Article Text

This article has a correction. Please see:

The role of lumbar puncture in suspected CNS infection—a disappearing skill?
  1. R Kneen,
  2. T Solomon,
  3. R Appleton
  1. The Roald Dahl EEG Unit, Department of Neurology, Royal Liverpool Children’s Hospital (Alder Hey), Liverpool L12 2AP, UK
  1. Correspondence to:
    Dr Appleton;

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Declining use of lumbar puncture may be detrimental to patient care

Introduced by Quincke in 1891 to diagnose and treat tuberculous meningitis, lumbar puncture (LP), has since been used extensively in the investigation of many neurological conditions.1 Cerebrospinal fluid (CSF) analysis and culture is the definitive method of diagnosing central nervous system (CNS) infections and, until the late 1970s, all or most patients admitted with suspected CNS infection in Western countries underwent LP.2 In many developing countries this remains the practice. However, in the West the use of LP has declined over the last 20 years, primarily because of increasing anxiety that the procedure may precipitate cerebral herniation. In this paper we review the reasons for the declining use of LP and examine whether its usage is now so infrequent that it may be detrimental to patient care.


It has been recognised since the early 1960s that raised intracranial pressure (ICP) is a serious complication of acute bacterial meningitis.3 Causes are thought to include occlusion of the arachnoid granulations, cerebritis, generalised brain oedema, inappropriate antidiuretic hormone secretion, subdural effusions or abscesses, seizures, and an increase in CSF production.4 Raised ICP in itself does not cause brainstem herniation—patients with idiopathic (benign) intracranial hypertension may have very high pressures without causing herniation or death. In other CNS infections, including cerebral malaria, high opening pressures are often found, but herniation is rare.5

Herniation can result when raised ICP leads to differences in pressure between different brain compartments.6 The uncus of the temporal lobe may herniate through the tentorial opening, or the cerebellar tonsils may herniate through the foramen magnum. These processes damage the brainstem either directly or indirectly, by compromising its blood supply resulting in ischaemia and haemorrhage. By careful examination, it is possible to recognise the different herniation syndromes and rostro-caudal progression from one to the next (see fig 1).7,8


Initial concern about the safety of LP in children with bacterial meningitis was raised in 1978 following the publication of a retrospective review of more than 300 children with proven bacterial meningitis, 10 of whom died.9 Eighteen children had an episode of cerebral herniation diagnosed clinically. Two of the 10 deaths were attributed to this mechanism and in one patient cerebral oedema and uncal herniation were confirmed at autopsy. One patient had clinical signs of herniation immediately following LP, which (consequently) led the authors to consider whether LPs contributed to herniation.

In 1982, a report from the Royal College of Physicians (UK) reviewed 88 of the 90 deaths from meningococcal infection (including septicaemia) in England and Wales during 1978.10 Coning was diagnosed clinically in six meningitis cases and confirmed at autopsy in five. Five of the six patients had undergone LP, but one patient (with autopsy evidence of coning) died before LP was performed. The authors concluded that LP could precipitate coning and suggested that LP may not be mandatory in all cases of clinically obvious meningococcal infection.


The debate on the safety of LP continued following the report of a 20 month old child with meningococcal meningitis who had a respiratory arrest two hours after LP, and died 48 hours later.11 Coning was not confirmed because there was no autopsy. Four experts were asked to comment on whether LPs should be delayed in all “sick” children; two considered that LP should be avoided if there was a clear clinical diagnosis and two felt that LP should still be undertaken.11 Subsequent correspondence highlighted the lack of available guidelines for undertaking LP and specifically citing indications and contraindications.12 A personal view on the contraindications to undertaking LP suggested that LP should be avoided in sick patients with signs of raised ICP who might be at imminent risk of cerebral herniation.13 These were similar to recommendations produced by a comprehensive review on meningitis from the USA.14 Although there was no evidence that LP caused herniation in patients who would not have herniated anyway, a working consensus was achieved, which appeared to change little over the past 15 years.7,13–17 In summary, it was generally recommended that LP should be undertaken on all patients with suspected CNS infection unless one of the following contraindications is present:

