Colonization of residents and staff of a long-term-care facility and adjacent acute-care hospital geriatric unit by multiresistant bacteria


Corresponding author and reprint requests: D. M. Livermore, Antibiotic Resistance Monitoring and Reference Laboratory, Health Protection Agency, Centre for Infections, 61, Colindale Avenue, London NW9 5EQ, UK


Clin Microbiol Infect 2010; 16: 934–944


Long-term-care facilities (LTCFs) are reservoirs of resistant bacteria. We undertook a point-prevalence survey and risk factor analysis for specific resistance types among residents and staff of a Bolzano LTCF and among geriatric unit patients in the associated acute-care hospital. Urine samples and rectal, inguinal, oropharyngeal and nasal swabs were plated on chromogenic agar; isolates were typed by pulsed-field gel electrophoresis; resistance genes and links to insertion sequences were sought by PCR; plasmids were analysed by PCR, restriction fragment length polymorphism and incompatibility grouping. Demographic data were collected. Of the LTCF residents, 74.8% were colonized with ≥1 resistant organism, 64% with extended-spectrum β-lactamase (ESBL) producers, 38.7% with methicillin-resistant Staphylococcus aureus (MRSA), 6.3% with metallo-β-lactamase (MBL) producers, and 2.7% with vancomycin-resistant enterococci. Corresponding rates for LTCF staff were 27.5%, 14.5%, 14.5%, 1.5% and 0%, respectively. Colonization frequencies for geriatric unit patients were lower than for those in the LTCF. Both clonal spread and plasmid transfer were implicated in the dissemination of MBL producers that harboured IncN plasmids bearing blaVIM-1, qnrS, and blaSHV-12. Most (44/45) ESBL-producing Escherichia coli isolates had blaCTX-M genes of group 1; a few had blaCTX-M genes of group 9 or blaSHV-5; those with blaCTX-M-15 or blaSHV-5 were clonal. Risk factors for colonization of LTCF residents with resistant bacteria included age ≥86 years, antibiotic treatment in the previous 3 months, indwelling devices, chronic obstructive pulmonary disease, physical disability, and the particular LTCF unit; those for geriatric unit patients were age and dementia. In conclusion, ESBL-producing and MBL-producing Enterobacteriaceae and MRSA were prevalent among the LTCF residents and staff, but less so in the hospital geriatric unit. Education of LTCF employees and better infection control are proposed to minimize the spread of resistant bacteria in the facility.


Resistant bacteria are a major public health concern. Current problems include Enterobacteriaceae with extended-spectrum-β-lactamases (ESBLs), derepressed or acquired AmpC cephalosporinases and metallo-β-lactamases (MBLs), and also methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococci (VRE). These organisms are most often discussed in the context of hospital infection, but are also important in long-term-care facilities (LTCFs), where they have been associated with increased morbidity, mortality and cost [1]. Colonized LTCF residents may act as vectors for the transfer of resistant bacteria into acute-care hospitals, where these may cause infection in their initial host or spread to other vulnerable patients. Donskey found that 15% of hospitalized patients colonized with resistant Gram-negative bacteria subsequently developed a bacteraemia due to the same strain [2].

Enterobacteriaceae with ESBLs, AmpC enzymes or MBLs are widely seen in clinical samples by the microbiology laboratory in Bolzano, Italy, as are MRSA strains, whereas VRE are rarer. Except for VRE, these organisms are recorded more frequently from LTCF residents than from acute-care hospital patients (R. Aschbacher, unpublished data). Such observations are, however, prone to sampling bias. To test their broader validity, we undertook a point-prevalence study of key resistant pathogens in residents and staff of the LTCF attached to the hospital and in patients of two of the hospital’s geriatric units. Particular attention was paid to MBL-producing and ESBL-producing Enterobacteriaceae, as these have caused clinical problems in the hospital [3].

Materials and Methods

Facility, patient characteristics, and survey design

In October 2008, we conducted a point-prevalence study concerning MRSA, VRE and of Enterobacteriaceae with ESBLs, MBLs or high-level AmpC activity in the 120-bed LTCF attached to the regional hospital in Bolzano, a city of 100 000 people in northern Italy. All residents of all five LTCF units were eligible to participate, and the study was approved by the Ethics Committee of the Bolzano Teaching Hospital. Informed written consent was obtained from the residents or, if they were unable to consent, from their relatives. LTCF staff members were also screened. A parallel 1-day point-prevalence study was performed involving all patients hospitalized in two of the three 24-bed units of the geriatric ward of the acute-care hospital, again with the patients’ written consent. The third unit was excluded, as it mostly serves cancer patients with advanced disease. The LTCF and geriatric unit staff move frequently among the various units, but not between the LTCF and the acute-care hospital.

