Calc Function

    • Calcs that help predict probability of a diseaseDiagnosis
    • Subcategory of 'Diagnosis' designed to be very sensitiveRule Out
    • Disease is diagnosed: prognosticate to guide treatmentPrognosis
    • Numerical inputs and outputsFormula
    • Med treatment and moreTreatment
    • Suggested protocolsAlgorithm





    Chief Complaint


    Organ System


    Patent Pending

    Bacterial Meningitis Score for Children

    Rules out bacterial meningitis.


    Use in patients aged 29 days to 19 years with CSF WBC ≥10 cells/μL. Do not use if patient is critically ill, recently received antibiotics, has a VP shunt or recent neurosurgery, is immunosuppressed, or has other bacterial infection requiring antibiotics (including Lyme disease).

    When to Use
    Why Use

    Pediatric patients (aged 29 days to 19 years) with suspected meningitis.

    Do NOT use if the patient:

    • Is critically ill, requiring respiratory or vasopressor support.
    • Received antibiotics <72 hours prior to lumbar puncture.
    • Has a VP shunt or recent neurosurgery.
    • Is immunosuppressed.
    • Has proof of another bacterial infection (e.g. UTI, bone infection, known bacteremia) that warrants inpatient antibiotic therapy.
    • Has known active Lyme Disease.
    • The Bacterial Meningitis Score (BMS) predicts bacterial vs. aseptic etiology in pediatric patients (aged 29 days to 19 years) with suspected meningitis.  
    • Can help determine if the patient will require admission for parenteral antibiotics while awaiting CSF culture results.
    • Higher score indicates higher likelihood of bacterial meningitis.
    • Sensitivity, specificity, and negative predictive value of the BMS decrease significantly for children under the age of 2 months.
    • Creators of the BMS advise against using the score in children who have already received antibiotics prior to LP, are ill-appearing, are <2 months of age, or have exam findings indicative of invasive bacterial infection such as petechiae and purpura.
    • Not effective at ruling out potentially harmful nervous system infections requiring antibiotics (e.g. herpes encephalitis, Lyme meningitis, tuberculous meningitis).
    • Meningococcal meningitis can present without CSF pleocytosis; thus, these patients can be misclassified as not having inclusion criteria for the use of the BMS. It is important to perform a thorough physical exam to assess for petechiae or purpura if there is suspicion for meningococcemia or meningitis, as CSF may be falsely normal.
    • Bacterial meningitis incidence has dramatically decreased since the advent of highly effective vaccines against the more common causes (H. flu, S. pneumo), making it more challenging to determine which patients should be admitted and observed while awaiting CSF culture results.
    • Helps stratify which patients do not necessarily require observation, due to higher likelihood of aseptic (i.e., spontaneously resolving) meningitis.
    • Helps avoid financial burden and health risk associated with hospitalization for observation and parenteral antibiotic administration.


    Please fill out required fields.

    Next Steps
    Creator Insights


    For patients at very low risk for bacterial meningitis (BMS 0):

    • Consider discharge with close follow-up (ideally within 24–48 hours) and return precautions for family, including new seizure activity, altered mental status, purpuric rash, or other concerning symptoms.
    • Patients may have received a dose of empiric antibiotics after LP was performed if concern for bacterial meningitis. If no antibiotics were administered, consider a single dose of long-acting antibiotics with good CSF penetration, such as ceftriaxone, prior to discharge.  

    For patients with at least 1 risk factor for bacterial meningitis or high clinical suspicion (BMS >0):

    • Consider admission for parenteral antibiotics and observation while awaiting CSF culture results.
    • Make sure CSF is sent for culture.
    • Consider continuous monitoring of vital signs and regular neurologic exams.
    • If not previously administered, start empiric broad spectrum antibiotics.
    • Consider expanding antimicrobial coverage:
      • If concern for herpes encephalitis, add acyclovir.
      • If high clinical suspicion for tuberculous meningitis, consult with infectious disease specialist and consider rifampin, isoniazid, pyrazinamide, and a fluoroquinolone or aminoglycoside.
    • Consider steroid administration based on clinical presentation, geographic area, and potential risk factors.

    Critical Actions

    • Physician gestalt, severity of illness and clinical presentation supersedes the application of the BMS prediction rule.
    • If significant suspicion for bacterial meningitis, err on the side of caution and admit for observation and empiric antibiotics.


