA14, and tet(A) (Supplementary Table 1). Mutations in ompK35, gyrA, and parC were observed, conferring resistance to carbapenems and fluoroquinolones.

The unrooted phylogenetic tree (Supplementary Fig. 1) confirmed the clustering patterns and highlighted the genetic relatedness within and between STs. The spatial distribution of clones across LTCF Sections supports localised transmission; potential environmental contributions are addressed in the Discussion.

Escherichia coli

E. coli isolates exhibited substantial genomic diversity (Fig. 2). Under the ≤ 15‑allele definition, only small clusters (including two‑isolate clusters) were observed, with most isolates remaining singletons, consistent with limited clonal transmission; cgMLST and phylogenetic analyses also revealed a fragmented population structure, with STs including ST131, ST1193, ST95, ST10, ST69, and ST155. ST131, the dominant extraintestinal pathogenic E. coli (ExPEC) lineage globally, was found in multiple clusters, consistent with the co-circulation of subclades. ST1193, an emerging clone, shared similar resistance and virulence profiles with ST131.

Fig. 2.

MST of E. coli isolates based on cgMLST allelic distances (missing alleles ignored). MST illustrating the genetic relatedness among isolates based on allelic profiles derived from cgMLST. Each node represents a unique isolate. Edge labels indicate pairwise allele differences. Clusters are defined as single‑linkage components with all edges ≤ 15 alleles (two‑isolate components included). Coloured shaded areas (clouds) group closely related isolates into clusters, suggesting potential transmission events or clonal expansion within the facility. Node colour denotes Sequence Type (ST); nodes with a unique ST are shown without fill. Isolate IDs appear inside the circles. Per-isolate ST and provenance are listed in Supplementary Table 2.

Resistance gene profiling revealed CTX-M-15 in 16 of 19 isolates, widely distributed across unrelated lineages. Other β-lactamase genes included bla_TEM−1B, bla_CTX−M−32, bla_CTX−M−55, and bla_DHA−1. Aminoglycoside resistance genes (aac(6’)-Ib-cr, aac(3)-IIa, aph(3’)-Ia, aadA1), chloramphenicol resistance genes (catB3, cmlA1), macrolide resistance gene (mph(A)), and sulfonamide genes (sul1, sul2, sul3) were variably present. Tetracycline resistance was supported by tet(A) (Supplementary Table 1).

The unrooted phylogeny (Supplementary Fig. 2) confirms the presence of small MST‑defined clusters alongside dispersed lineages, supporting a scenario dominated by multiple independent introductions.

Morganella morganii

M. morganii isolates showed high genomic fragmentation. Only one cluster was identified (strains 51_MM and 54_MM), with a distance of three alleles, suggesting recent transmission or a shared source (Fig. 3). All other isolates were genetically distinct, indicating sporadic acquisition.

Fig. 3.

MST of M. morganii isolates based on cgMLST allelic distances (missing alleles ignored). MST illustrating the genetic relatedness among isolates based on allelic profiles derived from cgMLST. Each node represents a unique isolate. Edge labels indicate pairwise allele differences. Clusters are defined as single‑linkage components with all edges ≤ 15 alleles (two‑isolate components included). Coloured shaded areas (clouds) group closely related isolates into clusters, suggesting potential transmission events or clonal expansion within the facility.

Resistance gene profiling revealed a low-frequency but diverse distribution of β-lactamase genes. bla_DHA−14 and bla_DHA−17 were the most frequently detected. One isolate (49_MM) harboured multiple bla_TEM variants, including bla_TEM−1B, bla_TEM−135, bla_TEM−141, bla_TEM−122, bla_TEM−1 C, bla_TEM−209, bla_TEM−29, bla_TEM−55, and bla_TEM−57. The β‑lactamase gene bla_CARB−2 was detected in isolate 52_MM. Other resistance genes included aac(3)-IIa, ant(2’’)-Ia, aph(6)-Id, aph(3’’)-Ib, floR, mph(A), qnrD1, sul1, sul2, and tet(G) (Supplementary Table 1).

The phylogenetic tree (Supplementary Fig. 3) confirmed the non-clonal distribution of resistance determinants.