Microbiological methods

To assess carriage of resistant bacteria, midstream or catheter urine samples and rectal, inguinal, oropharyngeal and nasal swabs from all participants were spread on ChromID ESBL, MRSA and VRE agars (bioMérieux, Marcy l’Etoile, France). The formulations of these media are proprietary, but ChromID ESBL is cefpodoxime-based and allows the growth of bacteria with cephalosporin resistance mechanisms other than ESBL production [4]; ChromID MRSA contains cefoxitin (4 mg/L) and ChromID VRE has vancomycin (8 mg/L). The ESBL plates were incubated for 18–24 h at 35°C, whereas the MRSA and VRE plates were incubated for 48 h. Gram-negative bacteria were subcultured on MacConkey agar, and Gram-positive bacteria on colistin–nalidixic acid sheep blood agar. The subcultured isolates were re-identified and tested for antibiotic susceptibility in the hospital’s clinical microbiology laboratory, using the Vitek 2 System (bioMérieux), calibrated against CLSI susceptibility criteria [5]. AST-GN015 susceptibility cards, with a direct ESBL test, were used for Gram-negative bacteria, AST-GP049 cards, with both oxacillin and cefoxitin, for MRSA, and AST-GP034 cards for enterococci. Phenotypic identification of β-lactamase types among the isolates growing on ChromID was based on the Vitek 2 results and those of β-lactamase-diagnostic Etests (cefotetan/cefotetan + cloxacillin for AmpC; imipenem/imipenem + EDTA for MBLs; and cefotaxime/cefotaxime + clavulanate, ceftazidime/ceftazidime + clavulanate and cefepime/cefepime + clavulanate for ESBLs; all three were from AB Biodisk, Solna, Sweden). In addition, Hodge/clover leaf plates were used for carbapenem-resistant isolates [6]. VRE were also confirmed by Etest, and equivocal MRSA isolates were tested for penicillin-binding protein 2′, using latex agglutination (Slidex MRSA Detection; bioMérieux). MICs for MBL producers and ESBL-producing E. coli isolates were reconfirmed by agar dilution, performed according to BSAC guidelines [7].

Molecular methods

Isolates were typed using pulsed-field gel electrophoresis (PFGE) of XbaI-digested genomic DNA [8], with banding patterns analysed using Bionumerics software (Applied Maths, Sint-Martens-Latem, Belgium); they were considered to be clonally related if there was ≥85% similarity in profile [9].

Multiplex PCR for blaCTX-M genes was performed with published primers [10]. Isolates with MBL phenotypes were tested by PCR with consensus primers for blaVIM-1/2 [11] and blaVIM-1 [12]. Published primers were also used to seek qnrA [13], qnrB [14] and qnrS [15] alleles and blaSHV genes [16]. Sequencing of PCR products derived from blaVIM and qnrS was performed with the same primers used for amplification. For blaSHV, primers SHV-c (ATGCGTTATATTCGCCTGTG) and SHV-d (CTTAGCGTTGCCAGTGCTCG) were used, together with other previously described primers [16]. CTX-M group 1 genes were amplified with the primers CTX-M1orfF (PROM+) and CTX-MorfR (PRECTX-M-3-B) [17], and CTX-M group 9 genes with primers CTX-M 9A and CTX-M 9C [18]; sequencing was performed using the same primers together with published internal primers [10]. Published primers were used to amplify the IS26 insertion element and to investigate linkage of blaCTX-M genes of group 1 with ISEcp1 and IS26 [19]. In all cases, products were first purified with the Geneclean Turbo for PCR Kit (Q-BIOgene, Cambridge, UK), with sequencing by the GenomeLab Dye terminator Cycle Sequencing system, using the Quick Start Kit (Beckman Coulter, High Wycombe, UK) with a Beckman Coulter CEQ 8000 Genetic Analysis System.

Mega X E. coli DH10B T1 electrocompetent cells (Invitrogen, Paisley, UK) were transformed by electroporation with a Bio-Rad Gene-Pulser II at 2.0 kV, 200 Ω and 25 μF, using plasmids extracted by the method of Kado and Liu, and precipitated twice with ethanol [20]. Transformants were selected on Luria–Bertani agar containing cefotaxime (2 mg/L). Plasmids bearing blaVIM-1 were extracted from transformants and digested with HpaI (Promega, Southampton, UK) and BamHI/SacI (Roche, Mannheim, Germany). The resulting fragments were separated by electrophoresis on 1% agarose.

Plasmid typing was by PCR (inc/rep PCR) for the major incompatibility groups [21]. Phylogenetic groups of E. coli were determined by PCR [22].

