PubMed Health. A service of the National Library of Medicine, National Institutes of Health.

National Collaborating Centre for Women's and Children's Health (UK). Bacterial Meningitis and Meningococcal Septicaemia: Management of Bacterial Meningitis and Meningococcal Septicaemia in Children and Young People Younger than 16 Years in Primary and Secondary Care. London: RCOG Press; 2010. (NICE Clinical Guidelines, No. 102.)

5Diagnosis in secondary care

5.1. Non-specific tests for meningococcal disease

Introduction

Meningococcal disease in childhood classically presents with a non-blanching rash in a feverish, ill child, although the rash may occur late in the illness or not at all in children who have meningococcal meningitis without septicaemia. Increased public awareness of meningococcal disease has meant that children may present earlier in the course of disease with fever and a petechial rash, although others may not yet appear unwell. Besides meningococcal disease, there are many other infective causes of petechial rashes in febrile children. Healthcare professionals assessing febrile children with rashes are, therefore, faced with deciding which children have invasive meningococcal disease and require immediate antibiotics and supportive therapy and which do not. Non-specific laboratory investigations are part of the diagnostic work-up of these children.

Clinical question

In children and young people up to 16 years of age with a petechial rash, can non-specific laboratory tests (C-reactive protein, white blood cell count, blood gas) help to confirm or refute the diagnosis of meningococcal disease?

Previous UK guidelines

‘Feverish illness in children’, NICE clinical guideline 4725 recommends that a full blood count and C-reactive protein should be performed as part of the initial laboratory investigations in:

  • infants younger than 3 months with fever
  • children older than 3 months with fever without apparent source with:

    one or more ‘red’ features (features suggestive of a high risk of serious illness); or

    one or more ‘amber’ features (features suggestive of an intermediate risk of serious illness).

The guideline recommends that the clinician should consider taking a blood gas sample in children with ‘red features’ as guided by the clinical assessment.

Studies considered in this section

All study designs evaluating the usefulness of white blood count, C-reactive protein (CRP) or blood gas for diagnosing meningococcal disease in children and young people with a petechial rash were considered for this section. Studies assessing the predictive ability of laboratory tests to diagnose invasive bacterial illness were included only if most cases of invasive illness were caused by N. meningitidis. Studies that included adults were excluded.

Overview of available evidence

Two prospective cohort studies [EL=2+] were found. One study involved children with petechiae and fever; the other study involved children with a non-blanching rash, 80% of whom had petechiae only. Both studies assessed the diagnostic value of white blood cell count and one study assessed the diagnostic value of CRP. No studies were found evaluating blood gas as an initial investigation for diagnosing meningococcal disease in children with a petechial rash.

Review findings

One prospective cohort study (USA, 1982–1983) [EL=2+] aimed to determine clinical and laboratory predictors of meningococcal disease in children with fever and petechiae admitted to a children’s hospital.43 Of 190 children aged 3 months to 15 years admitted with fever of more than 38°C and petechiae, 15 (8%) had invasive bacterial illness, 13 of whom had meningococcal disease. A total of 39 children (20.5%) had non-bacteraemic illness (S. pyogenes pharyngitis, urinary tract infection or viral infection). The remaining 136 children (71.5%) had no cause identified for their illness. Results were analysed for the 54 children with a confirmed microbiological diagnosis.

The study found that children with invasive bacterial disease had significantly higher mean peripheral white blood cell (WBC) counts and absolute immature polymorphonuclear neutrophil counts (band forms) than children with non-bacteraemic illness (mean white blood count: 17,600 cells/microlitre with invasive bacterial disease versus 11,600 cells/microlitre with non-bacteraemic illness, P = 0.005; peripheral band count: 3,717 with invasive bacterial disease versus 523 with non-bacteraemic illness, P < 0.001). The accuracy of initial laboratory tests as indicators of invasive bacterial illness in this subgroup weres: peripheral white blood count more than 15,000 cells/microlitre: sensitivity 67%, specificity 85%, positive likelihood ratio 4.5, negative likelihood ratio 0.39; peripheral absolute band form count more than 500 cells/microlitre: sensitivity 80%, specificity 74%; positive likelihood ratio 3.0, negative likelihood ratio 0.27.

If the peripheral white blood count, the peripheral absolute band form count and cerebrospinal fluid (CSF) white blood count were all normal, the likelihood of invasive bacterial illness was small (negative likelihood ratio of peripheral WBC more than 15,000 cells/microlitre or peripheral absolute band form more than 500 cells/microlitre or pleocytosis more than 7 cells/microlitre: 0.11). The high prevalence of invasive bacterial illness in the analysed subgroup (28%) affects the performance characteristics of the diagnostic tests under evaluation and limits the external validity of the study results to children seen in a secondary care setting.

A prospective cohort study (UK, 1998–1999) [EL=2+] assessed whether clinical features and laboratory investigations could predict meningococcal disease in 233 children admitted to a children’s Accident and Emergency Department with a non blanching rash.44 Fifteen children with an obvious alternative diagnosis were excluded and 218 children younger than 15 years were included in the final analysis. Of the 218 children, 11% (24) had laboratory proven meningococcal disease and 80% (175) presented with petechiae only (defined as new-onset, non-blanching spots in the skin, less than 2 mm in diameter), of whom 4 had meningococcal disease. Forty-three children (20%) presented with both petechiae and purpura (non-blanching spots more than 2 mm in diameter), of whom 20 had meningococcal disease. Children with meningococcal disease were more likely to have an abnormal neutrophil count than children who did not have meningococcal disease (OR 2.7, 95% CI 1.1 to 6.5).

As shown in table 5.1, 38% of children without meningococcal disease also had an abnormal neutrophil count and the diagnostic accuracy of an abnormal neutrophil count or an abnormal white blood cell count was low. No child with a CRP less than 6 mg/litre had meningococcal disease. However, the specificity of a CRP of more than 6 mg/litre for predicting meningococcal disease was low (see table 5.1). A CRP of more than 99 mg/litre had a high specificity but a low sensitivity for predicting meningococcal disease, with less than half of children in the study later diagnosed with meningococcal disease having an initial CRP above 99 mg/litre (see table 5.1).

Table 5.1. Accuracy of white blood cell count, neutrophil count and CRP for diagnosing meningococcal disease.

Table 5.1

Accuracy of white blood cell count, neutrophil count and CRP for diagnosing meningococcal disease.

Evidence statement

Evidence about the value of initial blood tests for predicting meningococcal disease in children with a petechial rash is limited by the small number of relevant studies.

There is evidence that children with meningococcal disease presenting to secondary care with a petechial rash are more likely to have a higher white blood cell count, a higher band count and an abnormal neutrophil count compared with children who do not have meningococcal disease. None of the above tests had sufficiently high sensitivity or specificity to accurately predict a diagnosis of meningococcal disease. One study found that a combination of normal peripheral white blood count, absolute band form count and CSF white blood count was associated with a low risk of invasive bacterial illness, including meningococcal disease.

There is evidence from one study that children presenting to secondary care with petechiae and fever with an initial CRP less than 6 mg/litre are unlikely to have meningococcal disease. A high CRP of more than 99 mg/litre can be used to identify children at a high risk of meningococcal disease. A high CRP is, however, poorly sensitive for predicting meningococcal disease and the absence of a high CRP cannot be used to rule out meningococcal disease.

No studies were found evaluating the usefulness of blood gas for diagnosing meningococcal disease in children and young people with a petechial rash.

GDG interpretation of the evidence

Children with invasive meningococcal disease may have a higher white cell count and CRP than those with viral infections and those with non-invasive bacterial infections. However, these tests alone cannot be relied on to predict which children have meningococcal disease. Children early in their illness or with rapidly advancing meningococcal disease may have a normal or low WBC count and a normal CRP.

The finding of a high CRP of more than 99 mg/litre is specific but not sensitive for meningococcal disease in children with fever and a rash. A low CRP does not exclude meningococcal disease.

The GDG concluded that a full blood count and CRP should be performed on children with fever (or history of fever) and a petechial rash and the results combined with a thorough clinical assessment for the signs of septicaemia and meningitis. Abnormal results may support the diagnosis where there is uncertainty but normal results cannot be used to exclude the diagnosis.

No evidence was identified in relation to the diagnostic accuracy of measuring blood gas in children and young people with petechial rash. However, ‘Feverish illness in children’ (NICE clinical guideline 47)25 recommends taking a sample of blood gas in children with features suggestive of a high risk of serious illness and this is reflected in the GDG’s recommendations.

The GDG highlighted the importance of starting empiric antibiotic treatment (with ceftriaxone) immediately in children with signs of bacterial meningitis or meningococcal septicaemia and this is reflected in recommendations included in this section. The clinical and cost effectiveness evidence relating to the choice of empiric antibiotics is presented in section 6.1.

The GDG noted that although polymerase chain reaction (PCR) is a specific test (see section 5.3 for a discussion of the clinical and cost effectiveness evidence relating to PCR), testing should be carried out using the initial blood sample, and so PCR testing is included in the recommendations in this section.

Recommendations

Diagnosis in secondary care

Perform a very careful examination for signs of meningitis or septicaemia in children and young people presenting with petechial rashes (see table 3.3).

Investigation and management in children and young people with petechial rash

Give intravenous ceftriaxone immediately to children and young people with a petechial rash if any of the following occur at any point during the assessment (these children are at high risk of having meningococcal disease):

If a child or young person has an unexplained petechial rash and fever (or history of fever) carry out the following investigations:

In a child or young person with an unexplained petechial rash and fever (or history of fever) but none of the high-risk clinical manifestations (see table 3.3):

  • Treat with intravenous ceftriaxone immediately if the CRP and/or white blood cell count (especially neutrophil count) is raised, as this indicates an increased risk of having meningococcal disease.
  • Be aware that while a normal CRP and normal white blood cell count mean meningococcal disease is less likely, they do not rule it out. The CRP may be normal and the white blood cell count normal or low even in severe meningococcal disease.
  • Assess clinical progress by monitoring vital signs (respiratory rate, heart rate, blood pressure, conscious level [Glasgow Coma Scale and/or APVU], temperature), capillary refill time, and oxygen saturations. Carry out observations at least hourly over the next 4–6 hours.
  • If doubt remains, treat with antibiotics and admit to hospital.

If the child or young person is assessed as being at low risk of meningococcal disease and is discharged after initial observation, advise parents or carers to return to hospital if the child or young person appears ill to them.

Be aware that in children and young people who present with a non-spreading petechial rash without fever (or history of fever) who do not appear ill to a healthcare professional, meningococcal disease is unlikely, especially if the rash has been present for more than 24 hours. In such cases consider:

5.2. Non-specific tests for bacterial meningitis

Introduction

If meningococcal meningitis presents with features of meningococcal sepsis (a non-blanching rash in a feverish, ill child) then non-specific laboratory blood tests will predominantly reflect inflammation in the bloodstream (see non-specific laboratory tests in children with suspected meningococcal disease in section 5.1). However, if a non-blanching rash does not accompany meningitis, the child will present with symptoms and signs suggesting meningitis. Non-specific laboratory investigations are part of the diagnostic work-up. The definitive test for meningitis is a lumbar puncture with laboratory examination of the CSF. In children with contraindications to lumbar puncture, or in clinical situations where medical staff are reluctant to undertake the procedure and CSF results are not available, the blood test results may then be consulted for evidence to confirm or refute the diagnosis of meningitis. The extent to which blood test results are informative about the presence or absence of bacterial meningitis will assist these decisions.

Clinical question

In children and young people under 16 years of age, are the results of non-specific laboratory tests predictive of bacterial meningitis?

Previous UK guidelines

No previous guidelines were identified in relation to this question.

Studies considered in this section

All study designs evaluating blood tests for procalcitonin, C-reactive protein or white blood cell count to discern meningitis from other diseases, or to discern bacterial meningitis from viral/aseptic meningitis, were considered for inclusion in this section. The majority of studies were retrospective and only those conducted in high income countries were included.

Studies of adults and children were included where data were presented separately for child participants. Findings are presented for three age groups: all children, infants and neonates.

Overview of available evidence

Predictive value of individual nonspecific blood tests for the differential diagnosis of bacterial meningitis from other illnesses

Procalcitonin

No studies evaluating procalcitonin were identified.

C-reactive protein

Two US studies were found that investigated the value of blood C-reactive protein (CRP) in aiding differentiation of bacterial meningitis from other illnesses. The first of these was a prospective cohort study58 [EL=II] which compared blood CRP of children with bacterial meningitis (n=10) with a control group which included children with: aseptic meningitis (n=14); extrameningeal bacterial infection (n=10); other febrile illnesses but presenting with symptoms suggestive of bacterial meningitis (meningeal signs and suggestive history (n=33); or suggestive history alone (n=102); or who were aged under 2 months and undergoing a ‘sepsis work-up’ (n=23). Significantly more children with bacterial meningitis had a blood CRP level of more than 1.0 mg/decilitre compared with children in the control group (8 out of 75 children with CRP more than 1.0 mg/decilitre versus 2 out of 85 children in control group; P = 0.047). This cutoff of CRP level of more than 1.0 mg/decilitre gave a sensitivity of 80%, specificity 55%, positive predictive value 0.11 and negative predictive value 0.98.

