Home / Clinical Guideline Hubs / Chapter 8.1
Section 8 — Infectious Disease & Sepsis Verify against local policy v1.0 · July 2026 Baylor Ed. 33 cross-checked Sept 2026

Chapter 8.1 — Early-Onset Neonatal Sepsis

Built on West Midlands Neonatal Guidelines 2025–28 (NICE NG195) · Kaiser Permanente EOS calculator · AAP EOS guidance · Baylor Guidelines for Acute Care of the Neonate

Educational guideline — verify locally. Antibiotic choices and doses differ between UK (NICE) and US (AAP) practice; verify all agents/doses against your Neonatal Formulary and local antimicrobial policy. Does not replace attending judgment.
BEDSIDE ACTION BOX — Suspected EOS

1. Overview & Definition

Overview

Early-onset sepsis (EOS) is invasive bacterial infection presenting in the first 72 hours of life, acquired around the time of birth. Group B streptococcus (GBS) and Escherichia coli are the predominant organisms. Because early signs are non-specific and deterioration can be rapid, care balances two goals: treat genuinely infected infants promptly, and avoid unnecessary antibiotics (and separation from mother) in the large number of at-risk but uninfected babies. Structured risk assessment — the NICE framework and/or a validated EOS calculator — supports that balance.

Why This Topic Matters

Untreated EOS is rapidly fatal, yet blanket antibiotic use exposes many well infants to harm (microbiome disruption, resource use, mother–baby separation). The commonest errors are delaying antibiotics in a sick baby and, conversely, prolonged "just-in-case" courses when cultures are negative and the baby is well.

Definition

EOS is culture-confirmed or clinically diagnosed invasive infection (bacteraemia, meningitis, or pneumonia) with onset within the first 72 hours of life. "Culture-negative/clinical sepsis" describes a symptomatic infant treated as infected despite sterile cultures.

2. Who This Guideline Applies To

Scope
  • All newborns being risk-assessed for, or treated for, suspected infection in the first 72 hours of life.
  • Well infants with maternal/perinatal risk factors (e.g., GBS, prolonged rupture of membranes, chorioamnionitis, intrapartum fever) requiring a decision on investigation, observation, or antibiotics.
  • Cross-references: GBS colonisation, late-onset sepsis (8.2), congenital infections (8.3), meningitis work-up, and the unwell/collapsing neonate.

3. Key Definitions

Core Terminology

Shared vocabulary for the EOS pathway.

TermDefinitionPractical note
EOSInvasive infection with onset <72 h of life.GBS and E. coli predominate.
Red-flag factorHigh-risk feature mandating investigation + antibiotics.e.g., confirmed infection in a co-twin; shock; need for CPR/ventilation; seizures.
IAPIntrapartum antibiotic prophylaxis (for maternal GBS).Reduces early-onset GBS; does not prevent late-onset GBS.
EOS calculator (KP-SRC)Validated tool estimating EOS probability to guide antibiotics in ≥34–35-week infants.NICE-endorsed alternative for eligible well babies; pair with observation.
Culture-negative sepsisSymptomatic infant treated as infected despite sterile cultures.Keep courses short if cultures negative and baby well.

4. Risk Factors

Maternal/Perinatal Risk Factors

A single red flag, or two or more other risk factors/clinical indicators, triggers investigation and antibiotics.

CategoryFactors
Red flagSuspected/confirmed infection in another baby of a multiple pregnancy.
Other risk factorsPrevious baby with invasive GBS; maternal GBS colonisation/bacteriuria/infection this pregnancy; preterm birth (<37 wk) after spontaneous labour; confirmed rupture of membranes >18 h before preterm birth; term pre-labour rupture >24 h; intrapartum fever >38°C with suspected/confirmed bacterial infection; clinical chorioamnionitis.

5. Clinical Indicators

Signs Suggestive of Infection

Red-flag indicators mandate immediate treatment; non-red-flag indicators are weighed with risk factors.

