EOS mainly develops from the transmission of bacteria that colonize the maternal genito-urinary and gastrointestinal system during the intrapartum period (39). Group B streptococcus (GBS) and Escherichia coli account for 70–80% of blood and cerebrospinal fluid cultures (40–42). Intra-partum antimicrobial prophylaxis (IAP) became the standard of clinical care to prevent GBS transmission in the 1990s (43). With its wide implementation, the incidence of EOS due to GBS in the United States has decreased from 1.5–1.7 cases per 1000 live births to 0.23 per 1,000 live births (44). Recent population surveillance studies in North America revealed that E. coli disease primarily occurs among preterm infants; GBS disease primarily occurs among term infants, with almost half occurring in infants born to mothers with negative GBS screening test results (45).

Gestational age (GA) is the strongest predictor of EOS (39). The incidence of EOS is inversely associated with GA, with 0.5 cases per 1000 neonates born at ≥34 weeks' gestation and up to 32 cases per 1,000 in those born at 22–24 weeks of gestation (39, 46). The EOS rate was reported to be 4.3–5.2% among infants born at a GA of <25 weeks in Canadian cohorts (47, 48). The risk factors associated with EOS in term infants include intrapartum GBS colonization or GBS bacteriuria during the current pregnancy, previous infant with invasive GBS disease, prolonged rupture of membranes ≥18 h, and maternal temperature ≥38°C (49, 50). In preterm infants (≤35 weeks' gestation), unexplained spontaneous preterm labor, premature rupture of membranes (PROM), acute onset of atypical fetal heart rate, or concern for intra-uterine infection (or chorioamnionitis) are risk factors for EOS (39).

Infants often have nonspecific clinical manifestations that can be indistinguishable from normal newborn transitions or prematurity conditions (38). Positive blood or cerebrospinal fluid culture is the hallmark of EOS. In general, blood cultures growing clinically significant pathogens are positive within 24–36 h of incubation (51, 52). Molecular assays like real-time PCR have been evaluated for the rapid detection of pathogens in neonatal EOS. They provide care providers with the critical information of organism identification and antimicrobial susceptibility from cultures flagged as positive, faster than with traditional microbiology methods, which may allow the clinicians to narrow empiric antibiotic choices more efficiently. They are also being studied as 'add-on' tests to identify pathogens that are not detected by conventional blood or CSF culture (53, 54).

Low white blood cell counts, low absolute neutrophil counts, and high immature-to-total neutrophil ratios are associated with increased odds of infection, but the sensitivities of complete blood count (CBC) indices were low (0.3–54.5%) to rule out EOS (55, 56). There is concern regarding falsely normal white blood cell screening tests among infants with fulminant sepsis, particularly when the CBC screening is performed between 1 and 7 h of life; consequently, repeated CBC screenings between 12 and 24 h of age are reasonable (57, 58).

C-reactive protein (CRP) is a non-specific inflammatory marker that typically begins to increase 6–8 h after infection onset. Levels are only elevated in 35–55% of neonates at the onset of illness, and for the first few hours after birth (59, 60). It is sensitive but not specific, so if it remains normal at 24–36 hours and the infant appears well, antibiotics can be safely discontinued (61, 62).

In the Neonatal Procalcitonin Intervention Study (NeoPInS), Procalcitonin-guided decision making was found to be superior to standard care in reducing antibiotic therapy in neonates with suspected EOS (intention to treat: 55.1 vs. 65.0 h, p < 0.0001) (63). Dongen et al. reported that umbilical cord blood procalcitonin (PCT) levels are increased in newborns ≥32 weeks with proven or probable EOS, but PCT might not be a reliable marker after maternal antibiotic treatment (64).

Cytokines, including interleukin 6 (IL-6), interleukin 8 (IL-8), gamma interferon (IFN-γ), and tumor necrosis factor alpha (TNF-α), and cell surface antigens such as soluble intercellular adhesion molecule (sICAM) and CD64, have been studied as measures for neonatal sepsis; however, the heterogeneity of available evidence prevents conclusions regarding their routine use as definitive diagnostic tools (38, 65, 66).

There is no additional benefit to include urine microscopy or culture as part of the investigations for early-onset neonatal infection in the first 24 h of life among both preterm and term infants because most infections of the urinary tract in this population are secondary to hematogenous seeding of the kidney by bacteremia (67, 68).

While lumbar puncture (LP) is an important means of obtaining CSF to rule out the presence of meningitis in infants, its routine use in neonates for EOS evaluation remains controversial (38). In any infants, the timely collection of CSF samples should balance the cardiorespiratory stability of the infant, the risk of EOS, and the potential harm associated with delayed antibiotic therapy. While antibiotics can render the CSF culture negative and no single CSF parameter can reliably exclude the presence of meningitis, antibiotic initiation should not be delayed by the LP procedure itself in critically ill infants (69).

