Interpretation of traumatic lumbar punctures: who can go home?

OBJECTIVE: To determine whether a ratio of observed to predicted (O:P) cerebrospinal fluid (CSF) white blood cells (WBCs) after a traumatic lumbar puncture (LP) can be used to predict which patients do not have meningitis and can safely be discharged from the hospital.
METHODS: A retrospective medical record review was performed on 2 cohorts of previously healthy children who had received an LP at Children's Memorial Hospital in Chicago, IL. All children were older than 1 month and had a red blood cell (RBC) count in the CSF >500/mm(3). Cohort 1 consisted of children who were examined in 1990 through 1999 and had CSF cultures positive for a bacterial pathogen. Cohort 2 consisted of children who were tested during January through December 1999 and had a CSF culture negative for any bacterial pathogen. Exclusion criteria included patients who received antibiotics within 72 hours before evaluation, patients with a previous neurosurgical procedure or CNS bleed, and patients whose complete blood count was not done within 6 hours of LP. The predicted CSF WBC count was calculated using the formula CSF WBC (predicted) = CSF RBC x (blood WBC/blood RBC). The O:P ratio was obtained by dividing the observed CSF WBC by the predicted CSF WBC. The simple ratio of WBCs to RBCs was also calculated. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated to predict the absence of disease. Receiver operator characteristic curves were generated for the O:P ratio and the WBC:RBC ratio. Continuous variables were analyzed with Mann-Whitney U test.
RESULTS: Among the 57 patients who fit all of the study criteria, 12 (21%) had positive CSF cultures for bacterial pathogens. The patients with meningitis were significantly older (median: 7.8 months; range: 1-106 months) than the patients without meningitis (median: 1.3 months; range: 1-139 months). The O:P ratio was significantly lower in the patients without meningitis (median: 0.064; range: 0.000054-1.09) as compared with patients with meningitis (median: 1.26; range: 0.045-4.72). The WBC:RBC ratio was significantly lower in the patients without meningitis (median: 0.001; range: 0-4.46) as compared with patients with meningitis (median: 1.98; range: 0.04-24.45). The specificity and positive predictive value of an O:P ratio <or=0.01 and a WBC:RBC ratio <or=1:100 were 100% predicting the absence of disease. The area under the curve for the O:P ratio (0.981) did not differ significantly from the area under the curve for the WBC:RBC ratio (0.970).
CONCLUSION: A WBC:RBC ratio of <or=1:100 (0.01) and an O:P ratio of <or=0.01 identified a large group of patients without meningitis. Using these methods in children younger than 1 month, the majority of patients without meningitis can be differentiated from those with meningitis despite the CSF abnormalities associated with a traumatic LP. However, the clinician should examine all clinical and laboratory information before opting not to treat a child after a traumatic LP.