Proteus mirabilis

A well-defined cluster of seven P. mirabilis isolates (58_PM, 60_PM, 61b_PM, 62b_PM, 63_PM, 64_PM, 65_PM) met the ≤ 15-allele definition. Distribution across LTCF sections is derived from Supplementary Table 2 (section/host per isolate); while Fig. 4 does not encode section-level metadata. Isolate 66b_PM, despite an MST edge of 11 alleles to 64_PM, is not assigned to the cluster owing to substantial missing allele calls. All these isolates carry bla_TEM−92. Their minimal allelic distance suggests active transmission or a persistent common source. Isolate 59b_PM, although genetically distant, shared an identical resistance profile. While this could be compatible with shared mobile genetic elements, no plasmid or contextual genomic analysis was performed, and therefore any inference of horizontal gene transfer remains hypothetical. All isolates within the main MST group met the ≤ 15‑allele threshold, whereas 57_PM and 67_PM were beyond this cut‑off, indicating a putative independent introductions; both carried broader resistance repertoires, including bla_CTX−M−14, bla_OXA−1, bla_TEM−1B, and multiple aminoglycoside, macrolide, and tetracycline resistance genes (Supplementary Table 1).

Fig. 4.

MST of P. mirabilis isolates based on cgMLST allelic distances (missing alleles ignored). MST illustrating the genetic relatedness among isolates based on allelic profiles derived from cgMLST. Each node represents a unique isolate. Edge labels indicate pairwise allele differences. Clusters are defined as single‑linkage components with all edges ≤ 15 alleles (two‑isolate components included). Coloured shaded areas (clouds) group closely related isolates into clusters, suggesting potential transmission events or clonal expansion within the facility. Because MST edges are computed on called loci and missing alleles are ignored, isolates with substantial missing allele calls are not assigned to clusters even if an MST edge ≤ 15 is displayed (e.g. 66b_PM).

One isolate (67_PM) did not harbour any ESBL-, AmpC-, or carbapenemase-associated genes according to ResFinder, despite having been classified as an MDRO in the 2022 PPS. As the PPS relied on phenotypic screening, whereas the present study applies genotypic detection, such discrepancies are possible and may reflect borderline phenotypes, alternative resistance mechanisms not captured by gene-based prediction tools, or technical variability inherent to point-prevalence screening.

The phylogenetic tree (Supplementary Fig. 4) confirmed the clustering of bla_TEM−92-positive isolates. Resistance gene profiles are detailed in Supplementary Table 1.

Discussion

This study builds upon previous epidemiological investigations⁴ in a LTCF in the Autonomous Province of Bolzano/Bozen, Northern Italy. The LTCF hosts a complex resident population, including non-autonomous individuals, patients with dementia or in a vegetative state, and those with multiple comorbidities. The facility is organised into five units, each accommodating residents with varying degrees of independence and care needs, which directly influence the frequency and nature of staff contact. While the previous study quantified colonisation rates and identified clinical risk factors, this genomic analysis provides high-resolution insights into the clonal structure, transmission dynamics, and resistance mechanisms of MDR organisms circulating within the facility.

In K. pneumoniae, the within‑section clustering and high CTX‑M‑15 prevalence observed are compatible with recent localised transmission. These contextual data support an interpretation of environmental persistence with opportunities for on section spread in LTCFs^13,14.

Section 5 accommodates residents with severe neurological impairment and high care needs. These residents require close assistance from stable staff and shared equipment for mobilisation. Although rooms in Sect.  5 are single or double with private bathrooms, residents are non-ambulatory and frequently require staff assistance to mobilise using shared devices. This results in repeated close-contact care and potential indirect environmental exposure. Despite being physically more isolated within the facility, these structural and organisational characteristics may have favoured indirect contact and contributed to the observed clustering of ST39. These contextual data are compatible with the hypothesis of environmental persistence and on-ward spread in LTCFs, although direct evidence is lacking. The finding of a specific cluster only in Sect.  5 may also be due to the fact that these patients are never moved to other sections.

The genomic data analysed in this study were obtained during the 2022-point prevalence screening, by which time the organisational structure of the facility had largely returned to its usual configuration following the extensive reorganisation caused by the pandemic. These organisational changes may have contributed to the dissemination and persistence of specific lineages within the facility. However, a direct temporal and causal relationship with the observed genomic clusters cannot be definitively demonstrated.

The presence of high‑risk clones (e.g., ST405 and ST307) reinforces clinical concern: ST405 typically combines ESBL production and porin alterations, whereas ST307 has repeatedly been linked to constrained treatment options. Even sporadic introductions may therefore have disproportionate impact in LTCFs, underscoring the need for sustained surveillance and targeted infection control programs¹⁵.