The primary focus is to identify infants at high risk for infection to warrant empiric antibiotic therapy. Three approaches currently exist in the literature for the use of risk factors to identify infants who are at increased risk of EOS (50).

In the first approach, the following threshold values for intrapartum risk factors are utilized: (i) any newborn infant who is ill appearing; (ii) a mother with a clinical diagnosis of chorioamnionitis; (iii) a mother with GBS colonization and who received inadequate IAP, with a duration of ROM of >18 hours or birth before 37 weeks of gestation; or (iv) a mother with GBS colonization and who received inadequate IAP but with no additional risk factors. Recommendations include observation in the hospital, laboratory testing, empirical antibiotic therapy for infants, or a combination of these (50, 70).

In the second approach of the multivariate risk prediction model, an individualized risk of EOS based on several risk factors is generated, and then that risk is modified depending on the infant's clinical condition specified by explicit clinical characteristics in the first 12 h after birth (34). The Kaiser Permanente Research group developed an online sepsis risk calculator (SRC) to evaluate the risk of neonatal EOS, which has been validated as a safe way to reduce the proportion of infants receiving empiric antibiotics without apparent adverse effects in infants born after ≥35 weeks' gestation (71).

The final approach consists of serial evaluations within the first 48 hours of age for clinical signs of illness among infants with risk factors to identify infants with EOS, which have been shown to significantly decrease with the use of laboratory tests, blood cultures, and empirical antibiotic agents (50, 72).

Canadian Pediatric Society has also developed an algorithm to provide updated recommendations for the care of term (≥37 weeks' gestation) newborns with risk factors for EOS during the first 24 h of life (73).

The above mentioned approaches and algorithms must be used in conjunction with meticulous history-taking, careful clinical examination, and serial assessment of the at-risk infant. Used in this way, clinicians can identify those infants who are at high risk of developing EOS and prevent unnecessary antibiotic exposure in a substantially larger number of infants who are uninfected.

Blood culture is the gold standard for diagnosis of EOS, but its sensitivity can be affected due to small volume of blood sample and antibiotics given to newborn before sampling. In the recent years, umbilical cord blood culture for the diagnosis of EOS in term infants has been advocated as a large volume can be readily obtained and shown to have >80% sensitivity and specificity compared to venous blood samples (74), though strict and proper sterile techniques are needed to prevent contamination (75).

GA is the strongest predictor of EOS, and two-thirds of preterm births are associated with preterm labor, PROM, or clinical concern for intrauterine infection (49). Clinical features and common laboratory tests, such as CBC and CRP, are impacted by maternal conditions and cannot be reliably used to predict EOS among preterm infants. The objective of EOS risk assessment among preterm infants is to determine which infants are at the lowest risk for infection such that they are eligible to be spared from the administration of empirical antibiotics, or at the very least to be administered a shorter duration of empirical antimicrobial coverage (39).

Preterm infants who are considered at a lower risk for EOS have the following characteristics at the time of delivery: (i) obstetric indications for preterm birth (e.g., maternal preeclampsia or other noninfectious medical illness or placental insufficiency); (ii) cesarean birth; and (iii) absence of labor, attempts to induce labor, or any ROM before delivery (76, 77). Infants fulfilling these criteria can be considered at the lowest risk of EOS and may be managed without empiric antibiotics at birth (39). In a single-center study evaluating the impact of such a risk-factor approach in EOS evaluation among preterm infants, the guideline implementation was associated with decreased antibiotic initiation among low-risk, extremely low birth weight (ELBW) infants, without any change in incidence of confirmed infection or death (78). A randomized controlled trial is currently being conducted to test if the incidence of adverse outcomes is higher in babies receiving empiric antibiotics in the first week of life compared to babies receiving placebo (79). This study targets a population of ELBW infants that are clinically stable that did not have a known exposure to intraamniotic infection and were not born preterm for maternal indications.

In contrast, infants born preterm because of maternal cervical incompetence, preterm labor, PROM, clinical concern for intra-amniotic infection, or acute onset of unexplained non-reassuring fetal status are at the highest risk for EOS. All should undergo EOS evaluation with blood culture and empirical antibiotic treatment (39). Clinicians often express concerns about poor sensitivity of blood cultures in neonates resulting from inadequate inoculant volumes, resulting in frequent diagnosis of “culture-negative sepsis”, and prolonged use of empiric antibiotics in neonatal units (80). To help avoid a “culture-negative sepsis” diagnosis and prolonged use of unnecessary empiric antibiotics in neonates, a minimum of 1 ml of blood culture volume has been recommended, based on the bench study and expert recommendations (80–83). Real-time PCR has the advantage of detecting organisms present in low concentrations using small volumes of blood and providing rapid results. However, the sensitivity and specificity for diagnosis of EOS and detection of antibiotic resistance remains under investigation (53, 84).