An earlier retrospective cohort study59 (USA, 1984) [EL=III] compared blood CRP of children with bacterial meningitis (n=21) with a control group which included children with aseptic meningitis (n=8), no meningitis (defined as suspected meningitis but with normal CSF findings) (n=50) and leukaemia (n=40). A serum CRP of more than 1 mg/decilitre was found for 20 out of 21 cases (95%) of children with bacterial meningitis, 1 out of 8 cases (13%) of children with aseptic meningitis, 24 out of 50 cases (48%) with no meningitis and 5 out of 40 cases (13%) with leukaemia. Removing the cases with leukaemia this gives 20 out of 21 cases (95%) with bacterial meningitis versus 25 out of 58 (43%) for controls; P < 0.0001 (Fisher’s Exact Test). Again removing cases with leukaemia, this cutoff of serum CRP of more than 1 mg/decilitre gave an overall sensitivity of 95% and an overall specificity of 57% (GDG analysis).

White blood cell count

Two US studies were identified that examined blood WBC counts, as shown in table 5.2.

Table 5.2. Blood white blood cell count – descriptive statistics (infants and children of all ages).

Table 5.2

Blood white blood cell count – descriptive statistics (infants and children of all ages).

Of these two US studies, one was a retrospective study60 (2003) involving 5375 infants aged 3 to 89 days with fever evaluated in the emergency department for serious bacterial infection [EL=III]. Twenty-two children had confirmed bacterial meningitis; the remainder made up a control group (n=5353). No details are given to describe the control group other than that they had a CSF and blood sample sent as part of their clinical evaluation for suspected serious bacterial infection while in the emergency department. Blood WBC count was found to be a poor discriminator of bacterial meningitis from other bacterial illnesses. Results from the study are presented in Table 5.2.

In terms of differential diagnostic accuracy, blood WBC count was not found to be useful (area under the curve for ROC=0.43). For the three cutoff values tested while specificity reached 96% for a threshold of less than 5000 cells/microlitre the sensitivity achieved was only 32%, thus making this cutoff useful for ruling out bacterial meningitis but not as a proof of the disease. At higher thresholds the specificity remained high but sensitivity was not significantly improved (see table 5.3 for details).

Table 5.3. Blood white blood cell count – diagnostic statistics (infants and children of all ages).

Table 5.3

Blood white blood cell count – diagnostic statistics (infants and children of all ages).

An earlier retrospective study61 described the white blood cell count of children (n=232) undergoing lumbar puncture for suspected meningitis [EL=III]. The study sample comprised: 46 children with bacterial meningitis (median age 11 months, range 0 to 157 months); 132 children with aseptic meningitis (median age 2 months, range 0 to 219 months); and 56 children with extrameningeal infection (median age 6.5 months, range 0 to 79 months). Extrameningeal infections included urinary tract infection (UTI) (n=22), occult bacteraemia (n=13), cellulitis/abscess (n=7), enteritis (n=7), otitis media (n=4), pneumonia (n=2) and septic arthritis (n=1). The values found for WBC counts and neutrophil counts for each study group are detailed in table 5.2. In children without bacteraemia the WBC count was similar in those with bacterial meningitis to those with extrameningeal bacterial infection (WBC/microlitre: median bacterial meningitis=14,500, extrameningeal bacterial infection=13,800; P = 0.57). A WBC count threshold of 1500/microlitre to differentiate between bacterial meningitis and aseptic meningitis or extrameningeal bacterial infection gave a sensitivity of 22% and a specificity of 73%.

Predictive value of individual nonspecific blood tests for differentiating bacterial versus aseptic meningitis

Procalcitonin

Three relevant studies were identified that examined the usefulness of blood procalcitonin assay in differentiating bacterial from aseptic meningitis.

A recent European multicentre study undertook a secondary analysis of retrospective cohort studies from six paediatric emergency or intensive care centres across five European countries62 [EL=III]. A total of 198 children were included in the analysis (BM=96, aseptic meningitis =102) aged 29 days to 15.9 years (mean 4.8 years). The median level of blood procalcitonin (ng/ml) was significantly higher in cases of bacterial meningitis compared with aseptic meningitis (see table 5.4). Meta-analysis using a pooled diagnostic odds ratio (DOR) showed a significant association between high procalcitonin levels and risk of bacterial meningitis (pooled DOR 139; 95% CI 39–498, I2=0%).

Table 5.4. Procalcitonin level – descriptive statistics (children of all ages).

Table 5.4

Procalcitonin level – descriptive statistics (children of all ages).

The area under the curve (AUC) for the ROC curve for procalcitonin was very high at 0.98 (compared with 0.89 for C-reactive protein, 0.88 for CSF protein and 0.87 for CSF neutrophil count; P = 0.001) (see table 5.5 for summary details). Blood procalcitonin was found to be more accurate than C-reactive protein, CSF protein level and CSF neutrophil count in differentiating bacterial from aseptic meningitis.

Table 5.5. Procalcitonin level – diagnostic statistics (children of all ages).

Table 5.5

Procalcitonin level – diagnostic statistics (children of all ages).

An earlier retrospective cohort study by the same author63 (2000–2004) [EL=III] reported similar findings. The study included blood samples from 167 children aged 28 days to 16 years (BM=21, aseptic meningitis=146). Blood procalcitonin was again much higher in bacterial meningitis than in aseptic meningitis. Procalcitonin was found to be the most accurate test in differentiating bacterial from aseptic meningitis with an ROC AUC of 0.95 (0.95 for C-reactive protein, 0.93 for CSF protein, 0.87 for CSF neutrophil count, 0.81 for CSF WBC count). See tables 5.4 and 5.5 for details.

A third European study64 [EL=III] (2000) compared blood parameters for differentiating between bacterial meningitis and viral meningitis. The study included 74 children aged 3 months to 13 years (for bacterial meningitis n=23, mean age 3.2 years and for viral meningitis n=51, mean age 2.1 years). The study only reports descriptive statistics for procalcitonin levels: again these are much higher in cases of bacterial meningitis compared with confirmed viral meningitis (bacterial meningitis: mean=60.9 microgram/litre (range 4.8 to 335 microgram/litre) versus viral meningitis: mean=0.32 microgram/litre (0 to 1.7 microgram/litre); P < 0.0001).

C-reactive protein

A systematic review with meta-analysis65 [EL=III] was identified, the aim of which was to evaluate published evidence relating to diagnostic accuracy of CSF and serum C-reactive protein (CRP) tests in the diagnosis of bacterial meningitis. Serum CRP had been measured in 14 of the 35 studies included in the systematic review (see table 5.6 for a summary of diagnostic accuracy data from these studies and the study characteristics). Many of the 35 studies included in the systematic review had fairly small sample sizes (66% included fewer than 100 children and 29% included fewer than 50 children); they had been conducted in different populations (three in the USA, two in Finland, and one each in France, Italy, Spain, Sweden, Poland, South Africa, Thailand, Indonesia and Chile); and they had used different study designs. The two main approaches used in the studies were to recruit either ‘patients suspected of having bacterial meningitis, irrespective of final diagnosis’ or ‘patients with confirmed meningitis’. On the basis of this information and whether recruitment was conducted prospectively, consecutively or selectively, the authors of the systematic review further characterised each study as reporting the ‘clinical performance’ of a CRP test or not, with studies that reported clinical performance of the CRP test being defined as prospective studies with patients recruited in clinical setting (see table 5.6).

Table 5.6. Summary of studies providing data on diagnostic accuracy of serum C-reactive protein (CRP) as a predictor of bacterial meningitis (as opposed to viral or aseptic meningitis; based on Gerdes 1998).

Table 5.6

Summary of studies providing data on diagnostic accuracy of serum C-reactive protein (CRP) as a predictor of bacterial meningitis (as opposed to viral or aseptic meningitis; based on Gerdes 1998).

The included studies were heterogeneous with respect to the cutoff values for CRP used to classify the patients as having bacterial meningitis or viral/aseptic meningitis and with respect to the participants’ ages. Seven studies (n=552 participants) included children under 18 years, three included adults and children reported separately (age range 16 to 83 years, n=144 participants), three included a mix of adults and children (age range 1 week to 60 years, n=265 participants) and one study did not reported details of the participants’ ages.

The systematic review reported the results of a meta-analysis, but caution should be exercised in interpreting the findings because of the heterogeneity of the included studies with respect to inclusion of low-income countries and dates of data collection. However, in conducting the meta-analysis, no statistically significant inter-study variance was reported by the authors of the systematic review and so the findings from the systematic review are reported here.

Of the 14 studies that examined serum CRP, one was excluded from the analyses because it included only three patients with bacterial meningitis. The total number of patients included in the 13 remaining studies comparing bacterial with aseptic meningitis was 749 (bacterial meningitis n=338, aseptic meningitis n=411). When serum CRP log true-positive fractions were regressed on log false-positive fractions for patients with bacterial and aseptic meningitis, these regression estimates were obtained: intercept 5.0 (95% CI 3.8 to 6.2) with corresponding OR=150 (95% CI 44 to 509); slope −0.17 (P = 0.6). The sensitivity for CRP measurement was 92.4% and the specificity was also 92.4% (standard error 0.068). When the analysis was restricted to the six studies that were classified as estimating ‘clinical performance’, the regression was intercept 5.0 with corresponding OR=143. The predictive values of serum CRP were reported as being ‘almost identical’ to those of CSF CRP. The post-test probability of bacterial meningitis given a positive CRP test depends upon the pre-test probability in an assumed clinically relevant range of 0.05 to 0.30. The post-test probability of not having bacterial meningitis given a negative test is high and declines only slightly in that range. At 5% prevalence, PPV=44.8% and NPV=99.7%, whereas at 30% prevalence PPV=86.3% and NPV=97.3%.

A further four studies that were published after the systematic review65 were identified for inclusion in the guideline review. Two of these studies have already been detailed above (Dubos, 200862 and Dubos, 200663). Findings for serum CRP from these studies are presented in table 5.7.

Table 5.7. CRP level – descriptive statistics (children of all ages).

Table 5.7

CRP level – descriptive statistics (children of all ages).

A third recent retrospective study35 [EL=III] involved 92 children aged 0 to 15 years (mean 5.6 years, median 5.0 years) admitted to a Belgian regional hospital from 1997 to 2005 for observation and with subsequent confirmed diagnosis of viral (n=71) or bacterial (n=21) meningitis. Children with bacterial meningitis were found to have significantly higher level of serum CRP than children with viral meningitis (see table 5.7). A threshold of 2.0 mg was found to have a high sensitivity and a high NPV but a low PPV (see table 5.8).

Table 5.8. Blood C-reactive protein – diagnostic statistics (children of all ages).

Table 5.8

Blood C-reactive protein – diagnostic statistics (children of all ages).

An earlier retrospective study66 [EL= III] included 237 children aged 3 months to 15 years, 55 with bacterial meningitis (recruited from 1984 to 1991 into two large Finnish studies) and 182 children with confirmed or presumed viral meningitis (recruited from one Finnish hospital from 1977 to 1992). As in other reported studies, children with bacterial meningitis were found to have significantly higher serum CRP levels than those with viral meningitis (see table 5.7). A CRP threshold of more than 20.0 mg/litre gave high sensitivity and specificity with an NPV of 99%. At a CRP threshold of more than 40.0 mg/litre the specificity and PPV rose to 100%, but this was at the expense of the sensitivity and NPV (see table 5.7).

White blood cell count

Six studies were identified that reported accuracy of blood WBC count for differentiating between bacterial meningitis and aseptic meningitis or viral meningitis. Five of these included studies have already been described in preceding sections.35;61–63;66 Findings from these studies in relation to blood WBC count are presented in table 5.9. The sixth study67 investigated blood WBC counts in neonates and is detailed below.

Table 5.9. Blood white blood cell count – descriptive statistics (children of all ages).

Table 5.9

Blood white blood cell count – descriptive statistics (children of all ages).

Blood white blood cell count – neonates

One study was identified that looked at blood WBC count in neonates67 [EL=III]. The study included 34 neonates (aged 28 days or younger) who underwent a complete sepsis evaluation (including lumbar puncture) in a US emergency department from 1982 to 1989, and who had a discharge diagnosis of meningitis (bacterial meningitis=10, aseptic meningitis=24). The total WBC count range was 2600 to 28000 cells/microlitre. No statistically significant differences were found between neonates with bacterial meningitis and aseptic meningitis.

