14.1 Early- and Late-Onset Sepsis

14.1 Early- and Late-Onset Sepsis

Quick Answer: Early-onset sepsis (EOS) is infection in the first about 72 hours of life, usually vertical: group B Streptococcus (GBS), Escherichia coli, and Listeria monocytogenes. Start ampicillin plus gentamicin conceptually while cultures are pending. Late-onset sepsis (LOS) presents after 72 hours and is often healthcare-associated: coagulase-negative staphylococci (CONS), Staphylococcus aureus, gram-negative rods, and Candida. HSV and enterovirus mimic bacterial shock. Blood culture is the anchor. CRP and procalcitonin cannot replace cultures. AACN Certification Corporation does not publish an official antibiotic-duration table.

The Neonatal CCRN Test Plan for exams on and after November 12, 2025 lists sepsis (early and late onset) as a multisystem patient problem (multisystem domain 20%). OpenExamPrep independent study material covers that leaf as a timing, pathogen, culture, and marker skill. Lung-focused infection as a respiratory problem lives with pneumonia and ventilator disease. This section is the whole-infant sepsis evaluation: why a day-of-life-1 crash is not the same disease as a day-14 PICC infant, and why inflammatory labs are supporting actors rather than directors.

CCRN Neonatal practice questionsPractice questions with detailed explanations

Why the clock matters

EOS is infection acquired shortly before or during birth. Many references use ≤72 hours of life; some research papers use 7 days. Exam stems are rarely ambiguous: chorioamnionitis, prolonged rupture of membranes, and illness before the third day point to EOS. LOS is infection after that window, usually in a hospitalized preterm infant, and tracks central lines, ventilation, parenteral nutrition, and bowel injury.

The same species can appear in both windows. E. coli causes devastating EOS in very low-birth-weight infants and also causes LOS. The source and the first empiric thought still differ. EOS thinking starts with maternal genital and enteric flora and with ampicillin/gentamicin. LOS thinking starts with the unit's skin, stool, water, and line flora.

Keep late-onset GBS in a third mental bucket. GBS disease at 7–89 days is often bacteremia or meningitis in a term or late-preterm infant who may never have lived in the NICU. It is not the same epidemiologic story as CONS growing from a week-old PICC. On the exam, read the age, the devices, and the maternal history before you name the organism class.

Worked picture: a 39-week infant at 14 hours of life has respiratory distress, a glucose of 32 mg/dL, and a heart rate of 185. Maternal membranes were ruptured 22 hours and the mother had a fever in labor. That is an EOS resuscitation, not a wait-for-the-CRP story. Contrast a 25-week infant on day 16 with a PICC for lipid-containing TPN who develops new apnea, a platelet drop, and a glucose swinging from 40 to 240 mg/dL. That is LOS until proven otherwise.

Well-appearing term infants with limited intrapartum risk are sometimes managed with enhanced observation or a multivariate risk estimate rather than automatic antibiotics. That is a well-baby pathway. An ill, poorly perfused, or preterm infant is not on that pathway. If the stem shows shock, apnea, or respiratory failure, you treat.

EOS pathogens and maternal risk

Risk factors that raise pre-test probability:

• GBS bacteriuria or colonization this pregnancy, or a previous infant with invasive GBS disease
• Inadequate IAP when prophylaxis was indicated (wrong drug, too close to delivery, or not given)
• Prolonged rupture of membranes, commonly discussed at ≥18 hours
• Maternal fever or diagnosed intra-amniotic infection / chorioamnionitis
• Prematurity and very low birth weight
• Fetal tachycardia and foul amniotic fluid as supporting—not standalone—clues

Exam trap: a negative GBS screen does not exclude EOS. The screen can miss colonization, and E. coli and Listeria never appeared on that swab. Another trap: treating a maternal GBS-positive, well, fully prophylaxed term infant as if that history alone mandated a week of NICU antibiotics without looking at the infant.

Clinical signs are nonspecific. Thermal instability—hypothermia is often more ominous than fever in a preterm infant—apnea, tachypnea, increased work of breathing, tachycardia, delayed capillary refill, mottling, glucose instability, feeding intolerance, lethargy, jaundice that arrives too early, and a rising lactate all belong on the list. Neutropenia or a high immature-to-total neutrophil ratio can support the story. AACN Certification Corporation does not publish a single official CBC cutoff table to recite as exam law. Treat the infant, not an isolated band count.

EOS meningitis can be present even when the fontanelle is soft. If the infant is stable enough, CSF belongs in the evaluation of ill EOS. If the infant is too unstable for a lumbar puncture, treat and tap later. Do not withhold antibiotics for a perfect CSF kit.

Empiric EOS therapy without inventing an AACN clock

Conceptual first-line coverage is ampicillin plus gentamicin. Ampicillin covers GBS and Listeria (and many enterococci). Gentamicin adds gram-negative coverage and synergistic killing of GBS and Listeria. Critically ill VLBW infants in units with resistant gram-negatives may need a broader partner—that is local epidemiology, not a hidden handbook sentence.

Ceftriaxone is not a casual neonatal choice: it displaces bilirubin from albumin and can precipitate with calcium-containing fluids. Ampicillin and gentamicin do not treat HSV. If the differential includes herpes, acyclovir is a separate decision, not a substitute for the bacterial pair.

Duration is not an AACN-official number. The test plan names sepsis; it does not issue a mandatory stop-hour or a mandatory meningitis-day count. You decide duration with culture results (including CSF when meningitis is possible), whether shock or pneumonia is present, how the infant looks at reassessment, and infectious-disease plus unit protocol.