The combination of ampicillin and gentamicin is most commonly administered to neonates at risk for EOS. This combination targets the most common pathogens pending culture results (39, 50). Aminoglycoside-based regimens should generally be used instead of cefotaxime-based treatments because of lower levels of susceptibility to cefotaxime and to potentially prevent the development of antibiotic resistance (85).

The empirical administration of more broad-spectrum agents may be indicated in infants who are critically ill with additional risk factors, particularly after prolonged antepartum maternal antibiotic treatment, or if the mother is known to be colonized with an antimicrobial resistant organism, until appropriate culture results are known (39, 50). In some places, benzylpenicillin is being used instead of ampicillin, which is a narrower-spectrum alternative. Decisions regarding the optimal empirical coverage should be guided by the local epidemiology and antibiogram.

The use of the narrowest-spectrum effective antimicrobials allows for the targeted killing of the infecting pathogen while minimizing undesirable impacts on the colonizing microbial flora of the infants. At the individual level, antibiotic use either perinatally or postnatally has been linked to disruptions in the microbiome (86, 87). Alteration in intestinal microbiota has been postulated to play a critical role in the development of significant neonatal morbidities, through the regulation of systemic inflammation (19). Antibiotic treatment in the first week after birth may also increase the subsequent risk of wheezing, infantile colic, and childhood obesity (88, 89). A study from the Canadian Neonatal Network demonstrated that a prolonged initial antibiotic exposure of >3 days within the first week after birth was associated with increased odds of mortality or significant morbidities, including chronic lung disease, patent ductus arteriosus, retinopathy of prematurity, and severe neurologic injury, even after adjustment for baseline differences in the characteristics of very-low-birth-weight (VLBW) infants (26).

Empirical cefotaxime use without justification should be avoided, as its overuse may increase the prevalence of MDRO, candidiasis, and possibly neonatal death (20, 21, 90, 91).

In the absence of positive blood culture or clear evidence of site-specific infection, infants that were administered empirical antibiotics because of risk factors for EOS or clinical indicators of possible infection who remain well should have their antibiotics discontinued after 36 h, particularly when the trends of CRP concentration are reassuring (39, 50, 65, 92). If antibiotics are continued for longer than 36 h despite negative blood cultures, the infant should be reexamined at least once every 24 h to determine whether it is appropriate to stop the antibiotic treatment (92).

Persistent cardiorespiratory instability is common among preterm infants with VLBW and this symptom is not by itself an indication for prolonged empirical antibiotic administration (39). Continuing empirical antibiotic administration in response to laboratory test abnormalities is rarely justified, particularly among preterm infants at a lower risk for EOS and those born with maternal obstetric conditions known to affect fetal hematopoiesis (39).

There have been concerns regarding the incomplete detection of low-level bacteremia and the effects of intrapartum antibiotic administration; modern blood culture systems can reliably detect low levels of bacteremia provided that 1 mL of blood is inoculated, and studies have shown no impact of IAP on time to detection of EOS (51, 93, 94).

When EOS is confirmed by blood culture, a lumbar puncture should be performed if not done previously (39). Antibiotic choice should be pathogen-specific with a narrow effective spectrum of activity and should be guided by sensitivity results. There are a few studies that have guided the duration of therapy. The duration of antimicrobial therapy for uncomplicated bacteremia without a focus is usually 7–10 days with a pathogen-specific antibiotic (95). Seven days may be considered for some low-risk cases of neonates over 32 weeks of gestation and weighing over 1500 g (96–98). When uncomplicated meningitis attributable to GBS is diagnosed, therapy is extended to 14 days (95). Extended courses of treatment are indicated for complications, including cerebritis and osteomyelitis. Antimicrobial therapy is usually recommended to be administered for a minimum of 3 weeks for gram-negative meningitis, provided sterility of the CSF is ensured 24-48 hours after the appropriate treatment commences (99–101).

IAP for at-risk women has become the standard of care in many countries over the past two decades and is an effective way to prevent EOS (102–104). A meta-analysis revealed that compared with a risk-based strategy, screening-based prophylaxis was associated with a reduced risk of GBS-EOS (105). The Society of Obstetricians and Gynecologists of Canada (SOGC), in 2018, reaffirmed the recommendation of offering all women screening for colonization with GBS at 35–37 weeks of gestation, by performing culture tests with one swab from the vagina and one from the rectum (106).

The two major risk factors that predispose neonates in the NICU to sepsis are preterm birth and low birth weight (111). In the Canadian Neonatal Network, a 5% incidence of LOS has been reported among neonates with a BW of 1000–1499 g, who survived for more than two calendar days, but the incidence increased to 30–36% in those with a BW of 750–999 g (47, 48). Other known risk factors for LOS include prolonged mechanical ventilation, use of intravascular devices, failure of early breast milk feeding, and prolonged hospitalization (111, 113, 114). Central line–associated bloodstream infections (CLABSIs) are the most common type of LOS among infants in the NICU, resulting from extraluminal or intraluminal catheter contamination after colonization of the neonatal skin, oropharynx, intestinal tract, and/or the catheter surface (115).