CategoryIndicators
Red flagApnoea; need for CPR; seizures; need for mechanical ventilation; signs of shock.
OtherAltered behaviour/tone (floppiness); feeding difficulty/intolerance (distension, vomiting, high aspirates); abnormal heart rate; respiratory distress; hypoxia/cyanosis; jaundice within 24 h; signs of encephalopathy or PPHN; temperature <36°C or >38°C unexplained; unexplained bleeding/thrombocytopenia/coagulopathy; hypo/hyperglycaemia; metabolic acidosis (base excess ≥10).

6. Management Algorithm

1
Risk-assess every newborn
Identify red-flag and other risk factors and clinical indicators.
2
Any red flag, or ≥2 non-red-flag factors/indicators?
Yes → take blood culture and start empirical antibiotics (within 1 h of the decision). For well ≥34–35-week babies, an EOS calculator may refine whether antibiotics are needed.
3
Only 1 risk factor OR 1 clinical indicator?
Use clinical judgement; consider withholding antibiotics and observe closely for ≥12 h (at 1 h, 2 h, then 2-hourly). New concern → investigate and treat.
4
Investigate before/with antibiotics
Blood culture in all (CRP/procalcitonin are not recommended for EOS decisions); LP if strong suspicion or meningitis signs (give antibiotics first if LP would delay them). Avoid routine urine culture and surface swabs.
5
Reassess at 36–48 hours
Blood culture sterile and baby clinically well → stop antibiotics (NICE units also require a CRP that is not raised). Positive culture or ongoing clinical/lab concern → continue and tailor.
6
Confirmed infection
Tailor to organism/sensitivities; treat bacteraemia ~7–10 days and meningitis ~14–21 days (organism-dependent); repeat cultures/LP as indicated.
7
⚠ Do-not-miss
A sick baby gets antibiotics now — never delay for a calculator, CRP, or LP. Consider HSV (aciclovir) if encephalopathy/vesicles/maternal history; consider meningitis if not improving.

7. Investigations

Before Starting Antibiotics
  • Blood culture in all before the first dose (do not delay antibiotics for other tests).
  • CRP and procalcitonin are not recommended for confirming or excluding EOS (box below); NICE units measure CRP at presentation and 18–24 h.
  • Lumbar puncture if there is strong clinical suspicion of infection or signs of meningitis, and it is safe — but give antibiotics first if the LP would delay them.
  • Do not perform routine urine culture in EOS; take skin/eye/umbilical swabs only if localised signs (purulent eye discharge → also send viral PCR for Chlamydia/gonococcus; umbilical infection → flucloxacillin + gentamicin).
  • FBC and other tests as clinically indicated; a normal white cell count does not exclude sepsis.

8. Antibiotics & Duration

Empirical Antibiotic Table

Default regimen follows Baylor Ed. 33 (Table 9-4); the UK alternative is listed last. Doses by gestation and postnatal age, and drug levels, are tabulated in 8.2.

SettingFirst-line regimenNotes
Default (Baylor Ed. 33)Ampicillin 100 mg/kg every 8 h + an aminoglycoside — Baylor uses amikacin; gentamicin (e.g., 4 mg/kg every 24 h at 35–43 weeks, ≤7 days old) is the alternative.Baylor's amikacin choice follows its local antibiogram, not a claim of superiority — pick the aminoglycoside by your own susceptibility data. 48-hour rule-out; 7–10 days of organism-specific therapy if the culture is positive.
Meningitis suspectedAmpicillin + ceftazidime at meningeal doses.Organism-specific duration; repeat CSF 24–48 h into therapy. Avoid ceftriaxone in neonates.
Umbilical infectionFlucloxacillin + gentamicin.Cover staphylococci; de-escalate on cultures.
Other sources: UK / NICEBenzylpenicillin 25 mg/kg 12-hourly (8-hourly if very ill) + gentamicin 5 mg/kg per local interval.Add cefotaxime for suspected/confirmed Gram-negative sepsis or meningitis; stop benzylpenicillin if Gram-negative confirmed.
Who gets evaluated, by gestation (Baylor Ed. 33)