Blood neutrophil count

Four of the previously described studies also included data for the diagnostic accuracy of blood neutrophil count in differentiating bacterial from aseptic or viral meningitis.35;61–63 Summary statistics for findings from these four studies, all of which were retrospective in design [EL=III], are given in tables 5.10, 5.11 and 5.12.

Table 5.10. Blood white blood cell count – diagnostic statistics (children of all ages).

Table 5.10

Blood white blood cell count – diagnostic statistics (children of all ages).

Table 5.11. Blood neutrophil count – descriptive statistics (children of all ages).

Table 5.11

Blood neutrophil count – descriptive statistics (children of all ages).

Table 5.12. Blood neutrophil counts – diagnostic statistics (children of all ages).

Table 5.12

Blood neutrophil counts – diagnostic statistics (children of all ages).

Evidence summary

Bacterial meningitis versus other infections

No evidence was identified that examined the diagnostic accuracy of procalcitonin for differentiating bacterial meningitis from other infections.

Findings from two small studies showed that at a cutoff of more than 1.0 mg/decilitre, blood C-reactive protein (CRP) levels have moderate to good sensitivity for differentiating bacterial meningitis from other infections but poor specificity.

Findings from two retrospective studies show that blood white blood cell (WBC) counts have poor sensitivity in differentiating bacterial meningitis from other infections. Findings for specificity varied widely.

Bacterial meningitis versus aseptic or viral meningitis

Findings from three retrospective studies showed that blood procalcitonin levels are significantly higher in children with bacterial meningitis compared with those with aseptic meningitis (two studies) or viral meningitis (one study). Findings from two of these studies also report good sensitivity and specificity for the diagnostic accuracy of procalcitonin in differentiating bacterial meningitis from aseptic meningitis.

Findings from four retrospective studies show that blood CRP levels are significantly higher in children with bacterial meningitis compared with aseptic meningitis (two studies) or viral meningitis (two studies). Findings from a meta-analysis involving 13 studies plus four more recent studies show that blood CRP has good sensitivity and moderate to very good specificity at differentiating bacterial meningitis from aseptic meningitis (meta-analysis and two studies) or viral meningitis (two studies).

Four of five retrospective studies that investigated the diagnostic accuracy of blood WBC count reported a significantly higher level in children with bacterial meningitis compared with aseptic meningitis (two studies) or viral meningitis (two studies). All five studies reported poor sensitivity for differentiating bacterial from aseptic or viral meningitis at a threshold of 15000 cells/microlitre or more or 25000 cells/microlitre or more and moderate to good specificities. At a cutoff of more than 25000 cells/microlitre, one study reported a specificity of 100% but a very low sensitivity of 20%.

One small retrospective study found no significant difference in the blood WBC count of neonates with bacterial meningitis compared with those with aseptic meningitis.

Findings from four retrospective studies reported conflicting findings regarding differences between blood neutrophil counts for children with bacterial meningitis compared with aseptic meningitis (three studies) or viral meningitis (one study). Findings from two of these studies show neutrophil count has moderate sensitivity and specificity for differentiating between bacterial and aseptic meningitis.

GDG interpretation of the evidence

CRP, WBC and procalcitonin levels in the bloodstream reflect inflammation in the bloodstream and are not directly informative about inflammation in the cerebrospinal fluid (CSF). Because bacterial infection in the bloodstream often precedes bacterial meningitis, CRP, WBC and procalcitonin levels may be elevated when bacterial meningitis is present.

CRP, procalcitonin and WBC counts have insufficient sensitivity and specificity to differentiate bacterial meningitis from other illnesses.

Raised procalcitonin, CRP and WBC counts and neutrophil count have reasonable specificity (67–93%) for bacterial meningitis in comparison to aseptic meningitis at commonly used cutoffs. Higher thresholds yield higher specificity (up to 100%) at the expense of lowering the sensitivity.

CRP levels of more than 20 mg/litre and procalcitonin of more than 0.5 nanograms/ml have greater than 83% sensitivity for differentiating bacterial meningitis from aseptic meningitis.

Total WBC count and neutrophil count have low sensitivity for differentiating bacterial meningitis from aseptic meningitis.

The evidence review indicates that non-specific laboratory blood tests cannot be used to distinguish bacterial meningitis from other illnesses (other illnesses are defined variously in the reviewed papers and include: febrile illnesses presenting with symptoms suggestive of bacterial meningitis; suspected meningitis but with normal CSF findings; suspected serious bacterial infection; and extrameningeal infections including urinary tract infection, occult bacteraemia, cellulitis/abscess and enteritis).

Where available, high procalcitonin (more than 0.5 nanograms/ml) may be useful to rule in bacterial meningitis (high sensitivity and specificity) but a low procalcitonin is insufficient to rule out the diagnosis. Up to 11% of children will have a low procalcitonin despite having bacterial meningitis.

High CRP (more than 20 mg/litre) may be useful to rule in bacterial meningitis (moderate sensitivity and moderate specificity) but a low CRP is insufficient to rule out the diagnosis. Up to 17% of children will have a CRP less than 20 mg/litre despite bacterial meningitis.

Although total white cell count and neutrophil count have low specificity and sensitivity for bacterial meningitis in comparison with aseptic meningitis, children with a high WBC count (more than 15 cells/microlitre) or neutrophil count (more than 10 neutrophil/microlitre) are three to seven times more likely to have bacterial meningitis.

Although none of the tests allow bacterial meningitis to be ruled out, the GDG felt that they are useful to add to other variables when making the decision about the management of suspected bacterial meningitis.

Recommendations

Investigation and management in children and young people with suspected bacterial meningitis

In children and young people with suspected bacterial meningitis, perform a CRP and white blood cell count:

5.3. Polymerase chain reaction tests for bacterial meningitis and meningococcal disease

Introduction

Confirming the diagnosis of bacterial meningitis and meningococcal disease is essential to ensure that the correct antibiotic therapy is used for the correct duration of time and to support decisions about the long-term follow-up of the child. Traditionally, the confirmation of the diagnosis of these diseases has relied on microscopy and culture of blood and cerebrospinal fluid (CSF). With the advent of DNA based diagnostic tests, such as polymerase chain reaction (PCR), it is important to decide which are the most effective and cost-effective diagnostic tests to support management of the child.

Clinical questions

What is the diagnostic value of blood and CSF PCR in children and young people with suspected meningococcal meningitis or meningococcal septicaemia?

Previous UK guidelines

The SIGN guideline on ‘Management of Invasive Meningococcal Disease in Children and Young People’27 recommends that all children with suspected invasive meningococcal disease should have blood taken for meningococcal PCR to confirm the diagnosis. The guideline recommends that if lumbar puncture is performed, CSF should be sent for PCR analysis.

Studies considered in this section

The review included studies of any design assessing the diagnostic value or accuracy of real-time PCR assays that target meningococcal or pneumococcal-specific genes as these types of assay are most widely used in the UK. Laboratory studies that primarily assessed the accuracy of PCR using bacterial isolates and that included only small numbers of clinical samples were excluded from the review. Studies without a well-defined reference standard were excluded.

Overview of available evidence

Three clinical diagnostic studies [one EL=Ib and two EL=II], one retrospective review [EL=III] and one laboratory diagnostic study [EL=III] were found.

Review findings

Blood PCR for suspected meningococcal disease

One prospective study (Australia, 2000–2001) [EL=Ib] compared the diagnostic accuracy of Taqman™ real-time PCR targeting the N. meningitidis capsular transfer gene (ctrA) with culture of blood or CSF in 118 children with possible meningococcal septicaemia or meningitis admitted to a tertiary care paediatric hospital.68 The reference standard for diagnosis of meningococcal disease was a clinical diagnosis reached by consensus of the attending clinician and an infectious diseases physician plus a confirmatory laboratory test in the case of suspected meningococcal meningitis (positive CSF Gram stain, CSF culture or PCR). In total, 24 children were diagnosed with meningococcal disease using the reference standard. The study found that blood PCR was more sensitive than blood culture for diagnosing meningococcal disease. Blood PCR was positive for 21 out of 24 cases (sensitivity 88%, 95% CI 68 to 97) and blood culture was positive for 14 out of 24 cases (sensitivity 58%, 95% CI 37 to 78). Both PCR and culture were 100% specific (95% CI 96 to 100) (see table 5.13).

Table 5.13. Diagnostic accuracy of real-time polymerase chain reaction (PCR) versus culture in studies with a clinical gold standard.

Table 5.13

Diagnostic accuracy of real-time polymerase chain reaction (PCR) versus culture in studies with a clinical gold standard.

Of the 24 children with gold standard confirmed meningococcal disease, blood PCR was positive for 8 out of 8 with clinical signs of septicaemia alone, 9 out of 11 with clinical signs of septicaemia and meningitis, and 4 out of 5 children with clinical signs of meningitis alone.

All children with a positive blood culture had positive PCR results. Blood PCR was positive but blood culture negative in 7 out of 24 cases (29%). Blood PCR remained positive for longer than blood cultures after parenteral antibiotics: for a third of patients tested, PCR remained positive up to 72 hours after parenteral antibiotic administration.

One prospective study (UK, 2000–2001) [EL=II] evaluated the diagnostic accuracy of ctrA whole-blood Taqman PCR (WB-Taqman) in 196 children with suspected meningococcal disease admitted to a children’s hospital.69 The reference standard was a clinical diagnosis of meningococcal disease made by the attending physician in the absence of alternative positive microbiological investigations. In total, 98 children were diagnosed with meningococcal disease using the gold standard. The study found that whole-blood PCR performed better than blood culture for confirmation of clinically diagnosed meningococcal disease. Whole-blood PCR was positive for 84 out of 95 clinical cases (sensitivity 88%, 95% CI 81 to 95) and blood culture was positive for 32 out of 98 children (sensitivity 33%, 95% CI 24 to 42). Both techniques were 100% specific (see table 5.13). All children with a positive blood culture had positive PCR results. PCR was positive, but blood culture negative in 52 out of 95 children (55%) with clinically diagnosed meningococcal disease. Of 22 children with clinical signs of meningitis, blood PCR was positive in 16 (sensitivity 78%). The positivity of whole blood PCR in children with clinical signs of meningitis but not septicaemia was not reported. The sensitivity of PCR was not decreased by preadmission antibiotics (PCR sensitivity 93% for 14 children given preadmission antibiotics). The sensitivity of blood culture in children given preadmission antibiotics decreased to 21%.

The study69 compared the performance of WB-Taqman PCR with that of serum Taqman PCR (S-Taqman) assessed in an earlier cohort study (1997–1999) [EL=II] conducted at the same hospital involving 319 children with suspected meningococcal disease.70 The earlier study used the same clinical gold standard described above to define meningococcal disease: 166 children were diagnosed with meningococcal disease using the gold standard. Comparative analysis found that case confirmation increased from 47% with S-Taqman70 to 88% with WB-Taqman, P < 0.001.69 Rates of blood culture positivity were similar for the two studies at P = 0.8 (see table 5.13). Both PCR–ELISA and real-time PCR were used in the earlier study, which also used two screening assays, one targeting IS1106 and one targeting ctrA. When all laboratory tests (including blood and CSF culture, PCR and rapid antigen testing) were used for diagnosis, case confirmation increased from 72% (with S-Taqman PCR) in the earlier study to 94% (with WB-Taqman PCR).

CSF PCR for suspected bacterial meningitis (including meningococcal meningitis)

One retrospective review of case notes (Belgium, 2002–2006) [El=III] compared the performance of duplex CSF real-time PCR with CSF Gram stain and culture in 70 patients admitted to a tertiary care hospital with suspected bacterial meningitis.71 The PCR assay targeted ctrA for N. meningitidis and the pneumolysin gene (ply) for S. pneumoniae. The age of the patients was not recorded. The gold standard for diagnosis of bacterial meningitis was a composite of clinical features of meningitis plus a confirmatory laboratory test (positive CSF Gram stain, positive CSF or blood culture, or positive blood or CSF PCR). Twenty-three patients were diagnosed with meningococcal meningitis and 14 patients were diagnosed with pneumococcal meningitis using the gold standard.

The study found that CSF PCR was more sensitive than Gram stain or CSF culture for diagnosing meningococcal meningitis and pneumococcal meningitis. For meningococcal meningitis: CSF PCR was positive in 20 out of 23 cases (sensitivity 87%) compared with 6 out of 23 (sensitivity 27%) for CSF Gram stain and 4 out of 23 (sensitivity 17%) for CSF culture (see table 5.13). CSF culture was 100% specific, whereas CSF PCR was 96% specific: the two patients with false positive CSF PCR results had meningococcal septicaemia with probable contamination of CSF by blood. For pneumococcal meningitis, the sensitivity of CSF PCR for detecting S. pneumoniae was 100% (14 out of 14 cases) compared with 62% (8 out of 14) for CSF Gram stain and 36% (5 out of 14) for CSF culture. All techniques were 100% specific. CSF PCR was the only positive confirmatory laboratory test in 11 out of 23 patients with meningococcal meningitis and in 5 out of 14 patients with pneumococcal meningitis. Information about prior antibiotic use was not available from the medical notes.