LOS can result from viral, bacterial, or fungal microorganisms (111). In infants admitted to NICUs, coagulase-negative Staphylococcus (CoNS) is the most common pathogen in LOS (116). Other common bacteria include Staphylococcus aureus, E. coli, Klebsiella spp, Enterococcus spp, and GBS (116). Neonates are also susceptible to invasive infections from organisms found in the hospital environment, such as Enterobacter spp, Pseudomonas spp, and Serratia spp. Fungi, specifically yeasts as Candida spp., can also cause systemic infections; however, its incidences varied across different NICUs (117).

The gold standard for the diagnosis of LOS is a positive blood culture. However, limitations to positive blood culture include the delay between collection and positivity, false positive results from contamination, and false negatives resulting from antibiotic therapy prior to collection or submission of inadequate blood volume (116). A minimum of 1 mL of blood should preferably be obtained from two different venipuncture sites (83, 118). This minimum volume is recommended due to the potential for paucibacillary infection (1–10 colony-forming units (CFU) per milliliter), with a reported sensitivity of 98% for volumes >1 mL (119). In adults, if an intravascular catheter is in place, peripheral and central blood cultures can improve the diagnosis of central line-associated bloodstream infection based on differences in time to positivity of culture from the two sites of >2 h and may also aid in the identification of sample contamination (120). This concept has not been studied in the neonatal population, likely because of the relatively small-sized lumen of their central venous catheters.

Neonatal meningitis frequently occurs in the absence of bacteremia in patients with LOS (69). Conducting LP is important in the diagnosis of meningitis, although its interpretation can be affected by the high frequency of traumatic LPs and the variability of cytologic and biochemical indices in the neonatal population (121). Repeated CSF fluid parameters may provide additional insights into the treatment response and adverse effects from the outcomes of bacterial meningitis (122). Bacterial nucleic acid detection in CSF (such as 16s ribosomal RNA) can be useful if CSF culture results are negative because of preceding antibiotic therapy. Urine examination is also an important component of LOS evaluation, as UTI is not uncommon, particularly among preterm infants [Please refer to Section 6 on “urinary tract infection” for diagnostics, interpretation, and management].

Due to non-specific symptomatology, biomarkers are used in addition to microbiologic cultures to guide the initiation and duration of antibiotic therapy (123, 124). There are numerous sepsis biomarkers like chemokines, cytokines, leucocyte cell surface antigens, and acute phase proteins described, however, none of these are considered standard of care at the bedside (125). Procalcitonin and CRP are the most extensively studied biomarkers and appear to have the greatest diagnostic utility at 12–24 h (mid-phase) and >24 h (late phase) from initial clinical suspicion of LOS (123, 124). Single values of CRP or procalcitonin are neither sensitive nor specific to guide care decisions in the diagnosis of sepsis. Two CRP levels (<10 mg/L obtained at 24 hours apart and at 8–48 h after onset of possible LOS) indicated that bacterial infection is unlikely (60).

Empiric antimicrobial therapy should be guided by local epidemiology (e.g., methicillin-resistant Staphylococcus aureus (MRSA) incidence) and antibiogram data, including coverage for common gram-negative and gram-positive pathogens (111). One must also consider the pharmacokinetic and pharmacodynamic properties of antibiotics in neonates (126). Cloxacillin, a narrow-spectrum antibiotic, is recommended to treat gram-positive LOS, especially if S. aureus infection is suspected. Vancomycin—a broad-spectrum antibiotic—is now discouraged as empiric LOS therapy, if the prevalence of MRSA in the community remains low, as it then becomes difficult to determine whether blood cultures positive for CoNS are contaminated (127). No survival benefit with empirical vancomycin therapy has been demonstrated for CoNS bloodstream infections in infants from a retrospective review including 4,364 infants from 348 NICUs, even though the median duration of bacteremia was 1 d longer in infants who received delayed vancomycin therapy (128).

Unless meningitis is suspected, gram-negative coverage with an aminoglycoside is preferred over third-generation cephalosporins because of the potential for the latter to induce antimicrobial resistance. In most jurisdictions, aminoglycosides are more likely to cover nosocomial gram-negative bacteremia (e.g., Enterobacter spp, Serratia spp, Pseudomonas spp) (129).

Currently, there are no evidence-based guidelines on the duration of antimicrobial treatment for neonatal bacteremia and meningitis. The duration is determined by each patient's risk factors, clinical evaluation, laboratory tests, specific pathogens that are identified, and their susceptibility to antimicrobials (111, 130).

