On this page
While biomarkers and rapid tests show high accuracy, evidence is conflicted regarding the sensitivity of procalcitonin across different sample types and the reliability of clinical prediction rules in neonates.
- Diagnostic sensitivity of Procalcitonin (PCT) for bacterial meningitisSupporting: Serum PCT is highly accurate for differentiating bacterial from viral meningitis in children, with a pooled sensitivity of 0.96. [3]Opposing: CSF procalcitonin demonstrates poor diagnostic sensitivity (68.6%) for bacterial meningitis, despite high specificity. [13]
- Reliability of clinical triage for meningitis in febrile infantsSupporting: The PECARN prediction rule accurately identifies febrile infants aged 28 days or younger at low risk for invasive bacterial infections, including meningitis. [14]Opposing: Clinical features in neonates are non-specific and overlap significantly with sepsis, making it challenging to reliably identify meningitis based on clinical signs alone. [12]
- Reliability of CSF culture as the diagnostic reference standardSupporting: Conventional microbiological assays and CSF cultures are treated as the definitive reference standard for evaluating new rapid antigen tests. [2]Opposing: CSF culture is an imperfect reference standard, particularly in partially treated cases, and inconsistent reference standards contribute to a high risk of bias in diagnostic studies. [1, 20]
Deep Dive — Evidence Details
Clinical Bottom Line
- ▸Serum procalcitonin (PCT) is a highly effective rule-out tool for pediatric bacterial meningitis with an LR- of 0.08 [3].
- ▸CSF rapid antigen tests (RAT) allow for immediate triage of Streptococcus pneumoniae, the leading cause of bacterial meningitis [2].
- ▸Biomarker assessment in both blood and CSF is essential for accurate diagnostic differentiation in pediatric CNS infections [1].
Serum (PCT) is a highly accurate biomarker for differentiating bacterial from viral in children, yielding a pooled sensitivity of 0.96 (95% CI 0.92-0.98) and specificity of 0.89 (95% CI 0.86-0.92) [3]A1a. With a negative likelihood ratio (LR-) of 0.08 (95% CI 0.04-0.14), a low serum PCT is clinically sufficient to rule out bacterial etiology in pediatric patients [3]A1a. The pooled positive likelihood ratio (LR+) for PCT is 7.5 (95% CI 5.6-10.1) [3]A1a.
Rapid antigen tests (RAT) in the (CSF) provide immediate triage for Streptococcus pneumoniae, facilitating earlier than conventional cultures [2]A1a. Although cultures remain the reference standard, the use of CSF and blood biomarkers is essential for accurate diagnosis in children suspected of central nervous system infections [1]A1a.
| Test | Target Population | Sensitivity (95% CI) | Specificity (95% CI) | Clinical Utility |
|---|---|---|---|---|
| Serum PCT | Pediatric | 0.96 (0.92-0.98) | 0.89 (0.86-0.92) | Rule-out (LR- 0.08) [3]A1a |
| CSF RAT | General | Not reported* | Not reported* | Rapid triage for S. pneumoniae [2]A1a |
| CSF Biomarkers | Pediatric | Variable | Variable | Essential for accuracy [1]A1a |
*Meta-analysis [2]A1a confirmed utility for triage without providing pooled sensitivity/specificity in the abstract.
Clinical Scenario and Population
- ▸Neonatal sepsis prevalence is high in low-resource NICUs, frequently involving XDR Gram-negative pathogens.
- ▸Tuberculous meningitis diagnosis often relies on 'possible' or 'probable' criteria, as only one-third of cases are definitive.
- ▸Neuroleptic malignant syndrome is a critical ICU mimic for CNS infection in patients with fever and altered consciousness.
Prevalence of in low-resource intensive care units (NICU) is approximately 29.92% [6]B3b. Within this population, culture-positive sepsis occurs in 11.5% of neonates, and 38.4% of Gram-negative episodes involve extensively drug-resistant (XDR) pathogens [9]B3b. Risk is further stratified by maternal factors; (PPROM) is a primary driver of early-onset sepsis [5]B2b, while significantly increases the risk of eclampsia (aOR 2.5) and acute kidney injury (aOR 4.11) [4]B3b. Very preterm infants (<1,500 g) represent a high-risk spectrum, with birth asphyxia rates reaching 42.3% [7]B3b.