The multiplex real-time Taqman PCR simultaneously targeting N. meningitidis (ctrA), Haemophilus influenzae (bexA) and S. pneumoniae (Ply)72 detected N. meningitidis in 89% of the 36 CSF samples from culture-confirmed cases of meningococcal meningitis [EL= III]. It detected S. pneumoniae in 91% of 23 CSF samples from culture-confirmed cases of pneumococcal meningitis. Specificity was not assessed using clinical samples.

Evidence statement

Blood PCR for suspected meningococcal disease

There is evidence from well conducted clinical studies that real-time PCR of blood samples is more sensitive than blood culture for confirming a clinical diagnosis of meningococcal disease and is highly specific. Whole blood PCR performs significantly better than serum or plasma PCR. In two clinical studies 29% to 55% of children with meningococcal disease had a negative blood culture and a positive blood PCR. The sensitivity of PCR was less affected by antibiotic administration than the sensitivity of blood culture. There is insufficient evidence from these studies to determine the diagnostic accuracy of whole-blood PCR in children with meningococcal meningitis without septicaemia.

CSF PCR for suspected bacterial meningitis (including meningococcal meningitis)

There is limited evidence about the diagnostic accuracy of CSF real-time PCR.

One small retrospective study found that duplex CSF real-time PCR was more sensitive than Gram stain or CSF culture for diagnosing meningococcal or pneumococcal meningitis in a clinical setting. CSF PCR was highly specific (96% to 100%). One small laboratory study found that CSF multiplex real-time PCR detected N. meningitidis in 89% and S. pneumoniae in 91% of culture-positive CSF samples.

Cost effectiveness

There is variation in the use of PCR for the diagnosis of meningococcal disease and bacterial meningitis in England and Wales. Therefore, the GDG identified this as an important priority for economic analysis in order to inform guideline recommendations. A summary of this analysis is presented here, with full details given in appendix I.

A model was developed for a population of children presenting to secondary care with a suspicion of meningococcal disease. In this population three diagnostic strategies were compared:

  1. routine PCR and blood culture to all
  2. blood culture to all followed by PCR only if the blood culture is negative
  3. routine ‘rapid’ PCR and blood culture to all.

The first two strategies were thought by the GDG to represent current practice. At present, the GDG does not consider that the NHS has the necessary infrastructure to offer a rapid PCR strategy and in that sense it can currently be considered only as a hypothetical option. Nevertheless, it was considered useful to include it in the model as it is a strategy for which the technology exists and could plausibly be available in the future.

Antibiotic treatment is generally initiated on admission in those with suspected meningococcal disease or bacterial meningitis. This is a conservative approach to minimise adverse outcomes in actual cases. Therefore, confirmation of the diagnosis may sometimes be used as a basis for discontinuation of treatment and hospital discharge but not to initiate treatment. Therefore, it was not thought that the different diagnostic strategies would lead to differences in outcomes, and consequently the model took the form of a cost-minimisation analysis.

The results for the base-case analysis are shown in table 5.14.

Table 5.14. Base-case costs for alternative diagnostic strategies.

Table 5.14

Base-case costs for alternative diagnostic strategies.

While strategy 2 produces some savings in terms of a reduction in PCR tests ordered, this saving is of a relatively small magnitude because most blood culture results are negative which means that a PCR is then needed to confirm the diagnosis. The model assumes that PCR results would be available three days after admission in strategy 1 compared to five days in cases where PCR was ordered in strategy 2. In strategy 3 the PCR result is available 24 hours after admission. Therefore, the strategies with routine PCR are cheaper overall because the earlier availability of the PCR result facilitates earlier hospital discharge and discontinuation of treatment in some cases which generates a saving which more than offsets the additional PCR costs.

However, there was considerable uncertainty around some of the model parameters particularly with respect to the proportion of patients where an earlier negative PCR result would result in earlier discharge. Therefore, sensitivity analysis was undertaken to explore scenarios in which strategy 2 might be considered cost effective; for example, increasing the proportion of patients who were relatively well and would no longer be suspected of meningococcal disease following a negative blood culture. This subset of patients in the model would be discharged after the negative blood culture, obviating the need to order a PCR in strategy 2. While the sensitivity analysis showed that there were scenarios in which strategy 2 was cheaper, the GDG considered the parameter values to make this happen were outside their plausible ranges.

GDG interpretation of the evidence

PCR testing of blood samples for suspected meningococcal disease

There is high level evidence to support the use of real-time whole blood PCR using ethylenediaminetetraacetic acid (EDTA) for the diagnosis of meningococcal septicaemia in children and young people. There is evidence that PCR remains positive even if taken after antibiotics have been given, when blood culture is likely to be negative. However, a negative PCR test result for N. meningitidis does not rule out meningococcal disease. An economic analysis suggested that routine PCR was cheaper than a strategy in which ordering a PCR was conditional on a negative blood culture. As noted above, the GDG felt that a strategy of rapid PCR and blood culture to all was only a hypothetical option as the NHS currently lacks the necessary infrastructure to provide it.

PCR testing of CSF for suspected bacterial meningitis (including meningococcal meningitis)

Although there is no high level evidence to support the use of CSF real-time PCR for the diagnosis of meningococcal or pneumococcal meningitis in children and young people, most of the evidence was gathered in an era before the routine use of such tests. Emerging low level evidence supports the utility of these tests in establishing a diagnosis of bacterial meningitis and identifying the causative organism.

Further evaluation of this test in supporting a diagnosis of meningitis will be necessary. Real-time PCR may help to confirm the diagnosis in those children in whom microscopy and culture of CSF has not shown an organism. Limited evidence suggests that PCR may remain positive for up to 72 hours after antibiotics have been administered. The consensus view of the GDG was that samples retrieved from other blood sciences laboratories may be useful and CSF samples taken up to 96 hours after admission to hospital may give useful results.

Confirmation of the diagnosis helps to determine the appropriate antimicrobial chemotherapy and its duration. Confirmation of the diagnosis of meningococcal disease and bacterial meningitis is also important in assessing the effectiveness of current vaccine policy and will assist the assessment of the need for future vaccines.

Recommendations

Polymerase chain reaction (PCR) tests for bacterial meningitis and meningococcal disease

Perform whole blood real-time PCR testing (ethylenediaminetetraacetic acid [EDTA] sample) for N. meningitidis to confirm a diagnosis of meningococcal disease.

The PCR blood sample should be taken as soon as possible because early samples are more likely to be positive.

Use PCR testing of blood samples from other hospital laboratories if available, to avoid repeating the test.

Be aware that a negative blood PCR test result for N. meningitidis does not rule out meningococcal disease.

Submit CSF to the laboratory to hold for PCR testing for N. meningitidis and S. pneumoniae, but only perform the PCR testing if the CSF culture is negative.

Be aware that CSF samples taken up to 96 hours after admission to hospital may give useful results.

5.4. Skin samples and throat swabs for meningococcal disease

Introduction

Diagnostic tools that have been used historically in children and young people with suspected meningococcal disease are microscopy and culture of skin scrapings and nasopharyngeal (throat) swabs. With the advent of real-time PCR testing for N. meningitidis it is important to decide whether examination of skin lesions or throat swabs remains useful for confirming the diagnosis of meningococcal disease.

Clinical question

What is the diagnostic value of microscopy and culture of skin aspirates in children and young people with meningococcal septicaemia?

In children and young people with suspected meningococcal disease what is the diagnostic value of throat swabs?

Previous UK guidelines

The SIGN guideline on ‘Management of Invasive Meningococcal Disease in Children and Young People’ states that in three studies examination of aspirates or scrapings from skin lesions was useful in providing rapid diagnosis of invasive meningococcal disease. The guideline states that, because of the lack of a consistent gold standard and differences in the nature of lesions and techniques, it was not possible to show if examination of skin lesions is more effective in diagnosing invasive meningococcal disease than other tests.27

The SIGN guideline found insufficient evidence to form recommendations on the use of throat swabs.

Studies considered in this section

All study designs evaluating the role of laboratory examination of skin lesions and throat swabs in the diagnosis of meningococcal disease were considered for this section. Diagnostic accuracy studies without a defined gold standard were excluded.

Overview of available evidence

Two retrospective studies [EL=III] and one prospective cohort study [EL=III] evaluating the role of laboratory examination of skin lesions were included in the review.

No studies were found evaluating the role of laboratory examination of throat swabs.

Review findings

One retrospective study (1988–1994) [EL=III] evaluated the diagnostic usefulness of Gram stain of films made from petechial scrapings by reviewing data from 52 children admitted to a children’s hospital in Ireland with laboratory confirmed meningococcal disease.73 Meningococcal disease was defined using these laboratory criteria: positive blood culture, positive CSF Gram stain and culture, or positive microscopy of skin scrapings. Petechiae were found in 35 of 52 children, of whom 30 had scrapings taken by the attending clinician. Of these children, 11 had received preadmission antibiotics. Gram-negative diplococci were detected in petechial scrapings from 24 out of 30 children (80%); blood culture was positive in 11 of these 30 children (37%); and CSF microscopy and culture were positive in 6 out of 26 children (23%). Seventeen children had a negative blood culture but positive petechial scraping microscopy (57%). Of the 26 children who had a lumbar puncture and petechial scrapings, 17 had a negative CSF examination and positive petechial scraping microscopy (65%). In 14 cases, diagnosis of meningococcal disease was based on positive petechial scraping results alone. Previous antibiotic treatment did not seem to affect petechial scraping microscopy results (P = 0.372) but was associated with significantly fewer positive blood cultures (P = 0.04) and significantly fewer positive CSF Gram stain and cultures (P < 0.05).

When all 52 cases of confirmed meningococcal disease were taken into account, Gram stain of petechial scrapings was not significantly more effective than blood culture or CSF examination in detecting meningococcal infection. Blood culture was positive in 19 out of 52 children (37%), CSF Gram stain was positive in 23 out of 48 (48%) and CSF culture was positive in 22 out of 48 (46%).

Positive skin film microscopy was included in the reference standard, which may lead to an overestimation of the diagnostic accuracy of this technique. The specificity of petechial scraping microscopy was not assessed.

A prospective cohort study (2001–2003) [EL=III] conducted at a university hospital in the Netherlands assessed the diagnostic value of skin biopsy of petechiae or purpura in 31 patients with suspected meningococcal disease and skin lesions.74 Skin biopsy was performed by a dermatologist. Of the cases, 72% were 16 years or younger. Meningococcal infection was defined as: positive culture of blood, CSF or skin biopsy, positive CSF Gram stain, or identification of Gram-negative diplococci in a skin biopsy plus no alternative microbiological diagnosis and response to antibiotics. Of the 31 patients, 25 had confirmed meningococcal infection according to these criteria. An additional 12 skin biopsy specimens from the dermatology department (taken from adult patients with suspected nevus nevocellularis or skin malignancy) were included as negative controls. Of the children, 92% had received antibiotics before skin biopsy. Blood culture was performed before starting intravenous antibiotics.

Gram stain of skin biopsy was positive in 10 out of 25 cases (40%). Gram stain of CSF was positive in 8 out of 14 cases (57%). Comparison of culture results found that a greater proportion of blood or CSF specimens were positive compared with skin biopsy specimens: blood culture was positive in 14 out of 25 cases (56%); CSF culture was positive in 7 out of 14 cases (50%); and skin biopsy culture was positive in 9 out of 25 cases (36%). When results of culture and Gram stain were combined, the proportion of positive results among the different types of specimen was similar: CSF examination was positive in 9 out of 14 cases (64%) and skin biopsy examination was positive in 14 out of 25 cases (56%). In 14 patients the diagnosis was based on positive microbiology from one type of sample: CSF in 7 patients, blood in 4 patients and skin biopsy in 3 patients. There were no false positive results for the 6 clinical controls and the 12 dermatology specimens.

A retrospective study (2000–2006) [EL=III] aimed to determine the diagnostic usefulness of meningococcal real-time PCR performed on biopsy of skin lesions in patients with clinical purpura fulminans (defined as septic shock, extensive purpura and disseminated intravascular coagulation).75 In total, 34 patients (27 children aged 5 months to 15 years) were admitted with purpura fulminans to the intensive care units of a university hospital in France. Real-time ctrA Taqman PCR and culture was performed on biopsy specimens taken from ‘necrotic or ecchymotic lesions’ or from ‘petechial purpura’ after cleaning with local antiseptic. Results of skin biopsies from nine patients with purpuric lesions who did not fulfil all the criteria for purpura fulminans were used as negative controls. Blood culture was performed on all 34 patients; 17 patients had serum PCR. Most patients had been given pre-hospital antibiotics. Skin biopsy was carried out within 24 hours of antibiotic administration.

