Pediatric Lumbar Puncture and Cerebrospinal Fluid Analysis
The most common complication of performing an LP is a traumatic tap. The incidence of traumatic LP is approximately 20% for all children undergoing the procedure, and can be as high as one-third in neonates. The frequency of traumatic LP varies inversely with the age of the patient and the experience level of the physician. LP success rate has been shown to be doubled in young infants who received pre-LP local anesthesia. A prospective study of 1459 children receiving LP showed that procedures performed without local anesthetic were more likely to be traumatic or unsuccessful vs. those performed with local anesthesia.
Traumatic LP is often the result of inserting the needle too deeply and piercing the extensive vascular plexus lining the ventral epidural space of the vertebral column. When peripheral blood elements contaminate the subarachnoid compartment, it can obscure accurate CSF analysis. A review of 100 consecutive cases of traumatic LP in pediatric patients (defined as RBC count > 1000/mm) showed the distribution of RBC counts was < 10,000/mm in 65% of cases, < 50,000/mm in 80% of cases, and > 100,000/mm in 12% of cases.
Physicians have searched in vain for an accurate method to "correct" the impact of RBCs contaminating the CSF after traumatic LP to determine whether there was pre-LP CSF pleocytosis. Although not factoring variations in differing CBC cell counts, a commonly quoted rule of thumb is to expect the addition of 1 WBC/mm for every 1000 RBC/mm added. More specifically, it is traditional teaching that when traumatic LP causes peripheral blood contamination of normal CSF, a similar ratio of WBC:RBC should characterize both the CBC and CSF cell counts. In contrast, when CSF pleocytosis precedes traumatic LP, the CSF ratio of WBC:RBC should exceed that of peripheral blood. Thus, when CSF pleocytosis is associated with traumatic LP, it should be possible to use this relationship to distinguish the contribution of contaminating blood elements, and to determine if the pretap CSF WBC was abnormally elevated. The common formula for this relationship is:
The predicted CSF WBC count is compared to the CSF count actually observed; when pleocytosis precedes traumatic LP, the observed CSF WBC count should exceed that predicted.
Although intuitively logical, multiple studies have demonstrated that this prediction is not correct in all instances of culture-negative CSF associated with traumatic LP. For obscure reasons, the predicted CSF WBC count consistently overcorrects the observed count when using this formula. The clinical implication is that underlying CSF pleocytosis (possibly accompanying bacterial meningitis) could escape detection.
A study of mostly adult patients with traumatic LP compared CSF profiles of those with and without bacterial meningitis to determine the accuracy of this correction in distinguishing the two groups. The ratios of observed:predicted CSF WBC counts in the 2 groups were calculated; in nearly 90% of cases of bacterial meningitis, the ratio was > 10. Yet there was considerable overlap between groups, as 7% of those with bacterial meningitis had a ratio of < 1, and 62% with culture-negative CSF had a ratio > 1.
A similar study compared the CSF characteristics of children with and without bacterial meningitis who had traumatic LP. There was a significant difference between groups in rate of CSF %–PMN predominance, hypoglycorrhachia, and bacteria visualized on Gram-stained smear. A ratio of observed:predicted CSF WBC > 1 characterized all cases of bacterial meningitis, whereas 63% of culture-negative cases had a ratio < 1 (none exceeded 25). This study concluded that bacterial meningitis risk should be presumed when traumatic LP is associated with CSF pleocytosis and an observed:predicted ratio of CSF WBC counts > 1.
Others have performed artificial contamination of CSF of known composition with varying amounts of blood of known composition, yet were unable to consistently show a reliable relationship such that the clinician can accurately "correct" for traumatic LP and exclude risk for bacterial meningitis.
With respect to the other common indices, prior studies have shown CSF glucose is usually unaffected by blood contamination from traumatic LP, whereas CSF protein can be abnormally elevated due to the addition of RBCs from peripheral blood. There is an approximate correlation of 1 mg/dL of CSF protein increase for every 1000 RBCs/mm.
From these data, it is apparent that no fully reliable method exists to eliminate with certainty the potentially life-threatening infection of bacterial meningitis in all cases of traumatic LP associated with CSF pleocytosis. When the LP is traumatic, and the data obtained are insufficient to exclude bacterial meningitis, management should include hospitalization with empiric parenteral antibiotic therapy pending culture results.