The epidemiology of healthcare-associated infections among neonates in NICUs has been described from the National Nosocomial Infection Surveillance (NNIS) system of the Centers for Disease Control and Prevention (CDC) from January 1992 to June 2004 (12). Device-associated bloodstream infections, followed by ventilator-associated pneumonia, are the most common healthcare-associated infections in the NICU. These infection rates increase with decrease in birth weight and age (13,14,15 and 16). Reports published by the CDC-NNIS provide information on pooled means and percentiles of the distribution of device-associated infection rates by birth weight category and device utilization ratios (Tables 49-1 and 49-2) (12). Pediatric Prevention Network’s point prevalence study of NICUs also reports frequency of other infections in the lower respiratory tract(12.9%), urinary tract (8.6%), and ear-nose-throat (8.6%) (11). NICU surveillance is conducted in various methodologies including reporting of site-specific rates by birth weight in the United States (17). The frequency of healthcare-associated infections in individual NICUs varies from 6% to 25% in the United States and 8% to 10% in Europe based upon multicenter studies data (18,19,20,21 and 22).

In a multicenter study from National Institutes of Child Health and Human Development (NICHD) Neonatal Research Network of 5447 patients, the increased use of intrapartum antimicrobial therapy, for prevention of group B streptococcal disease, was associated with a decrease in the incidence of early-onset (<3 days) group B Streptococcal sepsis (from 5.9 to 1.7 per 1000 live births) but was associated with a proportionate increase in sepsis with ampicillin-resistant Escherichia coli isolates (from 3.2 to 6.8 per 1000 live births) (23). In high-risk infants there may also be an increase in the incidence of healthcare-associated infections due to multi-drug resistant gram-negative organisms. This increase in gram negative organisms has been seen in many national and international centers (24,25,26,27,28,29,30,31,32,33 and 34).

Identifying neonates who are at greatest risk of developing nosocomial sepsis can be complex, but is important so that prompt, accurate diagnosis can be made, appropriate management modalities can be instituted and unnecessary use of broad-spectrum antimicrobial agents can be avoided. Various scoring systems have been developed in an attempt to provide objective criteria to diagnose
nosocomial sepsis. Recently, the Nosocomial Sepsis Prediction score for neonates (NOSEP) attempts to do this, based on clinical and laboratory parameters. This model has been validated in five regional NICUs in Europe (35). The NOSEP-NEW- II score includes: (a) body temperature >38.1°C, (b) C-reactive protein ≥30 mg/L, thrombocy topenia, ≤190×10⁹, and neutrophil fraction >63%, and (c) total parenteral nutrition, ≥15 days. The incorporation of additional factors in this score such as surgery, maternal hypertension, and ventilation at the time of sepsis work up, increased the discriminative performance (A_z of receiver operating characteristic curve from 0.71 to 0.82). The NOSEP-NEW-II score (range 0-19 points) was significantly higher in patients with healthcare associated sepsis (median, 9) than with patients without sepsis (median, 6), p= 0.008. The calibration curve of the NOSEP score shows a very good correlation between predicted risk and observed prevalence of nosocomial sepsis (Spearman’s correlation coefficient p <.005). No discrepancy was seen between the predicted and observed outcomes (Hosmer-Lemeshow goodness-of-fit chi squarep=.48). There was also very good uniformity of fit (similar fit across different strata). A model to predict a future event such as the risk of developing a healthcare associated infection should incorporate both intrinsic and extrinsic risk factors as was done in the pediatric nosocomial infection risk model (36). Limiting factors of these models are applicability differences in case-mix, the diversity of the patient population with varying risk factors (e.g., access to care) and availability to limited laboratory resources developing countries (37).

Surveillance of healthcare-associated infection rates is used to evaluate the infection trends in NICUs and to compare different patient populations. These rates are the basic statistical tools for infection control. A comprehensive infection surveillance, prevention, and control program can be a part of the performance-improvement program in the hospital and annual planning for strategies to meet specific hospital needs and priorities (38). Many of the difficulties in comparing intra-hospital and inter-hospital rates can be decreased using standardized definitions and taking risk factors into account (15).

