Neonatal sepsis occurs in 0.5 to 8.0/1000 births. The highest rates occur in
Onset of neonatal sepsis can be early (≤ 3 days of birth) or late (after 3 days).
Early-onset neonatal sepsis usually results from organisms acquired intrapartum. Most infants have symptoms within 6 hours of birth.
Most cases are caused by group B streptococcus (GBS) and gram-negative enteric organisms (predominantly Escherichia coli). Vaginal or rectal cultures of women at term may show GBS colonization rates of up to 35%. At least 35% of their infants also become colonized. The density of infant colonization determines the risk of early-onset invasive disease, which is 40 times higher with heavy colonization. Although only 1/100 of infants colonized develop invasive disease due to GBS, > 50% of those present within the first 6 hours of life. Nontypeable Haemophilus influenzae sepsis has also been identified in neonates, especially premature neonates.
Other cases tend to be caused by gram-negative enteric bacilli (eg, Klebsiella species) and certain gram-positive organisms (Listeria monocytogenes, enterococci [eg, Enterococcus faecalis, E. faecium], group D streptococci [eg, Streptococcus bovis], alpha-hemolytic streptococci, and staphylococci). Also, S. pneumoniae, H. influenzae type b, and, less commonly, Neisseria meningitidis have been isolated. Asymptomatic gonorrhea occurs occasionally in pregnancy, so N. gonorrhoeae may rarely be a pathogen.
Late-onset neonatal sepsis is usually acquired from the environment (see Neonatal Hospital-Acquired Infection). Staphylococci account for 30 to 60% of late-onset cases and are most frequently due to intravascular devices (particularly central vascular catheters). E. coli is also becoming increasingly recognized as a significant cause of late-onset sepsis, especially in extremely LBW infants. Isolation of Enterobacter cloacae or Cronobacter sakazakii (formerly Enterobacter sakazakii) from blood or cerebrospinal fluid may be due to contaminated feedings. Contaminated respiratory equipment is suspected in outbreaks of hospital-acquired Pseudomonas aeruginosa pneumonia or sepsis.
Although universal screening and intrapartum antibiotic prophylaxis for group B streptococcus have significantly decreased the rate of early-onset disease due to this organism, the rate of late-onset GBS sepsis has remained unchanged, which is consistent with the hypothesis that late-onset disease is usually acquired from the environment.
The role of anaerobes (particularly Bacteroides fragilis) in late-onset sepsis remains unclear, although deaths have been attributed to Bacteroides bacteremia.
Candida species are increasingly important causes of late-onset sepsis, occurring in 12 to 18% of extremely LBW infants.
Certain maternal perinatal and obstetric factors increase risk, particularly of early-onset neonatal sepsis, such as the following:
Hematogenous and transplacental dissemination of maternal infection occurs in the transmission of certain viral (eg, rubella, cytomegalovirus), protozoal (eg, Toxoplasma gondii), and treponemal (eg, Treponema pallidum) pathogens. A few bacterial pathogens (eg, L. monocytogenes, Mycobacterium tuberculosis) may reach the fetus transplacentally, but most are acquired by the ascending route in utero or as the fetus passes through the colonized birth canal.
Though the intensity of maternal colonization is directly related to risk of invasive disease in the neonate, many mothers with low-density colonization give birth to infants with high-density colonization who are therefore at risk. Amniotic fluid contaminated with meconium or vernix caseosa promotes growth of group B streptococcus and E. coli. Hence, the few organisms in the vaginal vault are able to proliferate rapidly after PROM, possibly contributing to this paradox. Organisms usually reach the bloodstream by fetal aspiration or swallowing of contaminated amniotic fluid, leading to bacteremia.
The ascending route of infection helps to explain such phenomena as the high incidence of PROM in neonatal infections, the significance of adnexal inflammation (amnionitis is more commonly associated with neonatal sepsis than is central placentitis), the increased risk of infection in the twin closer to the birth canal, and the bacteriologic characteristics of early-onset neonatal sepsis, which reflect the flora of the maternal vaginal vault.
