Late-onset sepsis (LOS) is a common complication of preterm infants during prolonged hospitalization in the neonatal intensive care unit (NICU).1,2 The incidence of LOS in the NICU varies according to birth weight (BW) and gestational age (GA) and ranged from 25–30% in very low birth weight (VLBW: ≤1500 g) infants1,3 to 6.2–10% in late-preterm (GA: 34–37 weeks) infants.4 Several neonatal factors, including patent ductus arteriosus, necrotizing enterocolitis (NEC), mechanical ventilation, administration of total parental nutrition and use of central venous catheters, have been identified as risk factors of acquiring LOS.5–8 LOS is associated with significant mortality, neurodevelopmental impairment among survivors and increased healthcare costs.9–11
Most studies regarding neonatal LOS focused on VLBW infants and failed to provide an overview of all high-risk neonates who experienced LOS in the NICU.1,3,12–14 Up-to-date information available on the clinical manifestations of LOS, its relationship with underlying diseases and predictors of final mortality are scarce, as few reports have addressed these issues. In addition, most studies only analyzed the first episode of LOS in every neonate, which does not represent the real incidence rate of LOS in the NICU.1,3 We therefore undertook a study on neonates with LOS in order to update our knowledge of the clinical manifestations, epidemiology and risk factors of adverse outcomes in present-day clinical practice.
MATERIALS AND METHODS
Subjects and Setting
We identified neonates admitted to the NICU of Chang Gung Memorial Hospital between January 1, 2004, and December 31, 2011, who had at least 1 episode of LOS through our databases. The NICU of Chang Gung Memorial Hospital provides care from primary- to tertiary-level in a university-affiliated teaching hospital in northern Taiwan, which contains 3 units and has a total capacity of 49 beds equipped with mechanical ventilator and 28 beds with special care nurseries. Our NICU database was recorded by a full-time nurse specialist, who dedicated to follow up neonates from birth or admission if transferred from other hospital, until discharge or death since more than 10 years ago. This study was approved by the institutional review board of Chang Gung Memorial Hospital, with a waiver of informed consent.
Patient characteristics were extracted from medical records and administrative databases. Laboratory databases were queried to identify all positive results of blood cultures. Our NICU database contained information available on basic demographic data, summary of hospital courses, discharge diagnosis and complications of prematurity. Medical records were reviewed to characterize all detailed courses of every episode of LOS, including the clinical manifestations, progression of septic conditions, concurrent infectious focus, treatment, infectious complications and outcomes. Severity of illness was evaluated at the most severe period during the course of LOS using the neonatal therapeutic intervention scoring system.15 All recorded data describing the LOS episodes were reviewed by two investigators (S.-M.C. and J.-F.H.) for face validity.
LOS was defined as a positive result on 1 or more blood cultures obtained after 6 days of life. Blood cultures positive for following microorganisms generally considered to be contaminants, including corynebacterium, propionibacterium, penicillium and diphtheroids, were excluded from analysis. The diagnosis of sepsis due to coagulase-negative staphylococci (CoNS) was based on the criteria of the Vermont Oxford Network Database3,14,16,17 and required clinical signs of sepsis, blood culture positive for CoNS and intravenous antibacterial therapy for at least 5 days after performing blood culture, or until death. Whenever CoNS and another pathogen were identified in the same blood culture, only the other pathogen was identified as the pathogen.
An episode of sepsis was defined if a patient had a positive blood culture treated with antibiotics therapy for 5 or more days or treated for a shorter period if the patient died and the presence of at least 2 of the following clinical symptoms of sepsis: fever or hypothermia, hyper- or hypoglycemia, apnea or tachypnea, frequent desaturation with increased requirement of ventilator support, bradycardia and/or cyanosis, feeding intolerance, abdominal distension, seizure, decreased activity, skin mottling and hypotension. If the same organism was identified after a 14-day course of appropriate antibiotic therapy or 1 or more negative blood culture, or if a different organism was identified from a subsequent culture 7 days after the first one, it was considered as a recurrent episode of LOS. Polymicrobial bacteremia episode was defined as more than 1 microorganism identified from a single set of blood culture or from different sets within 72-hour period.