The CDC-NNIS System defines a nosocomial infection as a localized or systemic condition (1) that results from an adverse reaction to the presence of an infectious agent(s) or its toxin(s) and (2) that was not present or incubating at the time of admission to the hospital (39). For determination of the presence and classification of healthcare-associated infections, a combination of clinical findings and results of laboratory tests should be used. These definitions have been developed by CDC to provide uniform criteria to compare infections within hospitals and between different data systems. For infections in infants and neonates, whose clinical manifestations differ from those in older persons, specific criteria are used. These criteria are used according to the listing of major and specific site codes and descriptions. For example laboratory-confirmed primary bloodstream infection in infants or neonates must meet at least one of the following criteria:

The evaluations and comparisons of healthcare-associated NICU sepsis caused by CONS have failed partly due to the lack of uniformity in the definition of sepsis (40). Recently using mathematical modeling, the predictive value of a single and two positive blood cultures for CONS was estimated to be 55% and 98% respectively (41). From a surveillance point of view, this may require improving current CDC definitions as follows:

Findings from this mathematical modeling are also consistent with recommendations from the Vermont Oxford Neonatal National Evidence-based Quality improvement collaborative for Neonatology (42).

The accuracy of reporting healthcare-associated infections data also varies according to the type of infection (43). In nine participating NNIS hospital intensive-care units, the CDC epidemiologists determined the reported bloodstream infections, pneumonias, surgical-site infections and urinary tract infections. Primary bloodstream infection was the most accurately identified with sensitivity of 85% and specificity of 98.3%, in comparison to the other infections, pneumonia, surgical-site infection and urinary tract infection with a sensitivity of 68%, 67%, and 59% and specificity of 97.8%, 97.7%, and 98.7% respectively. When these CDC-NNIS hospitals reported a healthcare-associated infection, the infection most likely was a true infection.

The CDC-NNIS System collects data from infection control professionals on nosocomial infections in NICU patients. All body sites are included in this data. Site-specific infection rates can be calculated by using as a denominator the number of patients at risk or patient-days, and/or device days (umbilical catheter/central line or ventilator use) for each of the four birth-weight categories (≤1000 g, 1001 to 1500 g, 1501 to 2500 g, and >2500 g) (12). Birth weight specific rates of bloodstream infection and ventilator-associated pneumonia are shown in Tables 49-1 and 49-2. The incidence of bloodstream infections and ventilator-associated pneumonia is inversely proportional to the birth weight. Birth weight can also be a marker of severity of illness and a predictor of mortality in very low birth weight infants (44).

The device utilization days are widely used in calculating risk-adjusted infection rates in NICU. Device utilization is calculated as follows: Number of umbilical and central line-associated bloodstream infections per number of umbilical and central-line days multiplied by 1000. Another useful measure of device use in a NICU is the device utilization ratio (12). This is calculated as number of device-days per number of patient-days. Device utilization stratified by birth weight group and device utilization ratio are essential for valid inter-hospital NI rate comparisons (17). Increase in device utilization can reflect either higher severity of illness in infants requiring increased use of invasive devices or healthcare provider’s patient care practice. Under these circumstances targeted efforts to decrease infection rates may require efforts to decrease use of devices. CDC-NNIS rates can serve as an external benchmark and NICUs can be compared using the percentiles of site-specific infection rate and device utilization ratios. If the infection rates are above the 90^th percentile, it may indicate a problem in the infection control or higher severity of illness in patients requiring more device use. If they are below 10^th percentile, it may reflect of underreporting of infections or infrequent or/and short duration of device days (12).

The effect of infection control interventions for a NICU should be evaluated at least annually using a risk-adjusted healthcare-associated infection rate. Risk should be adjusted for device use such as central venous catheters and birth weight category (45). Overall impact of an infection control program can be easily estimated within the hospital over a period of time (38). Continuous quality control improvement processes by multidisciplinary teams in a NICU can be evaluated using bloodstream infections rates over the time period in focus (46). Reduction of infection rates will encourage the efforts and show the right direction of control measures.

