Nursing home (NH) residents are known to be reservoirs of multidrug-resistant (MDR) bacteria, mainly due to their frequent hospitalizations, recurrent use of invasive medical devices and high antibiotic consumption [Reference Cassone and Mody1]. Variable rates of intestinal colonization by extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae (ESBL-E) (6–41%) have been reported in European countries, while carbapenemase-producing Enterobacteriaceae (CPE) have not yet been identified [Reference Cochard2–Reference Saegeman5]. CTX-M-producing Escherichia coli Sequence Type (ST) 131 clone dominates by far the population of MDR Enterobacteriaceae colonizing the intestine of NH residents [Reference Nicolas-Chanoine, Bertrand and Madec3], but detailed analysis of ST131 subclonal structure has been scarce. In Portugal, ESBL-E (and particularly CTX-M-15-producing E. coli ST131 or Klebsiella pneumoniae ST15 clones) are endemic for several years in the clinical setting [Reference Rodrigues6–Reference Gonçalves, Cecílio and Ferreira8], whereas CPE (mainly KPC-3-producing K. pneumoniae) are quickly penetrating in our geographic region since the end of 2015, especially on susceptible populations [Reference Rodrigues9]. The aim of this work was to perform a pilot study to assess the current faecal carriage rate of ESBL-E or CPE among NH residents in Portugal, and the clonal and subclonal structure of these isolates.

Fresh rectal swabs from 20 residents at four NHs located in the North of Portugal (5–6 km distance between them) were collected in July 2014 and analysed. Five residents per NH (ten females, ten males) were recovered, representing 9·5% of the total residents’ population. Eighty-five per cent of residents were ⩾65 years old (mean age of 75 years), 70% were previously hospitalized and all of them received antibiotic treatment during the 3 months preceding sampling (Supplementary Table S1). Samples were suspended in 2 ml of saline and screened for Enterobacteriaceae resistant to third-generation cephalosporins and/or carbapenems by seeding 0·2 ml of the suspension on CHROMagar™ Orientation plates supplemented with vancomycin (4 mg/l) plus ceftazidime (1 mg/l) or ertapenem (0·25 mg/l), respectively, and further incubation (37 °C/24 h) [Reference Rodrigues10]. Presumptive Enterobacteriaceae isolates (oxidase negative, each different morphotype per plate) were selected for further studies. ESBLs and/or carbapenemases were identified by the DDST and Blue-Carba test, respectively, followed by polymerase chain reaction (PCR) and sequencing [Reference Rodrigues10]. Susceptibility testing to non-β-lactam antibiotics was performed by the disk diffusion method (http://www.eucast.org/clinical_breakpoints/) and presumptive E. coli ESBL producers were identified by species-specific PCR [Reference Rodrigues10]. The clonal structure of ESBL-producing E. coli was analysed by identification of E. coli phylogenetic groups and MLST (http://mlst.ucc.ie/mlst/dbs/Ecoli) [Reference Rodrigues10]. Subclonal typing of B2-ST131 isolates was performed by PCR or PCR and sequencing of markers for ST131 serogroups (O25b:H4, O16:H5), clades (A, B, C1, C2) and fimH _TR allele, and virulence genes (ibeA, iroN, sat, afa/draBC, papG allele II/III, cnf1, hlyA, cdtB, K1) [Reference Rodrigues10, Reference Matsumura11]. Plasmid analysis included replicon typing and subtyping (IncF plasmids) by PCR and sequencing (http://pubmlst.org/plasmid/primers/incF.shtml).

Intestinal colonization by ESBL-E was detected in 4/20 (20%; 95% confidence interval (CI) 5·7–43·7) of the residents (Table 1), a colonization rate similar to that (24·5%) reported previously in our country in a larger sample from residents at NHs and long-term care facilities (LTCFs) [Reference Gonçalves, Cecílio and Ferreira8]. These NH are managed by the same institution, share the nursing team and are served by the same hospital (H1). However, the asymmetry in the colonization rates observed (varying from 0% in NH1/NH2, 20% in NH3 and 60% in NH4) might be explained by the higher number of bedridden residents in NH3 and NH4 at sampling, which are at a higher risk of acquisition of MDR bacteria.

