Diagnostic Stewardship in the Evaluation of a Child With a Potential Central Nervous System Infection

OVERVIEW Bacterial meningitis can cause significant morbidity and mortality, especially when treatment initiation is delayed. Young children with bacterial meningitis, especially infants, frequently present with nonspecific symptoms, making timely diagnosis challenging.1 The overall incidence of bacterial meningitis has been declining over the past 2 decades due to the widespread uptake of the conjugate vaccinations for pneumococcus and meningococcus, the most common causes of bacterial meningitis beyond the neonatal period.2 This clinical review will provide an evidence-based approach to diagnostic evaluation of immunocompetent children with potential central nervous system (CNS) infections, including bacterial meningitis, with the goal to identify treatable causes without over-testing low-risk children (Case, Supplemental Digital Content 1, https://links.lww.com/INF/F478). DIAGNOSTIC EVALUATION Physical Examination When evaluating a child with a potential CNS infection, physical examination has a crucial role in the identification of children at risk. Clinical symptoms of meningeal irritation (ie, meningismus) include headache, nuchal rigidity (neck stiffness) and photophobia. The Kernig and Brudzinski signs provide physical examination evidence of meningeal irrigation but may be absent early in infection. In 2 systematic reviews of signs,3,4 the Kernig's sign positive likelihood ratio (LR) range 3.5–10.3, Brudzinski sign (positive LR 2.3–2.5) and any nuchal rigidity (positive LR 1.4–13.3) increased the likelihood while the absence of clinical signs of meningism lowered the likelihood of bacterial meningitis (negative LR 0.3–0.57).3 Blood Tests Routine blood testing should include a white blood cell count (WBC) with differential as well as inflammatory markers (C-reactive protein and procalcitonin).5 Blood cultures should be routinely obtained as the causative bacterial pathogencan frequently be isolated from the blood, especially in the youngest children.6 Lumbar Puncture A lumbar puncture (LP) is required to confirm the diagnosis of bacterial meningitis, but antibiotics should be not delayed when LP performance is delayed due to clinical instability, suspected elected intracranial pressure or overlying skin infection.1 Given the declining incidence of bacterial meningitis, the decision to perform an LP frequently involves shared decision-making with the patient and caregivers to balance the risk of bacterial meningitis against the inconvenience, discomfort and risks of the procedure. Cerebrospinal Tests Routine cerebrospinal fluid (CSF) tests include WBC, red blood cell count, glucose and protein. Traumatic LPs occur commonly, especially in young infants,6 introducing peripheral blood into the CSF and complicating the interpretation of results. After a traumatic LP, the CSF WBC count can be corrected by subtracting the CSF red blood cell count divided by 1000 from the measured CSF WBC count.7 Although bacterial cultures take several days to return results, CSF Gram stain can detect bacteria with results available in hours. Although the specificity of CSF gram stain is high 99.9%, 95% confidence interval (CI): 99.8–99.9%, the sensitivity is modest (67%, 95% CI: 54–78%) with a positive predictive value of 60% (95% CI: 48–71%).8 A negative CSF Gram stain results should not be used in isolation to exclude bacterial meningitis. DISTINGUISHING VIRAL FROM BACTERIAL MENINGITIS Children with bacterial meningitis require prompt initiation of parenteral antibiotics, while those with viral meningitis need supportive care. For children with meningitis, available multivariate meningitis clinical prediction rules combine clinical and laboratory factors to assist emergency department decision-making. Bacterial Meningitis Score The Bacterial Meningitis Score is a highly accurate clinical decision rule that identifies children at very low risk for bacterial meningitis with none of the 5 high-risk clinical predictors (Table 1).9 Low-risk children who are otherwise well-appearing may not require inpatient care or parenteral antibiotics. The Bacterial Meningitis Score has been evaluated in a wide variety of clinical settings. In a patient-level meta-analysis that aggregated data from 8 published studies that included more than 5000 children,11 the Bacterial Meningitis Score accurately identified children with bacterial meningitis (sensitivity 99.3%, 95% CI: 98.7–99.7%; specificity 62.1%, 95% CI: 60.5–63.7% and negative predictive value 99.7%, 95% CI: 99.3–99.9%]. TABLE 1. - High-risk Clinical Predictors for Two Clinical Prediction Rules to Distinguish Bacterial vs. Viral Meningitis Bacterial Meningitis Score9 Meningitis Score for Emergencies10 Positive CSF Gram StainCSF ANC ≥ 1,000 cells/µLCSF protein ≥ 80 mg/dLPeripheral blood ANC ≥1000 cells/µLSeizure at or prior to presentation Serum procalcitonin >1.20 ng/LSerum CRP >4.0 mg/dLCSF ANC >1000 cells/µLCSF protein ≥80 mg/dL ANC, absolute neutrophil count; CRP, C-reactive protein; CSF, cerebrospinal fluid. Meningitis Score for Emergencies The Meningitis Score for Emergencies is a more recent clinical prediction rule that identifies children at low risk for bacterial meningitis who have none of the 4 high-risk laboratory predictors (Table 1).10 The rule was derived in a 25-center retrospective cohort of children diagnosed with meningitis over a 5-year period and validated in a prospective cohort over 2 years performed at the same institutions. The Meningitis Score for Emergencies accurately identifies children with bacterial meningitis (sensitivity 100%, 95% CI: 95.0–100%; specificity 83.2%, 95% CI: 80.6–85.5% and negative predictive value 100%, 