All comorbidities of prematurity, including respiratory distress syndrome, intraventricular hemorrhage, bronchopulmonary dysplasia and NEC, were based on the latest updated diagnostic criteria in the standard textbook of neonatology.18 Congenital anomalies in this study included all neonates with either documented or undocumented syndrome, chromosome abnormalities, genetic or metabolic disorders, but not simple cleft palate or polydactyly. Persistent bacteremia or fungemia was defined as 3 or more consecutive positive blood cultures, at least 48 hours apart, during a single sepsis episode.19 All concurrent infectious focus, including NEC, ventilator-associated pneumonia, catheter-related bloodstream infection or meningitis, were also recorded and based on strict diagnostic criteria of Centers for Disease Control and Prevention and previous official publications.20,21 Infectious complications were defined as a newly onset infectious focus, venous thrombosis or vegetation, which is directly related to bacteremia or major organ dysfunction within 1 week after onset of bacteremia. For patients who died during hospitalization, the cause of death was recorded according to the clinician’s presumption, and sepsis attributable mortality was defined as neonates who expired within 3 days after onset of sepsis, those who died of infectious complications or clinically progressive deterioration since onset of LOS.
For clinical characteristics of LOS, microorganisms were divided into gram-positive, gram-negative, fungus and polymicrobial organisms. Specific microorganisms, such as Pseudomonas, Klebsiella, Serratia or extended-spectrum â-lactamase–producing gram-negative bacilli, which have been found to be significantly associated with mortality, were also analyzed separately.
Statistical significance for unadjusted comparisons was determined by using the x2 test, Fisher’s exact test or Wilcoxon rank-sum tests. Generalized logit regression models were used for comparisons involving categorical variables with >2 levels. Student’s t test and the Mann–Whitney test were used to compare continuous variables. All P values were 2-tailed, and P values <0.05 were considered to be statistically significant. Logistic regression methods were used to analyze the risk factors for final in-hospital mortality in neonates with LOS. All event-level covariates, which were the microorganisms subgroups or some specific pathogens, and infant’s level covariates including all perinatal and neonatal risk factors were analyzed by multivariable logistic regression analysis to assess the independent risk for final adverse outcome after the variables that were significantly associated with intrahospital mortality at a P value of <0.05 by univariate analysis. Because of the strong correlation between BW and GA, the risk factor of BW was excluded from the analysis. Statistical analyses were performed using SPSS version 15.0 (SPSS, Chicago, IL).
Incidence of LOS and Basic Demographics
From January 2004 to December 2011, 5010 infants from our NICU survived at least 5 full days and composed the study population. A total of 942 episodes of LOS were identified in 713 (14.2%) infants (Table 1); of them, 102 (14.3%) and 48 (6.7%) infants experienced 2 and more than 2 episodes of LOS, respectively. BW and GA were strongly associated with risk of LOS. More than one-third of neonates born before 28 weeks of gestation had at least 1 episode of LOS (37.0% and 36.3% at ≤25 weeks and 26–27 weeks, respectively) compared with only 7.3% and 11.1% among late-preterm infants (born at 33–36 weeks) or term born neonates, respectively (P < 0.001). Similarly, the infection rate decreased with increased BW. Approximately one-third (33.6%) of neonates with BW less than 1000 g had at least 1 episode of LOS, and the rate decreased to 23.2% for infants 1001–1500 g, 10.9% for those 1501–2000 g and 7.5% for those more than 2000 g (P < 0.001). More than half of LOS occurred in VLBW (birth body weight ≤1500 g) infants (60.1%) and GA ≤30 weeks (56.4%).
