# Neonatal Sepsis

Neonatal sepsis requires immediate culture-directed evaluation and empiric treatment when illness is plausible, while structured early-onset risk assessment, serial examination, and timely culture review prevent avoidable antibiotic exposure in uninfected infants.

**Clinical question:** How should clinicians evaluate, empirically treat, and safely de-escalate suspected early- or late-onset neonatal sepsis?

Updated: 2026-08-21T00:10:19.605912Z

## What matters in practice
- For infants born at 35 weeks' gestation or later, acceptable early-onset sepsis strategies are categorical risk assessment, multivariate EOS-calculator assessment, or serial structured examinations; no approach identifies every case at birth. [10][11]
- Obtain a blood culture before empiric antimicrobials whenever this does not materially delay treatment; collect at least 1 mL to optimize pathogen recovery. [11][12]
- CBC indices, CRP, and procalcitonin cannot independently establish neonatal sepsis. Serial clinical assessment and culture results drive treatment decisions; serial CRP can support discontinuation in a clinically improving infant. [9][11][12]
- For suspected early-onset bacterial sepsis, ampicillin plus gentamicin remains the standard empiric regimen; broaden only for severe illness, meningitis concern, or epidemiologically credible resistant Gram-negative risk. [11][12]
- Stop empiric antibiotics by 36 to 48 hours when cultures remain sterile and there is no convincing site-specific infection or persistent clinical evidence of infection. [9][11][12]

## Separate early-onset from late-onset disease

Timing, acquisition route, and local epidemiology determine the initial evaluation and empiric regimen.

Early-onset sepsis (EOS) is generally defined as infection presenting before 72 hours of life, although some surveillance definitions use the first 7 days. It is usually vertically acquired. In the United States, group B streptococcus (GBS) predominates in term infants, whereas Escherichia coli is relatively more important in preterm and very-low-birth-weight infants. [11][12][22]

Late-onset sepsis (LOS; 72 hours or later) is commonly healthcare-associated in hospitalized preterm infants. Coagulase-negative staphylococci, Staphylococcus aureus, Gram-negative bacilli, and Candida are relevant pathogens, but the empiric choice should follow unit-specific susceptibility data and the infant's device, operative, and colonization history. [9][11][12]
- Treat shock, respiratory failure, seizures, or rapidly progressive illness as an emergency: obtain cultures promptly but do not defer antimicrobials for a lengthy diagnostic workup. [11][15]
- A sepsis-like syndrome also warrants consideration of HSV, enterovirus, fungal infection, congenital infection, metabolic disease, and noninfectious cardiopulmonary disease when the presentation or trajectory is discordant with bacterial sepsis. [11][12][17]

## Choose and operationalize one EOS assessment strategy

For well-appearing term and late-preterm infants, maternal risk factors alone should not mandate antibiotics.

For infants born at 35 weeks' gestation or later, the American Academy of Pediatrics recognizes three EOS risk-assessment approaches: categorical maternal and neonatal risk thresholds, multivariate assessment using the Neonatal EOS Risk Calculator, and serial physical examination based on the infant's evolving condition. Each requires a local protocol defining vital-sign frequency, escalation criteria, and documentation. [10][11]

The EOS calculator incorporates gestational age, highest maternal intrapartum temperature, maternal GBS status, rupture-of-membranes duration, and intrapartum antibiotics, then combines these with the infant's clinical examination. In one prospective implementation cohort of 204,685 infants, calculator-guided care reduced blood-culture testing by 66% and empiric antibiotic treatment by 48% compared with a categorical CDC-based strategy. [11]

Calculator-derived risk of at least 1 per 1,000 births supports blood culture plus clinical observation; a risk of at least 3 per 1,000 supports empiric antibiotics in the cited review. This tool is not a substitute for reassessment: an infant who becomes ill during observation requires immediate evaluation and treatment. [11]
- For preterm infants at 34 6/7 weeks' gestation or less, delivery circumstances are central. Lowest-risk characteristics are noninfectious indication for delivery, cesarean birth, and no labor, attempted induction, or membrane rupture before delivery; selected infants may receive observation alone or culture plus observation. [10][11]
- Preterm birth after cervical insufficiency, preterm labor, preterm premature rupture of membranes, suspected intra-amniotic infection, or unexplained acute nonreassuring fetal status confers high EOS risk and generally warrants blood culture plus empiric treatment. [10][11]

*EOS approaches for infants born at 35 weeks' gestation or later. [10][11]*

| Approach | Core action | Operational requirement |
| --- | --- | --- |
| Categorical risk assessment | Treat clinically ill infants and those with significant intrapartum infection concern; observe or test other risk groups according to local thresholds. [10][11] | Avoid ambiguous definitions; specify surveillance and treatment triggers. [11] |
| Multivariate EOS calculator | Use maternal variables plus newborn examination to recommend observation, culture, or empiric antibiotics. [11] | Use only within validated gestational-age range and pair with serial assessment. [11][15] |
| Serial examination | Withhold empiric antibiotics in initially well-appearing infants while performing structured repeated examinations through 48 hours. [11] | Require reliable staffing, predefined escalation criteria, and documented examinations. [11] |

## Culture first; use laboratory tests as adjuncts

The diagnostic task is to identify invasive infection without extending antibiotics for nonspecific abnormalities.

