Investigators from multiple institutions conducted a retrospective study to assess the additional utility of measuring cerebrospinal fluid (CSF) protein and glucose levels in diagnosing infectious meningitis (IM), compared to CSF white blood cell count alone, in an era of declining incidence of bacterial meningitis and with the availability of multiplex PCR assays. For the study, the laboratory database at a children’s hospital in the US was reviewed to identify all patients <18 years old on whom CSF studies were performed for evaluation of IM between October 2015 and August 2017. PCR testing for enterovirus and parechovirus was available throughout the study period; multiplex testing for 14 viral, bacterial, and fungal pathogens was available beginning in January 2017. Study participants were classified as having confirmed IM if culture and/or PCR testing were positive, clinical IM if the discharge diagnosis was meningitis despite no microbiologic confirmation, or no IM. CSF WBC counts >5/cm2, CSF protein levels >80 mg/dL, and CSF glucose values <20 mg/dL were classified as abnormal. The sensitivity, specificity, and positive and negative predictive values (PPV and NPV, respectively) of each test as a predictor of confirmed IM was determined. In addition, the sensitivity, specificity, PPV, and NPV of the addition of CSF protein, or CSF glucose, levels to CSF WBC counts were calculated and compared to CSF WBC alone as a predictor of IM using McNemar tests.
Data were analyzed on 735 patients with CSF studies for evaluation of IM. Of these, 446 children (60.7%) were <2 months old. A total of 45 patients (6.1%) were diagnosed with confirmed IM, including 12 with bacterial meningitis, and 24 (3.3%) were diagnosed with clinical IM. There were 5 cases of viral meningitis (all caused by enterovirus or parechovirus) identified by PCR in which CSF WBC count and CSF protein and CSF glucose levels were normal. Overall, the sensitivity, specificity, PPV, and NPV of CSF WBC count as a predictor of IM were 88.9%, 63.5%, 13.7%, and 98.9%, respectively. Similar values for CSF protein levels were 44.4%, 74.6%, 10.3, and 95.4%, respectively, and for CSF glucose 13.3%, 99.7%, 75.0%, and 94.6%, respectively. The addition of CSF glucose to CSF WBC did not change the test characteristics compared to CSF WBC alone. However, when CSF protein was added to CSF WBC, the specificity (54.1%) and PPV (11.2%) were lower compared to when CSF WBC alone was assessed, and the inclusion of CSF protein significantly decreased the overall accuracy of testing (P <0.0001). Three cases of bacterial meningitis were caused by pathogens not included in PCR testing; in each case the CSF WBC count was abnormal, thus all cases of bacterial meningitis were identified by an abnormal CSF WBC count and/or PCR testing.
The authors conclude the...