Epidemiological investigation and statistical analysis

To examine risk factors for colonization with resistant organisms, in-house physicians reviewed hospital records and, using an Epi Info questionnaire, recorded demographic data as follows: patient age, gender, length of stay in the LTCF or geriatric ward, diagnosis at admission, Barthel immobility score, comorbidities (dementia, urinary incontinence, diabetes, cancer, vascular diseases, chronic obstructive pulmonary disease, decubitus ulcers and immunosuppressive therapy), and indwelling medical devices (urinary catheters, percutaneous enteral gastrostomy tubes, tracheostomy tubes and nasogastric tubes). For the LTCF patients, the hospital ward of the previous acute admission, antibiotic treatments in the preceding 3 months and other possible risk factors (urinary incontinence, decubitus ulcers and immunosuppressive therapy) were also recorded. For statistical analysis, three groups of subjects were defined: (i) LTCF residents; (ii) LTCF staff; and (iii) geriatric unit patients. Associations were investigated using the chi-square test or Fisher’s exact method. Group-specific mean ages were compared by the generalized linear model procedure, after testing for homoschedasticity of variances with Levene’s test. In the case of heteroschedasticity, Welch’s test was performed. Logistic regression analyses were developed to investigate colonization with resistant bacteria in general and ESBL producers and MRSA in particular, first as univariate and then as multivariate models through stepwise selection. Analyses were performed using the SAS statistical package, release 9.1.3, and with Epi Info v. 3.5.


Epidemiological investigation

Among 120 LTCF residents present in October 2008, 111 (92.5%) agreed to participate in the point-prevalence survey; among staff members, 69 of 79 (87.3%) agreed to participate, mainly nurses and physicians. Sixty-one (55.0%) of the participating residents were women, with a median age of 84 years (range 22–96 years); the median age of the 50 male residents was 77 years (range 22–94 years). The median length of LTCF stay was 21 months (range: <1–96 months). The most recent hospital admission of 36 of the 111 participating LTCF residents was to one of the two acute-care geriatric units that were also surveyed. Forty-five of the 48 geriatric unit patients also agreed to participate in the investigation: 32 (71.1%) were women (median age 85 years, range 75–97 years) and 13 were men (median age 79 years, range 74–91 years); none was ordinarily a resident in the LTCF.

The isolation frequencies for the various resistance types in different patient groups are shown in Table 1. Among the LTCF residents, 74.8% (83/111) were colonized with at least one resistant organism, and 31.5% (35/111) with at least two. ESBL-producing Enterobacteriaceae, together with MRSA strains, were isolated from 28.8% (32/111), and at least two different ESBL-producing enterobacterial species from 21.6% (24/111). Twenty-seven of 111 LTCF residents, two of 69 LTCF staff members and two of 45 geriatric unit patients were also colonized with Acinetobacter baumannii (which grew on the ESBL ChromID agar), but none of these isolates was resistant to carbapenems, and they were not studied further. Similarly, colonization with Pseudomonas aeruginosa was not investigated.

Table 1.   Long-term-care facility (LTCF) residents, staff and geriatric unit patients colonized with bacteria of various resistance types
Bacteria and resistance typeScreening samples; % colonizedLTCF residents vs. geriatrics unit patients: p-value of differences
LTCF (residents), n = 111LTCF (staff), n = 69Geriatrics unit (patients), n = 45
  1. ESBL, extended-spectrum β-lactamase; MRSA, methicillin-resistant Staphylococcus aureus; VRE, vancomycin-resistant enterococci.

  2. aEnterobacteria producing both an ESBL and an MBL are included in both totals.

  3. bUrine sample of one further patient tested positive for two MBL-producing enterobacterial species (Providencia stuartii, M. morganii) shortly before the screening period; inclusion of these samples brings the MBL-positive rate to 6.3% (7/111).

  4. cAll three VRE were Enterococcus faecalis.

All resistance groups74.827.522.2<0.0001
All enterobacteria, ESBL-positivea64.014.58.9<0.0001
 Escherichia coli, ESBL-positive41.411.66.7<0.0001
 Proteus mirabilis, ESBL-positive24.31.52.2<0.0001
 Klebsiella pneumoniae, ESBL-positive11.71.5 0.034
 Morganella morganii, ESBL-positive7.2 2.20.45
 Citrobacter koseri, ESBL-positive3.6   
 Klebsiella oxytoca, ESBL-positive1.8   
 Enterobacter cloacae, ESBL-positive0.9   
 Serratia fonticola, ESBL-positive 1.5  
All enterobacteria, MBL-positive5.4b1.52.20.67
 E. coli, MBL-positive1.81.5  
 K. pneumoniae, MBL-positive1.8   
 K. oxytoca, MBL-positive1.8   
 Citrobacter freundii, MBL-positive  2.2 
All enterobacteria, high-level AmpC4.
E. cloacae, high-level AmpC3.6 2.21.0
 E. coli, high-level AmpC0.9 2.20.49
 Serratia marcescens, high-level AmpC  2.2 
 Hafnia alvei, high-level AmpC 1.5  
VREc2.7  0.56

ESBL-producing Enterobacteriaceae and, more surprisingly, MRSA isolates were most often recovered from rectal samples. The best combinations of screening measures to recognize patients colonized with key resistant bacteria were a rectal swab and a urine sample for ESBL producers, and both inguinal and oropharyngeal swabs for MRSA (Table 2). Testing only nasal swabs would have resulted in substantial underestimation of colonization with MRSA.