Above 35 0/7 weeks: any infant with signs of sepsis gets a CBC, blood and CSF cultures, and antibiotics. If the mother had a fever above 38 °C before delivery or within 24 hours after, review the obstetric history — and where intra-amniotic infection, endometritis or another systemic maternal infection is judged present, use the Neonatal Early-Onset Sepsis Risk Calculator to decide. Empiric antibiotics are not given unless the calculator recommends them, but vital signs every 4 hours are, even in a well-looking infant. With rupture of membranes beyond 18 hours but no signs of infection in mother or infant: observe in hospital for 48 hours. Healthy-appearing infants need paediatric follow-up within 2–5 days of discharge.

Below 34 6/7 weeks: evaluate and treat for prolonged rupture of membranes, maternal fever, intra-amniotic infection, unexplained premature labour, maternal antibiotics for suspected bacterial infection, or any sign of sepsis. If none of these is present and delivery was by caesarean without labour and with intact membranes, no evaluation is needed unless sepsis is suspected clinically. Well-appearing at-risk preterm infants get a CBC and blood culture — the lumbar puncture is at the attending's discretion — and meningeal doses of antibiotics. A VLBW infant whose course and evaluation make sepsis very unlikely may not need an LP at all.

Inflammatory markers do not help in early-onset sepsis

CRP and procalcitonin have little to no value in confirming or excluding early-onset infection and are not recommended for that purpose. (They retain a limited role in late-onset sepsis — a CRP below 1 mg/dL at 24 hours after the workup began makes bacterial infection unlikely and can shorten antibiotic exposure.) Treating culture-negative sepsis beyond 36–48 hours is discouraged; in term infants, stop once the blood culture is sterile at 24–48 hours of incubation. Meningeal doses if the CSF is abnormal, could not be obtained, or a Gram-negative organism is suspected; non-meningeal doses if the CSF is normal. Repeat a positive blood culture every 24 hours until two consecutive sterile cultures, and repeat the CSF 24–48 hours into therapy to document sterility.

Duration & Stopping Rules
  • Stop at 36–48 hours if the blood culture is sterile and the baby is clinically well (NICE: at 36 h, also requiring a CRP that is not raised).
  • Culture-confirmed bacteraemia: usually ~7–10 days; meningitis: ~14–21 days (organism-dependent).
  • Review and de-escalate to the narrowest effective agent once sensitivities are known.
Evidence & Source Differences
  • Framework vs calculator: the NICE risk-factor/indicator framework and the Kaiser Permanente EOS calculator are both endorsed; the calculator (for well ≥34–35-week infants) safely reduces antibiotic exposure when combined with structured observation.
  • Empirical regimen: default ampicillin + an aminoglycoside; UK practice uses benzylpenicillin + gentamicin. Both cover GBS and E. coli.
  • Doses and drug levels: Baylor Tables 19-3 and 19-4 in 8.2.

9. Monitoring

Monitoring Table

Serial observation is central — the trajectory guides both starting and stopping.

ParameterFrequencyTarget / watchAction if abnormal
Vital signs / clinical status≥12 h if observing (1 h, 2 h, then 2-hourly)Stable, feeding wellNew indicator → investigate and treat.
Blood cultureAt 36–48 h and as results returnNegativePositive → continue/tailor; consider LP.
CRPNot used for EOS decisions by default (NICE units: baseline + 18–24 h)—Base stopping on cultures and the clinical course.
Gentamicin levelsPer TDM protocol. Baylor Ed. 33: when treatment will exceed 48 h or renal function is abnormal — peak and trough around the third dose (sooner with acute kidney injury), then at least weeklyTherapeutic/safe trough. Peak 30 min after the infusion ends (efficacy); trough just before the next dose (safety); follow BUN and creatinineAdjust dose/interval. Doses are chosen by PMA, postnatal age and weight, and should be updated for weight weekly; involve pharmacy in renal dysfunction and ID for complicated infections. Vancomycin: trough 8–20 mcg/mL by organism, MIC and site; check levels only with renal dysfunction, treatment failure, or on specialist request — at appropriate doses alone it is not nephrotoxic in children.
Glucose / acid–baseAs indicatedNormalSupport; reconsider severity.