Blood culture remains the reference test for bacterial sepsis. Obtain it before the first antibiotic dose whenever feasible. A minimum 1 mL blood volume improves detection of low-density bacteremia. Modern continuously monitored systems detect more than 90% of untreated bacteremia by 36 hours in one review, supporting early antibiotic reassessment when the infant improves and cultures remain negative. [9][11][12]

CBC abnormalities have limited positive predictive value. Leukopenia below 5,000/mm3 and severe neutropenia are more concerning than leukocytosis, but normal values do not exclude sepsis. An elevated immature-to-total neutrophil ratio is nonspecific; CBC findings should not independently justify prolonged empiric therapy. [10][11][12]

CRP rises after infection onset and has limited utility at initial presentation. Serial rather than single measurements can contribute to a decision to stop therapy when cultures are negative and the infant is clinically well. Procalcitonin rises earlier but is also affected by noninfectious neonatal physiology and should not independently diagnose infection. [9][11][12]
- EOS: do not routinely obtain urine culture in an otherwise standard EOS evaluation. [11][12]
- LOS: obtain urine testing when evaluating infants outside the neonatal unit; urine culture is generally recommended in LOS evaluations in the StatPearls review. Local pathways may differ for infants already hospitalized in a NICU. [11][12][15]
- Chest radiography is appropriate when respiratory findings raise concern for pneumonia or when an alternative pulmonary diagnosis needs assessment. [12]

### Lumbar puncture and meningitis

Perform lumbar puncture before antibiotics when it is safe and does not significantly delay treatment in an infant with strong clinical concern for sepsis or meningitis. Defer until stabilization in respiratory compromise, shock, uncontrolled seizures, or bleeding risk. [11][15]

Lumbar puncture is particularly important with positive blood culture, CNS signs, failure to improve, or strong persistent suspicion. Meningitis may occur despite a negative blood culture; CSF evaluation should include cell count and differential, protein, glucose with paired blood glucose, Gram stain, culture, and pathogen-directed PCR when available. [11][12][15]
- A clinically well infant assessed only because of maternal EOS risk factors does not routinely need lumbar puncture. [12][17]
- If bacterial meningitis is confirmed, arrange audiologic and neurodevelopmental follow-up; NICE recommends audiologic assessment within 4 weeks of the infant being well enough for testing. [15]

## Start empiric therapy promptly, then narrow or stop decisively

Exact neonatal dosing must follow gestational age, postnatal age, renal function, and local neonatal formulary guidance.

For suspected EOS, intravenous ampicillin plus gentamicin provides coverage for GBS, E. coli, enterococci, and Listeria and remains the recommended empiric combination in AAP-derived guidance. The supplied sources do not provide a U.S. neonatal dosing table sufficient to safely specify ampicillin or gentamicin dose intervals across gestational and postnatal ages; use an institutional neonatal dosing reference and therapeutic drug monitoring protocol. [10][11][12]

Avoid routine empiric third-generation cephalosporins for uncomplicated EOS because broader exposure is associated with antimicrobial resistance and invasive fungal infection. Consider expanded Gram-negative coverage for an infant who is critically ill despite ampicillin-gentamicin, has credible resistant Gram-negative risk, or has suspected Gram-negative meningitis; obtain infectious diseases or microbiology input and use local susceptibility data. [11][12]

For hospital-acquired LOS, empiric regimens should cover local Gram-positive and Gram-negative epidemiology. Narrow-spectrum antistaphylococcal therapy plus an aminoglycoside is a reasonable approach where methicillin-resistant S. aureus prevalence is low; reserve empiric vancomycin for units or infants with a credible resistant Gram-positive risk. Suspected necrotizing enterocolitis requires anaerobic coverage, such as metronidazole. [9][11][15]
- Suspected neonatal meningitis: use a regimen with reliable CSF activity. NICE recommends intravenous amoxicillin plus cefotaxime when the pathogen is unknown in a neonatal unit. [15]
- Suspected HSV disease, particularly with vesicles, seizures, hepatitis, liver failure, or compatible CSF pleocytosis, warrants urgent HSV PCR testing and consideration of empiric acyclovir while results are pending. [11][17]
- Suspected invasive candidiasis in high-risk preterm infants requires antifungal-directed evaluation and treatment; amphotericin B deoxycholate is cited as first-line empiric therapy when Candida is suspected. [11]