95% CI: 99.4–100%). In a large retrospective validation of children with meningitis treated at one of the 184 Hospital Corporation of America Healthcare hospitals, the Meningitis Score for Emergencies missed 2 of 28 children with bacterial meningitis (sensitivity 92.9%, 95% CI: 83.3–100%; specificity 65.2%, 95% CI: 51.5–79% and negative predictive value 93.4%, 95% CI: 85.4–100%).12 The Meningitis Score for Emergencies includes newer biomarkers (ie, C-reactive protein and procalcitonin) and appears promising. However, additional validation studies are needed to demonstrate the accuracy in a broad range of clinical settings to ensure children with bacterial meningitis will not be misclassified. Polymerase Chain Reaction Tests CSF polymerase chain reaction (PCR) tests can identify bacterial or viral pathogens more quickly than traditional bacterial cultures. Rapid diagnosis of enteroviral meningitis has been associated with shorter duration of parenteral antibiotics and hospitalization.13 CNS herpes simplex virus (HSV) infections are uncommon but cause substantial morbidity and mortality without appropriate antiviral treatment. In infants ≤60 days of age who had an LP performed for evaluation of infection, only 0.4% (95% CI: 0.4–0.5%)14 had an HSV infection, with a peak in the second week of life. The following high-risk predictors accurately identified infants ≤60 days of age with an invasive HSV infection: age <14 days, prematurity (<37 weeks gestation), seizure, ill-appearance, abnormal temperature (≥38.0 °C or <36.4 °C), vesicular rash, platelet count <150,000 cells/mm3 and CSF pleocytosis (CSF WBC ≥15 cells/mm3 <28 days and ≥10 cells/mm3 for 29–60 days).15 Low-risk infants may not require HSV testing or empiric treatment: sensitivity of 95.6% (95% CI: 84.9–99.5%), specificity of 40.1% (95% CI: 36.8–43.6%) and negative predictive value of 0.11 (95% CI: 0.03–0.43). Outside the neonatal period, HSV can rarely cause aseptic meningitis or meningoencephalitis associated with a primary rather than reactivation infection. The commercially available Biofire PCR panel simultaneously tests CSF for 6 bacteria (Escherichia coli, Haemophilus influenzae, Listeria monocytogenes, Neisseria meningitidis, Streptococcus agalactiae and S. pneumoniae), 7 viruses (cytomegalovirus, enterovirus, HSV 1/2, human herpes virus 6, human parechovirus and varicella zoster virus) and 1 fungus (Cryptococcus) with results available in as little as 1 hour. Test implementation decreased the time-to-optimal antimicrobials (from 28 hours to 18 hours) for children with bacterial meningitis16 and shortened the duration of antibiotics for children with viral meningitis.17 Of 460 children undergoing evaluation of CNS infection, 12 had a positive Biofire panel for a bacterial pathogen not identified by culture.17 Depending on the clinical scenario, discordant results could be a false negative bacterial culture (potentially due to antibiotic pretreatment) or a false positive PCR result. Approach to Child with Meningitis For children with meningitis, validated meningitis clinical prediction rules (ie, Bacterial Meningitis Score or the Meningitis Score for Emergencies) identify children who can be safely managed outpatient without further testing. Children with one or more high-risk predictors should have a CSF multiplex PCR test obtained while awaiting bacterial culture results.18,19 Additional studies are needed to understand the real-world impact of multiplex CSF PCR tests. However, in a single-center study, the Biofire assay for infants ≤2 months with meningitis increased the costs of microbiologic testing without increasing overall costs.16 Regional Considerations In Lyme disease endemic areas (eg, Northeast, Mid-Atlantic and upper Midwest regions of the United States, as well as Northern Europe), Lyme meningitis can present similarly to aseptic meningitis. The diagnosis requires positive 2-tier Lyme disease serology: a first-tier enzyme immunoassay followed by an IgG/IgM immunoblot standard 2-tier test (STTT) or a second enzyme immunoassay modified two-tier test (MTTT). MTTT tests have higher sensitivity in early disease, but similar specificity when compared with STTT tests.20 Neither CSF Lyme PCR testing nor ratio of intrathecal to peripheral Borrelia antibodies have sufficient sensitivity to replace peripheral serology,21,22 and should not be routinely obtained. The rule of 7's is a clinical prediction rule that identifies children at low risk of Lyme meningitis with the absence of 3 high-risk predictors: ≥7 days of headache, ≥70% CSF mononuclear cells (lymphocytes plus monocytes) and seventh or other cranial neuritis. In a patient-level meta-analysis, the rule of 7's had a sensitivity of 98% (95% CI: 89–100%), specificity of 40% (95% CI: 30–50%) and a negative predictive value of 100% (95% CI: 95–100%).23 CONCLUSIONS Clinicians should take a stepwise approach to the diagnostic evaluation of children with suspected CNS infections, with a focus on treatable infections. In the era of widespread conjugate vaccinations against H. influenzae, S. pneumoniae and N. meningitidis, bacterial meningitis has become uncommon but continues to have significant morbidity and mortality. Bacterial meningitis clinical prediction rules (both the Bacterial Meningitis Score and the Meningitis Score for Emergencies) combine peripheral and biomarkers to identify children at a very low risk for bacterial meningitis. Multiplex CSF PCR panels can shorten the time to optimal antibiotics for children with bacterial meningitis and reduce unnecessary treatments for those with viral meningitis. Additional studies are needed to understand the real-world impact of multiplex PCR assays compared to available pediatric meningitis clinical prediction rules.