Late-Onset Bloodstream Infections (LO-BSI) continue as a critical complication associated with hospitalization of very low birth weight (VLBW) infants. LO-BSI contributes to morbidity, mortality, and other long-term adverse outcomes [1–5]. Surveillance of these infants, especially blood stream infections (BSI) was introduced in intensive care units in Poland [6, 7] however, the epidemiology of these infections has not previously been collected systematically in Polish VLBW infants.

The epidemiology of infections among neonatal intensive care units (NICUs) in the USA has been explored through the National Healthcare Safety Network (NHSN) system of the Centers for Disease Control and Prevention (CDC); however, limited information has been available from the European countries. The biggest one, German “Krankenhaus-Infections-Surveillance-System”, NeoKISS (http://www.nrz-hygiene.de/en/surveillance/hospital-infection-surveillance-system) began in January 2000 as a prospective surveillance system for VLBW infants [8, 9]. The similar surveillance systems have been implemented in France: Epidemiologie des Petits Ages Gestationnels, EPIPAGE study [10] and in the United Kingdom: the Neonatal Infection Surveillance Network, neonIN [11].

The aim of this study was to assess the epidemiologic factors and microbiological spectrum of primary LO-BSIs associated with or without intravascular devices together with identification of risk factors and the distribution of causative pathogens. A second aim was to implement uniform definitions, specimen acquisition, and culturing techniques for infection surveillance: continuous, systematic collection, analysis and interpretation of health-related and infections data.

Utilization of data collected in the Polish Neonatology Surveillance Network (PNSN) for the scientific purpose was approved by the Bioethics Committee of Jagiellonian University Medical College – no. KBET/221/B/2011.

Continuous prospective target surveillance of infections was conducted from 1/1/2009 through 12/31/2011 at six Polish NICUs (only teaching hospitals) which participated in PNSN. These tertiary NICUs provided care for 20% of all VLBW infants born in Poland annually. The surveillance included infants hospitalized in these NICUs whose birth weights were <1501 grams (VLBW) at birth until they achieved a weight of 1800 grams or died. The general fatality case rate was 16.3%. The fatality case rate in subgroup of infants with a birth weight <500 grams was 75%; thus on 1 infant reached a weight of >1800 grams with a LO-BSI, and 3 were transferred to other units and were not further included in the surveillance within the PNSN. All VLBW infants of suspected or documented infected were subject to registration regardless of the time of occurrence according to criteria of Gastmeier et al. (see Appendix 1) when they had clinical signs of a bloodstream infection (BSI) [8]. Pneumonia and necrotizing enterocolitis were defined also according to criteria of Gastmeier et al. [8]. LO-BSI was defined when diagnosed >72 hours after delivery. Clinical sepsis represented an infant where signs of infection existed but on blood culture a causative organism was not identified. Central venous catheter (CVC)-associated BSI (CVC-BSI) and peripheral intravenous catheter (PVC) associated BSI (PVC-BSI) were infections associated with the use of a central or peripheral venous catheter within the preceding 48 hours prior to the onset of the infection [8, 12]. The CVC utilization rate was 0.45 and the PVC utilization rates was 0.16 (calculated by dividing the number of days of CVC/PVC by the total number of patient days).

A surgical procedure was defined as surgery performed within an operating theater except for insertion of CVC involving a surgical incision for tissue removal or repair or to treat a surgical condition. The analysis was performed in infants prior their attaining a weight of 1800 grams in all cases. In the analysis of the relationship between LO-BSI and need for surgery, infants were included when the initial signs were observed between 3 to 30 days after operative intervention. Infants with more than one episode of infection were included; however, the addition of or change in pathogen was not necessarily indicative of a new episode of infection.

All blood specimens of at least 1 milliliter were injected into an aerobic blood culture bottle (Bactec Plus 26 Aerobic; BD Microbiology Systems) and cultured on MacConkey agar, horse blood agar (at 37 degrees C, each for 24 h) and Sabourand agar (at 37 degrees C for 38 hours. The isolates were characterized using biochemical tests, bioMerieux identification kit API system (bioMerieux, France).

Our results of LO-BSI are the first to be reported by the Polish Neonatology Surveillance Network (PNSN) and from Central Europe (except Neo-KISS) based on a national program for infection surveillance and control among NICUs. Our previous report on this population within the PNSN focused on early onset infection [15].

LO-BSI incidence in our population are lower than those observed in other national surveillance projects. For example, in a report from the Netherlands the incidence in their groups of 742 VLBW infants was 14.9/1000 pds [16], and in the German NeoKISS study in their first year of reporting their incidence was 8.3/1000 pdys that included 303 VLBW infants. In the USA, the reported incidence for infants born at 28 weeks gestation or earlier was: 36% [1], while in Israel a 39% rate [17].