### Gentamicin monitoring

Gentamicin requires dose-interval adjustment and drug concentration monitoring. NICE recommends obtaining a trough immediately before the second dose if a second dose is administered, targeting trough concentrations below 2 mg/L and, when treatment exceeds three doses, below 1 mg/L. This is international guidance; U.S. centers should follow local neonatal pharmacokinetic protocols. [15]

*Empiric therapy principles by syndrome; final selection should incorporate local antibiograms and culture results. [9][11][12][15]*

| Clinical setting | Initial approach | Escalation or tailoring |
| --- | --- | --- |
| Suspected EOS | IV ampicillin plus gentamicin. [11][12] | Add broader Gram-negative coverage only for severe illness, meningitis concern, or credible resistant-pathogen risk. [11][12] |
| Hospital-acquired LOS | Use local susceptibility data; consider antistaphylococcal therapy plus aminoglycoside, with vancomycin reserved for appropriate resistant Gram-positive risk. [9][11] | Add anaerobic activity if necrotizing enterocolitis is suspected. [11][15] |
| Meningitis concern | Use agents with CSF activity; obtain CSF when safe. [15][17] | Narrow to pathogen-directed therapy and extend duration according to organism, CSF sterilization, and complications. [15][17] |

## Use culture time-to-positivity and clinical trajectory to stop unnecessary therapy

Culture-negative illness is common; prolonged therapy requires a documented indication.

For suspected EOS, discontinue empiric therapy by 36 to 48 hours of sterile culture incubation unless there is clear site-specific infection or compelling persistent clinical evidence. Persistent cardiorespiratory instability in very-low-birth-weight infants, or isolated laboratory abnormalities, should not alone drive prolonged empirical treatment. [9][10][11][12]

For suspected LOS, NICE recommends reassessment at 48 hours and discontinuation when cultures are negative, initial suspicion was not strong, the infant is clinically reassuring, and CRP trends are reassuring. If antibiotics continue despite sterile cultures, review daily for a stop decision. [15]

For uncomplicated culture-positive bacteremia without meningitis, 7 days is recommended in NICE guidance for EOS and LOS, with longer treatment for incomplete recovery, Gram-negative or S. aureus infection, central-line infection, osteomyelitis, intra-abdominal disease, or other focal infection. Duration must be individualized to organism, infection site, source control, CSF findings, and clearance cultures. [15]
- Document the presumed syndrome, microbiologic evidence, source evaluation, and planned reassessment time at antibiotic initiation. [16]
- De-escalate to the narrowest active agent when organism identification and susceptibility data return. [11][12]
- In culture-negative cases, avoid using the label "sepsis" as the sole rationale for prolonged therapy without serial clinical and microbiologic justification. [9][11]

## Prevent vertical transmission and device-associated infection

Prevention changes both EOS burden and the probability that broad empiric therapy is needed.

GBS screening and intrapartum antibiotic prophylaxis have substantially reduced GBS-associated EOS. ACOG-based recommendations cited in the supplied review support maternal rectovaginal screening at 36 0/7 to 37 6/7 weeks' gestation and intrapartum prophylaxis when indicated. [11]

For LOS prevention, reduce invasive-device exposure, adhere to central-line insertion and maintenance practices, practice hand hygiene, and use antimicrobial stewardship to limit selection pressure and invasive candidiasis risk. [11][12]

In NICUs with substantial invasive candidiasis risk, antifungal prophylaxis may be considered for high-risk preterm infants. NICE recommends oral nystatin for infants receiving antibiotics for suspected LOS who weigh 1,500 g or less or were born before 30 weeks; intravenous fluconazole is an off-label alternative when enteral administration is not possible. Applicability to U.S. practice depends on local invasive candidiasis incidence and formulary policy. [15]

## Common questions

### When should empiric antibiotics be stopped in suspected EOS?

Stop by 36 to 48 hours if blood cultures remain sterile and there is no site-specific infection or convincing ongoing clinical evidence of infection. Do not prolong therapy for laboratory abnormalities alone. [9][10][11][12]

### Can a normal CBC or CRP exclude neonatal sepsis?

No. CBC and single inflammatory-marker values are insufficient to exclude sepsis. Serial clinical examination, culture results, and—in selected cases—serial CRP trends are more useful for safe de-escalation. [9][11][12]

### When is lumbar puncture required in neonatal sepsis evaluation?

Perform lumbar puncture when meningitis is suspected, blood culture is positive, the infant has CNS signs, fails to improve, or clinical suspicion remains strong, provided stabilization and timely treatment are not compromised. [11][12][15]