The study found that PCR of skin biopsy was significantly more sensitive than culture of skin biopsy or blood culture for detecting N. meningitidis (P < 0.0001). Skin biopsy PCR was positive in 34 out of 34 cases (100%) whereas culture of skin biopsy was positive in 5 out of 34 cases (15%). Blood culture was positive in 4 out of 34 cases (12%). Skin biopsy PCR was significantly more sensitive than serum PCR in detecting N. meningitidis (P = 0.023): skin biopsy PCR was positive in 17 out of 17 cases (100%); serum PCR was positive in 10 out of 17 cases (59%). There were no false positive PCR results for the negative controls.

Evidence statement

One retrospective study found that in children with suspected meningococcal disease and petechiae, Gram stain of petechial scrapings was positive more frequently than blood culture or CSF Gram stain or culture and was the only positive microbiological result in approximately 50% of cases. Prior antibiotic treatment was associated with fewer positive blood and CSF cultures but did not affect the positivity of petechial scraping microscopy.

One prospective study found that microscopy and culture of biopsy specimens taken from petechiae and purpura was as effective as blood culture in detecting meningococcal infection.

One retrospective study found that in patients with purpura fulminans who had received antibiotics, real-time PCR of biopsy specimens taken from ecchymoses or petechiae detected N. meningitidis more frequently than culture of skin biopsy or blood culture. Skin biopsy PCR was more sensitive than serum PCR.

Each of these small studies assessed different techniques used on different types of skin lesion. Two studies reported no clinical gold standard with inclusion of the index test in the reference standard. Specificity of skin lesion examination was not adequately addressed. Because of these limitations, the value of skin lesion examination for diagnosing meningococcal disease cannot be reliably assessed from these studies.

No evidence was identified in relation to the effectiveness of throat swabs.

GDG interpretation of the evidence

The laboratory examination of skin scrapings is not widely used in England and Wales as a diagnostic tool in children and young people with suspected meningococcal disease and in the modern NHS it is unlikely to be undertaken in settings other than an intensive treatment unit (ITU). Practice is unlikely to change in the foreseeable future, making it unlikely that skin scrapings will be undertaken to support the diagnosis of meningococcal disease in children.

There is no high-level evidence to support the use of microscopy and culture of skin lesions for the diagnosis of meningococcal disease. Limited evidence (mostly prior to the routine availability of whole-blood real-time PCR) indicates that in children with petechiae in whom meningococcal disease is suspected, particularly those given prior antibiotic treatment, Gram stain of petechial scrapings may help to confirm the diagnosis.

One small study suggests that PCR of skin biopsy specimens in purpura fulminans is more sensitive than PCR of serum. However, the available evidence is not sufficient to recommend routine use of microscopy and culture or PCR of skin scrapings for the diagnosis of meningococcal disease, particularly in the absence of data comparing the usefulness of skin scraping examination with whole-blood PCR.

The whole-blood PCR test in clinical practice has replaced skin scraping examination and the evidence does not support a return to the use of skin scraping for the diagnosis of meningococcal disease.

The GDG is aware that the SIGN Guideline on ‘Management of Invasive Meningococcal Disease in Children and Young People’27 found insufficient evidence on which to base a recommendation about the usefulness of throat swabs for the diagnosis of meningococcal disease. Meningococci are organisms that colonise the human nasopharynx asymptomatically in up to 10% of the population, with higher rates among adolescents and much lower rates in younger children. For this reason it follows that isolation of the organism from a throat swab cannot indicate invasive disease. In view of these observations and the lack of evidence on which to base a recommendation, the GDG came to a consensus that there could be no justification in undertaking throat swabs as a diagnostic test. Diagnosis should be made by isolation/detection of the organism in a normally sterile site (for example blood or CSF).

A review of patients on the Public Health Laboratory Service Meningococcus Reference Unit (MRU) database between 1994 and 1997, where both nasopharyngeal and systemic isolates were submitted, showed the organisms from both sites were identical in 97% (134 out of 138) of cases. However, in 3% of cases they were different, and a nasopharyngeal isolate in the absence of a systemic isolate does not confirm invasive disease.15 This suggests that if the diagnosis of meningococcal disease is confirmed by blood PCR, then a meningococcal isolate obtained from the throat is likely to be the cause of the systemic infection (at least in 97% of cases). However, the clinical application of this is limited, and the GDG consensus remained that throat swabs should not be used for diagnosis of meningococcal disease.

Recommendations

Skin samples and throat swabs for meningococcal disease

Do not use any of the following techniques when investigating for possible meningococcal disease: skin scrapings, skin biopsies, petechial or purpuric lesion aspirates (obtained with a needle and syringe), or throat swabs.

5.5. Performing lumbar puncture and interpreting cerebrospinal fluid parameters for suspected bacterial meningitis

Introduction

In cases of suspected meningitis, cerebrospinal fluid (CSF) is routinely obtained by lumbar puncture and examined for the presence of white blood cells (WBCs), red blood cells (RBCs), and protein and glucose concentrations (the latter interpreted as a ratio using a laboratory-determined blood glucose taken at the same time as the CSF). Taken together, these CSF variables can provide a rapid early guide to the probability of the patient having bacterial meningitis, even when bacteria are not detected on CSF Gram staining. Normal ranges for CSF variables vary slightly between laboratories, but approximate values are shown below.

  • opening pressure: 10–100 mmH2O (age under 8 years); 60–200 mmH2O (over 8 years)
  • appearance to the naked eye: clear and colourless
  • total protein concentration: 0.15–0.45 g/litre
  • glucose concentration: 2.78–4.44 millimole/litre (approximately 60% of the plasma value)
  • cell count (per microlitre): 0–5 WBCs (0–20 in neonates), no RBCs (if RBCs are present and the blood WBC count is within the normal range, more than one WBC per 500–1000 CSF RBCs can be expected in a child or young person with meningitis and should not be ignored) 76

The difficulty in interpreting CSF samples containing red blood cells (traumatic lumbar punctures) is well recognised.77;78 It has been reported that there is no advantage of adjusting leukocytes and neutrophils in CSF containing blood cells, suggesting that absolute white cell counts should be used rather than adjusted counts.

An increased CSF opening pressure is common, but not invariable, in bacterial meningitis. A CSF opening pressure greater than 250 mmH2O indicates raised intracranial pressure. CSF containing a high number of WBCs or RBCs (more than 200 WBCs or more than 400 RBCs per microlitre) may appear turbid to the naked eye. Overt turbidity due to the presence of WBCs is usually an indication of bacterial meningitis.

An increased CSF protein concentration may be due to the presence of blood in the CSF, polyneuritis, tumour, injury or any inflammatory or infectious condition of the central nervous system (CNS), including bacterial meningitis. Protein concentrations seen in bacterial meningitis are usually higher than in viral meningitis, and CSF protein levels may be particularly high in TB meningitis. A decreased CSF glucose concentration (CSF plasma to glucose ratio of less than 0.6) may be due to bacterial meningitis, including TB.

An increased WBC count in the CSF is usually an indication of bacterial or viral meningitis, but may also be found in cerebral or spinal abscesses, encephalitis and acute disseminated encephalomyelitis, following seizures, and in some non-infectious disorders (such as acute leukaemia). RBCs in the CSF sample are commonly the result of a traumatic lumbar puncture, but may also indicate bleeding in the CNS.

Differentiation of the CSF WBCs can also be useful. A raised CSF polymorphonuclear (PMN) cell count is usually indicative of bacterial meningitis, whereas a lymphocytic CSF is more often associated with viral meningitis. However, it is important to note that a raised CSF PMN cell count can also occur with viral aetiologies (for example herpes simplex virus or enterovirus meningitis). In addition, lymphocytic CSFs (or a mixture of PMN cells and lymphocytes) are not uncommon in the early stages of bacterial meningitis, especially in cases where oral antibiotics have been given prior to lumbar puncture. Lymphocytes may also be the predominant cell type in TB meningitis.

CSF bacterial culture is routinely performed. However, it should be noted that staining and culture for Mycobacterium tuberculosis is only normally performed when specifically requested by the clinician and/or clinical details, including risk factors for TB, are provided. If TB meningitis is suspected on clinical grounds, approximately 5 ml of CSF should be sent for examination to enhance the sensitivity of staining for acid-fast bacilli (which is only rarely positive) and culture. TB PCR should also be considered, and the case should be discussed with a clinical microbiologist and an infectious disease specialist.

Meningococcal PCR testing of CSF (in addition to whole-blood PCR) should also be performed in cases of suspected meningococcal meningitis.

PCR testing for viruses (for example HSV, enteroviruses) should also be considered depending on the clinical presentation and CSF variables. It should be noted that a CSF WBC count in the normal range, or the presence of PMN cells in the CSF, does not exclude viral meningitis.

Clinical question

In children and young people with suspected meningitis, can CSF variables (white blood cell count, glucose, protein) distinguish between bacterial and viral meningitis?

Previous UK guidelines

No previous guidelines were identified in relation to this question.

Studies considered in this section

All study designs evaluating the diagnostic accuracy of tests for CSF white blood cell count, CSF protein or CSF glucose to discern bacterial meningitis from viral or aseptic meningitis were considered for inclusion in this section. The majority of studies were retrospective and only those conducted in high income countries were included. Studies of adults and children were included where data were presented separately for child participants. Findings were presented in three age groups: all children, pre-school children and neonates. Overview of available evidence

CSF white blood cell count

Eleven studies examined the value of CSF WBC count to differentiate between bacterial and aseptic or viral meningitis.35;62–64;66;79–84 Nine were retrospective studies [EL=III] that generally extracted relevant demographic, clinical and laboratory test data from emergency department admission notes and compared these for children who were subsequently given a confirmed diagnosis of bacterial, viral or aseptic meningitis. Two studies recruited participants and collected data prospectively80;81 [EL=II]. Seven studies detailed exclusion criteria.62;63;66;81–84 Nine included children of broad age groups, one study83 included younger children only (1 month to 3.5 years) and one84 included neonates.

Bacterial meningitis was compared to viral meningitis in six studies,35;64;66;81–83 to aseptic meningitis in two studies62;63 and to more than one non-bacterial type in three others.79;80;84

CSF protein

Ten studies examined CSF protein concentration to differentiate between bacterial and aseptic or viral meningitis.35;62–64;66;79–81;83;84 Eight collected data retrospectively from case notes [EL=III] while two recruited participants and recorded data prospectively80;81 [EL=II]. Eight included children of broad age groups, one study83 included infants only (1 month to 3.5 years) and one84 included neonates.

Bacterial meningitis was compared to viral meningitis in six studies,35;64;66;81;83 to aseptic meningitis in two studies62;63 and to more than one non-bacterial type in three others.79;80;84 Three of the studies described confirmation of the diagnosis of a viral causative agent.64;81;83

CSF glucose

Eight studies35;62;63;66;80;81;83;84 were identified that assessed the diagnostic value of CSF glucose tests to discriminate between bacterial and viral, aseptic and/or nonbacterial meningitis. Five studies included children of all ages, one study included infants83 and one included neonates.84 Two of these studies were prospective80;81 [EL=II and the remainder retrospective.

Bacterial meningitis was compared to viral meningitis in four studies,35;66;81;83 to aseptic meningitis in two studies62;63 and to more than one non-bacterial type in two others.80;84

Review findings

CSF white blood cell count

Children of all ages

Four studies35;64;66;82 [EL=II to III] reported that the mean or median82 CSF while blood cell (WBC) count was significantly higher in bacterial meningitis compared to viral meningitis (see table 5.15). Only two of these four studies64;82 reported that all samples were systematically tested for viral agents: in the other two studies35;66 diagnosis was based on a combination of chart review (for example no report of antibiotic therapy, recorded diagnosis of viral meningitis) and a proportion of samples having been tested for viral infection.

Table 5.15. Cerebrospinal fluid (CSF) white blood cell (WBC) count - descriptive statistics (children of all ages).

Table 5.15

Cerebrospinal fluid (CSF) white blood cell (WBC) count - descriptive statistics (children of all ages).

Two studies79;80 [EL=III and EL=II respectively] reported CSF WBC counts for children with bacterial, viral and undetermined meningitis. Although a P value was not given in either study, the findings for undetermined meningitis (UM) were of a similar magnitude across the two studies (UM: mean 431 WBCs/ml, SD 772 WBCs/ml and UM: 264 WBCs/ml, SD 204 WBCs/ml respectively). Results for the bacterial and viral groups were consistent across the two studies and with the previously mentioned studies (see table 5.15). Three older studies included children with meningitis where H. influenzae was the causative agent in at least 50% of cases (see table 5.15).

