Lumbar punctures in suspected bacterial meningitis: Too many or too few?

Children aged 1 month to 14 years admitted to the Royal Alexandra Hospital for Children during a 10 month period with suspected meningitis were studied prospectively. The aims were to determine how often lumbar puncture (LP) was delayed or never done, in relation to the outcome of all children, in order to determine the risks of LP and the risks of not doing LP. Of 218 children with suspected meningitis, LP was performed immediately in 195 (89.4%). Meningitis was diagnosed in 49 of these (bacterial 18, viral 31). No child developed cerebral herniation due to immediate LP. There were 11 traumatic taps and two children required repeated attempts. Lumbar puncture was delayed, but performed at a later time in 17 children, of whom three had proven bacterial meningitis, 1 had presumed bacterial meningitis but no organism was detected and 13 had alternative diagnoses. Six children never had an LP, although ventricular cerebrospinal fluid was obtained from two Four of these six children had presumptive bacterial meningitis, one had tuberculous meningitis presenting with acute hydrocephaius and diagnosed post‐mortem, and one had a very poor neurological outcome and no final diagnosis was reached. Of the 27 children with bacterial meningitis, LP was performed immediately in 18, or two‐thirds. There were only minor adverse effects of immediate LP. Delayed LP probably resulted in failure to identify the organism in one child with bacterial meningitis, but did not adversely affect outcome in any child. Of the six children in whom LP was never performed, in only one was no final diagnosis reached. There was no evidence of excessive morbidity attributable to performing too many or too few lumbar punctures.