Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

Should a neonate with possible late onset infection always have a lumbar puncture?

Three Part Question

In [neonates] what [is the incidence of meningitis] in [late onset infection]?

Clinical Scenario

A baby born at 28 weeks gestation initially has no respiratory disease and is breathing spontaneously in room air. On day 6 of life the baby develops increasingly frequent and severe apnoeas and episodes of bradycardia that are mostly self-limiting. In view of this, nasal continuous positive airway pressure is started, a blood culture taken, and broad spectrum antibiotics commenced. On the ward round the next morning there is a debate as to whether a lumbar puncture (LP) should also have been performed, as part of the investigations for bacterial infection. The registrar opines that this was considered, but that the baby was thought "too unstable" for the procedure.
If an LP is performed routinely as part of the investigations for infection, how often will it be informative?

Search Strategy

Primary sources: searched Medline, Pubmed, Embase and CINAHL databases via Dialog Datastar
Cochrane and Dare: no relevant results

Search Outcome

26 references of which five were relevant clinical studies.

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Hristeva et al,
736 babies underwent an LP, of which 225 had LPs performed late (at >48 h of age) 88 of these were <31 weeks gestationProspective (level 1b)Prevalence of positive CSF culture in study group1.3% (4/310) cultures were positive Of 88 babies <31 weeks, 6 had late meningitis (4 bacterial;2 fungal) i.e. 6.8% prevalence (personal correspondence with authors) No mention of number of concurrent negative blood cultures 1 case of probable post-traumatic meningitis reportedHigh proportion of babies had an LP (42% of all admissions) Babies with respiratory distress had LPs deferred, thereby possibly underestimating the adverse effect
Schwersenski et al,
826 neonates who underwent an LP, out of which 114 had LPs performed at greater than 1 week of age (late onset) Birth weight: less than 1500 g to greater than 2500 gProspective (level 1b)Prevalence of positive CSF culture in study group3.5% (4/114) had meningitis. There were 8 positive CSF cultures, but 4 were considered contaminants 25% (1/4) had concurrently negative blood cultures No reported adverse effectsIncluded LPs done for post haemorrhagic hydrocephalus with raised ICP (8) Under estimation of meningitis as babies >72 h but <1 week, included in early onset
Visser et al,
400 neonates of whom 193 babies had concurrent blood and CSF cultures for suspected late onset (>72 h) sepsis Gestation: 25–42/40 Birth weight: 634–5650 gRetrospective cohort (level 2b)Prevalence of positive CSF culture in the study group2.5% (5/193) CSF samples positive 10% (21/193) blood cultures positive 24% of septic babies had meningitis (5/21) 15% had concurrently negative blood cultures No reported adverse effectsMeningitis likely to be under diagnosed as retrospective study All babies with suspected sepsis had lumbar punctures performed routinely before antibiotics were started
Kumar et al,
169 neonates who underwent an LP for suspected sepsis (119 of which were late onset) Gestational: <33–>36/40 Birth weight: <1500 g–>2500 gProspective (level 1b)Prevalence of positive CSF culture in study group3.3% (4/119) had positive CSF findings 11% had positive blood cultures All babies with meningitis had negative blood cultures No adverse effects of LP reportedExtensive antibiotic use present due to the higher risk population and this may underestimate incidence of meningitis Late onset clearly defined. Appears to be >7 days
Stoll et al,
2877/9641 (30%) had LP>72 h and 6056 (63%) had blood cultures at >72 h Gestation: <25–>33/40 Birth weight: >400–1500 g Average age at LP: 22 days (median: 16 days, range: 4–120 days)Prospective multi-centre study (level 1b)Prevalence of positive CSF culture in study2.2% (134/6056) had positive CSF cultures 5% (134/2877) of all lumbar punctures were positive (not all septic babies had lumbar punctures) 7.2% of those with positive blood cultures had meningitis 30% (45/134) with meningitis had negative blood cultures Babies with meningitis (compared to those uninfected) were: – ventilated longer (28 v 18 days) – in hospital longer (91 v 79 days) – more likely to fit (25% v 2%) – more likely to die (23% v 2%) There was no difference in the risk of death between infants who did and did not have lumbar punctures [284/2877 (10%) v 661/6764 (10%)]Study included only very low birth weight babies (<1500 g). This might overestimate risk as the VLBW is the susceptible population 11% of LPs repeated within 10 days were positive for the same organism even though the babies were on treatment with antibiotics


