KEY POINTS
- 1.
Late-onset sepsis in the neonatal intensive care unit is defined by culture-confirmed infection ≥72 hours after birth.
- 2.
Incidence is highest among preterm infants.
- 3.
The pathophysiology involves colonization with perinatally and/or hospital-acquired organisms, with transition to invasive infection promoted by hospital devices and immature mucosa.
- 4.
Risk factors include prematurity, presence of a central venous catheter, prolonged parenteral nutrition, and lack of breast milk feeding.
- 5.
The most common causative organisms are coagulase-negative Staphylococci and Staphlycoccus aureus.
- 6.
Clinical manifestations are nonspecific and often difficult to distinguish from instability characteristic of prematurity.
- 7.
Identification of a pathogenic organism by culture of a normally sterile bodily fluid, primarily blood or cerebrospinal fluid, is currently the gold standard for diagnosis.
- 8.
The choice of empiric antimicrobial therapy should be based on local microbiology and antibiotic susceptibility patterns, and targeted therapy should be narrowed based upon the isolate susceptibility profile and clinical response.
Introduction
Late-onset sepsis (LOS) is an important contributor to morbidity and mortality among both term (≥37 weeks’ gestational age [GA]) and preterm (<37 weeks’ GA) newborn infants. LOS is defined by isolation of a pathogenic species from a normally sterile body fluid. Among preterm infants, LOS is most commonly defined by isolation from blood or cerebrospinal fluid (CSF) culture. Infection confined to the urinary tract, joints, or bones may also occur as part of LOS with specific organisms. Bacteria, fungi (primarily Candida species), and viruses (herpes simplex virus, cytomegalovirus) may cause LOS. A consensus, physiology-based definition for neonatal sepsis does not currently exist. LOS is also defined based on the timing of the infection relative to birth. Among continuously hospitalized infants cared for in the neonatal intensive care unit (NICU), LOS is defined as occurring in infants ≥72 hours or 3 days after birth. Risk factors for infection and organisms causing infection rapidly change from those reflecting perinatal risk and perinatally acquired flora to those reflecting nosocomial risk factors and hospital-acquired flora. In contrast, among infants discharged home from the birth hospital in the first week after birth, LOS is defined as occurring in infants 7 to 90 days of age. This chapter will focus on bacterial and fungal LOS among primarily preterm infants cared for in the NICU.
Epidemiology
The incidence of LOS is inversely related to the degree of maturity (GA and birth weight [BW]) and varies across populations. Up to 40% of very preterm infants (<28 weeks’ GA) admitted to the NICU have at least one episode of LOS, compared with 30% of moderately preterm infants (28–32 weeks’ GA) and 17% of late preterm and term infants (≥33 weeks GA). The peak incidence of LOS is between the 10th and 22nd postnatal day. , The incidence of LOS has increased over time, likely related to improved survival of extremely preterm infants with prolonged hospitalization and intensive care.
Pathophysiology
The pathogenesis of gram-positive LOS, particularly that due to coagulase-negative Staphylococci (CONS) and Staphylococcus aureus , is often related to adherence and proliferation of bacteria on indwelling plastic medical devices, whereas gram-negative LOS often occurs by transmission from healthcare workers and contamination of catheters, parenteral solutions, or enteral formulas. , Intestinal, nasal mucosal, and skin colonization with invasive pathogens may also promote LOS both in the presence and absence of invasive medical devices. ,
Risk Factors
Multiple studies provide evidence for both nonmodifiable and modifiable risk factors for LOS. Nonmodifiable risk factors include lower GA and lower BW. , , Among preterm infants, some studies suggest small-for-GA infants may be at increased risk for LOS, particularly CONS infection. , Potentially modifiable risk factors for LOS include prolonged parenteral nutrition, presence of a central venous catheter (CVC), and breast milk feeding. , , , The decreased risk of LOS with breast milk is likely attributable to the complex immunomodulatory and antiinfective components of breast milk; more rapid achievement of full enteral feeds and decreased exposure to parenteral nutrition time and CVCs; and potentially to differences in constitution of the infant gut microbiome. Multiple studies identify presence and duration of a CVC as a risk factor for LOS. This may be due to entrance of commensal skin organisms into the catheter track or contamination of the catheter hub. , Mechanical ventilation and bladder catheterization also increase the risk of LOS. There is conflicting evidence for the role of early antibiotic administration on subsequent risk of all-cause LOS, fungal LOS, and comorbidities associated with LOS, such as necrotizing enterocolitis (NEC). , , Racial disparity has been observed in LOS caused by group B Streptococcus (GBS), with maternal black race a significant risk factor. Additional specific risk factors for fungal LOS due to Candida albicans include treatment with cephalosporin antibiotics, steroids, intralipids, and gastric acid suppressing medications.