The widespread presence of CTX-M-15 among K. pneumoniae isolates (90%) reflects its dominance in European LTCFs and its association with epidemic plasmids and successful clonal lineages. This aligns with data from the European Antimicrobial Resistance Surveillance Network (EARS-Net¹⁶. The co-occurrence of bla_CTX−M−15 with bla_OXA−1 and bla_TEM−1D in several isolates suggests plasmid-mediated dissemination, consistent with previous reports of epidemic plasmids circulating among high-risk clones such as ST39, ST405, and ST307¹⁷.

In contrast, for E. coli, the pattern points to multiple introductions with limited on‑site clustering, consistent with the expanding role of ST131 and ST1193 in LTCFs and hospitals. This interpretation aligns with recent reports on their clinical relevance in older or frail populations^18,19.

The high diversity and presence of globally relevant clones such as ST131 and ST1193 suggest that E. coli colonisation in LTCFs is driven more by external introductions and gene mobility than by local transmission. Although some clusters were observed, the majority of isolates were genetically distant from each other, and all were collected simultaneously in 2022. The presence of numerous singletons and diverse STs confirms a scenario of multiple, unrelated introductions and complex circulation dynamics within the LTCF. This is consistent with recent longitudinal studies showing that LTCFs are increasingly becoming reservoirs of MDR E. coli, with colonisation rates exceeding 25% and intermittent colonisation associated with higher mortality risk^20,21.

E. coli isolates, exhibiting high genomic diversity, had bla_CTX−M−15 distributed across unrelated lineages. This pattern, supported by the MST, is suggestive of gene mobility; however, since plasmid reconstruction was not performed, the contribution of horizontal gene transfer cannot be confirmed²². The presence of additional ESBLs such as bla_CTX−M−32, bla_CTX−M−55, and AmpC gene bla_DHA−1 in single isolates further supports the notion of multiple independent acquisition events.

M. morganii isolates were largely genetically unrelated, with only one cluster identified (strains 51_MM and 54_MM), with a distance of three loci, suggesting recent clonal transmission or a shared source. All other isolates were genetically distant, indicating sporadic introductions. The species is increasingly recognised for its resistance potential, particularly through high-level AmpC-type β-lactamases and mobile genetic elements. Recent genomic studies have shown that M. morganii harbours a wide array of resistance genes and that its evolution is driven by mobile genetic elements that enhance adaptability to selective pressures^17,23.

Beta-lactam resistance genes in M. morganii, including bla_DHA−14, bla_DHA−17 and various bla_TEM variants, were scattered across single isolates. Notably, the clustered isolates did not carry ESBL genes, reinforcing the hypothesis that resistance in M. morganii is shaped by mobile genetic elements rather than transmission chains²³. The scattered nature of resistance genes and the absence of ESBLs in the only clustered isolates suggest that M. morganii colonisation in the LTCF is driven by sporadic introductions and gene acquisition, rather than sustained transmission chains. This is consistent with its role as an opportunistic pathogen with increasing resistance potential in healthcare settings.

P. mirabilis presented a contrasting scenario, with a well-defined cluster of closely related isolates distributed across multiple LTCF sections. Their minimal genomic distance suggests active transmission or a persistent common source. In contrast, isolates such as 67_PM, 57_PM, and 59b_PM were genetically distant, consistent with independent introductions. The cluster isolates likely represent a circulating MDR clone. P. mirabilis is known for its ability to form biofilms, swarm, and produce urease, enhancing its persistence in healthcare environments. MDR clones may harbour ESBLs, plasmid‑mediated AmpC and, less frequently, carbapenemases.

The coexistence of a tight clonal cluster and distant isolates with similar resistance profiles in P. mirabilis, suggests that clonal expansion and gene mobility may operate in parallel. However, we emphasise that the plasmid context was not resolved here, so any inference of horizontal transfer remains speculative (see Limitations).

The integration of genomic data with epidemiological context highlights that LTCFs can function as reservoirs of MDROs. In this setting, the detection of the high‑risk clone K. pneumoniae ST307 and the globally disseminated E. coli lineage ST131, together with the presence of diverse and sporadic lineages, suggests opportunities for introduction and persistence within the facility, without implying any demonstrable directional flow between the LTCF and acute‑care hospitals. Frequent patient transfers, shared medical staff, and challenges in implementing rigorous infection control measures can contribute to the bidirectional flow of resistant organisms.

