Investigation of transmission rate of carbapenemase-producing carbapenem-resistant Enterobacteriaceae among contacted patients and healthcare workers in a tertiary hospital

To reduce transmission of carbapenemase-producing carbapenem-resistant Enterobacteriaceae (CP-CRE), it is recommended that patients and healthcare workers in close contact with CP-CRE-positive patients be screened. The aim of this study was to investigate the transmission rate of CP-CRE to patients and healthcare workers in close contact with CP-CRE-positive patients using whole-genome sequencing (WGS) with remeasurement of the minimum inhibitory concentration (MIC). This study was conducted in a between 2017 and 2019. The index patients were those found to be CP-CRE-positive during hospitalization and subsequently isolated in single rooms with contact precautions. Screening tests were performed on the patients who shared a with the index patients for at least one day and the healthcare workers who cared for them. When CRE was isolated, the presence and type of carbapenemase was determined using polymerase chain reaction, and WGS with remeasurement of the MIC was performed to determine the strain type and to verify the association between the CRE and that of the index patient. When the CP-CRE identied in patients or healthcare workers in contact with an index patient had a different type of carbapenemase, the pairs were excluded from genetic analysis.

In particular, CRE that produce carbapenemases, enzymes that hydrolyze carbapenems, are called carbapenemaseproducing CRE (CP-CRE) [8]. Carbapenem resistance, which is not related to carbapenemase production but is associated with porin de ciency or overexpression of e ux pumps, is practically non-transferable between bacteria [9,10]. In contrast, CP-CRE can easily transmit their resistant genes to other bacteria through plasmid transfer [9,10]. Therefore, CP-CRE is considered a greater hazard to public health than carbapenemase-non-producing CRE (CNP-CRE), and stricter infection control should be implemented when CP-CRE is detected [11].
The Korea Disease Control and Prevention Agency guidelines for controlling the transmission of CP-CRE in hospitals states that if CRE is cultured from a hospitalized patient's specimen, polymerase chain reaction (PCR) tests for detecting carbapenemase genes should also be performed. When carbapenemase genes are detected by PCR, the patient should be isolated in a single room with contact precautions. Moreover, all patients who shared a room overnight and all the healthcare workers who made contact with the patient are required to undergo CRE screening [11].
However, the actual rate of positivity in CP-CRE for patients and healthcare workers in close contact is not well known. A few studies have reported the rate to be between 2.8% and 3.2% [12,13]. One retrospective study, which reviewed 211 CP-CRE index patients and 1,369 contacted patients between 2010 and 2017 in a tertiary teaching hospital found that carbapenemases were transmitted to 44 contacted patients (3.2%) [13]. Another prospective study of household contacts between 2015 and 2018 found the rate of CP-CRE transmission to family members to be 2.8% (5 of 177 contacts) [12].
The aim of this study was to determine the actual rate of positivity in CP-CRE screenings of patients and healthcare workers exposed to CP-CRE-positive patients using whole-genome sequencing (WGS) with remeasurement of the minimum inhibitory concentration (MIC).

