71 Retrospective Study of Routine Lumbar Puncture in The Evaluation of Neonatal Sepsis

Abstract

Background: Lumbar puncture (LP) to exclude meningitis is part of an infection screen in many neonatal units. For best outcome, meningitis requires early initiation of treatment. Results of cerebrospinal fluid (CSF) microscopy may guide management before culture results are available. LP is, however, an invasive procedure that can be technically difficult in small infants. Blood staining of CSF is common making interpretation of microscopy impossible. Aim: To examine the value of LP in evaluation of suspected sepsis. Methods: Medical notes of infants having LPs between January 1999 and October 2001 were reviewed retrospectively. Data were collected for number of attempts to obtain CSF, procedural complications, results of CSF microscopy and culture and antibiotic treatment given. Results: 630 LPs were carried out in 527 infants. Median (range) gestational age and weight at birth were 36(23–42) weeks and 2580(500–5430) grams respectively. The median (range) age at time of LP was 2(1–96) days. Multiple attempts to obtain CSF were documented in 67 cases; 29 required ≥3 attempts. 314(50%) of the records reviewed contained no information about occurrence of associated complications. 280(44%) stated that no complications were seen. Adverse events such as apnoea, desaturation or bradycardia were noted in 36(5%) cases. In 7, these led to the procedure being abandoned. CSF was obtained at LP in 605(96%) cases. Of these, 431(71%) were suitable for cell count by microscopy. 169 samples (28% of those obtained) were too heavily bloodstained to determine a white blood cell count. Results of CSF culture were available for 616 LPs. 590(95%) samples yielded no growth. 26 samples from 24 babies were positive. 12 infants were treated for bacterial (9) or viral (3) meningitis. CSF microscopy aided diagnosis in only 5 cases. All those with bacterial disease had positive blood cultures. The other 12 infants did not display clinical evidence of meningitis and the samples were presumed contaminated. Positive and negative predictive values for CSF culture and microscopy in diagnosis are 35% and 99% respectively. Conclusion: In this group of infants, LP was often unhelpful. Many samples were unsuitable for microscopy and did not contribute to early exclusion or confirmation of meningitis. Multiple attempts and complications were common and may be underestimated in retrospective reviews. More selective identification of neonates at high risk of meningitis for investigation with lumbar puncture may avoid unnecessary intervention.