Colonization of long-term care facility residents in three Italian Provinces by multidrug-resistant bacteria

Rationale and aims of the study were to compare colonization frequencies with MDR bacteria isolated from LTCF residents in three different Northern Italian regions, to investigate risk factors for colonization and the genotypic characteristics of isolates. The screening included Enterobacteriaceae expressing extended-spectrum β-lactamases (ESβLs) and high-level AmpC cephalosporinases, carbapenemase-producing Enterobacteriaceae, Pseudomonas aeruginosa or Acinetobacter baumannii, methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococci (VRE). Urine samples and rectal, inguinal, oropharyngeal and nasal swabs were plated on selective agar; resistance genes were sought by PCR and sequencing. Demographic and clinical data were collected. Among the LTCF residents, 75.0% (78/104), 69.4% (84/121) and 66.1% (76/115) were colonized with at least one of the target organisms in LTCFs located in Milan, Piacenza and Bolzano, respectively. ESβL producers (60.5, 66.1 and 53.0%) were highly predominant, mainly belonging to Escherichia coli expressing CTX-M group-1 enzymes. Carbapenemase-producing enterobacteria were found in 7.6, 0.0 and 1.6% of residents; carbapemenase-producing P. aeruginosa and A. baumannii were also detected. Colonization by MRSA (24.0, 5.7 and 14.8%) and VRE (20.2, 0.8 and 0.8%) was highly variable. Several risk factors for colonization by ESβL-producing Enterobacteriaceae and MRSA were found and compared among LTCFs in the three Provinces. Colonization differences among the enrolled LTCFs can be partially explained by variation in risk factors, resident populations and staff/resident ratios, applied hygiene measures and especially the local antibiotic resistance epidemiology. The widespread diffusion of MDR bacteria in LTCFs within three Italian Provinces confirms that LTCFs are an important reservoir of MDR organisms in Italy and suggests that future efforts should focus on MDR screening, improved implementation of infection control strategies and antibiotic stewardship programs targeting the complex aspects of LTCFs.


Background
Life expectancy in Italy is rapidly increasing, with present values of 80.1 years for males and 84.7 for females [1]. Due to the ageing population, long-term care facilities (LTCFs), which provide ongoing skilled nursing care to residents and help meet both the medical and non-medical needs of elderly individuals with a chronic illness or disability, play an important role in the Italian healthcare system. Residents in LTCFs have a variety of risk factors for colonization with multidrug-resistant (MDR) bacteria; therefore, these facilities represent reservoirs of: i) Enterobacteriaceae expressing extended-spectrum β-lactamases (ESβLs), derepressed/acquired high-level AmpC cephalosporinases or carbapenemases, ii) Pseudomonas aeruginosa or Acinetobacter baumannii producing carbapenemases and iii) methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococci (VRE) [2][3][4].
To promote detailed studies of various microbiological aspects related to LTCFs in Italy, the Association of Italian Clinical Microbiologists (Associazione Microbiologi Clinici Italiani; AMCLI) in 2016 has set up a new working group consisting of Clinical Microbiologists (Gruppo di Lavoro per lo Studio delle Infezioni nelle Residenze Sanitarie Assistite e Strutture assimilabili; GLISTer); one of the main objectives of this working group is the study of the distribution and prevalence of MDR organisms in Italian LTCFs and therefore a multicenter point-prevalence survey, including the main MDR bacteria as described above, was performed in 2016 on residents of LTCFs, located in three Northern Italian cities.

The aim
Rationale and aims of the study were to compare colonization frequencies with MDR bacteria of LTCF residents in three different Northern Italian cities, located in different Italian regions, and to investigate their genotypic characteristics. Moreover, risk factors for colonization were compared between LTCFs and colonization prevalence was correlated with the local epidemiology of invasive MDR isolates.