Table 1. Epidemiological data of ESBL-producing E. coli isolates identified in faecal samples from NH residents in Portugal

^a F, Female, M, Male.

^b PhG, E. coli phylogenetic group.

^c ST, Sequence Type.

^d IncF plasmids were identified using the FAB formula (FII, FIA, FIB) as proposed in http://pubmlst.org/plasmid/

^e Variability among isolates is shown in parenthesis.

^f CIP, ciprofloxacin; NAL, nalidixic acid; STR, streptomycin; SUL, sulphonamides; TET, tetracycline; TMP, trimethoprim.

^g fimH161, one SNP to fimH22.

The four NH residents positive for ESBL-E had recognized risk factors for ESBL-E carriage, such as previous antibiotic exposure and hospitalizations, but there was no significant statistic association between colonization and demographic (age, gender) or clinical (previous antibiotic treatment or hospitalization) data (Supplementary Table S1). Besides the low sample size, the absence of CPE is noteworthy but might not reflect the current situation since sampling occurred before the burden of CPE producers in clinical settings [Reference Rodrigues9, Reference Manageiro12].

All the ESBL-E were identified as E. coli producing CTX-M-15 (n = 2; two samples) or CTX-M-14 (n = 2; 2 samples) from different residents (Table 1). The species and the ESBL-types detected in our study are in line with the recent epidemiological trends in Portuguese hospitals [Reference Rodrigues6], and with those observed in NHs from different European countries [Reference Nicolas-Chanoine, Bertrand and Madec3, Reference Willemsen4, Reference Gonçalves, Cecílio and Ferreira8]. All ESBL-producing E. coli belonged to the pandemic B2-ST131-O25b:H4 clone and different clades thereof (C2/H30-Rx and B/H22-like). For both of them, the previous hospitalization of residents in the same hospital suggests nosocomial acquisition (Table 1). The C2/H30-Rx clade producing CTX-M-15 (n = 2) was identified in two residents from the same institution (NH4). It belonged to virotype C (sat), presented a MDR pattern and harboured only N and X4 plasmid replicons, instead of the typical IncF plasmids (Table 1) [Reference Pitout and DeVinney13]. In fact, this clade corresponds to the most worldwide disseminated within E. coli B2-ST131 including in Portugal (Novais Â, unpublished results) [Reference Nicolas-Chanoine, Bertrand and Madec3]. Interestingly, B2-ST131-H30 virotypes A and B, previously associated with NH residents, were not detected in our sample [3]. Isolates from the less common clade B/H22-like (fimH161, differing in one SNP from fimH22) were identified in residents from NH3 and NH4, belonged to virotype D5 (ibeA, iroN, cnf1, hlyA), were MDR and produced CTX-M-14, and carried a higher diversity of plasmid replicons [I1, HI2, ColE, and an F2:A-:B1 virulence plasmid (resembling pAPEC-O2-ColV, GenBank accession number AY545598)] (Table 1). This clade (B/fimH22), firstly described in our country, is usually linked to community-acquired infections, but infrequently to fluoroquinolone resistance or ESBL production as reported in this study, which deserves further monitoring [Reference Nicolas-Chanoine, Bertrand and Madec3, Reference Pitout and DeVinney13]. This study, together with previous data, highlights circulation of different ST131 clades in diverse clinical and non-clinical settings in our country: (i) clade C2/H30-Rx in different Portuguese hospitals, NHs and LTCFs (this study, data not shown); (ii) clades C1-M27 and C1-nM27 in hospitals and healthy volunteers [Reference Rodrigues10]; and (iii) an atypical clade B in NHs (this study).

In summary, this pilot study among NH residents in our country pointed-out the role of this setting as a source of different CTX-M-producing E. coli B2-ST131 clades (CTX-M-15-clade C2/H30-Rx and CTX-M-14-clade B/H22-like). Our data underscore the importance of B2-ST131 subtyping in different settings and further evaluation of transmission dynamics of the different subclones.