The study was conducted in the municipality of Goiânia, capital of Goiás state, located in Central Brazil, a region with high rates of antibiotics consumption (Andrade 2012). In the 2010 census, the population of Goiânia was estimated at 1,302,001 inhabitants, with 20,014 live births, infant mortality at 14.6 deaths per 1000 live births, and 84,465 children under five years.14 Public health care is provided free of charge by the Brazilian Unified Health System, and an estimated 75% of the population use the public health system.15 In Goiania, children are first seen at the public healthcare centers (pediatric outpatient clinics) and referred to hospitalization whenever necessary. For this study, two pediatric populations were screened for MRSA, as described below.

From June 2007 through November 2008 were screened for MRSA carriage all neonates admitted to four NICUs, which attend children from public and private health insurance. Taking together these NICUs comprise the majority of beds (n=69) for neonates in the city. Data were collected on demographics (gender, date of birth, and age at NICU admission), each neonate's clinical history (length of hospital stay, congenital malformation, birth weight, preterm, timing of swab collection, nasogastric intubation, and diagnoses upon admission and at discharge), and variables related to their mothers (type of delivery, schooling, number of prenatal visits, and placental abruption).

The carriage survey was conducted from July 2007 through July 2008 in children less than five years of age attending the major outpatient clinic of the city, served by the Brazilian public health insurance (Brazilian Unified Health System). Data were collected on child (gender, age, congenital malformation, skin and soft-tissue infections, hospitalization in the last six months, antibiotic use in the last three months, day care attendance in the last six months, acute otitis media in the last 12 months, pneumonia in the last 12 months, and tonsillitis in the last 12 months) and the family (mother's schooling, smoking at home, number of family members, and health worker in the family). We calculated that a sample size of 2000 children would be necessary to ascertain MRSA prevalence assuming a 1.2% prevalence of MRSA nasal carriers16 with a 95% confidence interval (95% CI), a precision of 0.7%, design effect of 1.8, and a refusal rate of 10%. The number of children who were recruited at the outpatient clinics per day took into account the laboratory capability for processing nasal specimens (approximately 10swabs/day).

Pernasal flocked swabs were collected by rotating the swab twice in the vestibule of the anterior nostrils of each child. In the outpatients, a single sample was collected from each child. At the NICUs, two samples were collected per child; the first collected upon admission and the second at discharge or death. After collection, the swabs were placed in a liquid medium (ESwab, Copan Diagnostics, Inc., Murrieta, CA, USA) and sent to the Microbiology Laboratory of the Federal University of Goiás for immediate processing.

The Institutional Review Board approval was given by the Ethics Committee of the Federal University of Goiás (# 008/2007). All parents or legal guardians signed copies of the informed consent form before recruitment.

The specimens were plated onto mannitol salt agar (Difco® Laboratories Inc., Detroit, MI, USA) for S. aureus isolation. Trypticase soy agar plates, which were supplemented with 4% NaCl and oxacillin at 6μg/mL, were used to screen MRSA isolates. After 24–48h of incubation at 37°C, the colonies that were suspected of being S. aureus were confirmed by Gram staining, catalase, coagulase, and PCR amplification of the femB gene (Jonas 1999).

Susceptibility tests were determined by disk diffusion (Oxoid Ltd., Basingstoke, Hampshire, England) for oxacillin, cefoxitin, clindamycin, ciprofloxacin, erythromycin, rifampicin, tetracycline, trimethoprim–sulfamethoxazole, linezolid, and quinupristin/dalfopristin according to the Clinical Laboratory Standards Institute guideline.17 Isolates resistant to cefoxitin and oxacillin were submitted to the Etest® (AB Biodisk, Slona, Sweden) to determine the sensitivity to vancomycin. S. aureus strains ATCC 25923 and ATCC 29213 were used as quality controls.

Isolates showing inhibition zones of ≤10mm and ≤21mm around 1mg oxacillin (Oxoid, Cambridge, UK) and 30μg cefoxitin (Oxoid, Cambridge, UK) disks, respectively, and that were positive for mecA gene by PCR, were characterized as MRSA.

