Children's Bacterial Meningitis Score (BMS)

• Child with suspected meningitis who has CSF pleocytosis.
• Use alongside local pathways for empiric antibiotics and observation.
• Not validated for neonates, VP shunts, or immunocompromised hosts.

Predictors: CSF Gram stain positive; CSF ANC ≥1,000/µL; CSF protein ≥80 mg/dL; peripheral ANC ≥10,000/µL; seizure before/during presentation. Low risk applies only when all five predictors are absent; any predictor present → high risk.

Original derivation/validation (Nigrovic et al., Pediatrics 2002/2007) showed a very low probability of bacterial meningitis when all five predictors were absent among children with CSF pleocytosis. However, most centers have updated pathways post–Hib and pneumococcal conjugate vaccines; always verify local cut points and observation policies, and remember the score was not designed for infants without CSF pleocytosis.

Performance anchors (for context; confirm locally): negative predictive value near 100% in the validation cohort for children with CSF WBC ≥10/mm³ when all predictors were absent; sensitivity in the mid-to-high 90s for bacterial meningitis using the high-risk cut (≥1 predictor). Specificity was modest—so the score is built to confidently identify low risk, not to rule in disease.

1. Confirm eligibility: pediatric patient with CSF pleocytosis and no exclusion condition for BMS use.
2. Score all five predictors with exact measured values and units.
3. Classify risk (0 predictors low risk; any predictor high risk) and compare with clinical trajectory.
4. Set reassessment timing for repeat neuro exam, vitals, and pending CSF/culture/PCR data.
5. Document escalation trigger and handoff ownership before disposition.

• After initial score classification: repeat neurologic and hemodynamic assessment in the same care block.
• While cultures/PCR are pending: continue scheduled serial reassessment with explicit trend documentation.
• Any neurologic, perfusion, or trajectory worsening: immediate reassessment and escalation regardless of prior risk category.

• Use alongside age‑appropriate empiric coverage, local antibiograms, and neurologic monitoring protocols.
• Factor in prior antibiotics, immunization status, timing of lumbar puncture, and likelihood of partially treated meningitis when interpreting results.
• Document seizure history carefully; subtle events may satisfy the seizure predictor and reclassify risk.
• Interpret CSF ANC and protein in context of viral meningitis patterns, traumatic taps, and interval from symptom onset to LP.
• Apply only to children (typically 2 months–18 years in published cohorts) with CSF pleocytosis; neonates <60 days, VP shunts, or immunocompromised hosts were not included.
• When vaccines shift local epidemiology, reassess institutional performance of the score; adjunct PCR panels may change post-test probabilities.

• Treating low-risk status as a discharge rule without serial exams or reliable follow-up.
• Ignoring timing of prior antibiotics, which can blunt CSF findings and reduce score reliability.
• Applying BMS outside validated cohorts (neonates, immunocompromised patients, VP shunts).

• Clinical instability (shock, altered mental status, focal neurologic deficits) warrants aggressive management regardless of calculated band.
• Traumatic taps or delayed LPs reduce reliability; escalate to neurology/ID and rely on full clinical picture.
• Do not delay antimicrobials for scoring; ensure timely LP when safe and within local timing guidance.
• Immunocompromised patients or those with ventriculoperitoneal shunts are outside the original derivation—treat separately and escalate early.
• Persistent CSF leak, cochlear implants, or recent neurosurgery increase risk and supersede the low-risk band.

• Low risk (0 predictors): Discuss observation parameters, ensure reliable reassessment, and tailor further testing to clinical context; a changing exam should prompt re-evaluation regardless of initial band.
• High risk (≥1 predictor): Manage as bacterial meningitis until proven otherwise. Do not use the score to withhold therapy.
• Repeat neurologic exams, vitals, and review labs as they result; align with institutional pathways for admission, imaging, and consults.
• When viral panel/PCR results become available, reframe risk and antibiotic course; many centers still complete initial IV therapy until cultures are finalized.
• Document decision-making in the note (predictors present/absent, timing of antibiotics vs LP, communication with caregivers, and planned reassessment checkpoints).