Study design
This retrospective study was carried out from January 1, 2017 to December 31, 2019 at the Seoul National University Hospital in the Republic of Korea, a tertiary teaching hospital with 1,751 beds. The MIC of each antibiotic was determined using MicroScan WalkAway (Omron Microscan Systems Inc., Renton, WA, USA). Carbapenem resistance was de ned as reduced susceptibility to any carbapenem according to the Clinical and Laboratory Standards Institute criteria (imipenem, meropenem ≥ 2 μg/mL of MIC; ertapenem ≥ 1 μg/mL of MIC) [14]. When CRE was detected in the cultures of any clinical specimens, the Carba NP and PCR tests were conducted to detect carbapenemase genes. The PCR tests targeted carbapenemase genes such as Klebsiella pneumoniae carbapenemase (KPC), imipenemase (IMP), Verona Integron-encoded metallo-β-lactamase (VIM), New Delhi metallo-β-lactamase (NDM), and oxacillinase (OXA).
CP-CRE index patients were de ned as patients with positive tests for CP-CRE from any infected or colonized site during hospitalization who had contact with other patients in a shared room and healthcare workers before CP-CRE was identi ed. CP-CRE-contacted patients were those who had shared a room with an index patient for at least one day, and exposed healthcare workers were those that had contact with the patient at least once before they were identi ed. When CP-CRE was detected, the patient was isolated in a single room with contact precaution regardless of whether CP-CRE was a pathogen or a colonizer. Patients and healthcare workers who were in close contact also underwent CRE screening. The specimens from the contacted patients and healthcare workers were taken as rectal swabs. CRE screening tests utilized the disc diffusion method by using meropenem (10 μg) discs, with a meropenem-resistance cut-off value of < 25 mm, according to the European Committee on Antimicrobial Susceptibility Testing criteria [15]. When an contacted patient was positive for CRE, a PCR test for carbapenemase genes was performed to check for the presence and type of carbapenemase.
Whole genome sequencing for strain typing For strain typing and to verify the genetic association between the CP-CRE-positive index patients and secondary patients (contacted patients who had the same type of carbapenemases as the index patient), WGS was conducted for each CP-CRE index patient-secondary patient pair. Additionally, WGS was performed for each CP-CRE index patient-CNP-CRE detected patient pair to investigate any genetic associations.
Isolates from each CP-CRE index patient-secondary patient pair and CP-CRE index patient-CNP-CRE-detected patient pair were cultured in MacConkey agar plates. Bacterial DNA was extracted using InstaGene Matrix (Bio-Rad Laboratories, Hercules, CA, USA) and the concentration was measured using the QuantFluor ONE dsDNA System (Promega, Madison, WI, USA). The library was prepared using the Nextera DNA Flex kit (Illumina Inc., San Diego, CA, USA) and run on an iSeq TM 100 System (Illumina Inc.) following the manufacturer's instructions. The fastq data obtained were assembled using the Microbial Genomics Module of the CLC Genomics Workbench (Qiagen, Aarhus, Denmark). The same program was then used to align and compare the assembled readings for each pair. Bacterial strain typing was conducted by comparing each sequence with all reference sequences for four bacterial species (K. pneumoniae, K. aerogenes, E. cloacae, and E. coli), and phylogenies of the samples were drawn using the k-merbased tree construction method according to the neighbor-joining algorithm. To show consistency between genotyping results and phenotypic characteristics, the MIC for imipenem and meropenem of each bacterium was remeasured using Sensititre (Thermo Fisher Scienti c, Massachusetts, USA).
Additionally, multilocus sequence typing (MLST) analyses were conducted for each CP-CRE index patient-secondary patient pair and CP-CRE index patient-CNP-CRE detected patient pair to investigate whether plasmid transfer of carbapenemases had occurred. MLST was performed using the NGS-MLST tool in the CLC Genomics Workbench. Each PubMLST scheme was applied to each species for strain typing. Several housekeeping genes (gapA, infB, mgh, pgi, phoE, rpoB, and tonB for K. pneumoniae; dinB, icdA, pabB, polB, putP, trpA, and trpB for E. coli; and dnaA, fusA, gyrB, leuS, pyrG, rplB, and rpoB for E. cloacae) were selected for MLST, and several antibiotic resistance genes (KPC, OXA, Cefotaximase-Munich (CTX-M), Temoinera (TEM), and Sulfhydryl Variable (SHV)) were used for plasmid analysis. Primers were designed for these housekeeping and resistance genes, and nested PCR assays were conducted for each gene. Ampli ed sequences of each gene were also analyzed using the method described above.

Data collection and statistical analysis
Demographic and clinical information about the CP-CRE index and contacted patients was collected from the electronic medical records and infection control department reports of the study hospital. The data were reviewed for the following CP-CRE risk factors: (a) length of hospital stay, (b) previous hospitalizations, (c) previous need for intensive care, (d) presence of indwelling catheters, (e) bed-ridden state, (f) recent invasive procedures, (g) immunocompromised state, and (h) use of antibiotics with broad spectrum coverage [16][17][18][19][20].
All descriptive and statistical analyses were performed using Predictive Analytics Software for Windows (version 25.0; SPSS Inc., Chicago, IL, USA), and p-values less than 0.05 were considered signi cant.