Table 2.   Percentages of patients, residents or staff found to be positive for methicillin-resistant Staphylococcus aureus (MRSA) or extended-spectrum β-lactamase (ESBL) producers with various specimen type combinations
 MRSA (%)ESBL producers (%)
All study populations pooledLTCF residentsAll study populations pooledLTCF residents
  1. LTCF, long-term-care facility.

  2. aProportion of all positives with the group detected by indicated sample type.

Urine  2343
Rectal + inguinal75869797
Rectal + oropharyngeal77819695
Rectal + nasal8081  
Rectal + urine  9998
Inguinal + oropharyngeal82887382
Inguinal + nasal8081  
Inguinal + urine  8091
Oropharyngeal + nasal7769  
Oropharyngeal + urine  6068

MICs for, and molecular typing of, MBL-producing strains

Ten MBL-producing isolates were found; eight from LTCF residents (one with two different producers), one from a staff member, and one from a geriatric unit patient (Table 3). All had blaVIM-1 together with qnrS. The PFGE patterns of two blaVIM-positive Klebsiella oxytoca isolates from LTCF residents were identical, whereas the patterns of the two MBL-producing E. coli isolates from LTCF residents and the one from a staff member were >80% related. All three E. coli isolates belonged to phylogenetic group B2. MICs of imipenem for the MBL producers ranged from 2 to 16 mg/L, and were reduced to ≤0.5 mg/L in the presence of 320 mg/L EDTA, except in the case of Morganella morganii and Providencia stuartii isolates—species that are inherently less susceptible to imipenem. MICs of meropenem and ertapenem for the MBL producers generally were lower than those of imipenem, but above those for MBL-negative isolates. MICs of ciprofloxacin ranged from 1 to >8 mg/L. Two MBL producers were highly susceptible to aztreonam, with MICs ≤0.25 mg/L, whereas the aztreonam MIC for the P. stuartii isolate was 4 mg/L; values were ≥64 mg/L for the other seven isolates. Five of the seven aztreonam-resistant MBL producers, excluding only the two Klebsiella pneumoniae isolates, were examined by PCR and found to be positive for blaSHV. One blaVIM-1-positive E. coli isolate (aztreonam MIC >64 mg/L) also had a blaCTX-M gene belonging to group 1. Except for the P. stuartii isolate, the MBL-positive isolates had high-level resistance to oxyimino-cephalosporins (≥32 mg/L) and cefoxitin (≥32 mg/L), with weak synergy, at most, with clavulanate, generally reducing the cephalosporin MIC ≤4-fold.

Table 3.   MICs (mg/L) for the metallo-β-lactamase-producing Enterobacteriaceae, Escherichia coli DH10B, and a representative transformant
AntibioticGeriatric unitLTCF staffLTCF residentsRecipientTransformant
Citrobacter freundii, 208 Escherichia coli, 165 E. coli, 6 E. coli, 95 Morganella morganii, 445 Providencia stuartii, 446 Klebsiella pneumoniae, 20 K. pneumoniae, 57 Klebsiella oxytoca, 41 K. oxytoca, 105 E. coli DH10BDH10B (pKOX105)
  1. LTCF, long-term-care facility.

Amoxicillin–clavulanate 2 : 132323264>64>6432323232864
 +4 mg/L clavulanate>32>32>3216324>3216>32>32≤0.060>32
 +4 mg/L clavulanate>32>32>32>32324>32>32>32>320.25>32
 +4 mg/L clavulanate>32>32>3216883216>32>32≤0.06032
 +400 mg/L EDTA0.50.1250.1250.5420.250.1250.

Three MBL producers from LTCF residents (E. coli 6, K. oxytoca 105, and M. morganii 445) were used as plasmid donors for successful transformation of E. coli DH10B. All of the transformants were PCR-positive for blaVIM-1, qnrS, and blaSHV, and contained plasmids designated pECO6, pKOX105.1, and pMMO445, respectively. We also obtained a transformant with a derivative plasmid pKOX105.2 from K. oxytoca 105; this carried blaVIM-1 but lacked qnrS and blaSHV. The transferred plasmids pECO6 and pKOX105.1 (Fig. 1, lanes 1 and 2) were identical by digestion with HpaI (and BamHI/SacI; data not shown) and very similar to pKOX105.2 and pMMO445 (Fig. 1, lanes 3 and 4). All of the plasmids belonged to incompatibility group IncN.

Figure 1.

 Plasmids extracted from Escherichia coli DH10B transformants, digested with HpaI and separated on a 0.7% agarose gel: pECO6 (lane 1), pKOX105.1 (lane 2), pKOX105.2 (lane 3) and pMMO445 (lane 4) were from isolates collected in the surveillance of long-term-care facility residents. Lane M has molecular weight markers.