Facilities, patient characteristics and survey design
In October-November 2016, a multicenter pointprevalence screening study was conducted in four LTCFs concerning i) Enterobacteriaceae with ESβLs, carbapenemases or high-level AmpCs, ii) P. aeruginosa or A. baumannii with carbapenemases, iii) MRSA and VRE. The four facilities, located in the Northern Italian Provinces of Milan (n = 1), Piacenza (n = 2) and Bolzano (n = 1), offer high skilled 24 h nursing care.
Although the overall study was performed over a period of 2 months, the sampling interval in each facility lasted for a maximum of 1 week. All residents of the four LTCFs were eligible to participate, and the study was approved by the Ethics Committees of the three referring hospitals; informed written consent was obtained from the residents or, if they were unable to consent, from their relatives.

Microbiological methods
Sample processing, microbial identification and antibiotic susceptibility testing were carried out in the clinical microbiology laboratories of the referral hospitals. Microbiological methods for the LTCF screening study in Bolzano were previously described [5]. Similar methods were used in the epidemiological studies of Milan and Piacenza LTCFs, with minor modifications.
For the screening of MDR bacteria from LTCF residents in Milan midstream or catheter urine samples were cultured on Oxoid Brilliance™ ESβL plates (Thermo Scientific, UK), applying a 10 μg imipenem (IMP) disc (Oxoid, Thermo Scientific, UK), and on Oxoid Brilliance™ VRE (Thermo Scientific, UK). Inguinal, oropharyngeal and rectal swabs were seeded on Oxoid Brilliance™ ESβL, applying a 10 μg IMP disc, on Oxoid Brilliance™ VRE and on CHROMagar™ MRSA (BD Diagnostics, MD). Nasal swabs were plated on CHROMagar™ MRSA. All plates were incubated at 35 ± 2°C under aerobic conditions for 24-48 h. Isolate identification and antibiotic susceptibility testing were performed by the BD Phoenix™ System (BD Diagnostics, MD), according to European Committee on Antimicrobial Susceptibility Testing (EUCAST) criteria [6], using PHOENIX NMIC/ ID402 for non-urinary Gram-negative bacteria, PHOENIX UNMIC/ID403 for Gram-negative isolates from urine cultures, and PHOENIX PMIC/ID88 for MRSA and VRE. The strains were phenotypically confirmed for β-lactamase production by the ESBL+AMPC Screen Kit and the KPC + MBL Confirm ID Kit (Rosco Diagnostica A/S, Denmark).
Bacterial isolates collected from the LTCF in Milan were screened for bla KPC -, bla VIM -, bla OXA-48 -and bla NDM -type genes by the Cepheid GeneXpert System and confirmed by PCR. Check-MDR CT103 XL array (Check points Health B.V., Wageningen, The Netherlands) has been used to investigate the bla gene content of a carbapenem-resistant P. aeruginosa strain obtained from an oropharyngeal swab, which tested negative by previous molecular assays.
For gene sequencing, PCR products were purified using the quantum Wizard® SV Gel and PCR Clean-Up System (Promega, Madison, WT, USA) and subjected to double-strand Sanger sequencing. Sequences were analyzed according to the BLAST software [21].