Overall, 942 episodes of LOS occurred in a total of 253,644 neonate-hospital days. The incidence rate (IR) of LOS in the entire cohort was 3.71 per 1000 neonate-hospital days (95% confidence interval [CI]: 3.31–4.12 episodes per 1000 neonate-hospital days). However, the IRs did not decreased with increasing BW and GA; instead, the IRs were lowest in neonates with BW of 1250 g to 1500 g and GA of 29 to 30 weeks (date not shown), and both infants with increased BW and decreased BW or increased GA and decreased GA had higher IRs.
More than half of cases had underlying chronic complex condition (399/713, 56.0%) at onset of LOS (Table 1). Although 37.7% (n = 269) had symptomatic patent ductus arteriosus, which finally required treatment, very few cases had associated heart failure at onset of LOS. 21.3% neonates had multiple underlying chronic conditions. The median age at onset of first episode of LOS was 22 days (interquartile range: 14–34 days). Ninety of 713 (12.6%) infected infants died. The LOS attributable mortality rate was 7.2% (68 of 942 episodes).
Microbiological Characteristics and Clinical Manifestation
All causative microorganisms of these 942 episodes of LOS are summarized in Table 2. Gram-negative bacilli made up 32.6% of the total, and 17.9% (n = 55) of them were multidrug-resistant strain,22 including extended-spectrum â-lactamase–producing bacteria (n = 43), Acinetobacter baumannii (n = 4), Pseudomonas aeruginosa (n = 4), Stenotrophomonas maltophilia (n = 3) and Chryseobacterium meningoseptium (n = 1). Polymicrobial microorganisms were identified in 38 (4.0%) episodes, and 24 of them were caused by gram-negative bacilli. LOS caused by Candida spp. infections accounted for 5.5% of all episodes.
Table 3 shows the clinical manifestations and outcomes of LOS, stratified according to pathogens. Although gram-positive organisms were the most common pathogen, they had significantly lower rates of certain clinical manifestations, including septic shock, disseminated intravascular coagulopathy, leucopenia, thrombocytopenia, anemia and metabolic acidosis (all P < 0.01 after Bonferroni adjustment) than gram-negative LOS, fungemia or polymicrobial LOS. On the contrary, fungemia had the highest rates of septic shock (20/52, 38.5%) and disease severity among these 4 pathogen subgroups, judged by neonatal therapeutic intervention scoring system scores (19.1 ± 4.3, all P < 0.05).
Seven hundred eighty-four of 942 (83.2%) episodes of LOS were primary bacteremia, whereas others were combined with certain specific infectious focus. A total of 50 episodes LOS were combined with meningitis, with group B Streptococcus (15 episodes, 30%) as the most common pathogen and 57.7% (15/26) of group B Streptococcus LOS had meningitis. Other gram-positive cocci–associated meningitis occurred mostly in neonates with congenital or acquired hydrocephalus as the ventriculoperitonal shunt or extraventricular drainage complication. Gram-positive organisms remained the most common cause of catheter-related bloodstream infection (33/51, 64.7%), and gram-negative organisms were more commonly seen when the patients had NEC (10/15, 66.7%). For ventilator-associated pneumonia, almost equal numbers of gram-positive and gram-negative organisms were identified from endotracheal aspirates.
For outcome analysis, fungemia had significantly higher rates of infectious complications, persistent bloodstream infection and sepsis attributable mortality than gram-positive LOS, gram-negative LOS or polymicrobial microorganisms (all P < 0.01 after Bonferroni adjustment).
Predictors of In-hospital Mortality in Neonates With LOS
Univariate analysis and multivariate logistic regression analysis of potential risk factors for final mortality in neonates with LOS are displayed in Table 4. For the entire cohort, although the in-hospital mortality rates were highest among infants with GA ≤28 weeks (18.5%) and BW ≤1000 g (20.4%), they do not reach significant difference compared with term born neonates (12.7%, P = 0.182) or those with BW >2000 g (14.1%, P = 0.106), respectively. Patients who experienced more than 2 episodes of LOS had significantly higher rates of overall mortality (24/48, 50%) than patients with only 2 episodes of LOS (22/102, 21.6%) and than those with single episode of LOS (44/563, 7.8%) (all P value <0.001 by ÷2 test with Bonferroni correction and log rank test) (Fig. 1). Neonates with gram-negative sepsis, fungemia or Pseudomonas LOS had significantly higher risk of overall mortality, compared with those with only gram-positive LOS (P < 0.01 by log rank test) (Fig. 2). Other risk factors for overall mortality included patients with several underlying chronic conditions (Table 4).