(TBM) presents a broader diagnostic spectrum in endemic regions. In Thai cohorts, only 32.7% of patients meet "definite" criteria, while 44.2% are classified as "possible" TBM, necessitating high clinical suspicion despite negative initial testing [8]B3b. Differential diagnosis in the ICU must also account for (NMS), a rare mimic characterized by fever and altered mental status [10]C4.
| Population | Key Finding | Citation |
|---|---|---|
| NICU (Uganda) | 29.92% sepsis prevalence | [6]B3b |
| NICU (India) | 38.4% XDR Gram-negative sepsis | [9]B3b |
| TBM (Thailand) | 44.2% "possible" (non-definite) cases | [8]B3b |
| VPI (<1,500g) | 42.3% birth asphyxia rate | [7]B3b |
Index Test Performance vs. Reference Standard
- ▸Rapid antigen tests for S. pneumoniae in CSF provide a high LR+ (43.9), making them excellent for immediate triage while awaiting cultures.
- ▸CSF lactate (>3 mmol/L) and ferritin (>76 ng/mL) are highly accurate biomarkers for differentiating bacterial from viral meningitis in children.
- ▸Clinical features in neonates are too non-specific to rule out meningitis, necessitating CSF analysis in all cases of suspected neonatal sepsis.
Cerebrospinal fluid (CSF) culture serves as the reference standard, but its diagnostic delay often mandates the use of rapid index tests to guide empirical therapy [2]A1a. Rapid antigen tests (RAT) for Streptococcus pneumoniae demonstrate high diagnostic accuracy in CSF, with a pooled sensitivity of 96.6% (95% CI: 93.6–98.2) and specificity of 97.8% (95% CI: 95.6–98.9) [2]A1a. The positive likelihood ratio (LR+) of 43.9 (95% CI: 21.7–88.9) significantly shifts post-test probability; in a clinical scenario with a 10% pre-test probability, a positive RAT increases the probability of pneumococcal meningitis to 83%, while a negative result (LR- 0.03, 95% CI: 0.02–0.07) reduces it to 0.3% [2]A1a.
Biomarkers for Etiological Differentiation
Differentiating bacterial from viral etiologies in pediatric populations relies on biochemical markers when cultures are pending. CSF lactate at a threshold of 3 mmol/L provides a sensitivity and specificity of 91.4% [13]B2b. CSF ferritin (>76 ng/mL) performs with high accuracy, yielding a sensitivity of 91.4% and a specificity of 94.3% [13]B2b. While CSF procalcitonin (>0.5 ng/mL) lacks sensitivity (68.6%), its high specificity (100%) in specific cohorts makes it a powerful rule-in tool for bacterial infection [13]B2b. Serum C-reactive protein (CRP) at a 40 mg/L cutoff is less precise, with a sensitivity of 80% and specificity of 82.9% [13]B2b. The diagnostic utility of these biomarkers is superior to traditional white blood cell counts in the CSF, which can be confounded by traumatic taps or prior antibiotic use [13]B2b. Specifically, the high LR+ of CSF ferritin (16.0) and lactate (10.6) suggests they are more reliable than serum CRP for immediate clinical decision-making [13]B2b.
Neonatal Diagnostic Challenges
In the neonatal population, clinical features are often non-specific and overlap significantly with neonatal sepsis, making the diagnosis of bacterial meningitis particularly challenging [12]B3b. A cross-sectional study of 126 neonates with suspected sepsis emphasized that clinical signs alone cannot reliably distinguish meningitis from systemic infection [12]B3b. Consequently, and CSF analysis are mandatory for all neonates presenting with signs of serious bacterial infection, as blood cultures may not always capture the pathogen responsible for CNS involvement [12]B3b. The integration of rapid molecular or antigen-based tests alongside traditional biochemistry is essential to reduce the time to targeted antimicrobial therapy [2]A1a[13]B2b.
| Test (Index) | Sensitivity (95% CI) | Specificity (95% CI) | LR+ | LR- |
|---|---|---|---|---|
| CSF Rapid Antigen (S. pneumo) [2]A1a | 96.6% (93.6–98.2) | 97.8% (95.6–98.9) | 43.9 | 0.03 |
| CSF Lactate (>3 mmol/L) [13]B2b | 91.4% | 91.4% | 10.6 | 0.09 |
| CSF Ferritin (>76 ng/mL) [13]B2b | 91.4% | 94.3% | 16.0 | 0.09 |
| CSF Procalcitonin (>0.5 ng/mL) [13]B2b | 68.6% | 100% | ∞ | 0.31 |
| Serum CRP (>40 mg/L) [13]B2b | 80.0% | 82.9% | 4.7 | 0.24 |
Comparative Accuracy and Alternatives
- ▸Multiplex PCR (BioFire FilmArray) reduces antibiotic duration and hospital stay in pediatric meningitis cases [19].