Results
A total of 66 patients had CRE-positive culture studies, carbapenemase NP tests, and carbapenemase PCR tests between January 1, 2017 and December 31, 2019 ( Figure 1). Among them, 19 patients either had no contact with other patients, or all contacts had been discharged from the hospital before CP-CRE-positive results came out. The remaining 47 patients with CP-CRE were therefore considered to be index patients, with 152 patients and 54 healthcare workers potentially contacted. Among the 152 contacted patients, 135 (88.8%) did not have CRE, 10 (6.6%) had CNP-CRE, and 7 (4.6%) had CP-CRE. None of the healthcare workers had CRE. Of the 7 CP-CRE-positive patients, four had the same types of carbapenemase as the CP-CRE index patients they were in contact with (all of which were KPC).
The median hospital stay for contacted patients before CP-CRE screening was 12 days (interquartile range 7-23 days), and the median duration the room was shared with a CP-CRE index patient was 3 days (interquartile range 2-7 days) ( Table 2). Seventy-eight (51.3%) patients who were in close contact with an index patient had a history of previous hospitalization, and 27 (17.8%) had intensive care within the previous 3 months. Fifty-eight (38.2%) patients used anti-pseudomonal penicillins for more than 3 days within 1 month of CPE surveillance. For 56 (36.8%), 45 (29.6%), 35 (23.0%), and 30 (19.7%) patients, third or fourth generation cephalosporins, quinolones, glycopeptides, and carbapenems were administered, respectively.
Genomic evaluation in isolates from CP-CRE index and contacted patients Only 7 contacted patients had CP-CRE-positive surveillance tests, of whom 4 were secondary patients, and ten had CNP-CRE-positive tests. As described above, WGS with remeasurement of the MIC and MLST were conducted in the 4 CP-CRE index patient-secondary patient pairs with the same carbapenemase types and the 7 CP-CRE index patient-CNP-CRE-detected patient pairs. A total of 22 CP-CRE index patients, secondary patients, and CNP-CREdetected patients with available specimen cultures or CP-CRE surveillance results were subjected to genomic evaluation (Table 3). There were 10 clusters in total, and each cluster was composed of 2 or 3 patients.
From the 22 samples, each bacterium was genetically typed to all reference sequences of four bacterial species (K. pneumoniae, K. aerogenes, E. cloacae, and E. coli), and a phylogenetic tree was constructed based on the bacterial strain typing results by k-mer-based tree construction (Figure 2). CP-CRE index patient-CP-CRE-detected patient pairs, such as (I-2, C-2), (I-4, C-4), and (I-7, C-7) showed a genetic distance of less than 0.0005 from each other, and they were all reported to be K. pneumoniae. However, (I-3, C-3), another CP-CRE index patient-secondary patient pair, showed a genetic distance of 0.978 from each other, with I-3 classi ed as E. coli and C-3 as K. pneumoniae.
To con rm consistency between genotyping results and phenotypic characteristics, the MICs of imipenem and meropenem in each sample were remeasured (Table 4). Using the essential agreement de nition of ± 1 log 2 dilution error, (I-2, C-2), (I-4, C-4), and (I-7, C-7) showed similar MIC results between the CP-CRE index patient and the CP-CRE detected patients. The MIC results for other pairs, including (I-3, C-3), however, did not show accordance between the index patient and the detected patient. Although (I-3, C-3) was classi ed as a CP-CRE index patient-secondary patient pair, I-3 and C-3 were not in accordance with the carbapenem MIC, which could suggest a mismatch between their resistance genes or plasmids.
Results from the MLST, including plasmid analysis, also were similar to the WGS and MIC remeasurement results. The pairs of (I-2, C-2), (I-4, C-4), and (I-7, C-7) between the index and contacted patient were in accordance with the distribution of housekeeping genes and resistance genes in plasmids and the types of carrying plasmids (Supplementary Table 1). Based on the WGS and carbapenem MIC remeasurement results, with the additional MLST, the actual rate of CP-CRE transmission in CP-CRE surveillance was calculated as 1.5% (3/206) (95% con dence interval, 0.3%-4.2%).

Clinical characteristics of CPE-transmitted cases
There were only three CP-CRE transmissions con rmed by WGS and MIC remeasurement among the 206 patients and healthcare workers in contact with CPE-positive patients, so it was impossible to analyze the risk factors for CP-CRE transmission. Instead, the clinical characteristics of the three CP-CRE transmission cases are described in detail in a separate table (Supplementary Table 2). The median duration a room was shared with an index patient in the CP-CRE-transmitted patients and non-transmitted patients were 4 days (interquartile range 2-4 days) and 3 days (interquartile range 2-7 days), respectively. There was no signi cant difference between the two groups using the Mann-Whitney U-test (p-value = 0.857).

Discussion
The increasing prevalence of CP-CRE is a global concern, resulting in strict guidelines due to the possibility of plasmid transfer to other bacteria [21]. Therefore, CP-CRE transmission and screening guidelines have been implemented across the world. In Korea, when a source patient with CP-CRE is detected, the guidelines recommend isolation in a single room with contact precaution, and all potentially contacted healthcare workers and patients should undergo CP-CRE screening tests [11]. However, the rate of CP-CRE-positive screening tests has not been well studied, though a few studies have reported the rate to be between 2.0% and 3.2% [12,13].
In this study, the actual rate of CP-CRE transmission was investigated using WGS with MIC remeasurement. The CP-CRE transmission rate was 2.0% (3/152; 95% con dence interval, 0.4-5.6%) in patients who shared a room with an index patient. The 54 healthcare workers tested were all negative for CP-CRE (95% con dence interval, 0.0-0.1%). The calculated rate of CP-CRE transmission was relatively low, so the bene t of CP-CRE screening may not be high.
Considering that the rate of CP-CRE transmission was zero in healthcare workers, it can be inferred that not all CP-CRE contacted healthcare workers must undergo CP-CRE screening. Moreover, the three cases of CP-CREtransmission showed similar MIC results between the CP-CRE index patients and CP-CRE-detected patients. This might suggest that genotypic identity implies an accordance with phenotypic traits, such as antibiotic resistance and transmission of carbapenemase between index and contacted patients.