The most important risk factor in late-onset sepsis is
Other risk factors include
Gram-positive organisms (eg, coagulase-negative staphylococci and Staphylococcus aureus) may be introduced from the environment or the patient’s skin. Gram-negative enteric bacteria are usually derived from the patient’s endogenous flora, which may have been altered by antecedent antibiotic therapy or populated by resistant organisms transferred from the hands of personnel (the major means of spread) or contaminated equipment. Therefore, situations that increase exposure to these bacteria (eg, crowding, inadequate nurse staffing, inconsistent provider handwashing) result in higher rates of hospital-acquired infection.
Risk factors for candidal sepsis include prolonged (> 10 days) use of central IV catheters, hyperalimentation, use of antecedent antibiotics (especially 3rd-generation cephalosporins), and abdominal pathology.
Initial foci of infection can be in the urinary tract, paranasal sinuses, middle ear, lungs, or gastrointestinal tract and may later disseminate to meninges, kidneys, bones, joints, peritoneum, and skin.
Early signs of neonatal sepsis are frequently nonspecific and subtle and do not distinguish among organisms (including viral). Particularly common early signs include
Fever is present in only 10 to 15% of neonates but, when sustained (eg, > 1 hour), generally indicates infection. Other symptoms and signs include respiratory distress, neurologic findings (eg, seizures, jitteriness), jaundice (especially occurring within the first 24 hours of life without Rh or ABO blood group incompatibility and with a higher than expected direct bilirubin concentration), vomiting, diarrhea, and abdominal distention.
Specific signs of an infected organ may pinpoint the primary site or a metastatic site.
Most neonates with early-onset group B streptococcus (and many with L. monocytogenes) infection present with respiratory distress that is difficult to distinguish from respiratory distress syndrome.
Periumbilical erythema, discharge, or bleeding without a hemorrhagic diathesis suggests omphalitis (infection prevents obliteration of the umbilical vessels).
Coma, seizures, opisthotonos, or a bulging fontanelle suggests meningitis, encephalitis, or brain abscess.
Decreased spontaneous movement of an extremity and swelling, warmth, erythema, or tenderness over a joint indicates osteomyelitis or pyogenic arthritis.
Unexplained abdominal distention may indicate peritonitis or necrotizing enterocolitis (particularly when accompanied by bloody diarrhea and fecal leukocytes).
Cutaneous vesicles, mouth ulcers, and hepatosplenomegaly (particularly with disseminated intravascular coagulation [DIC]) can indicate disseminated herpes simplex.
Early-onset group B streptococcus (GBS) infection may manifest as a fulminating pneumonia. Often, obstetric complications (particularly prematurity, PROM, or chorioamnionitis) have occurred. In > 50% of neonates, GBS infection manifests within 6 hours of birth; 45% have an Apgar score of < 5. Meningitis may also be present but is not common. In late-onset GBS infection (at > 3 days to 12 weeks), meningitis is often present. Late-onset GBS infection is generally not associated with perinatal risk factors or demonstrable maternal cervical colonization and may be acquired postpartum.
Early diagnosis of neonatal sepsis is important and requires awareness of risk factors (particularly in LBW neonates) and a high index of suspicion when any neonate deviates from the norm in the first few weeks of life.
Neonates with clinical signs of sepsis should have a complete blood count (CBC), differential with smear, blood culture, urine culture (not necessary for evaluation of early-onset sepsis), and lumbar puncture (LP), if clinically feasible, as soon as possible. Neonates with respiratory symptoms require chest x-ray. Diagnosis is confirmed by isolation of a pathogen in culture. Other tests may have abnormal results but are not necessarily diagnostic. Infants should be given broad-spectrum empiric antimicrobial therapy.
Neonates who appear well are managed depending on several factors as discussed below under Prevention.
The total white blood cell count and absolute band count in neonates are poor predictors of early-onset sepsis. However, an elevated ratio of immature:total polymorphonuclear leukocytes of > 0.16 is sensitive, and values below this cutoff have a high negative predictive value. However, specificity is poor; up to 50% of term neonates have an elevated ratio. Values obtained after 6 hours of life are more likely to be abnormal and clinically useful than those obtained immediately after birth.