Statistical analysis
A significance level of p ≤ 0.05 was used. In-house physicians reviewed hospital records and, using a standard questionnaire, recorded demographic and clinical data as follows: patient age, gender, length of stay, Barthel immobility score, coma, comorbidities (dementia, urinary incontinence, diabetes, cancer, vascular diseases, chronic obstructive pulmonary disease, decubitus ulcer), presence of infection, antibiotic treatment in the preceding 3 months and the presence of indwelling medical devices. The significance of differences in risk factors and colonization proportions was calculated using the proportion comparison test. Logistic regression analyses were developed to investigate colonization of at least one site with ESβL producers and MRSA as dependent variables, first as univariate and then as multivariate models, including predictors with p < 0.05 in the univariate analysis, comprising the specific LTCF of residence, using stepwise logistic selection. Analysis were performed using the Medcalc® software version 15.11.4 (MedCalc software, Ostend, Belgium).  Table 1.
In total, ten carbapenemase-producing Enterobacteriaceae were detected: n = 7 KPC-producing K. pneumoniae and n = 1 VIM-1-producing E. cloacae complex were isolated from LTCF residents in Milan, and n = 2 VIM-1 producers (one E. coli and one Citrobacter amalonaticus) from residents in Bolzano. Two carbapenemase-positive P. aeruginosa were isolated from LTCF residents in Piacenza: in one case a bla GES-5 and in the other a bla VIM -like gene were identified. Moreover, two P. aeruginosa isolates collected in Milan and Piacenza presented a bla GES-1 ESβL. Nine bla OXA-23 -positive A. baumannii were isolated from two and seven LTCF residents in Milan and Piacenza, respectively.
MRSA strains were most frequently isolated from LTCF residents in Milan and Bolzano, whereas VRE isolates were highly prevalent in Milan (n = 21 Enterococcus faecalis), but rare in Piacenza (n = 1 E. faecalis) and Bolzano (n = 1 Enterococcus faecium).
Colonization of LTCF residents with ESβL-producing enterobacteria and MRSA was associated with several risk factors in univariate and multivariate analysis ( Table 3). In multivariate analysis, the LTCF of residence was an independent risk factor for ESβL (p ≤ 0.03 for all comparisons, except p = 0.53 for the comparison of Milan vs. Piacenza) and MRSA (p ≤ 0.02 for all comparisons) colonization. Risk factors for MRSA colonization were also associated with resident's gender; for the following risk factors significant differences between male (n = 226) and female (n = 114) residents were found:   blaCTX-M-group-9 6.7 5.