MICs of, and molecular typing of, ESBL-producing E. coli isolates

Forty-two of 46 ESBL-producing E. coli isolates from LTCF residents, all eight from staff members and all three from the geriatric unit patients were tested further. In all cases, MICs were ≥2 mg/L for at least one of cefotaxime, ceftazidime or cefpirome (Fig. 2), and were reduced eight-fold or more in the presence of clavulanate (4 mg/L) whereas cloxacillin (100 mg/L) had no significant effect. All were resistant to ampicillin and piperacillin, whereas amoxicillin–clavulanate (2 : 1) MICs varied between 8 and 32 mg/L. All but three (E. coli R35, S168 and S183 with MICs ≤0.125 mg/L) were substantially resistant to ciprofloxacin, with MICs ≥8 mg/L, whereas the MICs of aminoglycosides were bimodally distributed, with 23 of 53 isolates being susceptible to tobramycin (4 mg/L) and 28 of 53 being resistant at ≥32 mg/L; 37 of 53 were susceptible to gentamicin (4 mg/L) and 15 of 53 were resistant at ≥32 mg/L and 39 of 53 were susceptible to amikacin (8 mg/L), but 14 of 53 were resistant at ≥16 mg/L. All remained susceptible to imipenem (0.5 mg/L), meropenem (0.06 mg/L), ertapenem (1 mg/L), tigecycline (1 mg/L) and colistin (1 mg/L).

Figure 2.

 Pulsed-field gel electrophoresis (PFGE) dendrogram of the extended-spectrum β-lactamase (ESBL)-producing Escherichia coli isolates from the screening of the long-term-care facility (LTCF) residents (R) and staff members (S) and geriatric unit patients (G). Specific LTCF unit, ESBL type, cefotaxime (CTX) and ceftazidime (CAZ) MICs, the presence of IS26 adjacent to blaCTX-M-group-1 (‘IS26 link’), presence of qnrS and phylogenetic group are shown as A1, A2, B, C or D. PFGE clusters with >85% similarity are indicated. NA, not applicable; ND*, ESBL type not determined (SHV-negative and CTX-M negative); NT**, not typeable by the PCR method.

PFGE defined clusters A1, A2, B, C and D among 52 ESBL-producing E. coli isolates (one isolate was not typeable). Cluster A1 contained four blaSHV-5-positive isolates, whereas two further isolates with this enzyme formed cluster D. All but one of these blaSHV-5-positive isolates were from residents of LTCF unit 5; the exception was from a staff member. Clusters A2 and C contained 18 isolates with blaCTX-M genes of group 1 (one gene from each cluster was sequenced and identified as blaCTX-M-15), all with cefotaxime MICs >256 mg/L and ceftazidime MICs generally ≥64 mg/L. Another 14 isolates with blaCTX-M genes of group 1(one identified by sequencing as blaCTX-M-15) formed cluster B: cefotaxime and ceftazidime MICs for these were ≤128 mg/L and ≤32 mg/L, respectively, and PFGE patterns closely resembled that of UK strain A [19]. Two unique isolates had CTX-M-14-like enzymes, on the basis of partial sequencing and distinct PFGE patterns, whereas one, also with a unique PFGE pattern, had blaCTX-M-1 with insertion of ISEcp1 80 bp upstream. One ESBL-positive isolate had only blaTEM: cefotaxime and ceftazidime MICs for this isolate were 2 mg/L, decreasing to ≤0.25 mg/L in the presence of clavulanate; we did not seek to identify its enzyme.

PCR for the linkage of blaCTX-M genes of group 1 with upstream ISEcp1 gave an approximately 600-bp fragment for isolates in clusters A2 and C, and sequencing revealed an ISEcp1-like element 48 bp upstream of the β-lactamase start codon. A slightly larger fragment was seen in strain G183, which had blaCTX-M-1, whereas all 14 isolates belonging to cluster B had IS26 upstream of blaCTX-M genes of group 1, as in UK strain A [19]. Only four of the cluster B isolates gave amplification products in the linkage assay for ISEcp1and blaCTX-M genes of group 1, all with a fragment length compatible with insertion of IS26 in the inverted repeat sequence of ISEcp1, as in UK strain A [19]; the other ten cluster B isolates did not give a PCR product. Eight ESBL-positive blaVIM-negative isolates had qnrS; six of these belonged to cluster B, and none had qnrA or qnrB.