Microbiology
LOS is primarily (but not exclusively) associated with organisms acquired from the environment after birth ( Fig. 31.1 ). In a cohort of preterm infants born with BW <1000 g and GA <29 weeks from the Neonatal Research Network during 2000 to 2011, the majority of LOS was caused by gram-positive bacteria (73%). Gram-negative bacteria caused 17% of LOS, and fungal organisms caused 7%. CONS were isolated in 55% of the LOS cases. Other reports confirm that CONS are the most frequent blood culture isolates associated with neonatal LOS. , CONS are common commensal organisms that colonize human skin and mucosal membranes and can adhere to indwelling catheters and plastic medical devices in susceptible patients, forming multilayered biofilms. , CONS are also common blood culture contaminants in NICU patients, and determining true infection can be challenging. Other gram-positive bacteria causing LOS include S. aureus, Enterococci , and GBS, whereas Escherichia coli , Klebsiella , Pseudomonas , Enterobacter , and Serratia species are among the most common gram-negative LOS isolates. , , The majority of fungal LOS is caused by Candida species, primarily C. albicans and Candida parapsilosis. , , Although less common, viral infections can mimic bacterial LOS among preterm infants, and appropriate contextual consideration should be given to etiologies such as herpes simplex virus, seasonal respiratory viruses such as respiratory syncytial virus, and influenza, as well as cytomegalovirus among preterm infants fed their mother’s own breast milk. The prevalence of causative organisms may vary widely based on geographic region. , ,
Clinical Features
The signs and symptoms of LOS among newborns are often nonspecific and may be especially difficult to distinguish from physiologic instability characteristic of preterm infants. Respiratory decline from established baseline, feeding intolerance, and increased apnea are among the most common reasons for LOS evaluation. In a prospective cohort study of preterm infants with suspected infection, delayed capillary refill and gray skin were specifically associated with LOS. Infants may also present with hypothermia or hyperthermia (although temperature can be normal during isolette care), decreased activity, and tachycardia. Late signs may include cyanosis and/or hypotension; oliguria or anuria may be the first indication of significant hypotension. Jaundice may be a presenting symptom, particularly for urinary tract infection (UTI). Irritability and fontanelle bulging may accompany central nervous system infection. The combination of hyperglycemia and thrombocytopenia is a common feature of infants with fungal LOS.
Evaluation
Cultures
The diagnosis of LOS is made by culture of a pathogenic species from blood and/or CSF. Although meningitis is often a metastatic complication of bacteremia, late-onset meningitis can occur in isolation among preterm infants. Site-specific infection may occur without bacteremia: UTI, pneumonitis, and cellulitis may be diagnosed by site-specific culture (urine, tracheal fluid, drainage fluid). The optimal approach to diagnosis of late-onset infection includes blood and CSF culture prior to administration of empiric antibiotics. Urine culture should be obtained in older infants (especially those without CVCs), and site-specific cultures and radiographic studies should be obtained as clinically indicated. Appropriate blood culture technique can minimize concerns about the sensitivity of blood culture to detect bacteremia. Standard pediatric blood culture bottles should be inoculated with a minimum of 1 mL of blood to optimize organism recovery. When there is concern for gastrointestinal infection, the use of both aerobic and anaerobic culture bottles can optimize isolation of strict anaerobic species. Obtaining two blood cultures from separate sites, especially when a central line is present, aids in determination of contamination versus true infection. Modern blood culture systems use automated, continuous detection technologies with identification of a positive culture within 24 hours for most clinically relevant bacterial pathogens. Time-to-positivity data for neonatal cultures in one study suggest that 95% of bacterial and 84% of fungal pathogens are detected within 48 hours of incubation. Another study found that 71% of positive blood cultures were detected by 24 hours; CONS were detected after a mean of 21.7 hours. When bacteremia is identified, CSF analysis and culture should be performed if not done previously. Empiric antibiotic therapy should be adjusted as needed in response to isolate susceptibility data, using the narrowest spectrum of appropriate therapy. Repeat blood cultures (and CSF cultures, when meningitis is present) should be obtained to document sterility in response to therapy. Persistent bacteremia (>2–3 positive cultures on appropriate antimicrobial therapy) should prompt investigation for a site-specific complication such as venous thrombosis, abscess, osteomyelitis, or other organ-specific infection.