Two retrospective studies sought to discriminate between bacterial meningitis and aseptic meningitis62;63. The first study, which was a secondary analysis of multicentre data, included 96 cases of bacterial meningitis (from a total n=198). The second study (n=167) included 21 children with bacterial meningitis. Both studies reported that the median CSF WBC count was significantly higher in bacterial meningitis compared to aseptic meningitis (both P < 10−6). However, neither demonstrated that CSF WBC was a strong predictor for distinguishing bacterial from aseptic meningitis. The first study estimated Area Under Curve (AUC) as 0.81 and that a CSF WBC count above the threshold of 200 cells/microlitre was significantly associated with bacterial meningitis (sensitivity 76%, specificity 75%, OR=9, 95% CI 3 to 32, P < 10−5). The secondary analysis reported similar findings for the same threshold (sensitivity 79%, specificity 69%, OR=8.3, 95% CI 4.1 to 16.9).

Three studies66;81;82 gave estimates of sensitivity, specificity, PPV and NPV for different thresholds of CSF WBC counts to discriminate between bacterial and viral meningitis. One study80 combined viral and undetermined meningitis groups to compare the bacterial to a non-bacterial meningitis group (see table 5.16). The four studies that presented findings for a threshold of 500 cells/microlitre ranged in size (n=45 to 237), included locally available populations, variously included Gram-negative or both Gram-negative and Gram-positive bacteria, and used different data collection methods. No consistent findings for sensitivity, specificity, PPV and NPV were reported. One study66 compared three different thresholds but did not find any threshold value of CSF WBC count conferring high sensitivity, specificity and NPV to the test.

Table 5.16. Cerebrospinal fluid (CSF) white blood cell (WBC) count – diagnostic statistics (children of all ages).

Table 5.16

Cerebrospinal fluid (CSF) white blood cell (WBC) count – diagnostic statistics (children of all ages).

Pre-school children

One retrospective study83 [EL=III] of children aged 1 to 42 months found that the mean CSF WBC count was also significantly higher in bacterial meningitis than in viral meningitis in this younger age group (P < 0.0001).

Neonates

A retrospective study of neonates (defined as age under 4 weeks)84 (n=72 of whom 18 had bacterial meningitis) [EL= III] found that all viral and aseptic meningitis cases had a CSF WBC above a threshold of 22 cells/microlitre, compared to 83% of bacterial meningitis cases. However, this was a small study (bacterial meningitis n=18, viral meningitis n=13 and aseptic meningitis n=41), and neonates who had received antibiotic treatment of assessment were excluded, as were those whose lumbar puncture was ‘traumatic’ (more than 1000 RBC/mm3) unless the CSF culture tested positive for bacteria. This could explain why fewer neonates with bacterial meningitis had a CSF WBC more than 22 cells/microlitre.

CSF protein

Children of all ages

Three studies35;64;66 [EL=III] reported that the mean CSF protein concentration was significantly higher in bacterial meningitis compared to viral meningitis. Two studies79;80 [EL=III and EL=II, respectively] reported CSF protein concentration for children with bacterial, viral and undetermined meningitis. Although no P value was reported, the findings for undetermined meningitis were similar across both studies and results for the bacterial and viral groups were of similar magnitude to those studies where P values were reported (see table 5.17).

Table 5.17. Cerebrospinal fluid protein concentration - descriptive statistics (children of all ages).

Table 5.17

Cerebrospinal fluid protein concentration - descriptive statistics (children of all ages).

Three studies35;66;81 estimated the diagnostic accuracy of CSF protein concentration to discriminate between bacterial and viral meningitis providing estimates of sensitivity, specificity, PPV and NPV at different thresholds. One study80 compared bacterial to ‘non-bacterial meningitis (see table 5.18). The four studies that presented findings for a CSF protein concentration threshold of 100 mg/decilitre ranged in size (n=45 to 237), variously included bacteria which were Gram-positive or Gram-negative or both, and used different data collection methods. The best results for accuracy were reported in a small prospective study66 (n=45) but were not replicated elsewhere. No consistent findings for sensitivity, specificity, PPV and NPV were reported. One study66 presented findings for two different thresholds (1.0 g/litre and 1.5 g/litre) but did not find a threshold value of CSF protein concentration conferring high sensitivity and NPV to the test.

Table 5.18. Cerebrospinal fluid (CSF) protein concentration – diagnostic statistics (children of all ages).

Table 5.18

Cerebrospinal fluid (CSF) protein concentration – diagnostic statistics (children of all ages).

Two retrospective studies [EL=III] evaluated the predictive value of CSF protein concentration to discriminate between bacterial meningitis and aseptic meningitis62;63. The first study, which was a secondary analysis of multicentre data, recruited 96 cases of bacterial meningitis (n=198). The second study (n=167) included 21 children with bacterial meningitis. Both studies reported that the median CSF protein concentration was significantly higher in bacterial meningitis compared to aseptic meningitis (both P < 10−6). However, neither demonstrated that CSF protein concentration was a strong predictor for distinguishing bacterial from aseptic meningitis. The first analysis reported a lower area under the curve (AUC) estimate of 0.88, lower specificity and a lower OR for the same threshold (sensitivity 88%, specificity 65%, OR=14.2, 95% CI 6.3 to 32.7). The second study estimated the AUC as 0.93 and that a CSF protein concentration above the threshold of 0.5 g/litre was significantly associated with bacterial meningitis (sensitivity 86%, specificity 78%, OR=22, 95% CI 6 to 101, P < 10−8; adjusted OR=34, 95% CI 5 to 217, P < 10−3; adjustment for blood CRP, CSF WBC and neutrophil count).

Pre-school children

One retrospective study83 [EL=III] of children aged 1 to 40 months found that the mean CSF protein concentration was significantly higher in bacterial meningitis compared to viral meningitis in this younger age group (bacterial meningitis mean 1.5 g/litre, SD 1.0 g/litre versus viral meningitis 0.4 g/litre, SD 0.2 g/litre, P < 0.0001).

Neonates

A retrospective study of neonates84 [EL=III] (n=72) found that all viral and aseptic meningitis cases had a CSF protein concentration below a threshold of 1.70 g/litre, but only 56% of bacterial meningitis cases had a CSF protein concentration above this level. This threshold conferred high specificity and PPV, but a low sensitivity for identification of bacterial from non-bacterial meningitis (sensitivity 55.6%, specificity 100%, PPV 100%, NPV 87.1%).

CSF glucose

Children of all ages

Two studies [EL=III] compared the mean CSF glucose in children with bacterial meningitis to those with viral meningitis.35;66 Although the results showed that the mean CSF glucose was higher in viral than in bacterial meningitis in both studies, only one found that this was statistically significant [EL=III]. A third study [EL=II] that included children with viral and aseptic meningitis also reported that those with viral meningitis had a higher mean CSF glucose than those with bacterial meningitis although no P value was given (see table 5.19).80

Table 5.19. Cerebrospinal fluid (CSF) glucose concentration - descriptive statistics (children of all ages).

Table 5.19

Cerebrospinal fluid (CSF) glucose concentration - descriptive statistics (children of all ages).

Two retrospective studies [EL=III] compared the mean CSF glucose concentrations found in bacterial meningitis and aseptic meningitis62;63. Both studies reported that the median CSF glucose concentration was significantly higher in aseptic meningitis than in bacterial meningitis (both P = 0.01 and P < 10−6, respectively). A third study was a small (n=56) prospective study80 [EL=III] including children with viral and aseptic meningitis: this also reported that those with aseptic meningitis had a higher mean CSF glucose than those with bacterial meningitis, although no P value was reported.

Three studies35;66;81 gave details of the diagnostic accuracy of CSF glucose concentration in discriminating between bacterial and viral meningitis providing estimates of sensitivity, specificity, PPV and NPV at different thresholds (2.0 millimole/litre, 2.2 millimole/litre and 2.5 millimole/litre). Although optimal specificity was reached in one study at a cutoff value of 2.0 millimole/litre66, sensitivity was consistently low for this threshold and all others investigated. The best results were found in the study comparing bacterial to ‘non-bacterial meningitis80 (sensitivity=78%; see table 5.20).

Table 5.20. Cerebrospinal fluid (CSF) glucose concentration - diagnostic statistics (children of all ages).

Table 5.20

Cerebrospinal fluid (CSF) glucose concentration - diagnostic statistics (children of all ages).

Two retrospective studies [EL=III] estimated diagnostic accuracy of CSF glucose concentration to discriminate between bacterial and aseptic meningitis at a 2.5 millimole/litre threshold.62;63 The first analysis, which included a larger proportion of bacterial meningitis cases, reported slightly better results at the same threshold (sensitivity 67%, specificity 82%, OR=9.3, 95% CI 4.5 to 19.3). The second study estimated that a CSF glucose concentration above the threshold was significantly associated with aseptic meningitis (sensitivity 62%, specificity 78%, OR=6, 95% CI 2 to 17, P < 10−3). However, neither demonstrated that CSF protein concentration was a strong predictor for distinguishing bacterial from aseptic meningitis.

Pre-school children

One retrospective study83 [EL=2−] reported that the mean CSF glucose concentration was significantly higher in viral meningitis than in bacterial meningitis in a younger age group (1 month to 3.5 years) (bacterial meningitis: 1.6 millimole/litre, SD 1.3 millimole/litre versus viral meningitis: 3.2 millimole/litre, SD 0.7 millimole/litre; P < 0.0001).

Neonates

One study of neonates [EL=2−] reported estimates of diagnostic accuracy at a CSF glucose threshold of 1.87 millimole/litre.84 In this study, 11 out of 18 bacterial meningitis cases (61%) had results below this level, as did 7 out of 13 viral meningitis cases (54%) and 7 out of 41 aseptic meningitis cases (17%).. Comparing the results for bacterial meningitis to the combined results for non-bacterial meningitis did not result in clinically meaningful diagnostic accuracy estimates (sensitivity 61 %, specificity 74%, PPV 44%, NPV 85%).

Evidence statement

CSF white blood cell count

There is consistent evidence from eight studies of children of all ages and evidence from one study in pre-school children that CSF white blood cell (WBC) count was significantly higher in bacterial meningitis compared to viral, aseptic and non-bacterial meningitis. Because of the clinical need to reliably discriminate between children with bacterial meningitis and viral meningitis, the diagnostic accuracy of a test should include a high sensitivity. High sensitivity was not found in any study of children at a threshold of 500 cells/microlitre.

Results from a study in neonates suggested that a threshold of 22 cells/microlitre would not have sufficient diagnostic accuracy to discriminate non-bacterial from bacterial meningitis.

CSF protein

CSF protein concentration was consistently reported to be significantly higher in bacterial meningitis compared to viral, aseptic or non-bacterial meningitis in children. No clinically reliable threshold to discriminate between bacterial and viral or aseptic meningitis was determined for CSF protein concentration in children. In neonates, although a threshold was identified under which the CSF protein concentration for all non-bacterial meningitis cases occurred, 44% of bacterial meningitis cases also had these lower results.

CSF glucose

Evidence from two studies of children demonstrated that the mean CSF glucose concentration was significantly higher in aseptic meningitis compared to bacterial meningitis. There were inconsistent findings for the comparison between viral and bacterial meningitis for this age group, although in a study of infants, the mean CSF glucose concentration was significantly higher in viral meningitis than in bacterial meningitis. No clinically reliable threshold to discriminate between bacterial and viral, aseptic and/or nonbacterial meningitis was determined for CSF glucose concentration in children or in neonates.

GDG interpretation of the evidence

Although evidence has been found that there are significant differences in CSF WBC count and protein and glucose concentrations between bacterial and other forms of meningitis, no single variable has been shown to have sufficient diagnostic accuracy to confirm or exclude bacterial meningitis. The GDG is aware of some limited evidence that the presence of polymorphonuclear cells in CSF and the CSF plasma to glucose ratio are independent predictors of bacterial meningitis. In some of the included studies the absence of a positive CSF bacterial culture was used to indicate the absence of bacterial meningitis (‘aseptic meningitis’). In these studies, true cases of bacterial meningitis will be defined as ‘aseptic meningitis’ due to the low sensitivity of CSF bacterial culture.

Given that CSF variables cannot reliably exclude bacterial meningitis, the GDG was of the opinion that CSF WBC counts outside the accepted normal ranges should prompt the initiation of appropriate antibiotic therapy in cases of suspected bacterial meningitis (if antibiotics have not been started prior to the lumbar puncture). While a low CSF to plasma glucose ratio is also an indicator of bacterial meningitis in children aged over 28 days, no evidence was identified to indicate that this variable is commonly abnormal in the presence of a normal CSF WBC count. Recognising the lower sensitivity of the CSF WBC count for bacterial meningitis in neonates, the GDG was also of the opinion that bacterial meningitis should still be considered in neonates in whom the CSF WBC count is within the currently accepted normal range (less than 20 cells/microlitre). Furthermore, the GDG is aware of recent evidence that suggests that the CSF WBC count range in normal neonates is the same as that in older children and adults (less than 5 cells/microlitre) and that mild CSF pleocytosis (which may occur in symptomatic neonates without central nervous system infection) cannot be regarded as a normal finding.85

A particular problem is the interpretation of CSF findings in neonates: there is insufficient evidence to guide recommendations for defining the likelihood of bacterial meningitis in this age group. Performance characteristics of meningitis scoring systems based on blood test results and CSF findings have been studied in some populations and similar studies in the UK could improve the diagnosis or exclusion of bacterial meningitis. Studies are, therefore, needed to determine the ‘normal’ ranges of blood and CSF parameters in children and young people. The studies should include previously healthy children found to have aseptic meningitis as well as those in whom bacterial meningitis is confirmed.