Many risk factors are not accounted for in the CDC-NNIS High-Risk Nursery surveillance components (47,48 and 49). Besides birth weight and use of invasive devices, length of stay in NICU and severity of the co-morbidities and use of total parentral nutrition may contribute to increased incidence of healthcare-associated infections (50,51). Use of dexamethasone and neonatal neutropenia and maternal hypertension should also be considered important risk factors under specific maternal conditions (52,53). High-risk infants acquire nosocomial pathogens either endogenously or exogenously. Exogenously parents can also serve as a vector for transmission of organisms like Pseudomonas aeruginosa and Staphylococcus aureus (54). This vertical and horizontal transmission of microorganisms can be confirmed by use of molecular epidemiological tools (55).

The frequency of healthcare-associated or late-onset sepsis varies from 11% to 30% (56). The number of sepsis episodes increases with decreasing birth weight. Inter-center variability is also seen in the incidence of late onset sepsis in multi-institutional studies (6,57,58). CONS are the most common pathogens responsible for late-onset nosocomial sepsis in the NICU (6). It is still difficult to determine which blood culture isolates of CONS reflect true infections and which are contaminants. CONS are part of the patient’s endogenous flora. Oral mucosa and/or skin are colonized with CONS (59). The NICHD Neonatal Research Network reports that Gram-positive organisms specifically CONS were the most common pathogens causing late-onset >3 days) bloodstream infections in neonates. CONS was isolated from 43.4% of cases with late onset sepsis in very low birth weight infants in the NICHD Neonatal Research Network and 54% in Pediatric Prevention Network (6,60). In a prospective NICU incidence study CONS bacteremia are reported to be 12.4 per 1000 device days (61). Infants with late-onset sepsis have significantly increased length of stay in the hospital (79 days vs 60 days; p <.001) and mortality (18% vs 7%;<.001) (6). CONS infections usually present between 7 and 14 days of age (62). It is not a significant cause of mortality (less than 1%) (63).

Prolonged hospital stay and complications with prematurity, such as patent ductus arteriosus, prolonged ventilation, prolonged use of intravascular access, bronchopulmonary dysplasia and necrotizing enterocolitis are associated with an increased rate of late-onset sepsis (6). Persistent infections with CONS occur in significantly smaller and less mature infants than with non-CONS, but generally mortality is not higher. Infants with persistent infection should undergo aggressive evaluation for focal complications (64). Primary osteomyelitis and suppurative arthritis with CONS in preterm neonates have been also described in the absence of any indwelling central catheters (65). Most CONS isolates causing sepsis are frequently predominantly antibiotic resistant CONS types (66).

The role of CONS as causative pathogens or contaminants in cultures from blood and cerebrospinal fluid is difficult to determine (67). Procedures to help to differentiate CONS positive blood cultures from CONS contaminants include drawing of at least two blood cultures, using optimal skin antisepsis and catheter disinfection before drawing blood samples, and using adjunctive tests, such as IL-6 and CRP (42,60,68). CONS are the predominant organism, consisting of 44% versus approximately 25% of various Gram-negative bacteria in randomly collected environmental samples in NICU (3). The possible increase of CONS infections after twice-daily petrolatum ointment application indicates that routine ointment practices may colonize skin. Thus ointment routines remain controversial, especially in babies with intact skin (69).

Another cause for possible therapeutic failures in the NICU, is the emergence of vancomycin heteroresistant Staphylococcus capitis strains (70). Occurrences of endemic strains require careful evaluation of existing treatment regimens and therapeutic responses. Heteroresistance to vancomycin should be suspected in a NICU where CONS positive bloodstream infections treated with vancomycin and appropriate replacement of intravenous catheters fail to respond to treatment. There is an urgent need for evaluation of clinical practice guidelines for safely decreasing vancomycin use in NICUs (60).

In a recent CDC-NNIS surveillance report 89% of CONS were resistant to methicillin (12). Resistant Staphylococcus epidermidis clones to penicillin, gentamicin and erythromycin with the mecA gene can be significant nosocomial pathogens that can be transmitted between babies in the NICU (71). Usually predominant antibiotic resistant CONS types are detectable in neonates and staff of neonatal units, suggesting cross-contamination (66).