Detection of mecA gene and SCCmec types were investigated by multiplex PCR18 using COL (SCCmec type I), N315 (SCCmec type II), ANS46 (SCCmec type III), and MW2 (SCCmec type IV) as positive controls. Multilocus sequence typing was carried out as previously reported19 and submitted to the international database (http://www.mlst.net) to assign the sequence type (ST). Sequencing of the polymorphic X region of the protein A gene (spa) and the Panton-Valentine leukocidin (PVL) gene were performed as described elsewhere.20,21 The sequences were analyzed using the Ridom Staph Type software (Ridom, Germany) to determine the spa types. The following international MRSA clones were included as controls: ST5-II (New-York/Japan, USA100), ST239-III (Brazilian/Hungarian), ST5-VI (Pediatric, USA800), and ST22-IV (EMRSA-15). The clonal relatedness was investigated by pulsed-field gel electrophoresis (PFGE) with SmaI macrorestriction according to standard procedures.22 The PFGE patterns were interpreted according to the criteria proposed by Tenover et al.,23 and the resulting dendrogram was constructed using Dice coefficients with BioNumerics software (version 5.0). Isolates were defined as epidemiologically related if they shared ≥80% similarity on the dendrogram.

Prevalence of MRSA and the corresponding 95% confidence interval (95% CI) were calculated. Multidrug resistance (MDR) was defined for isolates with resistance to three classes of antimicrobials and to β-lactams. Moving-average was used to calculate the percentage of colonization by S. aureus; the value attributed to a given month was the average colonization in that month, and the ones of the months prior and after the given month. Differences in lengths of stay in NICUs between neonates colonized and non-colonized by MRSA strains were evaluated by the Mann–Whitney test.

During the study period, a total of 710 neonates were admitted to the four NICUs, and nasal swabs were obtained from 701 (98.7%) of them. The characteristics of the neonates and their mothers are described in Table 1.The overall prevalence of S. aureus and MRSA was 9.1% (95% CI 7.2–11.4) and 0.6% (95% CI 0.2–1.4), respectively, and varied among the NICUs (Table 2). Four neonates carried MRSA strains; three of them acquired MRSA during the hospitalization period and had higher lengths of stay (median=20 days) compared to neonates who were non-colonized by MRSA (median=10 days) (p=0.072).

A total of 2034 children were enrolled for nasal swabbing during the study period. The mean age of the children was 21.6 months (SD=16.0), and 31.8% of them had less than 12 months of age. The main clinical features of the participant children are shown in Table 1. Four hundred and eight children (20.1%; 95% CI: 18.4–21.9%) were colonized by S. aureus. Colonization by S. aureus showed an age-related distribution, with peak prevalence at 47 months of age (Fig. 1). Four children (0.2%; 95% CI: 0.1–0.5%) carried MRSA strains.

Fig. 1.

Age-related prevalence of Staphylococcus aureus colonization by age. Each dot represents a child colonized by S. aureus. Vertical arrows indicate the age of child colonized by the corresponding HA-MRSA and CA-MRSA clones.

(0.07MB).

We identified 16 MRSA isolates from eight children. Six children had more than one isolate. MDR MRSA isolates were found in three children. Eleven MRSA isolates did not meet criteria for MDR, but they were resistant to different classes of antibiotics. MRSA isolates showed resistance to erythromycin (43.8%), clindamycin (37.5%), ciprofloxacin (25.0%) and trimethoprim–sulfamethoxazole (18.8%). All MRSA strains were susceptible to vancomycin.

The majority of the MRSA strains clustered into two genomic groups; within each group we identified children from both, NICUs and outpatient clinics carrying MRSA strains genetically related to classical pandemic HA-MRSA Brazilian (SCCmec-III, ST239) and Pediatric (SCCmec-IV, ST5) clones (Fig. 2). CA-MRSA strains were only detected in outpatient children and were clonally related to Western Australian 1 clone/WA-1 (SCCmec-IV, ST128). Only the HA-MRSA Brazilian clone showed MDR pattern. None of the MRSA strains contained the PVL gene.

Fig. 2.

Molecular characteristics of the 8 MRSA strains from neonates admitted to neonatal intensive care units and children who visited the outpatient clinics. HDE288, representative Pediatric clone; HU25, representative Brazilian clone; NICU-1, neonatal intensive care unit of the Santa Bárbara Hospital; NICU-2, neonatal intensive care unit of the Hospital da Criança; NA, does not apply; MIC, minimum inhibitory concentration; MDR, multidrug-resistant; SCCmec, staphylococcal cassette chromosome mec; PVL, Panton-Valentine leukocidin; MLST, multilocus sequence type; SLV, single locus variant.

(0.28MB).