For bacteremia, in general, the recommended duration of antibiotics ranges from 7 to 14 days, depending on the pathogen isolated. Using CRP and blood cultures as a measure, it was found that a 10-day antibiotic duration was as effective as 14 days in neonatal bacteremia (97). In another study, 69 neonates with septicemia were randomized to receive a 7- or 14-day course of antibiotics (98). Although the 7-day antibiotic treatment appeared to be sufficient for other pathogens, neonates with S. aureus bacteremia experienced more treatment failure with 7 days of treatment compared to that with 14 days of treatment, which thus remains the recommended duration (98). Of note, retention of the central venous catheter in neonates with bloodstream infection is associated with delayed resolution of bacteremia and a higher incidence of recurrence (131).

The treatment duration for meningitis has traditionally ranged from 14 to 21 days, depending on the pathogen isolated. For example, 14 days of treatment is recommended for uncomplicated GBS meningitis, whereas a minimum of 21 days is recommended for uncomplicated meningitis caused by gram-negative bacteria (132, 133). However, it is important to note that the treatment duration should ultimately be individualized to the pathogen and the patient's clinical response.

Culture-negative sepsis implies that a neonate is clinically suspected to have sepsis, often due to nonspecific symptoms; therefore, a complete course of antibiotics is prescribed despite sterile cultures. As aforementioned, sterile cultures may be falsely negative owing to low-level bacteremia, antibiotic pre-treatment, or bacterial infection at a site that is not readily cultured (such as the lung) (134). As per recent reports, up to ten-fold higher antibiotic utilization for culture-negative sepsis is observed than that for culture-proven sepsis, (127, 135) however, this varies widely between institutions. Empirical therapy should be limited to 24–36 h if the infant shows no subsequent clinical or laboratory evidence of infection and cultures are negative (126).

Prolonged antibiotic courses (>5 days) in culture-negative settings have been associated with increased odds of necrotizing enterocolitis (NEC) and death (4% increased risk with each additional day) (136). Unnecessary antimicrobial use is increasingly recognized to pose significant harm to NICU patients, including increased risk of invasive candidiasis, prolonged hospitalization, bacterial resistance, (137) and predisposition to inflammatory dysregulation with long-term adverse health outcomes (138).

Historically, the CDC definition of VAP for infants up to one year of age was the most widely used definition for neonates, but this definition was criticized for its diagnostic inconsistency, especially for preterm neonates (148). Initiatives to refine the CDC definition of VAP have led to the introduction of the term pediatric ventilator-associated condition (VAC) by the National Health and Safety Network (NHSN). Pediatric VAC has an objective definition for respiratory deterioration defined as increase in mean airway pressure ≥4 mmHg and/or FiO₂ ≥0.25 for a specified duration of ≥2 calendar days (149). As pediatric VAC is not specific for infectious ventilator events, additional clinical and management criteria are required to fulfill the conditions for infection-related ventilator-associated complication (IVAC), and additional microbiological criteria are required to fulfill the criteria for possible VAP (150). Chest radiograph findings have been removed from the NHSN criteria for VAC, IVAC, and possible VAP for all populations due to concerns regarding reliability, which is particularly relevant in that the interpretation of radiological changes is especially challenging in neonates with pre-existing lung pathologies (148, 151). Quality improvement initiatives that target VAP, using internally consistent definitions, have been successful in improving practices (148–157) and are, therefore, encouraged (148).

Purulence defined as ≥25 polymorphonuclear leukocytes per low power field on tracheal aspirates (TA) and bronchioalveolar lavage (BAL) microscopy has been correlated with the duration of mechanical ventilation and the incidence of bronchopulmonary dysplasia, but not with clinical symptoms (146, 147, 158–161). Similarly, time since intubation is the strongest predictor of positive TA quantitative cultures, with approximately half of the bacterial growth occurring after 4–10 days of mechanical ventilation and almost 100% after 10 days (148, 161, 162). Most positive TA cultures represent endotracheal tube colonization and oropharyngeal contamination rather than infection (147, 148, 160) as supported by the fact that up to 58% of TA cultures are polymicrobial (139). Thus, a positive microbiological result is only one of several criteria for VAP. Certain groups have reported that less invasive BAL (“mini-BAL”) is safe in preterm infants and may decrease the false positive rates in VAP diagnosis (140, 158, 160).

The choice of empirical therapy should be made after considering the severity of illness, prior colonization, local antibiogram data, and risk factors for multidrug-resistant (MDR) pathogens (163). The most common pathogens for VAP are S. aureus and Enterobacteriaceae, and these organisms should be the target of empirical antimicrobials, whereas Pseudomonas aeruginosa should be targeted only in NICUs with a high prevalence of this pathogen. Narrow-spectrum therapy (e.g., cefazolin or a combination of cloxacillin plus either an aminoglycoside or cefotaxime) should be favored in non-critically ill neonates with no prior history of colonization with resistant bacteria, and in NICUs with low rates (<10%) of multidrug-resistant pathogens (164, 165).