A recent study by NICHD Neonatal Research Network revealed that only one-third of patients with late-onset meningitis had meningitis in the absence of sepsis. Cerebrospinal fluid cultures were performed only half as often as blood cultures, suggesting that meningitis may be under-diagnosed among very low birth weight infants. CONS was recovered from 29% of cases when meningitis occurred (72,77).

Enterococci have also been recognized as clinically important pathogens in high-risk, hospitalized children (73,74,75,76,77). Neonatal nosocomial enterococcal infections are being recognized with increased frequency (78,79,80,81,82,83,84,85). The Pediatric Prevention Network’s point prevalence study of NICUs also reports enterococci as the second most frequent (15%) nosocomial pathogen among neonates admitted to the NICU in 1999 and third most common (7.8%) in NICHD Neonatal Research Network studies (11). Most of these were bloodstream infections, but enterococci were isolated in 13.4% of cases with meningitis in NICHD Neonatal Research network studies (86). Enterococci are in digenous to the normal flora in humans and are known to colonize the intestines, gastrointestinal tract, and female genital tract. There are two major species of Enterococci that infect humans, Enterococcus faecalis(E. faecalis) and Enterococcus faecium (E. faecium). During the 1980s, E. faecalis was responsible for 80% to 90% of all isolates and E. faecium had accounted for only 5% to 15% (87).

In one case controlled study risk factors for nosocomial enterococcal infection in neonates are non-umbilical central lines (71% vs 32%), duration of indwelling central lines (26.5 vs 6.5 days), and bowel resection (29% vs 4%) (83). The mean gestational age was 27 weeks and the mean birth weight was 913 grams. Similarly, risk factors in another case-control study of young children also showed that central line and bowel resection are important risk factors, as well as duration of antimicrobial therapy (median duration of use was approximately one week) (88). A mortality of 14 to
17.6% and an attributable mortality of 8.0% were observed with enterococcal bacteremia patients (73,78). To prevent occurrence of nosocomial enterococcal infections, clinicians should carefully monitor the use and duration of antimicrobial agents. Rational delivery of antimicrobial therapy is critical, especially during an era of increasing rates of antimicrobial resistance in enterococci.

Critically ill, premature neonates in NICUs require ventilator support. The care of these hospitalized neonates contributes to increased rates of ventilator-associated pneumonia (11,13,15,17). Diagnosis of ventilator-associated pneumonia, however, is difficult in neonates with bronchopulmonary dysplasia due to underlying radiographic changes (132). A positive blood culture along with radiographic changes makes diagnosis of ventilator-associated pneumonia with greater specificity and sensitivity (39, 133). The rate of ventilator-associated pneumonia disproportionately affects very low birth weight neonates (Table 49-2) (134). In pediatric intensive care units, ventilator associated pneumonia is also the second most common nosocomial infection and P. aeruginosa is the most common cause of these infections, accounting for 22% of the cases (14).

Airway colonization of in-patients begins within 3 days to two weeks of the start of mechanical ventilation and may be due to either endogenous or exogenous organisms. In young children bacterial colonization and infection occurs more frequently in patients ventilated through a tracheotomy. During endotracheal ventilation, lower airway colonization occurs after two weeks of mechanical ventilation (135). Very low birth weight infants are at-risk of airway colonization with Gram-negative bacilli and also bloodstream infections (132,136). Airway colonization with Gram-negative rods may also be associated with severity of brochopulmonary dysplasia (137). Systemic antimicrobial treatment does not eradicate Gram-negative rod airway colonization. This may be in part, to the presence of foreign bodies (tracheal or endotracheal tube), low levels of antimicrobial concentration in respiratory secretions, or translocation of bacteria from the gastrointestinal tract (138). Topical antibiotic combinations of polymyxin E and tobramycin in a 2% paste applied four times a day to the tracheostoma in young children (median age 4 months) were found to be effective in reducing the exogenous route of colonization of the lower respiratory tract (139). High case-fatality rates (30-50%) with P. aeruginosa invasive infections has been seen in patients with compromised host defenses and in very low birth weight infants (140,141). Of note, tracheal culture can be negative preceding invasive infections. Ventilator-associated pneumonia is usually present when tracheal culture and blood culture simultaneously grow P. aeruginosa. In a retrospective study of 1571 very low birth weight infants, 751 patients were on a ventilator for longer than a week. P. aeruginosa colonization occurred in 33 patients and of these, 15 patients also developed bloodstream infections. Thirteen of the 15 infants with bloodstream infections and none of the 33 with tracheal colonization died within two days of positive culture (134).