After adjusting for all variables, patients with a GA ≤28 weeks had a significantly higher risk of final mortality, as opposed to those with GA >28 weeks (odds ratio [OR], 2.38; 95% CI: 1.20–4.55; P = 0.013). There was significant higher risk for final mortality for patients with fungemia (OR, 5.69; 95% CI: 2.48–13.01; P < 0.001) and Pseudomonas LOS (OR, 14.32; 95% CI: 3.87–53.01; P < 0.001), compared with the risk associated with only gram-positive LOS. Patients with more than 2 episodes of LOS had more than 3 folds increased risk of final mortality, compared with patients with single episode of LOS (OR, 3.34; 95% CI: 1.39–8.06; P = 0.007). After multivariate logistic regression analysis, neonates with underlying congenital anomalies (OR, 4.12; 95% CI: 1.60–10.60; P = 0.003), neuromuscular comorbidities (OR, 3.34; 95% CI: 1.66–6.73; P = 0.001) and secondary pulmonary hypertension with/without cor pulmonale (OR, 23.48; 95% CI: 5.96–92.49; P < 0.001) were found to be at independently increased risk for final mortality.
The present study evaluated a large cohort of all high risk neonates, representing all live-born infants in the NICU during the study period. Although it is well known that the incidence of LOS is reversely proportional to BW and GA,1,3 we found approximately one-third (303/942, 32.2%) of LOS in NICU occurred in late-preterm (GA ≥33 weeks) and term born infants, and their mortality composed approximately one-third (31/90, 34.4%) of all fatal cases, suggesting these non-VLBW infants deserve studies.1,3,12–14,23–27 We found the sepsis attributable mortality rate and overall mortality rate in our NICU were lower than previous studies,1,3,23,24 which may be owing to recent progress of neonatal care, early recognition of presumed sepsis and timely adequate empiric antibiotic therapy. Fungemia became the significantly most serious pathogen causing LOS. Besides, neonates with Pseudomonas LOS, fungemia and underlying chronic conditions of congenital anomalies, neuromuscular sequelae and secondary pulmonary hypertension with/without cor pulmonale were among independent predictors for final mortality.
In the subgroup of VLBW infants, the incidence of LOS in our cohort was lower than those of Stoll et al1 and Makhoul et al3 (less than 40% in most GA of our cohort, compared with approximately half of neonates born before 28 weeks of gestation in their studies), but the trend of decrease of both incidence and incidence rate was similar to theirs. However, we found the incidence rates of LOS increase as BW and GA increase in late-preterm or term born neonates, which suggested their underlying chronic conditions predisposing them at risk of LOS.
The distribution of pathogens in our NICU is also different from those of Stoll et al1,25 and Makhoul et al.3 Overall, we had a higher percentage of gram-negative bacteria and fewer gram-positive bacteria than those of Stoll et al and less fungal organisms than that of Makhoul et al. Specifically, we found less CoNS and more Klebsiella and some other rare pathogens. Makhoul et al3 had reported some of their CoNS infections may be contaminated specimens, thus all episodes of CoNS in our cohort were reviewed in order to reduce an overestimation of the true infection rate. However, a prospective study applying strict diagnostic criteria of CoNS sepsis is warranted to identify its real impact given it is well known as the most common pathogens of LOS in NICU.1–7
Our data showed gram-positive LOS had the least severe clinical manifestations, whereas fungemia was the most severe. Actually our results were consistent with previous studies demonstrating a strong association between clinical characteristics and the type of pathogen responsible for the LOS1,3,14,23 because LOS caused by group B Streptococcus was especially at high risk of meningitis (57.7%) and postinfectious neurological sequelae, such as hydrocephalus, empyema or encephalomalacia. Pseudomonas accounted for one-fourth of gram-negative LOS–associated sepsis attributed mortality (8/32, 25%). Thus, we concluded that great variations in clinical characteristics exist between different types of gram-negative or gram-positive organisms.