- ▸A CSF granulocyte proportion of ≥80% is 92% specific for bacterial meningitis in adults [15].
- ▸CSF inflammatory markers like CXCL10 and IL-6 can differentiate neonatal meningitis from systemic sepsis [17].
The PECARN prediction rule identifies febrile infants ≤28 days at low risk for invasive bacterial infections, potentially refining the use of routine lumbar punctures in this population [14]A1b. For rapid pathogen identification, multiplex PCR (e.g., BioFire FilmArray) significantly reduces the duration of empiric antibiotic therapy and hospital length of stay in pediatric patients compared to standard culture-based methods [19]B3b.
Rapid antigen tests (RAT) for Streptococcus pneumoniae in CSF offer high diagnostic accuracy for immediate triage while awaiting definitive culture results [2]A1a. In adults with CSF (≥5 cells/mm³), the proportion of granulocytes is a robust differentiator; a threshold of ≥80% granulocytes yields a specificity of 92% (95% CI 89–94%) for bacterial meningitis [15]A1b.
Novel CSF inflammatory markers, including CXCL10, IL-6, and IL-8, effectively differentiate neonatal bacterial meningitis from sepsis [17]B2b. While CSF bacterial load in pneumococcal meningitis predicts unfavorable outcomes (Glasgow Outcome Scale <5), it does not correlate with CSF white cell counts, suggesting independent mechanisms of injury [16]B2b. Additionally, serum anti-GM-CSF autoantibodies serve as a prognostic marker in 8.1% of non-HIV cases [18]B3b.
| Tool | Target Population | Clinical Utility |
|---|---|---|
| PECARN Rule | Infants ≤28 days | Identifies low-risk patients to potentially avoid LP [14]A1b |
| Multiplex PCR | Pediatrics | Rapid identification; reduces empiric antibiotic use [19]B3b |
| Rapid Antigen Test | General | Immediate triage for S. pneumoniae [2]A1a |
| CSF Granulocytes | Adults | ≥80% threshold is highly specific for bacterial etiology [15]A1b |
| CXCL10 / IL-6 | Neonates | Differentiates bacterial meningitis from sepsis [17]B2b |
Limitations and Certainty
- ▸CSF pleocytosis is absent in up to 80% of pediatric bacterial meningitis cases presenting within 24 hours of symptom onset.
- ▸Biomarker and prediction model accuracy is limited by high risk of bias and a lack of robust external validation across diverse populations.
- ▸Clinical features alone cannot reliably distinguish between bacterial and non-infectious etiologies, requiring reliance on imperfect reference standards.
Absence of in early-presentation pediatric cases significantly limits the sensitivity of initial CSF analysis [22]C4. In children presenting within 24 hours of fever onset, normal white cell counts and glucose levels occur in up to 80% of culture-proven bacterial cases, necessitating repeat evaluation or molecular testing if clinical suspicion remains high [22]C4. While (CrAg) testing demonstrates high summary sensitivity (99.7%; 97.4–100) and specificity (94.1%; 88.3–98.1) in HIV-positive adults, its utility is constrained by the reference standard of CSF culture, which may be imperfect in partially treated cases [20]A1a.
Systematic reviews using QUADAS-2 highlight significant risk of bias in biomarker studies due to spectrum effects and inconsistent reference standards [1]A1a[20]A1a. Furthermore, clinical prediction models often lack robust external validation, with performance varying significantly across different patient cohorts [23]A1b. The diagnostic yield is also complicated by co-existing infections; for instance, infants with require careful screening as meningitis co-occurs in a subset of this population, though no consensus exists on the threshold for lumbar puncture in these patients [24]B2a.