  • signs of raised intracranial pressure—altered pupillary responses, absent Doll’s eye reflex, decerebrate or decorticate posturing, abnormal respiratory pattern, papilloedema, hypertension and bradycardia

  • recent (within 30 minutes) or prolonged (over 30 minutes) convulsive seizures

  • focal or tonic seizures

  • other focal neurological signs—hemi/monoparesis, extensor plantar responses, ocular palsies

  • Glasgow Coma Score < 13 or deteriorating level of consciousness

  • strong suspicion of meningococcal infection (typical purpuric rash in an ill child)

  • state of shock

  • local superficial infection

  • coagulation disorder.

Although not previously defined, the definition of a prolonged seizure used here (more than 30 minutes) is the currently accepted definition of convulsive status epilepticus.18 Some debate exists as to whether a GCS of less than 13 or less than 12 is the most appropriate cut off as a contraindication to undertaking LP.7,15 Decerebrate posturing, misdiagnosed as tonic seizures, may be a manifestation of raised intracranial pressure and focal seizures may potentially indicate a structural abnormality including a tumour or an abscess. The occurrence of tonic-clonic, myoclonic, absence, and atonic seizures, in isolation, do not represent contraindications for LP.


In 1992 the issue became confused with the publication of a retrospective case review of 445 children with bacterial meningitis admitted to an Australian teaching hospital between 1984 and 1989, which examined the relationship between LP and herniation.19 Nineteen children had 21 clinically diagnosed episodes of coning of whom 14 died, two before undergoing LP. Of the remaining 19 coning episodes, three occurred before LP was undertaken, eight had an “early coning episode” (within three hours of LP) and eight occurred between three and 37 hours after LP. Autopsies in six patients revealed clear evidence of cerebral herniation in five and equivocal evidence in one patient. In view of the apparent cluster of early coning herniation episodes, the authors concluded that a clear temporal relationship existed between LP and cerebral herniation, and that LP could therefore cause herniation. However, as subsequent correspondence emphasised, a temporal relationship does not necessarily imply causality.20,21 An analysis of the relationship between blood cultures and herniation from the same study would also have shown a similar, and apparently positive relationship. More importantly, the study showed that many of these patients would not have undergone LP had the recommended contraindications been in practice. However, instead of being used to reinforce the recommendations, most clinicians appeared to interpret the findings of this study as further evidence against the use of LPs in any patient with suspected CNS infection.

The same year saw the publication of a review of 252 patients with meningococcal disease, including septicaemia, diagnosed in Gloucestershire over a 14 year period.22

Two of 17 deaths were attributed to a “neurological deterioration” after LP, with herniation confirmed at autopsy, although it was not stated whether there was clinical evidence of herniation before LP. Two other deaths were attributed to a delay in antibiotic treatment following normal initial CSF findings. The authors critically questioned the need for LP in patients with meningococcal meningitis but correspondents critically questioned the clinicans caring for the patients. It is well recognised that normal early CSF findings do not necessarily exclude bacterial meningitis and for this reason antibiotics should be given until CSF cultures are negative in those children who are felt to have a clinical diagnosis of meningitis.23