Discussion
The study evaluated the degree of colonization with drugresistant bacteria among residents of LTCFs located in three Northern Italian Provinces, finding high colonization of residents in Milan (75.0%), Piacenza (69.4%) and Bolzano (66.1%). Many residents had more than one target organism, underscoring the role of LTCFs as a reservoir for these isolates [2][3][4].
Colonization of LTCF residents with ESβL-producing enterobacteria was highly prevalent in all the surveyed LTCFs (60.5% in Milan, 66.1% in Piacenza and 53.0% in Bolzano), and group-1 CTX-M-type enzymes were highly predominant, especially in E. coli (80-97% of isolates). Notably, about 82% of K. pneumoniae and 32% of P. mirabilis isolates also harbored a bla CTX-M -type gene. In the same Bolzano LTCF, here screened for ESβL-producing enterobacteria, high colonization percentages, equal to 64.0 and 49.0%, were previously found in 2008 [22] and 2012 [23], respectively; the latter survey also screened a second LTCF in the Province of Bolzano, showing a colonization prevalence of 56.0%. In an Italian study carried out in 2006, a colonization prevalence of 54.0% was found in LTCF residents bearing a urinary catheter [24], while a more recent multicenter study, performed in 2015 and involving 12 Italian LTCFs, reported a mean ESβL colonization of 57.3% (range: 32.8-81.5%) [25]. In all these Italian studies, CTX-M enzymes were the predominantly produced ESβLs. The high ESβL colonization rates of > 50% in Italian LTCF residents are paralleled by high ESβL prevalence in invasive E. coli isolates [26]. Generally, ESβL carriage in most European countries is strikingly lower than that found in Italy [4], with exceptions reported from Ireland [27,28] and Portugal [29].
In our screening study, high-level AmpC-producing Enterobacteriaceae were rarely isolated in LTCF residents in Milan and Piacenza, but 24.3% of LTCF residents in Bolzano were colonized by M. morganii expressing a high-level DHA-AmpC phenotype; bla DHAtype genes in LTCF isolates have previously been found in a few E. coli and K. pneumoniae strains from Korea [30], but to our knowledge have not yet been reported in Italian LTCFs.
Carbapenemase-producing enterobacteria were not found in LTCF residents in Piacenza, rarely in Bolzano (1.6%) and more frequently in Milan (7.6%). As found in previous studies of carbapenemase-producing Enterobacteriaceae from Bolzano [22,23,31], the VIM-1producing E. coli and C. amalonaticus isolates from residents in this study were also positive for bla SHV-12 . In the present study, all carbapenemase producers from Milan, except an E. cloacae complex isolate expressing a bla VIM-1 gene, had KPC-type enzymes; similar results have been reported by other Italian studies in LTCF residents [25,32,33]. Carbapenemase-producing enterobacteria, especially KPC-producing K. pneumoniae, are epidemically spread in Italy [34] and the emergence of this MDR phenotype in LTCFs is worrying, expanding the reservoir of this health care threat. Nevertheless, as previously summarized [4], carbapenemase-producing Enterobacteriaceae are still rare in Italian LTCF residents; the reasons are probably multifactorial, comprising clinical characteristics of the enrolled residents [35] and the low carbapenem selective pressure in LTCFs. On average, only 1.1% of residents enrolled in our screening study received carbapenems within the previous 3 months (data not shown). Nevertheless, a carbapenemase-producing enterobacteria prevalence of 7.6% (mainly KPC-producing K. pneumoniae), reported here for the LTCF in Milan, gives rise to concern and has to be addressed by future hygiene and antibiotic stewardship measures.
This study shows the emergence of carbapenemaseproducing P. aeruginosa in LTCF residents in Piacenza, identifying single isolates with bla VIM -type and bla GES-5 determinants. P. aeruginosa expressing bla VIM -type determinants is widely spread in Italy [36], and an outbreak of GES-5-producing P. aeruginosa was reported from a LTCF in Japan [37]. Moreover, the ESβL genes bla GES-1 and bla BEL -like were found in two and one P. aeruginosa isolates, respectively; the latter rarely detected βlactamase was previously recovered in P. aeruginosa strains from Belgium [18]. A. baumannii producing OXA-23 carbapenemases have an epidemic diffusion in Italy [38], reflected in the present study by the isolation of this resistance type from LTCF residents in Milan (1.9%) and Piacenza (5.8%).
MRSA colonization prevalence here reported ranged widely in the surveyed LTCFs (5.7, 14.8 and 24.0% in Milan, Piacenza and Bolzano, respectively), similar to other Italian studies [25,39,40]. Varying MRSA  colonization prevalence, ranging from close to zero up to levels higher than 37%, has been reported in European studies [4]. Colonization by VRE in the present study was highly variable, ranging from 0.8 to 20.2%. VRE-carriage in European LTCF residents was found to be low, ranging from 0.0-3% [28,41,42].
For Enterobacteriaceae significant differences in colonization frequencies of LTCF residents were found: i) for CTX-M-type ESβL-producing E. coli between Piacenza (highest prevalence) and Bolzano, ii) for high-level AmpCproducing M. morganii (highest prevalence in Bolzano), iii) for carbapenemase producers, with highest prevalence in Milan, iv) for carbapenemase-producing A. baumannii, showing highest prevalence in Piacenza, and v) for MRSA and VRE, most prevalent in Milan. Therefore, no clear picture of general colonization differences can be deduced from overall colonization prevalence data.