Of note, K. pneumoniae isolate 24_KP was obtained from a patient in the hospital Geriatrics section and did not belong to any of the clusters identified within the LTCF; similarly, E. coli isolates 18_EC and 19_EC, recovered from two distinct patients in the same hospital section, were genetically unrelated to LTCF-associated strains. In addition, isolates 16_EC and 17ii_EC, obtained from LTCF staff members, did not cluster with any resident-derived isolates. The complete per‑isolate metadata (source and section) for the 2022 screening were reported previously⁴ and are not reproduced verbatim in this manuscript. Their genetic distinctiveness might indicate that staff colonisation did not contribute to the transmission chains observed within the LTCF, supporting the hypothesis that the identified clusters originated and propagated within the LTCF environment itself. However, these conclusions rely on a single point prevalence screening, which provides only a snapshot in time. Previous or intermittent transmission involving staff or hospital sources cannot be entirely ruled out. Longitudinal sampling would be required to clarify these temporal dynamics.

The high prevalence of bla_CTX−M in K. pneumoniae and E. coli and the detection of bla_DHA in M. morganii were reported previously by Nitti et al.⁴; our WGS results provide the genomic context for those observations. The single carbapenemase-producing K. pneumoniae isolate carrying bla_KPC−3 was confirmed both genotypically and phenotypically.

The shift from E. coli to K. pneumoniae in the 2022 screening study as the predominant ESBL-producing species, as described previously⁴, is further explained by the detection of high-risk clones and environmental persistence. This integrated approach strengthens the evidence for targeted infection control strategies in LTCFs.

Nevertheless, it should be underlined that the inference of horizontal gene transfer (HGT) and convergent acquisition in P. mirabilis and E. coli is based on the distribution of resistance genes among genetically distinct isolates. Because no plasmid reconstruction or contextual genomic analysis was undertaken, the genomic context of resistance determinants cannot be resolved, and any interpretation regarding horizontal gene transfer must be considered hypothetical and based on gene presence patterns rather than direct evidence of mobile genetic components. In comparable hospital settings, hybrid analyses of short and long reads have been used to reconstruct ESBL/carbapenemase plasmids and to distinguish clonal expansion from plasmid-mediated spread²⁴. This approach has also revealed links between ICU patients and the environment. Furthermore, recent evaluations suggest that, when contemporary chemistries and methylation-aware workflows are employed, long-read-only assemblies can achieve hybrid-level accuracy in plasmid recovery²⁵.

Issues related to infection control increased during the COVID-19 emergency. As a matter of fact, internal reorganisation of the LTCF led to frequent patient transfers between sections, including the relocation of patients in vegetative states to dementia units inside the LTCF. These movements, driven by emergency needs, likely facilitated cross-sectional transmission of MDROs. Additionally, limited availability of personal protective equipment (PPE) led to increasing risk of pathogen spread²⁶. In the pandemic period, several professionals (e.g., physicians, therapists, and cleaning staff) operated across multiple sections, promoting inter-section contact. Moreover, a growing number of foreign nurses and nurse auxiliaries shared accommodations in groups of 4–5, potentially increasing the risk of MDRO transmission through close contact. High staff turnover and the arrival of personnel from high-risk hospital departments further contributed to the introduction of colonised individuals into the facility^27,28.

Because all isolates were collected at a single time‑point, we cannot temporally align the emergence of specific clusters with operational changes during the COVID‑19 period. The observations on patient transfers, cross‑coverage of staff, and PPE constraints are therefore contextual hypotheses rather than time‑resolved causal inferences.

The LTCF’s social model, which encourages communal activities and interaction among residents, inherently increases opportunities for close contact. Family visits, often involving physical gestures such as hugs and caresses, represent additional transmission vectors, especially when visitors are unaware of or non-compliant with infection prevention measures²⁹.

While the previous study⁴ identified colonisation rates and associated risk factors, this work introduces a genomic dimension to the investigations. Importantly, it demonstrates how genomic surveillance can reveal complex transmission networks and inform targeted interventions within a single facility.

Whilst the genomic analysis provided valuable insights, it is important to acknowledge the limitations of the study. First, the study was conducted in a single LTCF with a relatively small sample of 67 isolates in total. This limited sample size may underestimate the true extent of the transmission network and may fail to capture rare clones or infrequent transmission events. This study was conducted in a single LTCF, which limits the generalisability of the findings to settings with different structural, demographic, or organisational characteristics. The isolates analysed were collected during a single point prevalence screening, which prevented longitudinal assessment of colonisation dynamics and limited the ability to capture temporal transmission patterns. Because sampling was restricted to a single time‑point, we could not temporally align cluster emergence with COVID‑related operational changes; such links should be considered contextual rather than causal.