Consistent with Bizzarro et al,26 we found no differences in outcome or mortality between polymicrobial and monomicrobial LOS, but our data were different from previous studies that multidrug-resistant gram-negative bacilli did not result in a higher risk of adverse outcome.27,28 Besides, many studies have found that fungal sepsis, gram-negative sepsis, especially Pseudomonas, Klebsiella or Serratia were significantly with a higher risk of mortality1,3,29–31 or early mortality14 (defined as death within 3 days of sepsis onset). Makhoul et al3 showed mortality is expected earlier with gram-negative pathogens than with gram-positive pathogens or Candida. These difference may be affected by appropriateness of antibiotic therapy that was administered on time at the onset of LOS,32,33 different treatment policies or intercenter variability,34,35 or recent progress in neonatal cares. Except Pseudomonas, we did not find significant worse outcome after Klebsiella or Serratia sepsis. Once appropriate antibiotics were administered, most LOS can be cured, but infectious complications, recurrent LOS and persistent bacteremia or fungemia need to be observed.
To our knowledge, this study is the first to enroll both pathogen types and neonatal characteristics for analyzing independent risk factors for overall mortality of neonates with LOS. We found Pseudomonas LOS and fungemia, even after adjusted to all neonatal characteristics, were still associated with a 14.3- and a 5.7-fold increase in risk for overall mortality after LOS. Previous studies used early mortality3,14 or the last positive culture before death1 for analysis. However, except for the three-fourths (68/90, 75.6%) of our death to be LOS attributable, most of the remaining (28/32, 87.5%) died of clinical sepsis or additional episode of nosocomial infections. Actually a vicious cycle existed that an episode of LOS requires interventions, and longer duration of mechanical ventilation and hospitalization is required, which predispose them at higher risk of another episode of infection. Because both infectious microorganisms and underlying neonatal conditions may possibly affect the final deadly infection, it is appropriate to apply our model considering all these characteristics. Besides, we did not enroll maternal or perinatal factors, such as maternal fever, premature rupture of membrane or early onset sepsis, in the model because they have been found unrelated to LOS12,30,36,37 and undoubtedly, the final mortality after LOS.
This study has some limitations. In addition to its retrospective nature and lack of 2 positive blood cultures for CoNS as documentation of sepsis, we did not evaluate the severity of all chronic conditions at the onset of LOS when they were analyzed for independent risk factors of overall mortality in the final logistic regression model. Besides, we did not include parameters of the susceptibility of pathogens to antibacterial agents or appropriateness of initial antibiotic therapy that was administered at the onset of sepsis in the final logistic regression model because we have found drug-resistant gram-negative LOS not to be associated with higher mortality.
All authors thank Mrs. Chiao-Ching Chiang for keeping the database of our NICU and all nursing staff working in our NICUs for keeping extremely detailed patient records, which contributed greatly to the completion of this research. The statistical analysis was helped by Miss. Chun-Chun Cheng and Mr. Yu-Jr Lin.
