Utilization of data collected in the Polish Neonatology Surveillance Network (PNSN) for scientific purposes was approved by the Bioethics Committee of Jagiellonian University Medical College – no. KBET/227/B/2012. All data entered into the electronic database and analyzed during this study were previously anonymised and de-identified. The PNSN is a prospective national surveillance system for the most relevant infections in VLBW infants (with birth weight < 1500 g) in Poland. Details of the following variables were collected for all VLBW newborns: birth weight and gestational age, gender, multiple pregnacy, type of delivery and information of the situation in time of delivery, for example chorioamnionitis, general status of newborns by Apgar score: at 1 and 5 minutes and Critical Risk Index for Babies, CRIB) and others. Participation in PNSN is voluntary and confidential, the surveillance covered over 20% of VLBW infants born in the regions where the NICUs were located.
Study population
Prospective surveillance of infections was conducted between 2009–2012 at five tertiary academic NICUs that lead the perinatal care, in hospitals (designed by letters A, B, C, E, F) that participated in the PNSN, and included 1768 newborns. The labor of 135 newborns was complicated by chorioamnionitis, while 429 infants were delivered after premature rupture of membranes (PROM). All episodes of infection were subjected to recording, regardless of the time of occurrence of the first symptoms. Case patients were defined according to Gastmeier et al. [3] with modification, as very low birth weight (VLBW) neonates who demonstrated clinical signs and/or symptoms of bloodstream infection (BSI) or pneumonia.
Early-onset infection was defined as infection diagnosed within 3 days after delivery. Device-associated infection was defined as infection diagnosed in newborns who had a device (central/peripheral venous catheter or mechanical ventilation or continuous positive airway pressure) placed within the 48-hour period before infection onset. Device utilization ratio (calculated by dividing the number of days with devices by the total number of patient/days, pds) was: 0.41 for CVC, 0.21 for PVC, 0.38 for MV, 0.28 for CPAP and 0.47 for TPN.
SA-infection was defined as culture-proven infection with isolation of SA. MSSA (or MRSA) infection was defined as culture-proven infection with isolation of MSSA (or MRSA).
Bacterial isolates
Collecting and identifying bacterial strains were performed in the local microbiology laboratories. Various diagnostic specimens were collected for culture and assessment of the microbial aetiology of infections. Altogether, 58 SA strains were isolated – and were considered to be the aetiological factor of infection. In this group, 49 strains were used for further analysis: most were from cases of pneumonia (20) and from BSI (19); nine were not stored. Strains in local laboratory were stored in −20°C, in laboratory of Chair of Microbiology were stored in −70°C. Isolates were identified by the automated identification system (VITEK 2, bioMerieux, Poland).
Susceptibility testing
All strains were tested using disk diffusion antimicrobial susceptibility methods according to current guidelines of the European Committee on Antimicrobial Susceptibility Testing (Clinical breakpoints tables v. 3.1; http://www.eucast.org v.3.1). E-tests for vancomycin and teicoplanin (bioMérieux, Paris, France) were also performed for all the isolates.
The MRSA resistance phenotype was detected using a cefoxitin disc (30 μg) according to the EUCAST guidelines. The macrolide-lincosamide (MLS) resistance phenotype of the isolates was determined according to a previously published protocol [4].
DNA isolation
DNA was extracted from isolates using the Genomic Mini kit (A&A Biotechnology, Gdynia, Poland) according to the manufacturer’s instructions.
Polymerase chain reaction (PCR) screening for resistance genes
PCR amplification was used to detect the mecA gene using previously published primers [5]. As controls, S. aureus ATCC 33591 (mecA+) and S. aureus ATCC 25923 (mecA−) were employed. Erythromycin resistance genes (ermA, ermB, ermC, and msr) were detected by multiplex PCR, and amplification of a 456 bp fragment of the mupA gene (mupirocin resistance gene) was performed by single PCR [6,7].
Antibiotic treatment
Data about antibiotic treatment were entered into the database by the ward personnel based on the physicians’ orders. Precise information about the type of drug, daily dose, and length of treatment for each antibiotic was collected and used for the calculation of two indicators: duration of treatment and defined daily dose in reference to each case of infection.
The aggregate sum of the number of days during which at least one dose of antibiotic was received for each antibiotic used (days of treatment, DOTs) was expressed in days and the defined daily dose (DDD), according to the ATC/DDD system of the World Health Organization (Anatomical Therapeutic Chemical, group “J01”). Only antibiotics for systemic use were taken into account.
Antibiotic usage for treatment (until cure) was assessed for 56 cases; two cases were excluded from analysis due to death of the infant within the first seven days of illness (unsuccessful treatment) and lack of data on treatment of one newborn.
Virulence factor screening
S. aureus isolates were checked for the presence of selected virulence genes: tsst (toxic shock syndrome toxin), sea, seb, sec, seg, seh, sei, sej (staphylococcal enterotoxins A, B, C, G, H, I, J), eta, etb (exfoliative toxins A and B), lukE (LukDE leucocidin), pvl (Panton-Valentine leucocidin, and hla (staphylococcal alpha haemolysin) using PCR with previously published primers [8-10]. The following S. aureus strains were used as positive controls: 2535/07 (eta+, etb+, seg+, sei+), 8977/99 (sea+, sec+, seg+, sei+), 6616/09 (seb+, tsst+, pvl+), 2027/06 (sea+, seh+), and 1034/05 (sea+, seg+, sei+, sej+). The strains used as controls were kindly provided by Prof. Marek Gniadkowski, National Medicines Institute, Warsaw, Poland.
Pulsed-Field Gel Electrophoresis (PFGE)
Analysis of the genetic similarity between SA isolates was performed by PFGE method in accordance with a previously published protocol [11]. Restriction enzyme digestion was performed with 25 U of SmaI enzyme in Tango buffer (ThermoScientific, USA). Electrophoresis was conducted in a CHEFIII PFGE unit applying the parameters: block 1- starting pulse 5 s, ending pulse 12 s, voltage 6 V/cm, run time 11 h, block 2- starting pulse 20s, ending pulse 60 s, voltage 6 V/cm, run time 13 h. The GelCompar (Applied Maths, Sint-Martens-Latem, Belgium) was used for cluster analysis using the Dice coefficient and unweighted pair group method with arithmetic mean. Isolates that clustered ≥ 95% were considered as an epidemic clone. S. aureus strain ATCC 11632 was used as reference.
Spa sequencing
spa typing was performed as described previously [12], using the spa typing website (http://www.spaserver.ridom.de/) that was developed by Ridom GmbH (Münster, Germany).
SCCmec typing
Staphylococcal cassette chromosome mec (SCCmec) typing was performed as described previously [13]. The following S. aureus strains were used as positive controls: ATCC-BAA 1762 (SCCmec IV), ATCC-BAA 2094 (SCCmec V), and ATCC-BAA 1681 SCCmec II).
Statistical methods
The statistical analysis was based on two main groups of techniques. Comparison of the ratio of MSSA/MRSA was performed with a contingency test supplemented by odds ratio analysis. To determine the relationship between continuous and categorical variables, Student’s T test and ANOVA were used. If the distribution of the continuous variable did not fit a normal distribution, the analysis was conducted with nonparametric equivalents of parametric tests: Wilcoxon test for Student’s T test, and Kruskal-Wallis test for ANOVA. The relationship between two continuous variables was analyzed with Pearson Regression or its non-parametric alternatives. Because of the categorical nature of the effect and combined – numeric as well as categorical – types of predictors, the model was constructed for binominal distribution of dependent variables and the logit-linked function. P-values of <0.05 were considered to be statistically significant.