### Is the EOS calculator appropriate for preterm infants?

It is intended for infants born at approximately 34 to 35 weeks' gestation or later, depending on the cited guidance. For more preterm infants, assess risk primarily from the circumstances of preterm delivery and clinical status. [10][11][15]

## References
1. A Quantitative Approach to Neonatal Sepsis Management | NEJM Clinician — clinician.nejm.org — https://clinician.nejm.org/nejm-jw.NA43555
2. Adherence to Perinatal Asphyxia or Sepsis Management ... — jamanetwork.com — https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2834206
3. Proposed Core Outcomes After Neonatal Sepsis — jamanetwork.com — https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2830560
4. 1140. Evaluation of Neonatal Sepsis Guidelines in a Neonatal ... — academic.oup.com — https://academic.oup.com/ofid/article/6/Supplement_2/S406/5605061
5. Neonatal sepsis: A review of current management strategies — www.sciencedirect.com — https://www.sciencedirect.com/science/article/abs/pii/S1355184124000553
6. Assessing the Use of Neonatal Sepsis Guidelines and ... — academic.oup.com — https://academic.oup.com/jpids/article/14/4/piaf017/8029324
7. Neonatal sepsis — www.sciencedirect.com — https://www.sciencedirect.com/science/article/abs/pii/S175172221930071X
8. Neonatal sepsis management in Africa: A rapid systematic ... — www.sciencedirect.com — https://www.sciencedirect.com/science/article/pii/S1875957225002062
9. Neonatal Sepsis - an overview — www.sciencedirect.com — https://www.sciencedirect.com/topics/medicine-and-dentistry/neonatal-sepsis
10. Quality assessment of clinical practice guidelines for neonatal sepsis using the Appraisal of Guidelines for Research and Evaluation (AGREE) II Instrument: A systematic review of neonatal guidelines — pmc.ncbi.nlm.nih.gov — https://pmc.ncbi.nlm.nih.gov/articles/PMC9424847
11. Neonatal Sepsis: A Comprehensive Review — pmc.ncbi.nlm.nih.gov — https://pmc.ncbi.nlm.nih.gov/articles/PMC11761862
12. Neonatal Sepsis - StatPearls - NCBI Bookshelf — www.ncbi.nlm.nih.gov — https://www.ncbi.nlm.nih.gov/books/NBK531478
13. Neonatal Sepsis: Aetiology, Pathophysiology, Diagnostic Advances and Management Strategies — pmc.ncbi.nlm.nih.gov — https://pmc.ncbi.nlm.nih.gov/articles/PMC11452898
14. Assessing the Use of Neonatal Sepsis Guidelines and Antibiotic Prescription With Large-Scale Prospective Data From Zimbabwe and Malawi - PMC — pmc.ncbi.nlm.nih.gov — https://pmc.ncbi.nlm.nih.gov/articles/PMC11976057
15. Neonatal infection: antibiotics for prevention and treatment - NCBI Bookshelf — www.ncbi.nlm.nih.gov — https://www.ncbi.nlm.nih.gov/books/NBK571222
16. Hospital Sepsis Program Core Elements — www.cdc.gov — https://www.cdc.gov/sepsis/hcp/core-elements/index.html
17. Neonatal Meningitis - StatPearls - NCBI Bookshelf — www.ncbi.nlm.nih.gov — https://www.ncbi.nlm.nih.gov/books/NBK532264
18. Neonatal Respiratory Distress Syndrome - StatPearls - NCBI Bookshelf — www.ncbi.nlm.nih.gov — https://www.ncbi.nlm.nih.gov/books/NBK560779
19. Newborn Health — www.who.int — https://www.who.int/teams/maternal-newborn-child-adolescent-health-and-ageing/newborn-health/essential-newborn-care
20. Neonatal Sepsis - PubMed — pubmed.ncbi.nlm.nih.gov — https://pubmed.ncbi.nlm.nih.gov/30285373
21. Sepsis Program Activities in Acute Care Hospitals — www.cdc.gov — https://www.cdc.gov/mmwr/volumes/72/wr/mm7234a2.htm
22. Early-Onset Neonatal Sepsis Surveillance and Trends | ABCs | CDC — www.cdc.gov — https://www.cdc.gov/abcs/reports/neonatal-sepsis.html
23. Tables of Antibacterial Drug Dosages — publications.aap.org — https://publications.aap.org/book/chapter-pdf/515318/rbo2021_s4_005_en.pdf
24. Cefprozil | Drug Lookup | Pediatric Care Online — publications.aap.org — https://publications.aap.org/pediatriccare/drug-monograph/18/5232/Cefprozil

## Editorial note

Prepared from cited clinical literature using Astra's research workflow. Verify recommendations against current guidance and patient-specific factors.