brand logo

Am Fam Physician. 2007;75(4):533-535

Clinical Question

Is it possible to predict the likelihood of bacterial meningitis in children with suspected meningitis?

Evidence Summary

Although the incidence of bacterial meningitis has dropped with increased immunization against Haemophilus influenzae, meningitis remains a devastating illness. For every child with meningeal signs and symptoms who is diagnosed with bacterial meningitis after lumbar puncture, many more undergo the procedure to rule out the disease. Therefore, researchers have attempted to identify predictors of bacterial meningitis to allow physicians to estimate risk and avoid unnecessary testing while correctly identifying children who need urgent intravenous antibiotics.

Clinical decision rules for diagnosing bacterial meningitis in children vary in several ways (Table 1).110 Some include complex multivariate models that require the use of a computer and do not allow the physician to easily see why a particular prediction is made. Others use a simple scoring system that assigns points to certain signs and symptoms and are easier to use at the point of care. Some rules require cerebrospinal fluid (CSF) findings, making them unsuitable for determining whether lumbar puncture should be performed. In addition, the rules have been prospectively validated to differing degrees.

StudyAge group studiedProspectively validatedClinical decision ruleCSF findings required
Lindquist, et al., 19881 Children and adultsNoMultivariate modelYes
Spanos, et al., 19892 Children and adultsNoMultivariate modelYes
Hoen, et al., 19953 3.5 years or youngerYesMultivariate modelYes
Jaeger, et al., 20004
Freedman, et al., 20015 2 months to 17 yearsYesPoint scoreYes
Nigrovic, et al., 20026 1 month to 19 yearsYesPoint scoreYes
Oostenbrink, et al., 20027 1 month to 15 yearsYesPoint scoreNo (includes follow-up score that requires CSF findings)
Oostenbrink, et al., 20048
Bonsu, et al., 20049 1 month to 18 yearsYesMultivariate modelYes
Nigrovic, et al., 200710 1 month to 19 yearsYesPoint scoreYes

The Oostenbrink rule (Table 211 ) is potentially the most useful for physicians because it is simple and can guide decisions about performing a lumbar puncture and about treating patients empirically with antibiotics.7,8 When originally developed, the rule stated that patients with a score of 9.5 or more should have a lumbar puncture.7 A prospective validation study in four Dutch hospitals applied the rule to 226 children.8 The study showed that two out of 25 children with bacterial meningitis had a score less than 9.5; the diagnosis would have been missed in these patients. One of the two children had meningeal signs and two protracted seizures, whereas the other was a 34-day-old premature infant who probably did not meet the minimum age cutoff of one month.8 The authors then used a lower cutoff of 8.5 points and applied it to the original study population (360 children) and the validation study population (226 children). None of the 205 children (95% CI, 0 to 2) with a score less than 8.5 points had bacterial meningitis; 13 percent (95% CI, 9 to 17) with a score of 8.5 to 14.9 had the disease; 52 percent (95% CI, 39 to 65) with a score of 15.0 to 19.9 had the disease; and 87 percent (95% CI, 79 to 95) with a score of 20.0 or more had the disease.8

Risk factorPoints
Duration of main problem in patient history1.0 per day (maximum 10)
History of vomiting2.0
Physical examination findings:
Disturbed consciousness8.0
Meningeal irritation*7.5
Serum C-reactive protein level, mg per dL (mg per L)
< 5.0 (50)0
5.0 to 9.9 (50 to 99)0.5
10.0 to 14.9 (100 to 149)1.0
15.0 to 19.9 (150 to 199)1.5
≥ 20.0 (200)2.0

A recent study attempted to validate five clinical decision rules3,57,9 using a retrospective database.12 The study showed that the Nigrovic,6 Freedman,5 and Bonsu9 rules had 100 percent sensitivity for detecting children with bacterial meningitis, although the Freedman rule5 was only 12 percent specific.12 The authors conclude that the Nigrovic rule6 (Table 36,10,12 ) is preferred because it is easy to use and has a good specificity of 66 percent. Two out of 12 children with a score less than 8.5 using the Oostenbrink rule had bacterial meningitis.12 However, retrospective chart reviews, such as the Oostenbrink rule, typically underestimate the accuracy of clinical decision rules because consistent documentation of symptoms in medical records is unreliable.

Risk factor
Positive CSF Gram stain
CSF protein level ≥ 80 mg per dL (800 mg per L)
Peripheral absolute neutrophil count ≥ 10,000 cells per mm3 (10 × 109 per L)
Seizure at or before presentation
CSF absolute neutrophil count ≥ 1,000 cells per mm3 (1 × 109 per L)

The Nigrovic and Oostenbrink rules provide the best balance between prospective validation and ease of use. These rules should not be applied strictly. For example, seizure is relatively rare and does not meet statistical criteria for inclusion in the Oostenbrink rule; nevertheless, any child with suspected meningitis and seizure should undergo lumbar puncture. All children with suspected meningitis should have close follow-up, whether or not they have received lumbar puncture or empiric treatment.

Applying the Evidence

A four-year-old boy with signs of meningeal irritation has been sick for two days. He has vomited twice but does not have cyanosis, disturbed consciousness, or petechiae. His serum C-reactive protein level is 1.0 mg per dL (10 mg per L). Should he undergo lumbar puncture? If so, how should the results be interpreted?

Answer: Using the Oostenbrink rule (Table 211 ), the patient receives 11.5 points (2 points for duration of illness, 2 for vomiting, and 7.5 for meningeal irritation) and, therefore, should undergo lumbar puncture. The lumbar puncture findings include a negative CSF Gram stain, a CSF protein level of 60 mg per dL (600 mg per L; normal range: 10 to 40 mg per dL [100 to 400 mg per L]), a peripheral absolute neutrophil count of 6,500 cells per mm3 (6.5 × 109 per L; normal range: 4,000 to 10,000 cells per mm3 [4 to 10 × 109 per L]), and a CSF absolute neutrophil count of 1,400 cells per mm3 (1.4 ×109 per L; normal range: 0 to 5 cells per mm3 [0 to 0.005 × 109 per L]). Therefore, the patient has a significant risk of bacterial meningitis based on the Nigrovic validation studies (Table 36,10,12 ) and should receive empiric treatment while awaiting culture results.

This guide is one in a series that offers evidence-based tools to assist family physicians in improving their decision-making at the point of care.

This series is coordinated by Mark H. Ebell, MD, MS, deputy editor for evidence-based medicine.

A collection of Point-of-Care Guides published in AFP is available at

Continue Reading

More in AFP

More in PubMed

Copyright © 2007 by the American Academy of Family Physicians.

This content is owned by the AAFP. A person viewing it online may make one printout of the material and may use that printout only for his or her personal, non-commercial reference. This material may not otherwise be downloaded, copied, printed, stored, transmitted or reproduced in any medium, whether now known or later invented, except as authorized in writing by the AAFP.  See permissions for copyright questions and/or permission requests.