• Age, immunization status, and comorbid risk factors (shunt, immunocompromise).
• LP timing, antibiotics administered, and traumatic tap details.
• Five predictors with measured values and units.
• Risk band decision, consults, and observation or admission plan.

• Neonates <60 days: not validated; use age-specific sepsis/meningitis pathways and broader empiric coverage.
• Immunocompromised or VP shunts: higher baseline risk and atypical organisms; escalate to ID/neurosurgery, consider hardware imaging and expanded cultures.
• Traumatic tap: red-cell correction formulas are imperfect; consider parallel tube counts and clinical trajectory rather than over-relying on corrected ANC.
• Partially treated meningitis: may show near-normal CSF glucose/protein and negative Gram stain; historical antibiotic timing is important to interpretation.
• Vaccine-era changes: PCV/Hib reductions shift pretest probability; some centers pair BMS with CSF PCR to refine risk.

When in doubt, anchor to illness severity and consult early. The score should never delay antibiotics or imaging in a decompensating child.

• Neurologic decline, hemodynamic instability, or focal deficits regardless of initial BMS category.
• Uncertain or confounded CSF profile (traumatic tap, partial antibiotics, delayed LP).
• High-risk host factors outside derivation cohorts (immunocompromise, VP shunt, neurosurgical history).
• Discordance between score and bedside course over serial reassessments.

• BMS predictor status with source values and collection times.
• Antibiotic timing relative to LP and current antimicrobial plan.
• Trend summary for neuro exam, vitals, and lab trajectory.
• Consult status (ID, neurology, neurosurgery/critical care) and outstanding actions.
• Explicit next reassessment time and escalation trigger.

• Explain that the score helps frame risk but does not replace cultures, serial exams, or clinician judgment.
• Set expectations for monitoring (vitals, neuro checks), timing of culture/PCR results, and when therapy might narrow or stop.
• Provide clear return precautions if observation or step-down care is chosen: new fever, vomiting, photophobia, stiff neck, lethargy, or seizures.
• Align on pain/comfort plans for serial LP attempts and discuss rationale for early antibiotics if instability or high-risk features are present.

Document shared decision-making, caregiver concerns, and any language/cultural needs. Rapid updates when labs finalize build trust and reduce premature discharge.

• Pediatric Vital Signs for monitoring trends during evaluation.
• Pediatric Blood Volume when planning resuscitation and labs.
• Attia Pediatric Pharyngitis Predictor and Centor/McIsaac for upper-airway infectious differentials.
• Pediatric Transfusion Volume when ongoing blood loss or hemolysis is a concern.
• Pediatric eGFR to support antimicrobial dosing and renal monitoring.

The original BMS derivation and subsequent external validations were conducted in children with CSF pleocytosis, largely before widespread PCV13 uptake. While the low-risk cohort demonstrated very low rates of culture-proven bacterial meningitis, the score was not powered to address atypical organisms, immune compromise, or shunt-associated infections. Many modern pathways therefore pair the score with molecular panels, rapid Gram stain interpretation, and strict return/observation criteria.

Specificity remains limited—any single positive predictor triggers high risk—so the tool is intended to safely identify children who may be observed rather than to distinguish viral from bacterial in isolation. If your site has more recent epidemiology, work with stewardship/ID to recalibrate cut points or to embed the score as a documentation aide rather than a binary decision-maker.

Use judgment where the BMS does not apply: neonates, postoperative states, VP shunts, immunosuppression, meningoencephalitis presentations, and traumatic taps. When labs, exam, or course diverge from expectations, prioritize serial exams and consulting teams over calculated bands.

• Nigrovic LE et al. JAMA. 2007;297(1):52–60.
• Nigrovic LE et al. Pediatrics. 2002;110(4):712-719.