Risk factors for colonization

Colonization of LTCF residents with resistant organisms, including ESBL producers and MRSA, was associated with several risk factors in univariate and multivariate analysis (Table 4). Age >86 years was an independent risk factor for ESBL producers among LTCF residents, as was administration of antibiotics within the previous 3 months. An association with previous antimicrobial treatment was marginally significant for MRSA (OR 2.31; 95% CI 1.00–5.34; p 0.050); prior treatment with fluoroquinolones was specifically associated with MRSA colonization in univariate analysis only. Univariate analysis also revealed that the presence of any invasive medical device was a significant risk factor for colonization with resistant bacteria, ESBL producers or MRSA, and indicated an association between colonization with ESBL producers and the presence of a percutaneous enteral gastrostomy tube, tracheostomy tube or indwelling urinary catheter, and also between colonization with resistant bacteria and a urinary catheter. MRSA colonization was weakly associated with a nasogastric tube (OR 8.82, 95% CI 0.99–78.3, p 0.051). Chronic obstructive pulmonary disease was an independent risk factor for MRSA, but the most significant risk factor for MRSA, ESBL producers and all resistant bacteria in LTCF residents was chronic immobility, based on a Barthel index score of 0.

Table 4.   Univariate and multivariate logistic regression analysis of risk factors for colonization of long-term-care facility (LTCF) residents with resistant bacteria
Variable111 Residents (%)Colonization with MRSAColonization with ESBL producersColonization with resistant bacteriaa
Univariate analysisMultivariate analysisUnivariate analysisMultivariate analysisUnivariate analysisMultivariate analysis
OR (95% CI)p-ValueOR (95% CI)p-ValueOR (95% CI)p-ValueOR (95% CI)p-ValueOR (95% CI)p-ValueOR (95% CI)p-Value
  1. ESBL, extended-spectrum β-lactamase; MRSA, methicillin-resistant Staphylococcus aureus; ND, not defined.

  2. aColonization with at least one of the following: MRSA, vancomycin-resistant enterococci, Enterobacteriaceae with ESBLs, metallo-β-lactamases, or derepressed or acquired AmpC.

Male sex45.00.95 (0.44–2.04)0.88  1.18 (0.54–2.57)0.68  0.93 (0.39–2.19)0.86  
Age ≥86 years32.40.67 (0.25–1.82)0.44  1.6 (0.57–4.49)0.374.63 (1.12–19.1)0.0340.84 (0.29–2.41)0.74  
Antibiotics in last 3 months28.82.31 (1.00–5.34)0.05  8.25 (2.40–30.3)0.00097.51 (1.48–38.0)0.015NDND  
 Fluoroquinolones28.16.42 (1.27–32.5)0.0255.81 (0.84–40.3)0.0754.95 (0.60–41.1)0.14  NDND  
 Penicillins59.41.18 0.43–3.23)0.74  3.59 (0.98–13.2)0.054  NDND  
 Cephalosporins6.31.6 (0.10–26.2)0.74  NDND  NDND  
Dementia66.70.64 (0.29–1.42)0.27  0.29 (0.11–0.74)0.01  0.25 (0.08–0.80)0.018  
Peripheral vascular disease75.71.7 (0.67–4.31)0.27  1.6 (0.66–3.87)0.29  2.16 (0.84–5.52)0.11  
Incontinence96.4NDND  NDND  NDND  
Diabetes271.07 (0.46–2.53)0.87  0.65 (0.28–1.54)0.33  0.72 (0.28–1.82)0.48  
Cancer19.81.41 (0.55–3.63)0.47  0.77 (0.30–2.01)0.590.22 (0.04–1.07)0.0590.66 (0.24–1.84)0.43  
Decubitus ulcer9.91.36 (0.39–4.76)0.63  1.57 (0.39–6.27)0.52  1.58 (0.32–7.79)0.57  
Chronic obstructive pulmonary disease6.310.8 (1.26–93.7)0.0311.7 (1.13–120.8)0.0393.6 (0.42–31.0)0.24  NDND  
Physical disability (Barthel score of 0)74.84.06 (1.28–12.9)0.017  16.7 (5.13–54.7)<0.000111.4 (2.34–55.5)0.00323.1 (7.43–71.8)<0.000119.2 (5.82–63.6)<0.0001
Any medical device48.63.09 (1.40–6.84)0.0052.26 (0.87–5.85)0.0936.14 (2.47–15.2)<0.0001  7.83 (2.50–4.6)0.00044.59 (1.24–16.9)0.022
 Percutaneous enteral  gastrostomy tube271.57 (0.67–3.67)0.29  12.3 (2.76–55.4)0.0013.2 (0.99–10.37)0.053NDND  
 Tracheostomy tube181.07 (0.40–2.87)0.89  14.2 (1.83–111)0.011  NDND  
 Urinary catheter31.52.16 (0.95–4.89)0.065  3.91 (1.46–10.5)0.007  3.58 (1.14–11.3)0.029  
 Nasogastric tube5.48.82 (0.99–78.3)0.051  NDND  NDND  
LTCF unit 120.80.05 (0.01–0.39)0.004  0.13 (0.04–0.36)<0.0001  0.07 (0.02–0.20)<0.0001  
LTCF unit 221.61.45 (0.58–3.63)0.4212.4 (1.26–122)0.0318.53 (1.89–38.5)0.0057.63 (0.91–64.0)0.0614.69 (1.03–21.4)0.046  
LTCF unit 321.62.8 (1.11–7.07)0.02930.9 (3.18–301)0.0030.38 (0.15–0.96)0.04  1.02 (0.36–2.88)0.97  
LTCF unit 418.90.97 (0.36–2.57)0.948.17 (0.79–84.2)0.0772.04 (0.69–6.06)0.2  4.6 (1.46–14.3)0.009  