The platelet count may fall hours to days before the onset of clinical sepsis but more often remains elevated until a day or so after the neonate becomes ill. This fall is sometimes accompanied by other findings of DIC (eg, increased fibrin degradation products, decreased fibrinogen, prolonged international normalized ratio [INR]). Given the timing of these changes, the platelet count is not typically helpful in evaluating a neonate for sepsis.
Because of large numbers of circulating bacteria, organisms can sometimes be seen in or associated with polymorphonuclear leukocytes by applying Gram stain, methylene blue, or acridine orange to the buffy coat.
Regardless of the results of the CBC or LP, in all neonates with suspected sepsis (eg, those who look sick or are febrile or hypothermic), antibiotics should be started immediately after cultures (eg, blood and CSF [if possible]) are taken.
There is a risk of increasing hypoxia during an LP in already hypoxemic neonates. However, LP should be done in neonates with suspected sepsis as soon as they are able to tolerate the procedure (see also Diagnosis under Neonatal Bacterial Meningitis). Supplemental oxygen is given before and during LP to prevent hypoxia. Because GBS pneumonia manifesting in the first day of life can be confused with respiratory distress syndrome, LP is often done routinely in neonates suspected of having these diseases.
Umbilical vessels are frequently contaminated by organisms on the umbilical stump, especially after a number of hours, so blood cultures from umbilical venous lines may not be reliable. Therefore, blood for culture should be obtained by venipuncture, preferably at 2 peripheral sites. Although the optimal skin preparation to do before obtaining blood cultures in neonates is not defined, clinicians can apply an iodine-containing liquid and allow the site to dry. Alternatively, blood obtained soon after placement of an umbilical arterial catheter may also be used for culture if necessary.
Blood should be cultured for both aerobic and anaerobic organisms. However, the minimum amount of blood per blood culture bottle is 1.0 mL; if < 2 mL is obtained, it should all be placed in a single aerobic blood culture bottle. If catheter-associated sepsis is suspected, a culture specimen should be obtained through the catheter as well as peripherally. In > 90% of positive bacterial blood cultures, growth occurs within 48 hours of incubation. Data on capillary blood cultures are insufficient to recommend them.
Candida species grow in blood cultures and on blood agar plates, but if other fungi are suspected, a fungal culture medium should be used. For species other than Candida, fungal blood cultures may require 4 to 5 days of incubation before becoming positive and may be negative even in obviously disseminated disease. Proof of colonization (in mouth or stool or on skin) may be helpful before culture results are available. Neonates with candidemia should undergo LP to identify candidal meningitis. Indirect ophthalmoscopy with dilation of the pupils is done to identify retinal candidal lesions. Renal ultrasonography is done to detect renal mycetoma.
Urine testing is needed only for evaluation of late-onset sepsis. Urine should be obtained by catheterization or suprapubic aspiration, not by urine collection bags. Although only culture is diagnostic, a finding of ≥ 5 white blood cells/high-power field in the spun urine or any organisms in a fresh unspun gram-stained sample is presumptive evidence of a urinary tract infection (UTI). Absence of pyuria does not rule out UTI.
Numerous tests are often abnormal in sepsis and have been evaluated as possible early markers. In general, however, sensitivities tend to be low until later in illness, and specificities are suboptimal.
Acute-phase reactants are proteins produced by the liver under the influence of IL-1 when inflammation is present. The most valuable of these is quantitative C-reactive protein. A concentration of ≥ 1 mg/dL (9.52 nmol/L) (measured by nephelometry) is abnormal. Elevated levels occur within 6 to 8 hours of developing sepsis and peak at 1 day. The sensitivity of C-reactive protein measurements is higher if measured after 6 to 8 hours of life. Two normal values obtained between 8 hours and 24 hours after birth and then 24 hours later have a negative predictive value of 99.7%.
Procalcitonin is being investigated as an acute-phase reactant marker for neonatal sepsis. Although procalcitonin appears more sensitive than C-reactive protein, it is less specific (1).