A variety of risk factors for MRSA and ESβL colonization have previously been reported [4]; many of these have also been analyzed in the present survey. Interestingly, male residents carried a more than double risk for MRSA carriage when compared with female residents, probably because of the higher frequencies of other risk factors in males (administration of any antibiotic within the previous 3 months, hospitalization within the previous 12 months and coma), predisposing men rather than women to MRSA acquisition. Moreover, in our study the trend for an inverse correlation (p = 0.09) between age > 85 years and MRSA prevalence was associated with a significantly lower percentage of male residents > 85 years, compared to females; similar results have been found by other authors [43]. In the present survey, administration of cephalosporins during the previous 3 months resulted to be an independent risk factor for ESβL colonization; the LTCFs in Piacenza registered the highest consumption of cephalosporins, correlating with highest ESβL prevalence in LTCF residents from Piacenza. Other independent risk factors for ESβL colonization were physical disability, the presence of any invasive medical device and cancer. Whereas no significant differences were found between residents in the three Provinces for cancer as risk factor, physical disability and the presence of any medical device showed highest prevalence in the LTCF in Bolzano; nonetheless, LTCF residents in Bolzano had the lowest ESβL prevalence in the present screening study.
Therefore, further factors may have contributed to the observed differences, comprising staff/resident ratio and practiced hygiene and infection control measures [44]. The LTCF in Bolzano showed the highest staff/resident ratio, and understaffing has been shown to be a risk factor for colonization of LTCF residents by MDR organisms [2]. All of the surveyed LTCFs in the present study follow hygiene, infection prevention and control measures according to guidelines of The Society for Healthcare Epidemiology of America (SHEA) and The Association for Professionals in Infection Control and Epidemiology (APIC) [45]. Nonetheless, the Bolzano LTCF had introduced enforced hygiene measures, according to the World Health Organization guidelines [46], after the 2008 screening study, showing an ESβL colonization prevalence of 64.0% in LTCF residents [22]; colonization frequency decreased significantly to 49.0% (p = 0.02) in 2012 [23], arriving at a slightly higher percentage of 53.0% in 2016, but other factors such as changed case mixes and risk factors may also have contributed to this decrease in ESβL prevalence [23].
Significant differences in antibiotic resistance epidemiology of blood culture isolates, used as a proxy for the general local antibiotic resistance epidemiology, were registered, as derived from European Antimicrobial Resistance Surveillance Network (EARS-Net) data for 2016 [26]. Specifically, we found the following antibiotic resistance data referred to the geographic regions of Milan, Piacenza and Bolzano, respectively: E. coli third generation cephalosporin-resistant: 22 Moreover, the snapshot approach used in this study might lead to the sudden increase in prevalence of a specific resistance phenotype, as shown for high-level AmpC-producing M. morganii detected in 2016 from Bolzano LTCF residents [5], which could be a transient phenomenon. Similarly, the high prevalence of VRE in LTCF residents from Milan could be due to a transitory local epidemic event.
Finally, the local circulation of highly transmissible clones, for example ESβL-producing E. coli, KPCproducing K. pneumoniae and OXA-23-producing A. baumannii could contribute to the explanation of the here reported screening results [38,47].
This study has some limitations. First, it has been done in only four LTCFs, located in three different Provinces in Northern Italy, and therefore data may not be extrapolated to other Italian LTCFs with differing characteristics. Second, the number of LTCF residents participating in the study was variable, ranging from 34% in Milan up to 100% in Bolzano. Third, we did not use an enrichment step during the laboratory analysis; this limitation is partially compensated by using 4-5 different specimen types for the screening of MDR bacteria. Fourth, different sample types, types of media and laboratory methodologies have been used in the three laboratories processing the samples from the different LTCFs. Fifth, molecular characterization and typing of isolates in the 2016 study was limited, not including pulsed-field gel electrophoresis (PFGE) and sequence typing (ST) of isolates and therefore not permitting the identification of epidemic clusters. Finally, screening of healthcare workers has been done only in one of the enrolled LTCFs [5], but not in the other surveyed facilities. Despite these limitations, the strength of our study is the comparison of colonization prevalence between LTCFs located in three different Provinces, comparing it also with differences in risk factors for colonization and in the local epidemiology of invasive isolates.

Conclusions
We performed a multicenter point-prevalence study in LTCFs located in three different Provinces in Northern Italy and found high colonization prevalence of LTCF residents for MDR organisms, especially ESβL-producing E. coli. Variability between the different facilities was noticeable also for other MDR organisms. Differences can be partially explained by i) differences in risk factors for colonization by MDR organisms, ii) changes in resident populations and staff/resident ratios, iii) applied hygiene measures and iv) differences in the local epidemiology of antibiotic resistance of clinical isolates. This widespread diffusion of MDR bacteria in LTCFs of three Italian Provinces confirms that these healthcare facilities are an important reservoir for MDR organisms. Future efforts should focus on screening activities, infection control strategies tailored on the complex aspects of LTCFs and implementation of antibiotic stewardship programs.

Additional file
Additional file 1: Table S1.