Test characteristics and interpretation of cerebrospinal fluid gram stain in children

Few data exist regarding the test characteristics of cerebrospinal fluid (CSF) Gram stain among children at risk for bacterial meningitis, especially the rate of false positive Gram stain.

We conducted a retrospective cohort study of children seen in the emergency department of Children's Hospital Boston who had CSF obtained between December 1992 and September 2005. Patients who had ventricular shunts, as well as those who received antibiotics before CSF was obtained were excluded. Test characteristics of CSF Gram stain were assessed using CSF culture as the criterion standard. Patients were considered to have bacterial meningitis if there was either: (1) growth of a pathogen, or (2) growth of a possible pathogen noted on the final CSF culture report and the patient was treated with a course of parenteral antibiotics for 7 days or more without other indication.

A total of 17,569 eligible CSF specimens were collected among 16,036 patients during the 13-year study period. The median age of study subjects was 74 days. Seventy CSF specimens (0.4%) had organisms detected on Gram stain. The overall sensitivity of Gram stain to detect bacterial meningitis was 67% [42 of 63; 95% confidence interval (CI): 54-78] with a positive predictive value of 60% (42 of 70; 95% CI: 48-71). Most patients without bacterial meningitis have negative Gram stain [specificity 99.9% (17,478 of 17,506; 95% CI: 99.8-99.9)] with a negative predictive value of 99.9 (17,478 of 17,499; 95% CI: 99.8-99.9).

CSF Gram stain is appropriately used by physicians in risk stratification for the diagnosis and empiric treatment of bacterial meningitis in children. Although a positive Gram stain result greatly increases the likelihood of bacterial meningitis; the result may be because of contamination or misinterpretation in 40% of cases and should not, by itself, result in a full treatment course for bacterial meningitis.