Inflammatory Markers
Multiple studies have evaluated the efficacy and utility of the complete blood cell count (CBC) for LOS diagnosis. Overall, the CBC and its components have poor predictive value, and no CBC index possesses adequate sensitivity to reliably rule out LOS. , Serial normal CBC values may, however, provide reassurance that LOS is not present. Inflammatory biomarkers may also be used, such as C-reactive protein (CRP), procalcitonin, and interleukin 6. CRP is an acute-phase reactant synthesized by the liver and reflects tissue damage, infection (bacterial, fungal, or viral), necrosis, or general inflammation. Sensitivity of CRP may be low during the early stages of infection, and elevation is nonspecific to LOS and may be related to other causes. These issues limit its utility, although serial levels may be of higher yield. , CBC and CRP may be of most predictive value in the individual infant if the clinician is able to compare baseline values with those obtained at the moment of clinical concern. Procalcitonin, an acute-phase reactant produced by hepatocytes and macrophages, has shown promise given its specificity for bacterial infections among neonates, early rise in levels during infection, and quick reduction in response to appropriate therapy, but wide variation in levels across uninfected infants has limited its use. , Interleukin 6 is a cytokine released in response to exposure to bacterial endotoxins and demonstrates an ability for early detection given its rise in the very early stage of infection, but thus far it has limited use in clinical care due to its very short half-life. , Other biomarkers and cytokine profiles have been studied to improve the diagnostic accuracy of LOS, but none are currently routinely used by neonatal providers. ,
Other Tests
Advanced molecular-based methods for identification of infectious organisms are emerging, including polymerase-chain reaction and microarray techniques. Advantages include small sample volume, quick turnaround time, and improved sensitivity and specificity. However, such techniques are currently limited by cost, inability to distinguish between colonization and true infection, inability to distinguish bacterial DNA from live bacteria, and lack of antimicrobial susceptibility testing. Heart rate characteristic variability has been studied in LOS prediction, both in isolation and in combination with laboratory tests and/or other physiologic data. Decreased heart rate variability and transient decelerations related to inflammation may indicate a high risk of LOS. Although one trial demonstrated decreased mortality in very preterm infants with displayed heart rate characteristics, the mechanism of decreased mortality was unclear. Available monitors are currently limited by a lack of specificity for LOS, meaning that frequent abnormal values may prompt additional evaluations for infection and create alarm fatigue among clinicians.