Page 7/19
The transmission rate of CP-CRE could be reduced by the implementation of standard infection control measures, especially strict hand hygiene [22]. Poor hand hygiene among healthcare workers could signi cantly increase CP-CRE transmission [22]; therefore, the compliance rate for hand hygiene among medical staff in hospitals should be reviewed. From 2017 to 2019, the rate of hand hygiene observation among healthcare workers for each year was approximately 95.0% in the study hospital. Our medical staff had good compliance with hand hygiene, and these rates were maintained for 3 years. Moreover, the three cases of CP-CRE transmission occurred in May, August, and October, 2018. While these cases occurred in similar periods, they were not linked epidemiologically with one another since they did not share the same pathways as each of the CP-CRE index patient-CP-CRE-detected patient pairs. Consequently, they were not considered CP-CRE outbreaks.
WGS analysis showed that CP-CRE from 75% (3/4) of the secondary patients was in accordance with CP-CRE from the index patients. Therefore, the transmission of plasmids carrying carbapenemase genes between index and secondary patients is possible. In addition, some pairs on the phylogenetic tree were not related to each other in terms of time and place, but they showed genetic identity with each other. This phenomenon could indicate that some bacterial strains may colonize the general environment of the hospital and be transmitted to inpatients.
Several studies have identi ed risk factors for CP-CRE in a single patient. One matched case-control study investigated 58 patients with CP-CRE among 621,623 admitted patients between 2011 and 2016, and found the following risk factors for CP-CRE: length of hospital stay more than 20 days, history of hospital admission within 1 year, and the use of antibiotics for more than 10 days [16]. Another retrospective study analyzing 303 CP-CRE patients and 5,929 control patients found that the risk factors for CP-CRE were long inpatient stays, mechanical ventilation, dialysis, and exposure to broad-spectrum antibiotics [20]. However, few studies have investigated the risk factors for transmission of CP-CRE in a shared room, and factors precipitating CP-CRE transmission have not been clari ed. Risk factors for the transmission of multidrug-resistant bacteria have been reported in several studies, suggesting the duration of stay in the same room and environmental contamination to be likely candidates [12,[23][24][25][26]. One prospective study was conducted on 29 methicillin-resistant Staphylococcus aureus (MRSA) index patients and 84 household contacts between 2005 and 2007, proposing that prolonged exposure to MRSA index patients at home was a signi cant risk factor for MRSA transmission [23]. In this study, the duration of sharing a room was not signi cantly different between the CP-CRE-positive and CP-CRE-negative patients. However, other previous studies have revealed that the contaminated environment of a shared room could in uence CP-CRE transmission in a hospital [24,26].
This study has several limitations. First, there were only a few CP-CRE-transmission cases, so it was impossible to identify signi cant risk factors for transmission. Second, environmental cultures in the environment around where CP-CRE index patients stayed was not conducted; therefore, information about environmental contamination could not be investigated. Moreover, several factors that could affect the spread of CP-CRE in a hospital, such as the number of antibiotic stewardship cases per 1000 patient-days, the ratio of caregivers to patients, and the amount of hydroalcoholic products used were not investigated [22]. Further studies considering environmental cultures, the enforcement rate of antibiotic stewardship, the ratio of caregivers to patients, and a greater number of CP-CRE transmitted cases should be conducted.

Conclusions
The rate of transmission of CP-CRE to patients who shared a room with CP-CRE-positive patients was calculated as 2.0% based on genetic and phenotypic analyses in a tertiary teaching hospital, while the rate of transmission to contacted healthcare workers was zero. Considering the low transmission rate of CP-CRE in shared rooms, screening tests for patients contacted to CP-CRE could be implemented in speci c cases only.     The phylogenetic tree of CP-CRE index patients and those in close contact who were CRE-positive. I, index; C, colony The phylogenetic tree of CP-CRE index patients and those in close contact who were CRE-positive, which was drawn using the k-mer-based tree construction method. The numbers below the horizontal lines indicate the genetic distance between the two bacterial colonies represented by dots. The strain typing results based on WGS are shown with the bacterial names. The colonies in the colored boxes represent the pairs considered to be the transmitted cases.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. CPETransmissionSupplementaryTablesARIC210424.docx