Recommendations relating to the interpretation of CSF parameters (white blood cell count, glucose, protein) are presented in section 5.7.

5.6. Contraindications to lumbar puncture

Introduction

Definitive diagnosis of meningitis requires microscopy, biochemical analysis and PCR analysis of a sample of CSF. Without a CSF sample, the resultant incomplete diagnosis detracts from clinical management. Bacterial meningitis may be clinically suspected but not confirmed, antibiotic use and selection may be inadequate, duration of antibiotic treatment cannot be optimised, development of complications cannot be anticipated and information to parents, prognostication and follow-up may be less well informed. Nevertheless, there are circumstances when a lumbar puncture is contraindicated, because of a risk of complications. Usually such risk is temporary and lumbar puncture can be deferred rather than abandoned completely.

Clinical question

When is lumbar puncture contraindicated in children and young people with suspected bacterial meningitis?

When is lumbar puncture contraindicated in children and young people with suspected meningococcal septicaemia?

Previous UK guidelines

The ‘Feverish Illness in Children’ guideline25 recommends the following:

‘Red’ group

Children with fever without apparent source presenting to paediatric specialists with one or more ‘red’ features should have the following investigations performed:

The following investigations should also be considered in children with ‘red’ features, as guided by the clinical assessment:

‘Amber’ group

Children with fever without apparent source presenting to paediatric specialists who have one or more ‘amber’ features should have the following investigations performed unless deemed unnecessary by an experienced paediatrician:

The SIGN guideline on ‘Management of invasive meningococcal disease in children and young people’27 recommends:*

Lumbar puncture is not recommended in the initial assessment of suspected IMD [invasive meningococcal disease] with features of septicaemia. Lumbar puncture may be considered later if there is a diagnostic uncertainty or unsatisfactory clinical progress, and there are no contraindications.

Lumbar puncture should be performed in patients with clinical meningitis without features of septicaemia (purpura) where there are no contraindications.’

The SIGN guideline notes the following contraindications to lumbar puncture:

Studies considered in this section

This systematic review looking at contraindications to lumbar puncture in children with suspected bacterial meningitis and children with suspected meningococcal disease includes five studies, four of which were surveys based on reviews of medical records [EL=3] and one of which was a case control study of poor quality [EL=2−].

Review findings

A prospective survey conducted in Australia (1991–1992) [EL=3]86 aimed to identify the risks of performing lumbar puncture and poor outcomes associated with not performing lumbar puncture. Of the 218 children admitted to hospital with suspected meningitis, 195 (89.4%) had a lumbar puncture performed immediately. Bacterial meningitis was diagnosed in 18 of these children (31 had viral meningitis). No child developed cerebral herniation following an immediate lumbar puncture. Eleven of the lumbar punctures were defined as traumatic and two children required repeated attempts. In nine of the 18 children with bacterial meningitis the lumbar puncture provided information that was defined by the authors as useful in deciding the appropriate management of the children.

Twenty-three children did not have an immediate lumbar puncture. The main reason for delaying lumbar puncture was severe obtundation, usually with a Glasgow Coma Scale score of 7 or less. Seventeen children had a lumbar puncture later. In seven children the lumbar puncture was delayed due to suspected raised intracranial pressure. A lumbar puncture was performed after a cranial computed tomography (CT) scan showed no abnormalities. Three children in the delayed lumbar puncture group had bacterial meningitis. Six children never had a lumbar puncture performed. Five of this group had bacterial meningitis diagnosed clinically and from blood cultures or urine antigen testing. No adverse outcomes were noted in relation to not having a lumbar puncture performed.

A UK retrospective survey87 was undertaken in 2000 [EL=3] to describe usual practice at the study hospital and identify the contribution of lumbar puncture to diagnosis and management of care. Medical records were examined of 415 children to identify those with suspected central nervous system (CNS) infections (n=52) or suspected meningococcal septicaemia (n=43). No lumbar puncture was performed in children with contraindications (as defined by the authors). Of the 47 children with suspected CNS infection and no contraindications, 25 (53%) received a lumbar puncture. Contraindications were defined as:

Forty-three children had suspected meningococcal septicaemia without CNS involvement. None of these children had a lumbar puncture performed. No patient in any group died or had sequelae. Sterile CSF cultures allowed 15 of the 25 children who had a lumbar puncture to have antibiotics discontinued compared with three of the 22 children who had no contraindications but did not have a lumbar puncture (P < 0.001).

A retrospective survey conducted in Australia (1984–1989) [EL=3] was undertaken to see whether the incidence of cerebral herniation was increased immediately following a lumbar puncture for children with bacterial meningitis.88 From 445 medical records reviewed, 19 children were identified as having cerebral herniation (a total of 21 episodes; two children had two episodes of herniation). The timing of herniation compared with lumbar puncture was:

At the time of lumbar puncture three children were unresponsive to pain, three were drowsy but rousable, one had a purpuric rash and clonus of the right ankle, another had neck stiffness, and one had decerebrate posturing and a rash. Outcomes for children who had cerebral herniation were very poor: 14 of the children died, two had no long-term sequelae reported, one had hearing loss and behavioural problems noted on follow-up (timing not noted) and two were discharged with serious neurological impairment.

A UK retrospective case control study (1974–1985) [EL=2−]aimed to identify features of meningitis associated with cerebral herniation and death.89 The study included 19 children who had been diagnosed with meningitis, who had had a lumbar puncture and who had subsequently died. This group was compared with a matched control group (n=19) of children who had also been diagnosed with meningitis, had had a lumbar puncture and subsequently recovered. The children were matched for: year of admission, gender, age and infecting micro-organism. However, the degree of matching achieved was quite poor with only one child being matched on all four factors and another seven matched on three factors.

Two features of raised intracranial pressure were found to be associated with a significantly increased risk of cerebral herniation: fits on admission (5 out of 17 versus 0 out of 17; RR 7.08, 95% CI 2.2 to 22.1, P = 0.02) and Glasgow coma scale score less than 8 (10 out of 17 versus 4 out of 17; RR 4.6, 95% CI 1.06 to 35.8, P = 0.03), although due to the small numbers of children involved these findings should be interpreted with caution.

A survey conducted in part prospectively (n=71 children) and in part retrospectively (n=52 children) in Nigeria90 [EL=3] (1999) sought to determine the frequency and outcomes of possible cerebral herniation in relation to lumbar puncture. The study compared incidence and timing of cerebral herniation in high- and low-risk patients as defined by a weighted scoring system based predominantly on clinical features associated with severe or mild to moderate illness (factors included: unrousable coma (3 points), hypothermia (2 points), convulsions (2 points), shock (1 point), age under 12 months (1 point) and symptoms persisting for more than 3 days (0.5 point).

A lumbar puncture was performed on presentation in 112 children (91%) and deferred in 11. The former group contained 18 children (16%) who were defined as being at high risk compared with seven (64%) of the latter group.

Four groups of children were described among those on whom a lumbar puncture was performed on presentation:

Seventeen children who had a lumbar puncture on presentation died, including seven within 24 hours. Eight children who had deferred lumbar puncture died, seven within 24 hours of the procedure.

Evidence statement

There is evidence that cerebral herniation occurs in bacterial meningitis.

There is evidence from two surveys that lumbar puncture is associated with a very low risk of cerebral herniation where it is undertaken on children without impaired level of consciousness or other signs of raised intracranial pressure. Evidence from another two surveys shows that where there are signs of loss of consciousness or other signs of raised intracranial pressure there is an increased risk of cerebral herniation, although there is evidence to suggest that the cerebral herniation noted after lumbar puncture may, in a number of cases, have been developing before the lumbar puncture was performed.

There is no evidence on which to conclude whether or not lumbar puncture causes cerebral herniation in bacterial meningitis.

GDG interpretation of the evidence

When used appropriately, lumbar puncture can provide important clinical information in suspected bacterial meningitis. Results can help to establish the diagnosis and effective management (choice of antibiotics, length of course of antibiotics, follow up arrangements and so on). Its proper use should not be neglected on the basis of over-interpretation of perceived risk.

There was no specific evidence found about the level of platelet count which would contraindicate a lumbar puncture. However, the GDG agreed by consensus that a platelet count below 100 × 109/litre was an appropriate cutoff for both neonates and older children and young people. The GDG’s view is that a platelet count below 50 × 109/litre is not safe in children and young people with disseminated intravascular coagulation and/or shock (but it is acceptable in haematology patients with no other morbidities).

If a lumbar puncture is contraindicated (for example in children and young people with a history of haemophilia), then data from a delayed lumbar puncture may still help to establish a diagnosis or influence management.

The GDG noted that seizures were a serious complication in cases of meningitis and could be particularly difficult to manage in some patients, including those with raised intracranial pressure. Nevertheless, confirmation of diagnosis by lumbar puncture is also important. Seizures are therefore a relative contraindication to lumbar puncture and appropriate management may neutralise that contraindication. The GDG was of the opinion that local or national protocols should be available for the management of seizures associated with bacterial meningitis (see section 6.3).

The GDG noted that a reduced or fluctuating level of consciousness would correspond to a Glasgow Coma Score (or Child’s Glasgow Coma Score in the case of children under 4 years) of less than 9 or a drop of 3 or more.

Recommendations relating to contraindications to lumbar puncture are presented in section 5.7.

5.7. Repeat lumbar puncture in neonates

Introduction

Neonatal meningitis differs from bacterial meningitis in older children in various ways. The most common bacteria that cause meningitis in neonates (Group B streptococcus, L. monocytogenes and E. coli) differ from those in older children, especially in the first week of life. Meningitis that occurs later may also be caused by organisms more commonly acquired in childhood (such as S. pneumoniae). Intracranial infection of the neonate is often associated with a poor developmental outcome making it crucial to initiate timely and appropriate treatment. Premature babies are at even greater risk of meningitis caused by a large spectrum of antibiotic-resistant pathogens and associated with a worse outcome than term babies: however, the sub-population of premature babies who develop meningitis while still in hospital is outside the scope of the guideline.

Historically it is known that, despite apparently adequate courses of antibiotics, neonatal meningitis can relapse or recrudesce. To document CSF sterilisation and thereby increase the chance of successful treatment, many paediatricians have adopted the practice of repeating a lumbar puncture in neonates either early on in treatment or at the end of a course of antibiotics. However, documentation of CSF sterilisation has not been shown to guarantee that the infection will not relapse. This section considers whether repeat lumbar puncture is a useful practice in ensuring treatment success for neonatal bacterial meningitis.

Clinical question

Should lumbar puncture be performed prior to stopping antibiotic treatment in children aged less than 3 months with bacterial meningitis?

Previous UK guidelines

No previous guideline has considered this clinical question in relation to neonates.

Studies considered in this section

Studies were included for consideration in this review if they included term neonates (that is, babies born at 37 weeks’ gestation or over, aged 28 days or less). Only studies from high-income countries were included. No limits were placed on study design, thus small case series were also included due to the limited number of eligible studies conducted in this area.

Overview of available evidence

No study was identified that directly addressed the clinical question that was posed. Two retrospective reviews of medical records were identified that were considered to contribute data to help inform the GDG.

Review findings

A retrospective review (USA) [EL=3] of medical records of 128 children with definite or suspected bacterial meningitis between 1992 and 1996 was conducted in order to define the time taken to achieve a sterile CSF after the initiation of antibiotic therapy.91 Twenty-one infants (median age 21 days, interquartile range [IQR] 9 days, 31 days) had Group B streptococcus meningitis. Following parenteral antibiotic treatment (usually with a third-generation cephalosporin), none of five samples tested within 24 hours was found to be sterile. Of four tested between 24 and 72 hours, three were sterile. All of the six tested after 72 hours were found to be sterile.

In a retrospective review of medical records (1981, USA) [EL=3] clinical and laboratory features of six children with recrudescence and 21 children with relapse were reviewed: nine of the children were neonates.92 These complications occurred mainly in infants aged less than 2 years and comprised less than 1% of all cases of bacterial meningitis. Neither the initial nor follow-up CSF findings were predictive of recrudescence or relapse. Prolonged or secondary fever was unrelated to these complications. Recrudescence was usually caused by inappropriate therapy whereas relapse after adequate therapy of bacterial meningitis was usually ascribed to persistence of infection in meningeal or parameningeal foci. Relapse did not become manifest until at least 3 days after discontinuation of therapy.