Despite the reservations about the Gloucester paper’s conclusions,23 and although it only considered patients with meningococcal disease (ignoring all other CNS bacterial and viral infections), its message became inappropriately applied to all patients with suspected CNS infection. Consequently, it would appear that many doctors have stopped undertaking LPs in patients with suspected CNS infection. In our own hospital a four month case note review of 448 acute medical admissions revealed that LP was performed in only 53% of 47 children who had clinical evidence or suspicion of CNS infection, and with no medical contraindications (Kneen R, Solomon T, Appleton R, unpublished observations). None of the patients who underwent LP experienced a neurological deterioration including recognisable cerebral herniation. In addition, only 17 of the 25 patients who had LP, CSF analysis was incomplete or inadequate—for example, no measurement of CSF glucose or protein levels, no simultaneous measurement of blood glucose levels, or no CSF cultures. Discussion with colleagues throughout the country has shown that our experience is not unique (Kneen R and Appleton R, personal communications). At some centres, no LP is performed without first consulting a paediatric neurologist, and at others, junior doctors believe LPs should never be performed without a CT scan—even in fully conscious patients with no neurological signs—despite the well recognised fact that a normal CT scan does not necessarily exclude raised ICP. Clinical signs remain the best predictors of incipient uncal herniation.24

In addition to concerns about safety, many doctors believe that because of newer diagnostic techniques (including specifically the polymerase chain reaction (PCR)) for bacterial antigens and because all children with possible meningitis or encephalitis receive antibiotics and/or aciclovir routinely, LP would contribute little to their management. However, our case note review of acute paediatric admissions showed that results of LPs contributed directly to patient management in 72% of cases, either by identifying an organism, allowing unnecessary antibiotics to be stopped after 24 hours, or by permitting an earlier discharge from hospital.

The reduction in the number of LPs being performed by junior doctors, with or without advice from their senior colleagues, may have a number of consequences. What was previously considered to be a routine, potentially important, and relatively safe investigation is now becoming almost obsolete. Whereas 10 years ago most doctors learnt to perform LPs as medical students or house officers, there are now many senior house officers who have never performed an LP, or even seen one being undertaken. The recent trend towards ward based and shift work, where juniors often do not follow up patients they admit, has meant they are less likely to see the benefits of investigating patients fully. It is easier to prescribe antibiotics and then “forget about the patient”, rather than perform LP, chase up the result, and then make an informed decision about the management. It is only when this strategy fails and patients have an unusual illness course or persisting non-specific symptoms and/or pyrexia that the benefit of an early LP is appreciated.

The advantages of obtaining a microbiological diagnosis extend beyond individual patient management. Identifying an organism allows appropriate prophylaxis to be recommended for close contacts, and public health service monitoring for disease outbreaks. Finally, there are potential health economic implications with reduced antibiotic and/or aciclovir usage and shorter hospital admissions for those children in whom CSF analysis has excluded meningitis or meningoencephalitis.

In summary, the circumstances in which LP should not be performed—possible incipient herniation, episodes of abnormal posturing including tonic seizures, and obvious meningococcal disease—have been clearly defined. There is no evidence that the procedure is harmful in most other patients with suspected CNS infection. We believe that in the majority of these patients LP is safe, and potentially beneficial. In contrast, the avoidance of LP would appear to be based on unsound and anecdotal evidence and may be potentially detrimental to patient care.


Clinical signs in brainstem herniation syndromes. Coronal (upper) and mid-sagittal (lower) sections of the brain highlighting the neuroanatomy and principal clinical manifestations of herniation syndromes. OCR, oculocephalic (doll’s eye) reflex; OVR, oculovestibular (caloric) reflex. (Reproduced with permission from reference 6.)

Declining use of lumbar puncture may be detrimental to patient care


View Abstract

Supplementary materials

    The role of lumbar puncture in suspected CNS infection�a disappearing skill?
    R Kneen, T Solomon, and R Appleton

    Publisher's Correction

    Figure 1 was incorrectly accredited to reference 6. Please note that Figure 1 is reproduced with permission from reference 8.
    Solomon T. Neurological Presentations. In: Beeching N, Gill G, eds. Lecture Notes on Tropical Medicine. Oxford: Blackwell Science, In Press 2002.

    The error is much regretted

Linked Articles

  • Atoms
    Harvey Marcovitch
    BMJ Publishing Group Ltd and Royal College of Paediatrics and Child Health