    Addition of the selected points:



    0 points

    1 point

    CSF Gram stain



    CSF absolute neutrophil count (ANC)

    <1,000 cells/μL

    ≥1,000 cells/µL

    CSF protein

    <80 mg/dL (800 mg/L)

    ≥80 mg/dL (800 mg/L)

    Peripheral blood ANC

    <10,000 cells/μL

    ≥10,000 cells/μL

    Seizure at (or prior to) initial presentation



    Facts & Figures


    Bacterial Meningitis Score

    Risk for Bacterial Meningitis


    Very low risk


    Not very low risk

    Evidence Appraisal

    The original Bacterial Meningitis Score was derived from a multicenter, retrospective cohort study published by Nigrovic et al in JAMA 2007.  Data were collected from 20 participating emergency departments of academic medical centers over a three year period. 3,295 patients aged 29 days to 19 years with CSF pleocytosis were scored using the BMS.

    Of the 1,714 who were categorized as very low risk, two were found to have bacterial meningitis. Both miscategorized patients were <2 months of age and had E. coli meningitis with an E. coli UTI but negative urinalysis.

    Sensitivity of BMS for bacterial meningitis was 98.3% (95% CI 94.2-99.8%). NPV was 99.9% (95% CI 99.6-100%). The investigators attempted to refine the score using recursive partitioning, which led to a simpler model with only three variables, but it also led to one additional patient with meningitis being misclassified as very low risk.

    Given the two misclassified patients were under the age of two months, the investigators analyzed the BMS for a subgroup of all patients under two months of age and found sensitivity was 92.3% (95% CI 74.9-99.4%), NPV 99.5% (95% CI 98.3-99.9%).

    The BMS was validated by Nigrovic et al in Archive of Disease in Childhood, 2012. This was a meta analysis of studies published between 2002 and 2012 and included 4,896 patients aged 29 days to 19 years. Sensitivity was 99.3% (95% CI 98.7-99.7%) for bacterial meningitis and NPV was 98.3% (95% CI 96.6-99.3%).

    Kulik et al in 2013 published a systematic review of several bacterial meningitis predictive rules, and of the studies reviewed, the BMS had the highest quality evidence and the best performance to date. They recommended that the score still be further evaluated with prospective trials.


    Dr. Lise Nigrovic

    From the Creator

    Why did you develop the Bacterial Meningitis Score? Was there a particular clinical experience or patient encounter that inspired you to create this tool for clinicians?

    When I was a resident in pediatrics, I found myself admitting well-appearing children who had meningitis, spent two days in the hospital, and then went home after their cultures were negative. Doing that over and over again made me wonder if we could do better and distinguish between children who have viral infections, which require only supportive care and are managed much better at home, and those who have the rare but serious bacterial meningitis. This question launched both the Bacterial Meningitis Score as well as my career in clinical research.  


    In this day and age, when hospital administrators are concerned with “unnecessary” admissions, this kind of score can be really helpful in stratifying those patients.

    Yes, of course we don’t want to spend unnecessary health-care dollars, but it’s also a major inconvenience for families, especially those who have other children at home, and the parents have to miss work. We want to reduce non-value-added care—hospitalization and antibiotics for a child with a viral infection doesn’t help them get better any more quickly.


    What pearls, pitfalls and/or tips do you have for users of the Bacterial Meningitis Score? Are there cases when it has been applied, interpreted, or used inappropriately?

    Clinical prediction rules are meant to assist judgment, not replace it. So they give you a risk estimate based on our best available evidence. But fundamentally, clinicians have to use their best judgment. Bacterial meningitis is a bad disease. The children who were misclassified by the Bacterial Meningitis Score were small babies with petechial rashes. So, I would be particularly cautious about applying the Score to babies under two months, especially those with a petechial rash.


    A long-acting parenteral antibiotic like ceftriaxone can further mitigate the risk. Even if the patient is low risk, I would strongly consider a single dose of antibiotics prior to discharge while you wait for bacterial cultures.


    We’ve always stressed that prediction tools shouldn’t be used in cases where you’re already pretty sure of the diagnosis clinically; they should be used to help you if you’re not sure.  