Being resident in unit 1 of the LTCF was associated with significantly less colonization by all resistant bacteria, ESBL producers and MRSA; residence in unit 3 was associated with significantly lower carriage of ESBL producers and MRSA than residence in the other three LTCF units. Of the eight MBL-producing Enterobacteriaceae from LTCF residents, five (from four residents) were from unit 4, two were from unit 2 and one was from unit 5, whereas four of the five Enterobacteriaceae with high-level AmpC production were isolated from residents in unit 4 and one was isolated from a resident in unit 5. The three VRE—all Enterococcus faecalis with the VanA phenotype—were from patients in units 3, 4 and 5. Dementia without chronic immobility was associated with a reduced risk for colonization with ESBL producers and all resistant bacteria in the univariate but not in the multivariate analysis, and this may explain the lower resistance rates in LTCF unit 1, which cares for residents with the behavioural and psychological symptoms of dementia (agitation/aggression, delusions, anxiety, aberrant motor behaviour, hallucinations, etc.).

Age ≥86 years was also a risk factor for colonization with all resistant bacteria in the geriatric units (Table 5); here, however (and in contrast to the LTCF), dementia was a positive risk factor for colonization with all resistant bacteria.

Table 5.   Univariate and multivariate logistic regression analysis of risk factors associated with colonization of geriatric unit patients with resistant bacteria
Variable45 Patients (%)Colonization with resistant bacteria of any or of the types studieda
Univariate analysisMultivariate analysis
OR (95% CI)p-valueOR (95% CI)p-value
  1. ND, not defined.

  2. aSee listing in Table 4.

Male sex28.90.55 (0.10–3.00)0.48  
Age ≥86 years37.83.27 (0.78–13.9)0.116.84 (1.11–42.3)0.039
Antibiotics in last 3 months33.32.5 (0.59–10.5)0.215.68 (0.62–34.9)0.061
 Fluoroquinolones26.71.19 (0.11–12.8)0.89  
 Penicillins46.71.5 (0.24–9.22)0.66  
 Cephalosporins26.71.19 (0.11–12.8)0.89  
Dementia33.34.33 (0.99–18.9)0.0519.52 (1.47–61.8)0.018
Peripheral vascular disease802.67 (0.29–24.3)0.38  
Incontinence57.81.97 (0.44–8.87)0.38  
Diabetes24.40.28 (0.03–2.49)0.25  
Cancer24.40.72 (0.13–4.05)0.71  
Decubitus ulcer8.94.13 (0.50–33.9)0.19  
Chronic obstructive pulmonary disease17.80.44 (0.05–4.12)0.47  
Any medical device22.21.67 (0.39–7.19)0.49  
 Urinary catheter28.91.07 (0.23–4.99)0.93  
 Nasogastric tube2.2NDND  


We evaluated colonization with resistant bacteria among residents of a LTCF and two linked geriatric units of the acute-care hospital in Bolzano. Overall, 74.8% of LTCF residents were colonized with at least one target organism, most frequently an ESBL producer (64.0%), MRSA (38.7%), MBL producer (6.3%) or AmpC hyperproducer (4.5%). VRE colonization was rare (2.7%). Many residents had more than one target organism, underscoring the role of LTCFs as a reservoir for these organisms [1,23,24]. MRSA colonization among LTCF residents has long been recognized, sometimes with colonization frequencies ≥50% [1,24,25].

Carriage of Enterobacteriaceae with VIM MBLs by 6.3% of LTCF residents is of special concern: MBL-producing Enterobacteriaceae are rare in Europe, except for Greece [26], where plasmids encoding the VIM-1 enzyme have spread among K. pneumoniae strains [27]. An Italian countrywide survey in 2004 identified only one MBL producer (an Enterobacter cloacae isolate) among over 12 000 Enterobacteriaceae screened [28], with further recent single isolates [29] or clusters being found in Genoa [30] and Bolzano [3]. The present data imply: (i) resident-to-resident transfer of MBL producers in the LTCF, on the basis of identical or similar PFGE patterns for a pair of K. oxytoca isolates and for three E. coli isolates, one of them from a staff member; and (ii) strain-to-strain plasmid transfer, on the basis of identical or related IncN plasmids with blaVIM-1 and qnrS in multiple Enterobacteriaceae species. The plasmids from the LTCF residents’ isolates had very similar restriction patterns to clinical strains collected up to 3 years previously [3].