After initiation of empirical therapy for possible VAP, antimicrobial agents should be reassessed 24–36 h after their initiation (148, 166). If an alternative diagnosis is established that explains the patient's clinical or radiological deterioration, or if TA cultures are negative, antimicrobial therapy for VAP should be discontinued. In neonates with probable VAP, the antimicrobial spectrum should be narrowed based on the culture results (164, 167). Combinations of antibiotics are not recommended as this does not hasten clinical recovery.

The optimal treatment duration for VAP in neonates is unknown (168). However, some key principles, such as time to clinical improvement, microbial burden/source control, and host immune status should be considered (169). A meta-analysis of four randomized controlled trials in immunocompetent adults suggests that a 7-to 8-day antibiotic treatment, irrespective of the causative pathogen, is as effective as a 15-day course in terms of clinical cure, mortality, and relapse risks, and is associated with a lower risk of subsequent MDR infections (164, 170). Treatment duration should be longer for those with complicated VAP (e.g., pneumatocele, lung abscess, or empyema), VAP with secondary bacteremia, and for those with underlying comorbidities impacting immunological response.

Multiple risk factors are involved in the pathogenesis of NEC, including intestinal inflammation, ischemia, and bacterial overgrowth in a highly vulnerable gastrointestinal tract (180). Bacterial overgrowth may progress to bacterial translocation across the intestinal wall and intra-abdominal infection with localized or diffused peritonitis, which can lead to bowel perforation (181, 182). Spontaneous intestinal perforation (SIP) is different from NEC. SIP generally occurs in the first week of age in very preterm infants and is not discussed in this narrative review (183).

Diagnosis and staging rely on a combination of clinical, laboratory, and radiologic findings, typically using the modified Bell's staging criteria (184). Bell stage I or suspected NEC presentation is non-specific, and symptoms include feeding intolerance, abdominal distension, and occult blood in the stool. Stage II NEC presents with ileus, dilated bowel loops, grossly bloody stools, and pneumatosis. Severe NEC or stage III is characterized by respiratory and metabolic acidosis, a fixed bowel loop, and may progress to intestinal perforation (184, 185).

Numerous pathogens are associated with NEC. For most infants, blood and peritoneal fluid cultures remain negative, but intra-abdominal polymicrobial infections are assumed to be present (186–189). Since NEC is associated with local bacterial overgrowth of endogenous intestinal flora, antibiotic empirical therapy should target enteric gram-negative aerobic and facultative anaerobic bacilli such as Enterobacteriaceae (e.g., E. coli and Klebsiella species), enteric gram-positive streptococci (e.g., Streptococcus anginosus), and in some situations, obligate anaerobic bacilli (e.g., Clostridium perfringens and Bacteroides fragilis) (190–192).

There is a wide variability in empiric antibiotic regimens used in clinical practice to treat NEC, reflecting the lack of clear guidelines on the type and duration of antibiotic treatment (175, 189, 193). The Surgical Infection Society and the Infectious Diseases Society of America (IDSA) published the only guideline addressing this issue in 2010 (190); it mainly discusses the treatment of complicated intra-abdominal infections (cIAI) in adults and older children. For NEC specifically, IDSA recommendations are based on experts' opinions and include the following combinations of broad-spectrum antibiotics: (i) ampicillin, gentamicin, and metronidazole, (ii) ampicillin, cefotaxime, and metronidazole, or (iii) meropenem (190).

To date, published studies have failed to demonstrate the optimal antimicrobial regimen for NEC in terms of efficacy and safety (190, 194). A Cochrane systematic review included two randomized controlled trials (RCT) conducted in the 1980s (194). The first RCT compared ampicillin-gentamicin to ampicillin-gentamicin-clindamycin in 42 premature infants with NEC (195); it showed no beneficial effect on mortality or intestinal perforation with the addition of clindamycin, but found a significantly longer time to successful reinstitution of enteral feeding and a higher incidence of stricture formation in the clindamycin group (195). The other RCT included 20 infants and showed no difference in outcomes with enteral vs. parenteral gentamicin when administered in combination with parenteral ampicillin and gentamicin (196).

More recently, a large multicenter, partly randomized controlled trial was completed with 128 infants selected randomly, and an additional 52 infants were enrolled on a non-randomized basis (197). All infants <121 days of age (≤33 weeks GA) had a cIAI and received one of the following three antibiotic combinations: ampicillin/gentamicin/metronidazole; ampicillin/ gentamicin/ clindamycin; or piperacillin-tazobactam/ gentamicin within 48 h of diagnosis. Mortality assessed at 30 days was similar in all three arms at 8–9%. Adverse events up to 90 days follow-up, events of special interest (including gastrointestinal surgeries, progression to a higher stage of NEC, intestinal stricture or perforation, feeding intolerance), and therapeutic success on day 30 were also similar among the three groups. This large RCT does not support the choice of one regimen over another for the treatment of infants with cIAI.