Besides, although patient transfers, shared staff and communal living are all potential drivers of transmission in LTCFs, no detailed epidemiological data were available for this study. Therefore, the proposed transmission routes should be interpreted as hypotheses, primarily supported by genomic clustering patterns.

Additionally, the lack of plasmid reconstruction or long‑read sequencing represents a major limitation. As a result, the genomic context of resistance determinants could not be resolved, and any interpretation regarding horizontal gene transfer must be considered hypothetical. Future work will prioritize long‑read sequencing and plasmid characterization on representative isolates.

Regarding cleaning protocols, the contracted cleaning company has its own procedures for cleaning the LTCF, which are in line with those of the South Tyrol Healthcare Authority (SABES). The Cleaning and Disinfection Service of the SABES regularly monitors cleaning procedures.

Furthermore, despite the potential role of environmental reservoirs might be relevant, no environmental sampling was performed, precluding direct confirmation of these hypotheses. Organisational factors, including but not limited to frequent patient transfers during the period of the pandemic, shared staff across multiple sections, communal living arrangements, and close contact with visiting family members, likely influenced the circulation of MDROs within the facility. However, these factors could not be systematically measured or controlled for, and as a result, the impact of these factors remains speculative.

These limitations underscore the necessity for further investigations incorporating longitudinal sampling, environmental and staff surveillance, and clinical correlation to fully elucidate the transmission dynamics and impact of MDROs in long-term care settings.

Conclusions

This study shows that genomic surveillance can uncover complex transmission dynamics of MDROs within LTCFs. Organisational factors such as patient transfers and shared staff likely contributed to dissemination. The predominance of CTX-M-15 and detection of carbapenemase-producing K. pneumoniae underscore the need for targeted infection control. Continuous genomic monitoring, integrated with stewardship and environmental interventions, is essential to mitigate MDRO spread in vulnerable care settings.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AMR

Antimicrobial Resistance

AMS

Antimicrobial Stewardship

BV-BRC

Bacterial and Viral Bioinformatics Resource Center

cgMLST

Core Genome Multilocus Sequence Typing

EARS-Net

European Antimicrobial Resistance Surveillance Network

ESBL

Extended-Spectrum Beta-Lactamase

ExPEC

Extraintestinal Pathogenic Escherichia coli

HGT

Horizontal Gene Transfer

LTCF

Long-Term Care Facility

MDROs

Multidrug-Resistant Organisms

MDR

Multidrug-Resistant

MLST

Multilocus Sequence Typing

MRSA

Methicillin-Resistant Staphylococcus aureus

MST

Minimum Spanning Tree

NCBI

National Center for Biotechnology Information

PATRIC

PathoSystems Resource Integration Center

PPE

Personal Protective Equipment

PPS

Point Prevalence Screening

VRE

Vancomycin-Resistant Enterococci

WGS

Whole-Genome Sequencing

Author contributions

I. B. designed the study, performed the whole-genome sequencing and the genomic analyses, prepared the figures and wrote the initial draft of the manuscript. R. A. contributed to the study’s design, carried out antimicrobial susceptibility testing, and participated in the interpretation of the results. P.C. contributed to the study’s design and acted as project manager. E.S., M.P. and F.L. managed patient-related activities. S.S. and D.R. secured funding. E.P. contributed to the study’s design and supervised the study. All authors reviewed and approved the final version of the manuscript.

Funding

This work was supported by the South Tyrolean Healthcare Service, Italy (grant number 2025 − 128).

Data availability

Whole-genome sequencing data supporting this study have been deposited in the National Center for Biotechnology Information (NCBI) database under BioProject accession number PRJNA1347242.To avoid duplication and respect copyright/data‑governance constraints, per‑isolate metadata for the 2022 cohort (source and section) are available in the previously published study ; in the present manuscript, provenance is referenced only where required for interpretation.All other relevant data are provided within the manuscript and its supplementary materials.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

The study was conducted using bacterial isolates collected as part of surveillance activities. No patient-identifiable data were used. The study was reviewed and approved by the Ethics Committee for Clinical Trials of the Autonomous Province of Bolzano/Bozen (approval issued during the 41st ordinary meeting, item 21, on November 16, 2022). No protocol ID or approval number was assigned. Informed consent was obtained from the participants. All procedures were carried out in strict accordance with the relevant guidelines and regulations, and a statement to this effect has been included in the Methods section.