Analyse rétrospective de 1331 échantillons de liquide céphalorachidien chez le nouveau-né suspect d'infection

Background.- Definitive diagnosis of a neonatal infection usually requires recovery of an etiologic agent from body fluids or tissues such as spinal fluid, blood and urine. Routine lumbar puncture (LP) may raise some problems in interpreting results. Population and methods.- A retrospective analysis of 1331 samples of CSF was made. LP had been performed on 1041 neonates including 569 prematures (54.6%) as a part of the evaluation for suspected bacterial infection. Results.- In 50.7% of the cases, the CSF was haemorrhagic. The frequency of traumatic taps increased with the degree of prematurity, low birth weight, precocity of LP, association with respiratory distress and disorders of coagulating factors. In addition, haemorragic taps modified biochemical and cytologic characteristics of CSF. One hundred and six (8%) CSF samples contained organisms but the diagnosis of meningitis was certain in only 23 cases (2.2%). In the other 83 cases, CSF was thought to be contaminated, mostly by coagulase-negative staphylococci. Since both blood and CSF cultures were positive for the same bacterial organism in 18 cases, it was concluded that the LP had been useful in identifying the pathogens in only five cases. The high frequency of contaminated CSF led to overestimation of the incidence of true bacterial meningitis (0.57% in our study). Conclusions.- The low incidence of meningitis in neonates, the risk of having an haemorragic tap associated with the possibility of clinical aggravation during LP and the fact that the same pathogen is frequently (78.2% of cases) identified in blood cultures suggest that the immediate and routine LP is of less value than expected in infants suspected to be infected. LP could be postponed when the neonate presents with a respiratory distress syndrome and/or a precarious haemodynamic state.