10. Contraindications & Precautions

Safety Cautions
  • Never delay antibiotics in a sick baby for a calculator result, CRP, or LP — treat within 1 hour of the decision.
  • A single normal CRP or a normal white cell count does not exclude sepsis — use the whole picture and trends.
  • Avoid prolonged empirical antibiotics when cultures are negative and the baby is well (antibiotic stewardship).
  • Avoid ceftriaxone in neonates (bilirubin displacement; calcium interaction) — use cefotaxime.
  • Consider and treat HSV (aciclovir) where features suggest it — bacterial cover alone is insufficient.
  • Monitor gentamicin levels (nephro/ototoxicity) per local protocol.

11. Escalation

Escalate When…
  • Signs of shock, respiratory failure, or clinical deterioration — resuscitate, escalate support, and involve seniors urgently.
  • Positive blood culture, suspected/confirmed meningitis, or failure to improve — tailor therapy, consider repeat LP, and seek microbiology/infectious-diseases advice.
  • Diagnostic uncertainty or possible HSV/atypical infection — start aciclovir and escalate.
Parent Counselling Points
  • "Because of risk factors around birth, we're checking your baby for infection and may start antibiotics as a precaution while we wait for the blood-culture results."
  • "If the cultures stay clear and your baby remains well, we usually stop the antibiotics after 36–48 hours."
  • "If an infection is confirmed, we continue a full course and, if needed, do a lumbar puncture to check for meningitis. We'll keep you informed and support feeding and bonding."

12. Key Pearls

High-Value Clinical Pearls
  • Treat the sick baby now; risk-stratify the well baby carefully — most at-risk babies are not infected.
  • Blood culture before antibiotics; antibiotics within 1 hour of the decision.
  • Empirical cover targets GBS and E. coli (ampicillin + an aminoglycoside; UK benzylpenicillin + gentamicin); avoid ceftriaxone.
  • CRP and procalcitonin don't confirm or exclude EOS; a normal CRP/WCC never excludes sepsis.
  • The 36–48-hour rule: stop if cultures are sterile and the baby is well.
  • EOS calculators cut antibiotic use in well ≥34–35-week babies — but pair with observation and never override a sick baby.

13. Common Mistakes to Avoid

Pitfalls & Better Practice

The recurring errors in EOS management.

MistakeWhy it harmsBetter practice
Delaying antibiotics in a sick baby.EOS can be rapidly fatal.Culture then antibiotics within 1 hour; don't wait for tests.
Relying on a single normal CRP/WCC to exclude sepsis.False reassurance.Use the whole clinical picture and cultures.
Prolonged antibiotics with negative cultures.Microbiome harm, resistance, separation.Apply the 36–48-hour stop rule.
Using ceftriaxone in neonates.Bilirubin/calcium risks.Use cefotaxime for Gram-negative/CNS cover.
Forgetting HSV.Untreated HSV is devastating.Add aciclovir when features suggest it.
Routine urine culture / surface swabs in EOS.Low yield, misleading.Blood culture ± LP; swabs only for localised signs.

14. Board-Style High-Yield Summary

Key Takeaways
  • EOS = invasive infection <72 h; GBS and E. coli predominate.
  • Any red flag, or ≥2 non-red-flag risk factors/indicators → blood culture + antibiotics within 1 hour.
  • Investigations: blood culture in all; LP for strong suspicion/meningitis (don't delay antibiotics); CRP/PCT not recommended for EOS decisions.
  • Empirical: ampicillin + an aminoglycoside (UK: benzylpenicillin + gentamicin); ampicillin + ceftazidime for meningitis; avoid ceftriaxone.
  • Stop at 36–48 h if cultures sterile and baby well; treat confirmed bacteraemia ~7–10 d, meningitis ~14–21 d.
  • EOS calculators reduce antibiotics in well ≥34–35-week babies; never override a sick baby.

15. References

Back to Clinical Guideline Hubs