Finally, several observational retrospective studies have compared antibiotic regimens for NEC management. Most were single-center studies with a limited sample size (189, 193, 198, 199). The largest of all observational retrospective studies involved 2,780 infants with <1500 g BW and diagnosed with NEC from 348 centers (175). After propensity score matching, the results suggest that for infants with NEC that were managed without surgery (stage II), anaerobic coverage was not associated with lower mortality, but resulted in an increased risk of intestinal strictures. However, in infants with NEC that was managed surgically (stage III), anaerobic antimicrobial therapy was associated with lower mortality. These results suggest that anaerobic coverage is beneficial in cases of severe and perforated NEC (stage III).

In summary, there is insufficient evidence to support one antibiotic regimen over another for the treatment of NEC. Empirical therapy should be effective against most pathogenic bacteria usually present in the intestinal flora (e.g., Enterobacteriaceae and gram-positive cocci). Clinicians should also consider local resistance rates, individual patients' microbiologic results, and clinical evolution to guide their antibiotic choice. Given that broad-spectrum antibiotics (including 3rd generation cephalosporins and carbapenems) increase the risk of adverse outcomes (IC, increased risk of colonization with antimicrobial-resistant organisms, and microbiota alterations), narrow-spectrum regimens and short therapy durations are encouraged.

There is no evidence supporting the optimal duration of antibiotic therapy for NEC (189). In stage I and stage IIA NEC (modified Bell's stage) (184), experts are generally in favor of discontinuing antibiotics after 3 days and after 5–7 days, respectively, if symptoms are resolved. In higher-stage NEC (IIB and III) or when the infant presents with bacteremia or sepsis, the duration of the antibiotics is generally extended to 10–14 days. If an intra-abdominal abscess has been identified, antibiotics should be continued until clinical and radiological responses are established. The recurrence rate of NEC after the first course of antimicrobial or surgical treatment does not appear to be reduced by a longer duration of treatment (200).

Finally, NEC is a risk factor for invasive candidiasis (IC), and fluconazole prophylaxis may be beneficial for high-risk infants. A study in very low BW infants, <6 weeks postnatal age, receiving broad-spectrum antibiotics for >2 days with 1 additional risk factor including NEC, showed that fluconazole prophylaxis reduced the incidence of IC, but failed to demonstrate a survival benefit (201).

The most common pathogen found in neonatal UTIs is E. coli, followed by Klebsiella pneumoniae and Enterobacter spp (204–206, 210, 211). Less common pathogens include: Pseudomonas aeruginosa, Enterococcus spp, Group B Streptococcus, S. aureus, viridans group Streptococci, Citrobacter freundii, Serratia marcescens, Klebsiella oxytoca, and Proteus vulgaris (205, 206, 210). Neonatal UTIs caused by fungal pathogens, specifically Candida species, could be associated with higher mortality (204, 208, 212, 213). Patients with vesicoureteral reflux may present with less common pathogens and are predisposed to pyelonephritis (204, 205, 211).

Suprapubic aspiration is the recommended method, but it is an invasive procedure that requires operators with skills in point-of-care ultrasound (POCUS) (206, 214, 215). Urinary catheterization has also been shown to be feasible in preterm infants, even with 24 weeks gestation, although it can also be technically difficult (214). Proper disinfection of the penis or perineum is essential to minimize the possibility of sample contamination. Sterile bag collection is not recommended because of its high contamination rate (204–206).

The thresholds for laboratory values diagnostic for UTIs vary across literature. The current gold standard for UTI diagnosis is a urine sample obtained by suprapubic aspiration or urinary catheterization that is positive for a single pathogenic organism (204–206). The most accepted threshold of pathogen growth is a concentration of ≥5x10⁷CFU/L (211). However, some studies have included a threshold of ≥10⁷CFU/L, especially in the presence of pyuria, to define a positive culture (204–206).

Elevated CRP and erythrocyte sedimentation rates were reported to have low sensitivity and specificity among neonates with UTI (204–206).

The initial treatment for neonatal UTI involves empirical parenteral broad-spectrum antibiotics (204–206). Most commonly, a combination of ampicillin and gentamicin or ampicillin and cefotaxime is administered to cover Enterococcus spp and gram-negative bacilli (204–206). Once a pathogenic organism is identified, treatment is tailored based on antimicrobial sensitivities (204–206). There is currently a lack of evidence on the optimal treatment duration. Traditionally, neonates were treated intravenously for 5–7 days (204–206). However, there is increasing evidence that term infants can complete therapy with oral antibiotics (216). The use of antibiotic prophylaxis for neonates with recurrent UTIs is not supported by evidence (205, 206).