| Marker | Population | Sensitivity (95% CI) | Key Limitation |
|---|---|---|---|
| CSF Pleocytosis | Pediatric (Early) | Low (Variable) | Absent in 80% of cases <24h [22]C4 |
| CrAg (Serum/CSF) | HIV+ Adults | 99.7% (97.4–100) | Dependent on fungal burden/culture [20]A1a |
| Clinical Models | General Suspected | Variable | Lack of external validation [23]A1b |
| Biomarkers | Pediatric | Variable | High risk of bias (QUADAS-2) [1]A1a |
References
- [1]
Groeneveld NS, Bijlsma MW, van de Beek D et al.. “Biomarkers in paediatric bacterial meningitis: a systematic review and meta-analysis of diagnostic test accuracy.” Clinical microbiology and infection : the official publication of the European Society of Clinical Microbiology and Infectious Diseases (2025). PMID: 39672463 ↗
L1SR_OBSCited in: Clinical Bottom Line, Comparative Accuracy and Alternatives, Limitations and Certainty - [2]
Someko H, Okazaki Y, Tsujimoto Y et al.. “Diagnostic accuracy of rapid antigen tests in cerebrospinal fluid for pneumococcal meningitis: a systematic review and meta-analysis.” Clinical microbiology and infection : the official publication of the European Society of Clinical Microbiology and Infectious Diseases (2023). PMID: 36503113 ↗
L1SR_OBSCited in: Clinical Bottom Line, Index Test Performance vs. Reference Standard, Comparative Accuracy and Alternatives - [3]
Henry BM, Roy J, Ramakrishnan PK et al.. “Procalcitonin as a Serum Biomarker for Differentiation of Bacterial Meningitis From Viral Meningitis in Children: Evidence From a Meta-Analysis.” Clinical pediatrics (2016). PMID: 26378091 ↗
L1SR_OBSCited in: Clinical Bottom Line - [4]
Ernawati E, Akbar MIA, Aryananda RA et al.. “HELLP syndrome as a major contributor to adverse maternal and neonatal outcomes among preeclamptic women: findings from a multicenter retrospective cohort study.” BMC pregnancy and childbirth (2025). PMID: 41430585 ↗
L3COHORTCited in: Clinical Scenario and Population - [5]
Mandelbrot L, Kennedy S, Rousseau J et al.. “Predicting neonatal infection in preterm premature rupture of membranes with vaginal microbiology and metagenomics: a prospective cohort study.” American journal of obstetrics and gynecology (2026). PMID: 41421745 ↗
L2COHORTCited in: Clinical Scenario and Population - [6]
Ssentongo SM, Ojok B, Opito R et al.. “Prevalence and risk factors of neonatal sepsis in the neonatal intensive care unit at Soroti Regional Referral hospital, Uganda: a retrospective study.” BMC pregnancy and childbirth (2025). PMID: 41254579 ↗
L3COHORTCited in: Clinical Scenario and Population - [7]
Wang H, Xu X, Zhang M et al.. “The delivery modes and morbidity/mortality in very preterm infants with birth weight <1,500 g: a retrospective cohort study.” Frontiers in public health (2025). PMID: 41211398 ↗
L3COHORTCited in: Clinical Scenario and Population - [8]
Taweephol T, Pongpitakmetha T, Boonwan N et al.. “Clinical characteristics and factors associated with tuberculous meningitis outcomes in Thailand: A 13-year retrospective cohort study.” International journal of infectious diseases : IJID : official publication of the International Society for Infectious Diseases (2025). PMID: 41173203 ↗
L3COHORTCited in: Clinical Scenario and Population - [9]
Yadav B, Gupta N, Sasidharan R et al.. “Extensively drug-resistant gram-negative sepsis in a neonatal intensive care unit from western India: A retrospective cohort study.” The Indian journal of medical research (2026). PMID: 42024915 ↗
L3COHORTCited in: Clinical Scenario and Population - [10]
Özdemir F, Birtane D, Yılmaz N et al.. “Intensive Care Management and Outcomes of Neuroleptic Malignant Syndrome: A Single-Center Retrospective Study.” Medicina (Kaunas, Lithuania) (2026). PMID: 41752777 ↗
L4COHORTCited in: Clinical Scenario and Population - [11]
Fabbri IS, Pagano T, Darin Y et al.. “Safety and Efficacy of Noninvasive Ventilation in Patients With Acute Exacerbations of Chronic Obstructive Pulmonary Disease and Decreased Level of Consciousness: A Retrospective Study.” The clinical respiratory journal (2026). PMID: 41603294 ↗