    Before applying a clinical prediction rule, I would consider the seriousness of the disease in question.  When you have a really deadly disease, like bacterial meningitis, you need that low risk group to be very low. The risk of bacterial meningitis in the low risk group of the Bacterial Meningitis Score is 1 in 1,000, which is very low, but it’s not zero, so if there’s uncertainty, there’s a strong argument for erring on the side of caution. If a rule is for low-risk ankle sprain, and somebody has delay in diagnosis of a nondisplaced fracture, it’s arguably not as serious as missing bacterial meningitis. Riley in Annals of Internal Medicine talks about tools assisting versus directing clinicians, and that’s how I think about the framing of clinical prediction rules in general. They shouldn’t say “you MUST do this,” but instead, “here’s a risk estimate,” and then ultimately, it’s up to the clinician and their experience and judgment.


    What recommendations do you have for doctors once they have applied the Bacterial Meningitis Score? Are there any adjustments or updates you would make to the score given recent changes in medicine?

    The countervailing thing to recognize is that vaccines have widespread uptake—we’ve moved from PCV7 to PCV13, and meningococcal vaccines, so the incidence of bacterial meningitis has gone down. The pre-test probability is declining, and knowing that, the negative predictive value is likely to be even better as the prevalence goes down.


    If the patient is otherwise well and the score is low risk with no factors, then I think that’s a very appropriate patient for outpatient management. If the patient has one or more factors, then 24 hours in the hospital and antibiotics is a reasonable approach. Some of those children may also be candidates for outpatient management in a shared decision making model, but I think the easiest interpretation would be to say those patients should be considered for inpatient management with parenteral antibiotics while awaiting preliminary culture results.


    The other thing is, we usually wait 48 hours for culture results to exclude bacterial growth, but the vast majority of cultures grow within the first 24 hours, so even if they had one factor but the cultures were negative, they might not have to stay two days.


    How do you use the Bacterial Meningitis Score in your own clinical practice? Can you give an example of a scenario in which you use it?

    If I identify meningitis in a patient and I think they’re low risk for bacterial disease, I feel comfortable giving them a dose of ceftriaxone and letting them go home. We’re a little bit confounded by being here in New England because many of the children who don’t look like they have bacterial meningitis have Lyme meningitis. And that’s a whole other kettle of worms, whether they need to be admitted to the hospital or not. So sometimes we think they’re low risk for bacterial meningitis, but we hospitalize them while awaiting the results of Lyme titers, which also take two days, like the culture results. It confuses things a little bit, but that’s a regional issue. (Editor’s Note: MDCalc is developing a calc for the Rule of 7s for Lyme Meningitis, derived by Dr. Aris Garro et al and validated in a large study by Dr. Nigrovic et al.)


    Does a history of febrile seizures or epilepsy alter the value of the prediction rule (which includes seizure at onset as a risk factor)?

    Using my clinician hat, a patient who has a seizure in the setting of fever and also has a history of seizures is much less concerning than a patient who’s never had a seizure before, but we don’t have the power to examine that, so the scientist in me says I can’t comment! I can’t scientifically say that my work has sorted that out.


    What are your thoughts on antibiotic overuse and misuse?

    I think it’s important to be judicious in the use of antibiotics. We want to have effective antibiotics for serious bacterial infections like meningitis. From a public health standpoint, though, the overtreatment of viral meningitis is probably not the major driver of bacterial resistance, though every little bit contributes—pharyngitis and otitis are much bigger-ticket items in terms of antibiotic prescriptions. I wrote another paper in Pediatrics where we just looked at the number of admissions to pediatric hospitals, and this is not the driver of resistance. But that being said, not everybody needs antibiotics, so if we can identify them, we can save unnecessary use of the big-gun antibiotics, like ceftriaxone, so that they’ll work when we really need them to.

    About the Creator

    Lise E. Nigrovic, MD, MPH, is an associate professor of pediatrics and emergency medicine at Harvard Medical School. She serves as co-director of Population Science for the Institutional Centers for Clinical and Translational Research (ICCTR) and the Boston Children’s Hospital Medical Research Office for Harvard Catalyst. Dr. Nigrovic’s research focus has been in the approach to diagnosis and management of children with infectious and traumatic emergencies.

    To view Dr. Lise Nigrovic's publications, visit PubMed

    Content Contributors
    About the Creator
    Dr. Lise Nigrovic
    Content Contributors