The risk of detecting bacteremia in an infant with a UTI is moderate, ranging from 3–17%, and is inversely related to age (204, 205, 211, 217). Renal ultrasonography is recommended for all neonates after the first episode of an UTI because of the high incidence of associated congenital anomalies in this population (204–206, 211, 218, 219). Early diagnosis of urological abnormalities is critical for managing recurrent infections and preventing renal scarring (205, 212).

Clean procedures—an incision in which no inflammation is encountered in a surgical procedure, without a break in sterile technique, and during which the respiratory, alimentary, and genitourinary tracts are not entered—include the insertion of a central venous line (CVL), patent ductus arteriosus (PDA) ligation, immediate repair of omphalocele, gastroschisis reduction, repair of congenital diaphragmatic hernia (CDH), inguinal hernia repair, and circumcision (228, 229). Data on the need for prophylactic antibiotics for peripherally-inserted central catheter placement is mixed (230–232). While some RCTs demonstrated reduced catheter-related bloodstream infection (230) and CoNS catheter-related sepsis, (231) others failed to show benefit (232). The rates of SSI with CVL can be as low as 0 per 100 procedures; therefore, the risks of antimicrobial exposure are unlikely to outweigh the benefits (222, 230).

Although omphalocele is considered a clean procedure, short-term preoperative antibiotics are recommended for primary closure of omphalocele, or antibiotics are continued until closure in staged repairs (233, 234).

Despite its classification as a clean procedure, wound infections in gastroschisis repair are common, leading to the argument that this procedure should be re-classified based on a higher level of contamination (222). Wound infections were more common in neonates undergoing delayed closure (21%) than in those undergoing immediate closure (8.2%) (235). Empiric coverage with a beta-lactamase-resistant penicillin derivative (i.e., cloxacillin) combined with an aminoglycoside initiated at birth and discontinued after wound closure is suggested (235).

Retrospective data demonstrated that SSI is an uncommon source of infection in children with congenital diaphragmatic hernia (CDH), occurring in less than 1% of 1,085 neonates, but prophylactic antibiotics are typically prescribed (236, 237).

There are no specific neonatal guidelines on SSI prophylaxis for PDA ligation, inguinal hernia, or circumcision, but apart from the above exceptions, prophylactic antibiotics are not recommended for clean procedures based on the expert opinion (228).

This class describes an operative wound in which the respiratory, alimentary, genital, or urinary tract is entered under controlled conditions and without unusual contamination. Clean-contaminated surgical procedures include esophageal atresia (EA), tracheoesophageal fistula (TEF) repair, duodenal atresia repair, gastrostomy G-tube insertion, choledochal cyst excision, and colorectal surgery.

A small retrospective review of 732 neonatal surgeries found no statistical difference in SSI among infants receiving greater than 24 h of postoperative antibiotics and those receiving less than 24 h for clean contaminated or contaminated gastrointestinal operations. Guidelines recommend a single dose of preoperative intravenous (IV) cefazolin prior to most clean-contaminated procedures (228).

A consensus on the role of antimicrobial prophylaxis for EA and TEF repair is lacking (228). A study on 48 patients with EA, isolated either normal oropharyngeal organisms or no organisms from the upper esophageal pouch of infants who underwent repair within the first 24 h despite half of them being on prophylactic antibiotics (238). All infants who underwent delayed closure demonstrated growth of micro-organisms, irrespective of whether they were on antibiotics, and only those on prophylactic antibiotics demonstrated growth of Pseudomonas and Serratia (238). Antibiotic prophylaxis for EA/TEF repair is often recommended with amoxicillin-clavulanate, or 24–48 h of cefazolin and metronidazole, piperacillin/tazobactam, or cefoxitin (228) in the absence of evidence of aspiration (238–240).

Despite a lack of pediatric data, the Surgical Care Improvement Project (SCIP) recommends SSI prophylaxis in children undergoing colorectal surgery (241, 242). This recommendation is based on a Cochrane review and meta-analysis of adult studies demonstrating significantly improved rates of SSI when prophylactic antibiotics were used compared to when no treatment was provided [risk ratio (RR): 0.34] (241, 242). There was a further 50% reduction in SSI when anaerobic coverage was added to agents targeting aerobic bacteria (RR: 0.55), and a 60% reduction when aerobic coverage was added to agents targeting anaerobes (RR: 0.41) (241, 242). SCIP guidelines recommend parenteral cefazolin + metronidazole within 1 h of incision and their discontinuation within 24 h of surgery completion (241).

There are no common class 3 procedures in the NICU population.

Antimicrobial recommendations for NEC are covered in Section 5 of this narrative review.