L3COHORTCited in: Clinical Scenario and Population - [12]
Pishori T, Furia FF, Manji K. “A cross-sectional study of clinical features of bacterial meningitis among neonates presumed to have sepsis in a tertiary hospital, Dar es Salaam, Tanzania.” The Pan African medical journal (2023). PMID: 38465011 ↗
L3OTHERCited in: Index Test Performance vs. Reference Standard - [13]
Sanaei Dashti A, Alizadeh S, Karimi A et al.. “Diagnostic value of lactate, procalcitonin, ferritin, serum-C-reactive protein, and other biomarkers in bacterial and viral meningitis: A cross-sectional study.” Medicine (2017). PMID: 28858084 ↗
L2OTHERCited in: Index Test Performance vs. Reference Standard - [14]
Burstein B, Waterfield T, Umana E et al.. “Prediction of Bacteremia and Bacterial Meningitis Among Febrile Infants Aged 28 Days or Younger.” JAMA (2026). PMID: 41359314 ↗
L1SR_OBSCited in: Comparative Accuracy and Alternatives - [15]
Staal SL, Olie SE, Ter Horst L et al.. “Granulocytes in cerebrospinal fluid of adults suspected of a central nervous system infection: a prospective study of diagnostic accuracy.” Infection (2024). PMID: 38520645 ↗
L1COHORTCited in: Comparative Accuracy and Alternatives - [16]
Chekrouni N, van Soest TM, da Cruz Campos AC et al.. “Bacterial load in cerebrospinal fluid predicts unfavourable outcome in pneumococcal meningitis: a prospective nationwide cohort study.” Clinical microbiology and infection : the official publication of the European Society of Clinical Microbiology and Infectious Diseases (2024). PMID: 38479702 ↗
L2COHORTCited in: Comparative Accuracy and Alternatives - [17]
Groeneveld NS, Olie SE, Visser DH et al.. “Cerebrospinal fluid inflammatory markers to differentiate between neonatal bacterial meningitis and sepsis: A prospective study of diagnostic accuracy.” International journal of infectious diseases : IJID : official publication of the International Society for Infectious Diseases (2024). PMID: 38395221 ↗
L2COHORTCited in: Comparative Accuracy and Alternatives - [18]
Jiang YK, Zhou LH, Cheng JH et al.. “Anti-GM-CSF autoantibodies predict outcome of cryptococcal meningitis in patients not infected with HIV: A cohort study.” Clinical microbiology and infection : the official publication of the European Society of Clinical Microbiology and Infectious Diseases (2024). PMID: 38295989 ↗
L3COHORTCited in: Comparative Accuracy and Alternatives - [19]
Kadambari S, Feng S, Liu X et al.. “Evaluating the Impact of the BioFire FilmArray in Childhood Meningitis: An Observational Cohort Study.” The Pediatric infectious disease journal (2024). PMID: 38190645 ↗
L3COHORTCited in: Comparative Accuracy and Alternatives - [20]
Temfack E, Rim JJB, Spijker R et al.. “Cryptococcal Antigen in Serum and Cerebrospinal Fluid for Detecting Cryptococcal Meningitis in Adults Living With Human Immunodeficiency Virus: Systematic Review and Meta-Analysis of Diagnostic Test Accuracy Studies.” Clinical infectious diseases : an official publication of the Infectious Diseases Society of America (2021). PMID: 32829406 ↗
L1SR_OBSCited in: Limitations and Certainty - [21]
Ellis J, Luintel A, Chandna A et al.. “Community-acquired acute bacterial meningitis in adults: a clinical update.” British medical bulletin (2019). PMID: 31556944 ↗
L5SR_OBSCited in: Limitations and Certainty - [22]
Zimmermann P, Curtis N. “Bacterial Meningitis in the Absence of Pleocytosis in Children: A Systematic Review.” The Pediatric infectious disease journal (2021). PMID: 33591075 ↗
L4SR_OBSCited in: Limitations and Certainty - [23]
van Zeggeren IE, Bijlsma MW, Tanck MW et al.. “Systematic review and validation of diagnostic prediction models in patients suspected of meningitis.” The Journal of infection (2020). PMID: 31794775 ↗
L1SR_OBSCited in: Limitations and Certainty - [24]
Poletto E, Zanetto L, Velasco R et al.. “Bacterial meningitis in febrile young infants acutely assessed for presumed urinary tract infection: a systematic review.” European journal of pediatrics (2019). PMID: 31473824 ↗
L2SR_OBSCited in: Limitations and Certainty