Bedside teaching you will hear: when blood (and CSF, if indicated) cultures are negative and the infant is well, many teams reassess at about 36–48 hours and stop if the probability of infection has collapsed. Proven bacteremia or meningitis is longer and organism-specific. If an option claims AACN requires 7 days for every EOS rule-out, reject it. If an option claims you may skip cultures because the CRP is low at birth, reject that too. If an option claims gentamicin duration is a secret handbook constant, that is the same error in a laboratory coat.

LOS: NICU organisms

Risk cluster: central venous catheters, parenteral nutrition and intravenous lipid, prolonged mechanical ventilation, recent surgery or necrotizing enterocolitis, prior broad-spectrum antibiotics, histamine-blocker exposure in historical series, crowded census, and extreme prematurity. This is why LOS and CLABSI teaching overlap. Hub hygiene and line removal are treatment, not paperwork. Healthcare-acquired-condition bundles are expanded in section 14.3; do not wait until that section to pull a useless PICC.

A CONS isolate from a meticulously prepped culture in a well late-preterm infant may be a contaminant. CONS from a symptomatic 600-gram infant with a PICC is true infection until you have a better explanation. Do not label every CONS result skin flora and leave the line untouched. Repeat cultures, examine the insertion site, and talk about removal.

Empiric LOS regimens follow the unit antibiogram. Many teams pair anti-staphylococcal coverage (often vancomycin when CONS or MRSA is likely) with gram-negative coverage (an aminoglycoside, cefepime, or another agent per local resistance). Add antifungal therapy when Candida is plausible—extreme prematurity, lingering central access, persistent thrombocytopenia, and hyperglycemia after broad antibiotics. This guide will not invent a single nationwide LOS cocktail as if AACN published one.

S. aureus LOS is not a 48-hour stop-and-see organism. Think osteomyelitis, septic arthritis, endocarditis, and pneumonia. Gram-negative LOS can produce rapid septic shock and meningitis. Candida may smolder for days, then declare with end-organ foci.

Viral and fungal disease that looks like bacteria

HSV clusters in the first 1–3 weeks of life. Maternal history of lesions is often absent; primary maternal infection near delivery is the highest-risk story, and you will not have that history reliably. Patterns: SEM (skin-eye-mouth vesicles, keratoconjunctivitis), CNS (seizures, often without a telling rash), and disseminated (hepatitis, DIC, pneumonitis, shock). Disseminated disease may have no vesicles. A week-old infant with liver failure, coagulopathy, and a sepsis exam needs acyclovir while PCR is pending. Surface cultures do not replace blood, CSF, and lesion PCR when lesions exist.

Enterovirus and parechovirus produce summer and fall sepsis-like illness, hepatitis, myocarditis, and meningitis. Antibiotics will not stop them. Myocarditis becomes cardiogenic shock in section 14.2. Isolation and supportive care matter; a high AST with a falling heart function is not a CRP problem.

Candida may smolder: slowly increasing oxygen, feeding intolerance, hyperglycemia, and falling platelets. Once candidemia is proven, teams look for endophthalmitis, endocarditis, and renal fungal balls. Amphotericin-class drugs are the conceptual backbone in many NICUs. Duration, again, is not an AACN-published hour count.

Cultures, CRP, and procalcitonin

Draw a blood culture before antibiotics when you can do so without delaying treatment of shock. Volume and sterile prep matter; a badly prepped bottle creates CONS confusion that haunts the next 48 hours. In shock, start antibiotics as soon as the bottles are (or cannot be) drawn—do not spend 20 minutes hunting a perfect sterile field while the lactate climbs.

CSF belongs in the evaluation of ill infants and in many LOS workups once the infant can tolerate a lumbar puncture. If the infant is too unstable, treat first and tap when safer. Document the reason for delay so the next nurse does not assume meningitis was excluded.

Urine culture is high-yield in LOS and low-yield as a primary EOS source in the first 72 hours, when hematogenous seeding of the kidney is uncommon and bag specimens mislead. Catheter or suprapubic samples beat perineal bags when you truly need urine.

Endotracheal aspirates show colonization more often than they prove pneumonia. Surface swabs of ear, umbilicus, or gastric fluid do not diagnose bloodstream infection. They describe what the infant swam through, not what is in the blood.

CRP rises over hours—often discussed around 6–12 hours—not minutes. Birth asphyxia, meconium aspiration, and surgery also raise it. A single CRP at delivery does not clear GBS. Serial CRPs sometimes support stopping antibiotics when cultures are negative; they do not replace the culture or the nurse's apnea and perfusion trend.

Procalcitonin is physiologically elevated in many healthy newborns over the first about 48 hours of life. After that window it is more informative and still imperfect (asphyxia and other inflammation move it). Neither marker is a sepsis on/off switch. If a stem offers CRP or procalcitonin as a reason to skip the blood culture, that option is wrong.

Putting it together at the bedside

1. Time the illness (EOS versus LOS versus late-onset GBS).
2. List plausible organisms and whether HSV belongs on the list.
3. Culture blood; add urine for LOS; add CSF when indicated and feasible.
4. Start conceptual empiric therapy (ampicillin/gentamicin for EOS; line-aware coverage for LOS; acyclovir if HSV is in the differential).
5. Support airway, glucose, temperature, and perfusion (shock details in 14.2).
6. Reassess when cultures and the infant declare themselves—without quoting a fake AACN stopwatch.

Exam traps: using a pediatric 20 mL/kg bolus as the identity of neonatal sepsis care; calling HSV not sepsis because vesicles are missing; treating CRP as a culture; forgetting that Listeria needs ampicillin; calling every CONS isolate a contaminant in an ELBW infant with a PICC; and inventing an official duration.