1. Stoll BJ, Hansen N, Fanaroff AA, et al. Late-onset sepsis in very low birth weight neonates: the experience of the NICHD Neonatal Research Network. Pediatrics. 2002;110(2 Pt 1):285–291
2. Schrag SJ, Cutland CL, Zell ER, et al.PoPS Trial Team. Risk factors for neonatal sepsis and perinatal death among infants enrolled in the prevention of perinatal sepsis trial, Soweto, South Africa. Pediatr Infect Dis J. 2012;31:821–826
3. Makhoul IR, Sujov P, Smolkin T, et al. Epidemiological, clinical, and microbiological characteristics of late-onset sepsis among very low birth weight infants in Israel: a national survey. Pediatrics. 2002;109:34–39
4. Sohn AH, Garrett DO, Sinkowitz-Cochran RL, et al.Pediatric Prevention Network. Prevalence of nosocomial infections in neonatal intensive care unit patients: Results from the first national point-prevalence survey. J Pediatr. 2001;139:821–827
5. Kawagoe JY, Segre CA, Pereira CR, et al. Risk factors for nosocomial infections in critically ill newborns: a 5-year prospective cohort study. Am J Infect Control. 2001;29:109–114
6. Geffers C, Gastmeier A, Schwab F, et al. Use of central venous catheter and peripheral venous catheter as risk factors for nosocomial bloodstream infection in very-low-birth-weight infants. Infect Control Hosp Epidemiol. 2010;31:395–401
7. Couto RC, Pedrosa TM, Tofani Cde P, et al. Risk factors for nosocomial infection in a neonatal intensive care unit. Infect Control Hosp Epidemiol. 2006;27:571–575
8. Perlman SE, Saiman L, Larson EL. Risk factors for late-onset health care-associated bloodstream infections in patients in neonatal intensive care units. Am J Infect Control. 2007;35:177–182
9. Stoll BJ, Hansen NI, Adams-Chapman I, et al.National Institute of Child Health and Human Development Neonatal Research Network. Neurodevelopmental and growth impairment among extremely low-birth-weight infants with neonatal infection. JAMA. 2004;292:2357–2365
10. Schlapbach LJ, Aebischer M, Adams M, et al.Swiss Neonatal Network and Follow-Up Group. Impact of sepsis on neurodevelopmental outcome in a Swiss National Cohort of extremely premature infants. Pediatrics. 2011;128:e348–e357
11. Atif ML, Sadaoui F, Bezzaoucha A, et al. Prolongation of hospital stay and additional costs due to nosocomial bloodstream infection in an Algerian neonatal care unit. Infect Control Hosp Epidemiol. 2008;29:1066–1070
12. Graham PL 3rd, Begg MD, Larson E, et al. Risk factors for late onset gram-negative sepsis in low birth weight infants hospitalized in the neonatal intensive care unit. Pediatr Infect Dis J. 2006;25:113–117
13. Stoll BJ, Hansen NI, Bell EF, et al.Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network. Neonatal outcomes of extremely preterm infants from the NICHD Neonatal Research Network. Pediatrics. 2010;126:443–456
14. Makhoul IR, Sujov P, Smolkin T, et al.Israel Neonatal Network. Pathogen-specific early mortality in very low birth weight infants with late-onset sepsis: a national survey. Clin Infect Dis. 2005;40:218–224
15. Gray JE, Richardson DK, McCormick MC, et al. Neonatal therapeutic intervention scoring system: a therapy-based severity-of-illness index. Pediatrics. 1992;90:561–567
16. Bizzarro MJ, Raskind C, Baltimore RS, et al. Seventy-five years of neonatal sepsis at Yale: 1928-2003. Pediatrics. 2005;116:595–602
17. Vermont Oxford Network Database Mannual of Operations, Release 2.0. 1993 Burlington, VT Vermont Oxford Network
18. Taeusch HW, Ballard RA, Gleason CA Avery’s Diseases of the Newborn. 20068th ed Philadelphia, PA Elsevier Saunders
19. Khashu M, Osiovich H, Henry D, et al. Persistent bacteremia and severe thrombocytopenia caused by coagulase-negative Staphylococcus in a neonatal intensive care unit. Pediatrics. 2006;117:340–348
20. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance definition of health care-associated infection and criteria for specific types of infections in the acute care setting. Am J Infect Control. 2008;36:309–332
21. Apisarnthanarak A, Holzmann-Pazgal G, Hamvas A, et al. Ventilator-associated pneumonia in extremely preterm neonates in a neonatal intensive care unit: characteristics, risk factors, and outcomes. Pediatrics. 2003;112(6 Pt 1):1283–1289
22. Magiorakos AP, Srinivasan A, Carey RB, et al. Multidrug-resistant, extensively drug-resistant and pandrug-resistant bacteria: an international expert proposal for interim standard definitions for acquired resistance. Clin Microbiol Infect. 2012;18:268–281
23. Gordon A, Isaacs D. Late onset neonatal Gram-negative bacillary infection in Australia and New Zealand: 1992-2002. Pediatr Infect Dis J. 2006;25:25–29
24. Ho JJ. Late onset infection in very low birth weight infants in Malaysian Level 3 neonatal nurseries. Malaysian Very Low Birth Weight Study Group. Pediatr Infect Dis J. 2001;20:557–560
25. Stoll BJ, Gordon T, Korones SB, et al. Late-onset sepsis in very low birth weight neonates: a report from the National Institute of Child Health and Human Development Neonatal Research Network. J Pediatr. 1996;129:63–71
26. Bizzarro MJ, Dembry LM, Baltimore RS, et al. Matched case-control analysis of polymicrobial bloodstream infection in a neonatal intensive care unit. Infect Control Hosp Epidemiol. 2008;29:914–920
27. Paterson DL, Ko WC, Von Gottberg A, et al. Antibiotic therapy for Klebsiella pneumoniae bacteremia: implications of production of extended-spectrum beta-lactamases. Clin Infect Dis. 2004;39:31–37
28. Jain A, Roy I, Gupta MK, et al. Prevalence of extended-spectrum beta-lactamase-producing Gram-negative bacteria in septicaemic neonates in a tertiary care hospital. J Med Microbiol. 2003;52(Pt 5):421–425
29. Makhoul IR, Bental Y, Weisbrod M, et al. Candidal versus bacterial late-onset sepsis in very low birth weight infants in Israel: a national survey. J Hosp Infect. 2007;65:237–243
30. Leigh L, Stoll BJ, Rahman M, et al. Pseudomonas aeruginosa infection in very low birth weight infants: a case-control study. Pediatr Infect Dis J. 1995;14:367–371
31. Cohen-Wolkowiez M, Moran C, Benjamin DK, et al. Early and late onset sepsis in late preterm infants. Pediatr Infect Dis J. 2009;28:1052–1056
32. Apisarnthanarak A, Holzmann-Pazgal G, Hamvas A, et al. Antimicrobial use and the influence of inadequate empiric antimicrobial therapy on the outcomes of nosocomial bloodstream infections in a neonatal intensive care unit. Infect Control Hosp Epidemiol. 2004;25:735–741
33. Gastmeier P, Sohr D, Geffers C, et al. Risk factors for death due to nosocomial infection in intensive care unit patients: findings from the Krankenhaus Infections Surveillance System. Infect Control Hosp Epidemiol. 2007;28:466–472
34. Carrieri MP, Stolfi I, Moro MLItalian Study Group on Hospital Acquired Infections in Neonatal Intensive Care Units. . Intercenter variability and time of onset: two crucial issues in the analysis of risk factors for nosocomial sepsis. Pediatr Infect Dis J. 2003;22:599–609
35. Chien LY, Macnab Y, Aziz K, et al.Canadian Neonatal Network. Variations in central venous catheter-related infection risks among Canadian neonatal intensive care units. Pediatr Infect Dis J. 2002;21:505–511
36. Perlman SE, Saiman L, Larson EL. Risk factors for late-onset health care-associated bloodstream infections in patients in neonatal intensive care units. Am J Infect Control. 2007;35:177–182
37. Auriti C, Ronchetti MP, Pezzotti P, et al. Determinants of nosocomial infection in 6 neonatal intensive care units: an Italian multicenter prospective cohort study. Infect Control Hosp Epidemiol. 2010;31:926–933