Many colonizing Enterobacteriaceae had ESBL genes: principally, blaCTX-M genes of group 1 (81%), blaCTX-M-14 (4%) and blaSHV-5 (13%). The host isolates of these genes included four major clusters as determined by PFGE, designated A1, A2, B and C. Cluster B isolates resembled the UK strain A, an ST131 variant, in: (i) PFGE profile (Fig. 2); (ii) having an IS26 element upstream of blaCTX-M-15; and (iii) antibiogram, with lower-level cephalosporin resistance, probably contingent on the separation of blaCTX-M-15 from its usual promoter in ISEcp1 [19]. Strain A is widespread in the UK and was previously identified from a patient in Austria who had visited southern Italy but not the UK [31]. All three MBL-producing E. coli isolates and 90% of ESBL-producing E. coli isolates belonged to phylogenetic group B2, whereas 3.8% of the ESBL producers belonged to group D. These phylogenetic groups—particularly B2—account for most virulent extra-intestinal strains of the species [22].

LTCF outbreaks of E. coli with CTX-M-15 (the commonest group 1 type) were reported in 2000–2002 in Canada [32], and a survey in northern Italy in 2006–2007 revealed them in 9.1–100% of urine samples from LTCF residents with indwelling catheters [33].

Movement of patients and staff between the LTCFs and hospitals may facilitate dissemination of resistant bacteria. It is unclear whether most de novo acquisition occurs in the LTCF or during occasional hospitalizations [1], but the lower colonization rates among the geriatric unit patients argue against the hospital as the main source, as does the fact that, during 2008, the prevalence of ESBL producers among routine clinical isolates from LTCF residents far exceeded that among geriatric unit patients (40% vs. 9%; p <0.0001). A striking feature was the colonization of 27.5% of LTCF staff with resistant bacteria: 14.5% with ESBL producers, 14.5% with MRSA, and one nurse with a blaVIM-1-positive E. coli strain. This carriage probably reflects resident-to-staff and, perhaps, staff-to-staff transmission. Elsewhere, MRSA colonization rates of LTCF staff of 7.5–22.7% [34,35] have been reported.

Risk factors found here for colonization with ESBL producers in LTCF residents included treatment with antibiotics within the preceding 3 months, whereas invasive medical devices were associated with both ESBL producers and MRSA. These findings support published data: prior antibiotic treatment is a well-recognized risk factor for colonization with resistant organisms [25], including ESBL producers [1] and MRSA [24], whereas invasive medical devices are considered to be important for ESBL producers [36] and MRSA [24,25]; chronic obstructive pulmonary disease was found to be an independent risk factor for MRSA colonization among LTCF residents in our study and by others [37]. However, the most important risk factor for carriage of resistant organisms, ESBL producers and MRSA was the particular LTCF unit of residence. This is explicable, because the five units manage residents with different levels of independence, basal disease, comorbidity and functional status, all of which influence the frequency and nature of staff contact. Residents in LTCF 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 large number of ambulant, and relatively autonomous, residents in unit 1 may explain the inverse association between dementia and resistant bacteria found here. A converse, positive, association between resistant bacteria and dementia in the geriatric units possibly indicates acute events leading to hospitalization, along with poor adherence of these patients to hygienic measures as compared with non-dementia patients.

To conclude, we found frequent colonization with resistant organisms among LTCF residents and staff, and much lower rates in acute-care geriatric units. This difference can probably be explained by different risk factors. Patients transferred from the LTCF to acute-care hospitals should be considered to be at high risk of carrying resistant organisms and should be screened on admission. Screening should include a rectal swab and a urine sample for ESBL and MBL producers, and inguinal and oropharyngeal (not just nasal) swabs for MRSA. Because LTCFs are also homes, hospital infection control guidelines are unrealistic, but education of the employees, improvement of hand hygiene, necessary use of gloves and gowns, monitoring of diseases and treatment are all keys to better management, as is ongoing surveillance.


We wish to thank M. Doumith, M. Warner and R. Pike for valuable advice and technical assistance; our thanks also go to E. Pagani and V. Pasquetto for help in the molecular confirmation of MBL-producing isolates. Part of the microbiological work was presented as an undergraduate research project by F. Truzzi.

Transparency Declaration

The microbiological work in Bolzano was partly financed by research grant number 2129 of the Bolzano Hospital. D. M. Livermore has shareholdings, or acts as enduring attorney for a shareholder, in AstraZeneca, Dechra, EcoAnimal Health, GlaxoSmithKline, Pfizer and Schering-Plough; he has had research contracts, or conference finance, in the past 3 years from AstraZeneca, Calixa Cerexa, Johnson & Johnson, Merck, Novartis, Novexel, Pfizer, Phico, Theravance and Wyeth. He is employed by the Health Protection Agency, and is also influenced by their views on antibiotic usage. Other authors have no conflicts of interest to declare.