Care processes in ventilated neonates vary and account for the different rates of ventilator-associated pneumonia. Most neonates are subject to frequent suctioning to maintain patent airways that can easily be blocked by excessive bronchial secretions. There has been increased use of closed endotracheal suctioning systems that permit continuous ventilation of patients instead of open systems that require disconnection from ventilators. In adults, closed endotracheal suctioning systems are associated with increased rates of bacterial colonization without an increased incidence of nosocomial pneumonia (142). Closed endotracheal suctioning systems have a theoretical advantage for use due to fewer physiologic disturbances and suction-induced complications (143). An outbreak of a healthcare-associated acute Gram-negative rod pneumonia suggestive of P. aeruginosa on autopsy was seen in one NICU concomitant with the increase use of closed endotracheal suctioning systems with a mortality rate of 33% (10/30) in patients (144). It is possible the catheters of closed endotracheal suctioning systems were not changed every 24 hours in this outbreak, as is recommended. This report differs from that of Cordero et al., who failed to show increased colonization and nosocomial P. aeruginosa pneumonia and bloodstream infections among patients undergoing closed endotracheal suctioning systems (145). Guidelines for prevention for nosocomial pneumonia has been recently published (146).

P. aeruginosa is also ubiquitous in the hospital environment and is found on moist surfaces such as sinks, toilets, floor mops and respiratory equipment (141,147,148). Environmental reservoirs have been implicated as sources of transmission of P. aeruginosa to patients in NICUs via the use of hand lotion and on contaminated fingernails of health care workers (147,149). Warm tub bath during a term labor was associated with neonatal P. aeruginosa meningitis and bacteremia in an 11-day-old neonate in Germany. Genotyping revealed clonal similarity between the blood culture isolate and the shower tubing used to clean the bathtub (150).

Healthcare-associated urinary tract infection in the neonate continues to be a source of morbidity and prolonged
hospital stay. Prevalence of urinary tract infection in the hospitalized neonate ranges from 4 to 25% (151). While the prevalence of urinary tract infections in non-hospitalized infants was only 0.1 to 1% (152). These studies demonstrated that prematurity, low birth weight and male gender contributed to increased risk for urinary tract infection in the hospitalized neonate. With improved survival of low birth weight infants, increased risk of urinary tract infection may also be seen in neonates <1000 grams (153). Incidence of vesicourethral reflux is also low in very low birth weight infants (154). In earlier studies neonatal urinary tract infection often occurred in association with sepsis (151). The exact mechanism of urinary tract infection in the very low birth weight is unknown, as they usually do not have in-dwelling catheters.

Multi-drug resistant organisms such as methicillin resistant Staphylococcus aureus (MRSA), vancomycin-resistant enterococcus (VRE), and antimicrobial-resistant Enterobacteriaceae have emerged as major causes of healthcare-associated infections. According to the CDC-NNIS, recent analysis shows a continuing increase in the incidence of multi-drug resistant organisms in intensive care units in hospitals in the United States. As of 2004, almost 28.5% of enterococcal isolates were VRE, 59% of Staphylococcus aureus isolates were MRSA, and 89% of CONS were resistant to methicillin (12). The increase is continuing despite the implementation of standard infection control measures. There are no specific resistance rates for neonates and young children admitted to intensive care units. Clinical manifestations of healthcare-associated infections with multi-drug resistant organisms or susceptible organisms are indistinguishable, which makes it more challenging to diagnose these infections and treat them promptly. The next section discusses the historical perspective of specific multi-drug resistant organisms, epidemiology and interventions.