1. Pontrelli G, De Crescenzo F, Buzzetti R, et al: Accuracy of serum procalcitonin for the diagnosis of sepsis in neonates and children with systemic inflammatory syndrome: A meta-analysis. BMC Infect Dis 17(1):302, 2017. doi: 10.1186/s12879-017-2396-7
The fatality rate is 2 to 4 times higher in LBW infants than in full-term infants. The overall mortality rate of early-onset sepsis is 3 to 40% (that of early-onset GBS infection is 2 to 10%) and of late-onset sepsis is 2 to 20% (that of late-onset GBS is about 2%). Mortality in late-onset sepsis highly depends on the etiology of the infection; infections caused by gram-negative bacilli or Candida species have rates of up to 32 to 36%. In addition to mortality, extremely LBW infants who develop bacterial or candidal sepsis have a significantly greater risk of poor neurodevelopmental outcome.
Because sepsis may manifest with nonspecific clinical signs and its effects may be devastating, rapid empiric antibiotic therapy is recommended (see Overview of Antibacterial Drugs : Selection and Use of Antibiotics); drugs are later adjusted according to sensitivities and the site of infection. Generally, if no source of infection is identified clinically, the infant appears well, and cultures are negative, antibiotics can be stopped after 48 hours (up to 72 hours in small preterm infants).
General supportive measures, including respiratory and hemodynamic management, are combined with antibiotic treatment.
In early-onset sepsis, initial therapy should include ampicillin plus an aminoglycoside (see table Recommended Dosages of Select Aminoglycosides for Neonates). Cefotaxime may be added to or substituted for the aminoglycoside if meningitis caused by a gram-negative organism is suspected. Antibiotics may be changed as soon as an organism is identified.
In late-onset sepsis, previously well infants admitted from the community with presumed late-onset sepsis should also receive therapy with ampicillin plus gentamicin or ampicillin plus cefotaxime. If gram-negative meningitis is suspected, ampicillin, cefotaxime, and an aminoglycoside may be used. In late-onset hospital-acquired sepsis, initial therapy should include vancomycin (active against methicillin-resistant S. aureus; see table Vancomycin Dosage for Neonates) plus an aminoglycoside. If P. aeruginosa is prevalent in the nursery, ceftazidime, cefepime, or piperacillin/tazobactam may be used in addition to, or instead of, an aminoglycoside depending on local susceptibilities.
For neonates previously treated with a full 7- to 14-day aminoglycoside course who need retreatment, a different aminoglycoside or a 3rd-generation cephalosporin should be considered.
If coagulase-negative staphylococci are suspected (eg, an indwelling catheter has been in place for > 72 hours) or are isolated from blood or other normally sterile fluid and considered a pathogen, initial therapy for late-onset sepsis should include vancomycin. However, if the organism is sensitive to nafcillin, cefazolin or nafcillin should replace vancomycin. Removal of the presumptive source of the organism (usually an indwelling intravascular catheter) may be necessary to cure the infection because coagulase-negative staphylococci may be protected by a biofilm (a covering that encourages adherence of organisms to the catheter).
Because Candida may take 2 to 3 days to grow in blood culture, empiric initiation of amphotericin B deoxycholate therapy and removal of the infected catheter before cultures confirm yeast infection may be lifesaving.
Exchange transfusions have been used for severely ill (particularly hypotensive and metabolically acidotic) neonates. Their purported value is to increase levels of circulating immunoglobulins, decrease circulating endotoxin, increase hemoglobin levels (with higher 2,3-diphosphoglycerate levels), and improve perfusion. However, no controlled prospective studies of their use have been conducted.
Fresh frozen plasma may help reverse the heat-stable and heat-labile opsonin deficiencies that occur in LBW neonates, but controlled studies of its use are unavailable, and transfusion-associated risks must be considered.
Granulocyte transfusions (see Blood Products : White blood cells (WBCs)) have been used in septic and granulocytopenic neonates but have not convincingly improved outcome.
Recombinant colony-stimulating factors (granulocyte colony-stimulating factor [G-CSF] and granulocyte-macrophage colony-stimulating factor [GM-CSF]) have increased neutrophil number and function in neonates with presumed sepsis but do not seem to be of routine benefit in neonates with severe neutropenia; further study is required.