Meningitis is an important complication of late onset neonatal infection. In the five studies cited, CSF culture was positive in 1.3–3.5% of babies with suspected infection. These studies covered the period 1980 to 2004, during which time the incidence of meningitis does not seem to have changed. There appear to be epidemiological differences, with a lower prevalence of late onset meningitis in neonates in the UK (Hristeva). This has implications for treatment, such as the increased length of therapy or the choice of antimicrobial agent, where agents with higher CSF penetration may need to be considered. The mortality and morbidity in late onset meningitis is higher than in early onset meningitis. Blood cultures may often be negative in these babies, causing antibiotics to be discontinued too soon. The studies are in agreement that 15–30% of babies with meningitis (CSF culture positive) have negative blood cultures. Some of this may be due to the presence of viral or fungal infections, while others are due to meningitis without bacterial infection. This highlights the importance of routinely performing an LP in the investigation of late onset infection. Except for the study by Visser and colleagues, LPs were not performed routinely for investigation of infection. They were more likely to be performed if the blood cultures were positive, and even then only 60% of such babies had LPs performed. Therefore, it is likely that there is an underestimation of the prevalence of meningitis. Visser et al found that nearly one third of septic babies had coexisting meningitis. Other studies, except for that of Kumar and colleagues, have not investigated this proportion, though all studies showed a similar overall incidence of meningitis. Late onset meningitis is associated with a variety of organisms, though they are predominantly Gram negative. There is a higher incidence of viral and fungal (commonly Candida albicans) organisms than in early onset infection. Episodes of group B streptococcal infection can present late and are more likely to be associated with meningitis. All studies excluded diphtheroids and Staphylococcus epidermidis as being contaminants, unless they were grown in multiple cultures and there was a clinical indication of infection. The low rates of LP in the studies are attributed to the perceived adverse effects of the procedures, where babies are considered "too sick to tap". Complications that have been described in the literature include trauma, introduction of infection, spinal epidermoid tumours, brain stem herniation, and hypoxic stress for the baby. In the studies reviewed, none of the above complications were reported. In the study by Stoll and colleagues there was no difference in the risk of death between infants who did and did not have an LP. However, meningitis increased the risk of death substantially (23% mortality in babies with meningitis versus 9% in those who had an LP but no meningitis). The study by Stoll and colleagues found that among patients who had an LP performed, there was no significant difference across centres in the rate of positive CSF cultures (confirmed in this review). This finding suggests that, although there are LP practice differences across centres, they probably are not explained by better clinical acumen. The study also found that 10 of 90 repeat LPs grew the same organism as the original CSF culture, emphasising the importance of a repeat LP to determine that meningitis has been appropriately treated. Although the number of babies to investigate for possible bacterial infection in order to diagnose one case of meningitis may seem high, lumbar puncture in this population seems to be safe, the treatment (intravenous antibiotics) is effective, and the event (meningitis) and the implications of missing it are potentially very serious.

Clinical Bottom Line

In neonates with late onset infection the prevalence of meningitis varies from 1.3% to 3.5% (depending on the patient population). (Grade B) 30–90 babies (depending on the patient population) who are already being investigated for serious bacterial illness would need to have an LP to detect one baby with meningitis. (Grade A) At least 15% of neonates with meningitis may have a negative blood culture. (Grade A) Lumbar puncture should be considered as part of the routine investigation of late onset infection (after 48 hours) in neonates. (Grade B)


  1. Hristeva L, Bowler I, Booy R. et al. Value of cerebrospinal fluid examination in the diagnosis of meningitis in the newborn. Arch Dis Child 1992;69:514–17.
  2. Schwersenski J, McIntyre L, Bauer C. Lumbar puncture frequency and cerebrospinal fluid analysis in the neonate. Am J Dis Child 1991;145:54–8.
  3. Visser VE, Hall RT. Lumbar puncture in the evaluation of suspected neonatal infection. J Pediatr 1980;96:1063–6.
  4. Kumar P, Sarkar S, Narang A. Role of routine lumbar puncture in neonatal infection. J Paediatr Child Health 1995;31:8–10.
  5. Stoll BJ, Hansen N, Fanaroff AA. et al. To tap or not to tap: high likelihood of meningitis without infection among very low birth weight infants. Pediatrics 2004;113:1181–6.