Prevention
Primary prevention of LOS should focus on the sources of infection. Nosocomial LOS is largely due to central-line–associated bloodstream infections (CLABSIs). Guidelines promoting strict hand-hygiene practices, barrier precautions, skin antiseptics, optimized daily care practices, management of infused fluids and infusion sets, and prompt removal of CVCs have been shown to reduce the incidence of LOS. Centers should develop local CVC care guidelines informed by best practice. Optimizing feeding strategies and interventions to reduce the need for CVCs may also contribute to CLABSI prevention. Secondary bloodstream infections can result from organ-specific diseases such as NEC, UTI, pneumonia, or cellulitis. Prevention of secondary LOS should focus on strategies to reduce the incidence of the primary condition. The risk of some conditions (e.g., NEC and UTI) may be impacted by specific practices such as breast milk feeding for NEC and minimal indwelling bladder catheters for UTI, but neither of these conditions is entirely preventable among preterm infants. There are no currently known approaches to the prevention of GBS-specific LOS. Probiotics have been studied as a potential means of preventing LOS by populating the intestinal microbiota with low-virulence organisms that normally constitute the microbiome of healthy term infants. Trials of probiotic supplementation for the prevention of LOS have inconsistent results, and in meta-analyses, the intervention was not shown to decrease the incidence of the disease. , Lactoferrin supplementation has also been studied for the prevention of LOS; a review of six randomized controlled trials suggests that the intervention decreases LOS without adverse effects but found the current evidence quality is low. A recent randomized controlled trial of lactoferrin supplementation among 2203 infants born at <32 weeks’ gestation found no impact on LOS. Fluconazole prophylaxis reduces the incidence of invasive candidiasis among extremely low BW infants. Centers with high rates of Candida colonization/infection should consider empiric therapy for such infants. Among centers with a relatively low burden of Candida infection, evidence does not support universal, BW based administration of fluconazole prophylaxis, although such centers may opt to administer targeted prophylaxis to extremely low BW infants (<1000 g) at highest risk, such as those receiving prolonged antibiotic therapy. National surveillance on incidence rates can indirectly reduce LOS by providing contemporary data to facilitate interhospital comparison, benchmarking, and continuous improvement efforts.
Empiric Therapy
Empiric therapy should be based on local center data informing the most commonly isolated organisms and antibiotic susceptibility patterns among LOS cases. Typical regimens should include coverage for gram-positive and gram-negative bacteria, although there is a lack of clinical consensus on the most appropriate regimen. Vancomycin is often used for gram-positive coverage because the majority of CONS are not susceptible to penicillinase-resistant β-lactams, and vancomycin provides coverage for methicillin-resistant S. aureus (MRSA) and ampicillin-resistant Enterococci . Empiric vancomycin prescription may not always be necessary, however, given that LOS with CONS has low virulence and is rarely fatal. Empiric use of penicillinase-resistant β-lactam antibiotics such as oxacillin or nafcillin may be appropriate in centers with low rates of MRSA infection, particularly if the center also utilizes prospective screening for MRSA colonization. Empiric gram-negative therapy typically includes an aminoglycoside such as gentamicin, amikacin, or tobramycin. Alternative gram-negative coverage options may include a cephalosporin, although emergence of colonization and infections with cephalosporin-resistant gram-negative organisms has been reported in units that routinely use these medications. , Amphotericin B should be used when empiric antifungal therapy is indicated. When blood cultures are sterile, antimicrobials should be discontinued after 36 to 48 hours based on local time-to-positivity data unless additional diagnoses (e.g., NEC) are present. Persistent symptoms in the face of sterile cultures and empiric broad-spectrum antibiotic therapy should prompt additional investigations such as renal and brain imaging for signs of focal fungal infection, cytomegalovirus testing among preterm infants fed their mother’s own milk, and respiratory viral testing among infants with respiratory symptoms and/or decline.
Targeted Therapy
The duration of therapy for infants with culture-confirmed bacterial LOS is typically 10 days after documenting sterile cultures but may vary from 7 to 21 days depending on the pathogen, type of infection, and time to culture clearance. Longer treatment duration is appropriate for a site-specific infection such as meningitis, septic arthritis, or osteomyelitis, and longer treatment may also be indicated for fungal infection. Once a pathogen is identified, antimicrobial therapy should be narrowed based upon the susceptibility profile. Antimicrobial therapies for the most common causative organisms of LOS are shown in Table 31.1 . Repeat blood cultures should be obtained every 1 to 2 days until sterility is achieved. When persistent bacteremia occurs (generally defined as >2 positive cultures on appropriate therapy), an isolated source of infection (abscess, septic thrombus, septic arthritis, osteomyelitis, or endocarditis) should be considered and appropriate imaging obtained. Consultation with infectious disease specialists is recommended for cases involving prolonged bacteremia, meningitis, or organ-specific fungal infection.