Neonates who appear well may be at risk of group B streptococcus infection. The Centers for Disease Control and Prevention (CDC) and the American Academy of Pediatrics (AAP) currently recommend managing these infants depending on several factors (1, 2), including
Presence of chorioamnionitis
Whether maternal group B streptococcus prophylaxis was indicated and given appropriately
Gestational age and the duration of membrane rupture
If there is neither chorioamnionitis nor indication for group B streptococcus prophylaxis, no testing or treatment is indicated.
If chorioamnionitis is present or strongly suspected, preterm and term neonates should have a blood culture at birth and begin empiric broad-spectrum antibiotic therapy. Testing should also include white blood cell count and differential and C-reactive protein at 6 to 12 hours of life. Further management depends on the clinical course and results of the laboratory tests.
If maternal group B streptococcus prophylaxis was indicated and given appropriately (ie, penicillin, ampicillin, or cefazolin given IV for ≥ 4 hours), infants should be observed in the hospital for 48 hours; testing and treatment are done only if symptoms develop. Selected patients ≥ 37 weeks gestation who have reliable caretakers and ready access to follow-up may go home after 24 hours.
If adequate group B streptococcus prophylaxis was not given, infants are observed in the hospital for 48 hours without antimicrobial therapy. If membranes ruptured ≥ 18 hours before birth or gestational age is < 37 weeks, blood culture, CBC with differential, and perhaps a C-reactive protein level is recommended at birth and/or at 6 to 12 hours of life. The clinical course and results of the laboratory evaluation guide management.
Alternative approaches to risk-stratifying neonates with respect to early-onset sepsis based on both maternal risk factors and the serial newborn examination are becoming more widely implemented but are not currently recommended by the CDC or AAP (3).
Giving IV immune globulin to augment the neonate's immune response has not been shown to help prevent or treat sepsis.
All pregnant women should be screened for GBS colonization late in gestation using both vaginal and rectal culture.
Women with a positive GBS screen should be given intrapartum antibiotic prophylaxis unless they are undergoing cesarean delivery before labor starts and before membrane rupture.
Women with a negative GBS screen should receive intrapartum antibiotics if they previously gave birth to an infant with GBS disease.
Women whose GBS status is unknown (eg, because they were not tested or results are unavailable) should receive intrapartum antibiotics if ≥ 1 of the following factors are present:
Women who had a positive GBS screen in one pregnancy have a 50% probability of having GBS colonization in a subsequent pregnancy (4).
Antibiotics typically used include penicillin, ampicillin, or cefazolin and should be given IV for ≥ 4 hours before delivery. Selection should take into account local GBS antimicrobial resistance patterns.
1. Brady MT, Polin RA: Prevention and management of infants with suspected or proven neonatal sepsis. Pediatrics 132:166-8, 2013. doi: 10.1542/peds.2013-1310
2. Polin RA and the Committee on Fetus and Newborn: Management of neonates with suspected or proven early-onset bacterial sepsis. Pediatrics 129:1006-1015, 2012. doi: 10.1542/peds.2012-0541
4. Puopolo KM, Lynfield R, Cummings JJ, et al: Management of infants at risk for group B streptococcal disease. Pediatrics 144(2):e20191881, 2019. doi: 10.1542/peds.2019-1881
Neonatal sepsis can be early onset (≤ 3 days of birth) or late onset (after 3 days).
Early-onset sepsis usually results from organisms acquired intrapartum, and symptoms appear within 6 hours of birth.
Late-onset sepsis is usually acquired from the environment and is more likely in preterm infants, particularly those with prolonged hospitalization, using IV catheters, or both.
Early signs are frequently nonspecific and subtle, and fever is present in only 10 to 15% of neonates.
Do blood and cerebrospinal fluid cultures and, for late-onset sepsis, also do urine culture.
Treat early-onset sepsis initially with ampicillin plus gentamicin (and/or cefotaxime if gram-negative meningitis is suspected), narrowed to organism-specific drugs as soon as possible.