Evidence statement

No evidence was found relating directly to the clinical question.

GDG interpretation of the evidence

Neonates who have persistent or re-emergent fever, deterioration in condition, new clinical findings (especially neurological findings) or persistently abnormal inflammatory markers should have imaging of the CNS and a repeat lumbar puncture as these abnormalities may signify a focus of infection. Positive imaging or a positive lumbar puncture should prompt a discussion with local microbiology specialists about choice of antibiotics and duration of treatment. Healthcare professionals could consider the use of cranial computed tomography and/or magnetic resonance imaging before repeating lumbar puncture in neonates who have persistent or re-emergent fever, deterioration in clinical condition, new clinical findings (especially neurological findings) or persistently abnormal inflammatory markers.

By consensus, the GDG considers that routine repeat lumbar puncture is not justified in neonates who are on the correct type and dose of antibiotics (based on identification of the causative organism) and are otherwise making a good clinical recovery.

The GDG considered repeat lumbar puncture before stopping antibiotic therapy is not routinely necessary, while acknowledging that some authorities suggest this should be considered. The argument for this is that the CSF white cell count, neutrophil count (or percentage), glucose concentration or protein concentration at the end of therapy may predict those who will relapse or have other complications. However, no published evidence was found to support this.

Recommendations

Performing lumbar puncture and interpreting CSF parameters for suspected bacterial meningitis

Perform a lumbar puncture as a primary investigation unless this is contraindicated.

Do not allow lumbar puncture to delay the administration of parenteral antibiotics.

CSF examination should include white blood cell count and examination, total protein and glucose concentrations, Gram stain and microbiological culture. A corresponding laboratory-determined blood glucose concentration should be measured.

In children and young people with suspected meningitis or suspected meningococcal disease, perform a lumbar puncture unless any of the following contraindications are present:

In children and young people with suspected bacterial meningitis, if contraindications to lumbar puncture exist at presentation consider delaying lumbar puncture until there are no longer contraindications. Delayed lumbar puncture is especially worthwhile if there is diagnostic uncertainty or unsatisfactory clinical progress.

CSF white blood cell counts, total protein and glucose concentrations should be made available within 4 hours to support the decision regarding adjunctive steroid therapy.

Start antibiotic treatment for bacterial meningitis if the CSF white blood cell count is abnormal:

  • in neonates at least 20 cells/microlitre (be aware that even if fewer than 20 cells/microlitre, bacterial meningitis should still be considered if other symptoms and signs are present – see table 3.3)
  • in older children and young people more than 5 cells/microlitre or more than 1 neutrophil/microlitre, regardless of other CSF variables.

In children and young people with suspected bacterial meningitis, consider alternative diagnoses if the child or young person is significantly ill and has CSF variables within the accepted normal ranges.

Consider herpes simplex encephalitis as an alternative diagnosis.

If CSF white cell count is increased and there is a history suggesting a risk of tuberculous meningitis, evaluate for the diagnosis of tuberculous meningitis in line with ‘Tuberculosis: clinical diagnosis and management of tuberculosis, and measures for its prevention and control’ (NICE clinical guideline 33).

Perform a repeat lumbar puncture in neonates with:

  • persistent or re-emergent fever
  • deterioration in clinical condition
  • new clinical findings (especially neurological findings) or persistently abnormal inflammatory markers.

Do not perform a repeat lumbar puncture in neonates:

  • who are receiving the antibiotic treatment appropriate to the causative organism and are making a good clinical recovery
  • before stopping antibiotic therapy if they are clinically well.

Research recommendations

Diagnosis in secondary care

Performing lumbar puncture and interpreting CSF parameters for suspected bacterial meningitis

What are the normal ranges for blood and CSF parameters in children and young people in the UK?

Why this is important

Bacterial meningitis is a rare disease that is not easily distinguishable clinically from aseptic meningitis. It is, however, important to recognise those children who are most likely to have bacterial meningitis to direct appropriate management of the condition and to avoid inappropriate treatment of aseptic meningitis. Since the introduction of vaccines to protect against Hib, meningococcus serogroup C and pneumococcus, no high-quality studies involving previously healthy children and young people have been conducted in the UK to determine normal ranges for blood test results or CSF findings in bacterial and aseptic meningitis. Such studies are needed to provide reference values to help interpret blood test results and CSF findings in children (especially neonates) and young people with suspected bacterial meningitis.

Does repeat lumbar puncture in neonates with bacterial meningitis alter the prognosis?

Why this is important

Bacterial meningitis in neonates differs from bacterial meningitis in older children in several ways, including the causative organisms and the risk of relapse even after a long course of antibiotics (with the risk being greater in neonates). This has led some healthcare professionals to repeat lumbar puncture before stopping antibiotic treatment to ensure that the CSF is sterile. The GDG found no evidence from which to evaluate the effectiveness of repeat lumbar puncture for preventing relapse of bacterial meningitis in neonates. A study is required in neonates with documented bacterial meningitis to determine what factors are associated with relapse and whether repeat lumbar puncture alters the prognosis. All neonates included in the study would need to receive a specified antibiotic regimen (tailored to the causative pathogen), involving similar dosages, dosing intervals and duration of treatment. The following data should be collected for each neonate in the study: signs and symptoms, blood test results (inflammatory markers), CSF findings (microbiology and chemistry) and central nervous system imaging. All variables should be measured at the start and end of treatment. Follow up should continue for 1 month after stopping antibiotic treatment, and longer-term follow-up (at 2 years) should also be conducted. Any deterioration in clinical condition should prompt a full clinical assessment, blood analysis, lumbar puncture, and imaging, from which it will be possible to evaluate the risk of relapse according to whether or not repeat lumbar puncture is undertaken.

5.8. Cranial computed tomography for suspected bacterial meningitis

Introduction

Identifying a causative organism in children and young people with suspected bacterial meningitis by examination of cerebrospinal fluid (CSF) obtained by lumbar puncture is essential to ensure optimal management.

Undertaking a lumbar puncture in children with raised intra-cranial pressure may result in cerebral herniation. Cranial computed tomography (CT) scanning prior to lumbar puncture has been advocated for children with a depressed conscious level to help determine the presence or extent of raised intracranial pressure to identify those at risk of cerebral herniation. CT scanning is also used to identify other potential causes of depressed conscious level, such as intracranial mass lesions. However, performing a CT scan might delay treatment in children with suspected meningitis and could be dangerous if undertaken in clinically unstable children. Therefore, it is essential to ensure the appropriate use of CT scanning, together with the accurate interpretation of scan results.

The ability of a CT scan to reliably detect raised intracranial pressure in children with suspected bacterial meningitis was the subject of this evidence review.

Clinical question

In children and young people with suspected or confirmed bacterial meningitis, can a cranial computed tomography (CT) scan reliably demonstrate raised intracranial pressure?

Previous UK guidelines

No previous UK guideline was identified that addressed this clinical question.

Studies considered in this section

All study designs assessing the role of CT scans in diagnosing raised intracranial pressure in children and young people with suspected or confirmed meningitis were considered for this section. Studies involving adults and children were considered for inclusion if outcomes were reported separately for children. Studies involving adults only were not considered.

Overview of available evidence

Three retrospective studies [EL=3] were found.

Review findings

One retrospective study (Sweden, 1994–1997) [EL=3] reported CT scan results of patients admitted to secondary care with bacterial meningitis and raised intracranial pressure (ICP).93 Of 53 patients with a diagnosis of bacterial meningitis, 12 (seven patients aged 2 to 16 years) had clinical evidence of increased ICP, confirmed by invasive ICP monitoring (ICP more than 20 mmHg). A cranial CT scan was performed in 10 patients prior to insertion of the ICP monitoring device. Cranial CT showed radiological signs indicating brain swelling in only 5 out of 10 patients (50%).

One retrospective review of medical records (Australia, 1984–1989) [EL=3] aimed to determine if the incidence of cerebral herniation increased immediately after lumbar puncture in children with bacterial meningitis admitted to a paediatric referral centre. The study also assessed whether any children with herniation had normal results on CT scan.88 Herniation was judged to have occurred if clinical or post-mortem findings were compatible with the diagnosis. CT scans of children with herniation and an equal number of scans from children without herniation were reviewed by a paediatric radiologist. From 445 medical records reviewed, 19 children aged 4 months to 15 years were identified as having cerebral herniation and 14 cranial CT scans were performed. Scans were performed from 1.5 hours before herniation to 18 hours after herniation. Cranial CT scan was normal in 5 out of 14 episodes of herniation (36%). The five normal scans were from four children (one child had two episodes of herniation). Two of the children with normal CT scans died: herniation was confirmed on necropsy.

One retrospective review of medical records (UK, 1986–1989) [EL=3] evaluated the role of cranial CT scan in the detection of raised intracranial pressure in 15 children transferred to a tertiary care centre with bacterial meningitis and clinical signs of raised intracranial pressure.94 Signs of raised intracranial pressure included: depressed level of consciousness with or without pupillary abnormalities, cranial nerve palsies, hyperventilation, Cheyne Stokes respiration and decorticate or decerebrate posturing. Of the 15 children with suspected raised intracranial pressure, six (40%) had a normal cranial CT scan. Scans of five children (approximately 30%) showed radiological signs of cerebral oedema. ICP measurements and the clinical outcome of children were not reported. The accuracy of CT scan for excluding raised intracranial pressure can therefore not be accurately assessed from these data.

Evidence statement

There is limited evidence from three small retrospective studies that CT scan is an insensitive technique for detection of raised intracranial pressure in children with suspected bacterial meningitis. In two studies, the clinical diagnosis of raised intracranial pressure was mostly presumptive. Studies were conducted in the 1980s and 1990s: no studies using recent CT scanning technology were found.

GDG interpretation of the evidence

Three retrospective studies were found addressing the use of CT scanning in the detection of raised intracranial pressure in children and young people with suspected or confirmed bacterial meningitis.

Although the available evidence was limited and not recent, it indicated that some children with raised intracranial pressure may have a normal CT scan. Due to the reported unreliability of CT scan for detecting raised intracranial pressure in children with suspected bacterial meningitis, the GDG saw no advantage in using CT scanning to aid in the decision regarding the safety of lumbar puncture. The decision to perform a lumbar puncture should be made on clinical grounds (see sections 5.6 and 5.7).

The GDG recognised that children with suspected bacterial meningitis who have a reduced conscious level or focal neurological signs may have alternative diagnoses, for which CT scan detection may be useful.

The GDG stressed that undertaking a CT scan should not delay appropriate treatment and that children should be stabilised clinically prior to transfer for scan.

The GDG note that Advanced Paediatric Life Support (APLS) guidance identifies that in a previously well, unconscious child (Glasgow Coma Scale score less than 9) who is not postictal, clinical signs of raised intracranial pressure may be evident.95 The GDG also noted that a reduced or fluctuating level of consciousness would correspond to a Glasgow Coma Scale score (or Child’s Glasgow Coma Scale score in the case of children under 4 years) of less than 9 or a drop of 3 or more.

Recommendations

Cranial computed tomography in suspected bacterial meningitis

Use clinical assessment and not cranial computed tomography (CT) to decide whether it is safe to perform a lumbar puncture. CT is unreliable for identifying raised intracranial pressure.

If a CT scan has been performed, do not perform a lumbar puncture if the CT scan shows radiological evidence of raised intracranial pressure.

In children and young people with a reduced or fluctuating level of consciousness (Glasgow Coma Scale score less than 9 or a drop of 3 or more) or with focal neurological signs, perform a CT scan to detect alternative intracranial pathology.

Do not delay treatment to undertake a CT scan.

Clinically stabilise children and young people before CT scanning.

If performing a CT scan consult an anaesthetist, paediatrician or intensivist.

Footnotes

*

See SIGN guideline at www​.sign.ac.uk/pdf/sign102.pdf

Copyright © 2010, Royal College of Obstetricians and Gynaecologists.

No part of this publication may be reproduced, stored or transmitted in any form or by any means, without the prior written permission of the publisher or, in the case of reprographic reproduction, in accordance with the terms of licences issued by the Copyright Licensing Agency in the UK [www.cla.co.uk]. Enquiries concerning reproduction outside the terms stated here should be sent to the publisher at the UK address printed on this page.

The use of registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant laws and regulations and therefore for general use.

Cover of Bacterial Meningitis and Meningococcal Septicaemia
Bacterial Meningitis and Meningococcal Septicaemia: Management of Bacterial Meningitis and Meningococcal Septicaemia in Children and Young People Younger than 16 Years in Primary and Secondary Care.
NICE Clinical Guidelines, No. 102.
National Collaborating Centre for Women's and Children's Health (UK).
London: RCOG Press; 2010.

Download

NICE (National Institute for Health and Care Excellence)

PubMed Health Blog...

read all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...