Our data focusing on infections associated with vascular lines do differ substantially from those published by other centers e.g. NeoKISS [18]. The CVC-BSI in our total study population was greater than 8.6/1000 CVC days, while in the NeoKISS surveillance it was 13.8/1000 CVC days during the initial year of reporting. However, in our population we did not identify CVC-BSI in extremely low birthweight infants (ELBW) infants <499 grams, because of their high fatality rate. However, in the NeoKISS surveillance program this group had the highest risk of CVC associated infections, as well as, PVC-BSI. Unfortunately, this severe morbidity in Polish NICUs had the highest incidence of both CVC-BSI and PVC-BSI among infants with birth weights <1000 gms. Our surveillance found a 60-80% increase over than reported by the NeoKISS program [9], and higher than that reported by Sengupta et al. [18].

Therefore, our results indicate the need to reduce the risk of infections associated with vascular line in Polish NICUs; Gastmeier et. al. found that after 3 years of surveillance: German NICUs reported a significant reduction in the of CVC-BSI infants from 13.8 to 10.6 cases/1000 CVC days (RR. 0.77, p=0.008) [9, 19]. The German experience suggests that participation in ongoing surveillance of nosocomial infections in NICUs, requiring individual units to provide data may lead to a reduction in BSI rates when there is through participation in a national surveillance system with feedback and quality improvement [19]. Historical data from the American National Healthcare Surveillance Network [20] from 1992–2001 found that CVC- BSI was 11.3/1000 CVCdays among infants with birth weights <750 gm and 6.9/1000 CVCdays among infants 1001–1500 gm. However, the most current data from this surveillance network reported a 4 fold reduction in catheter associated infections in 2012 [21], and similar to rates reported from Canada [22].

The next important issue: our data also indicate a high CVC utilization rate in Polish NICUs: almost twice as likely and often (or: for a longer duration) than in the German NICUs [9].

Gram positive cocci organisms were the dominant microorganisms isolated in LO-BSI, including CNS, and especially associated with catheter associated infections. The predominance of these organisms were associated with catheter related LO-BSI in the USA [18], Japan [23], Taiwan [24] and Israel [25]. Freeman et al. first described dominance of CNS with increasing survival of ELBW infants in the 1970s [26] with longer duration of use of intravenous catheters [26]; fortunately, LO –BSI with CNS have not been characterized by an increased mortality [26, 27].

Gram-negative rod infections were mostly observed among LO-BSI in patients without prolonged use of vascular lines; however, we found a lower frequency of gram-negative rod infections than have been reported in the UK [11], Taiwan [24], Germany [9], or U.S.A. [28]. In our surveillance yeast LO-BSI infections were found in 3.8% of infants <1000 gm and lower than that reported by Greenberg [3, 29], but higher than that reported by Mazoni (1.4%) [30].

VLBW infants constitute a small group of patients undergoing surgical procedures. The literature on postsurgical in VLBW newborns has not been widely publicized. Many researchers studying the epidemiology of nosocomial infections exclude the group of VLBW infants undergoing surgery citing the diversity of this specific population that would interfere with more generalizable knowledge. In our studies the dominant problem were LO-BSI associated with infant surgery. In a publication from the U.S., risk factors for infections are late PDA closure procedures (6 or more days after diagnosis) and very low birthweight [31]. Importantly, the number of PDA closures (81 in 1243 infants who survived > 72 hrs) is consistent with the report by Evans, who found about 10% of infants born before 27 weeks (and 3% of infants born at 27–29 weeks) required PDA surgical ligation [32]. Small number reports about the state of post-operative infants is insufficient [33, 34].

A limitation of this study is that there were not independent auditors at each site to valid submitted data. Furthermore, VLBW infants frequently have multisystem diseases resulting from immaturity such as respiratory distress syndrome and apnea or bradycardia episodes that require interventions with medical devices (MV or CPAP). For the first weeks of life limited enteral nutrition must be supplement with parenteral nutrition provided through vascular catheters that also increase their risk for infection. Infant born after maternal chorioamnionitis may receive initial antibiotic therapy that is of insufficient duration to treat neonatal sepsis, or may not have sufficient spectrum of coverage to treat organisms more prevalent in LO-BSI. As a group VLBW infants are immunologically "immature" with limited and variable transplacental antibody transfer from their mothers further increasing their risk for LO-BSI. Nonetheless by introducing common definitions, technique and quantity of blood specimen acquisition, and uniform microbiological techniques at each center in the PNSN, our surveillance provide for baseline data for quality improvement initiatives. Our focus will be to reduce the duration of CVC and PVC use to minimize the risk of LO-BSI in this population, and to meticulously care for these catheters once inserted.

In VLBW infants in whom use of CPAP or mechanical ventilation is nearly universal and in whom the presence of vascular catheters is nearly routine, we found that coagulase-negative staphylocci were the most common cause of LO-BSI. While rates of LO-BSI especially associated with CVC- BSI were higher in the Polish NICUs than those of other national networks, we now have evidence to inform the neonatologist into more precise selection of antimicrobial coverage when an LO-BSI is suspected. Furthermore, based on these findings it is necessary to implement a national program of infectious disease monitoring while promoting the shortest duration of CVC and PVC use as possible. There is also an ongoing need to monitor all VLBW undergoing surgery in the neonatal period for post-surgical infections to reduce the LO-BSI incidence.