PEDIATRICS Vol. 111 No. 3 March 2003, pp. 525-528
Interpretation of Traumatic Lumbar Punctures: Who Can Go Home?
* Children’s Memorial Hospital, Department of Pediatrics, Division of Emergency Medicine, Chicago, Illinois
Children’s Hospital, Department of Pediatrics, Section of Emergency Medicine, Denver, Colorado
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Objective. To determine whether a ratio of observed to predicted (O:P) cerebrospinal fluid (CSF) white blood cells (WBCs) after a traumatic lumbar puncture (LP) can be used to predict which patients do not have meningitis and can safely be discharged from the hospital.
Methods. A retrospective medical record review was performed on 2 cohorts of previously healthy children who had received an LP at Children’s Memorial Hospital in Chicago, IL. All children were older than 1 month and had a red blood cell (RBC) count in the CSF >500/mm3. Cohort 1 consisted of children who were examined in 1990 through 1999 and had CSF cultures positive for a bacterial pathogen. Cohort 2 consisted of children who were tested during January through December 1999 and had a CSF culture negative for any bacterial pathogen. Exclusion criteria included patients who received antibiotics within 72 hours before evaluation, patients with a previous neurosurgical procedure or CNS bleed, and patients whose complete blood count was not done within 6 hours of LP. The predicted CSF WBC count was calculated using the formula CSF WBC (predicted) = CSF RBC x (blood WBC/blood RBC). The O:P ratio was obtained by dividing the observed CSF WBC by the predicted CSF WBC. The simple ratio of WBCs to RBCs was also calculated. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated to predict the absence of disease. Receiver operator characteristic curves were generated for the O:P ratio and the WBC:RBC ratio. Continuous variables were analyzed with Mann-Whitney U test.
Results. Among the 57 patients who fit all of the study criteria, 12 (21%) had positive CSF cultures for bacterial pathogens. The patients with meningitis were significantly older (median: 7.8 months; range: 1–106 months) than the patients without meningitis (median: 1.3 months; range: 1–139 months). The O:P ratio was significantly lower in the patients without meningitis (median: 0.064; range: 0.000054–1.09) as compared with patients with meningitis (median: 1.26; range: 0.045–4.72). The WBC:RBC ratio was significantly lower in the patients without meningitis (median: 0.001; range: 0–4.46) as compared with patients with meningitis (median: 1.98; range: 0.04–24.45). The specificity and positive predictive value of an O:P ratio 
0.01 and a WBC:RBC ratio 1:100 were 100% predicting the absence of disease. The area under the curve for the O:P ratio (0.981) did not differ significantly from the area under the curve for the WBC:RBC ratio (0.970).
Conclusion. A WBC:RBC ratio of 1:100 (0.01) and an O:P ratio of 0.01 identified a large group of patients without meningitis. Using these methods in children younger than 1 month, the majority of patients without meningitis can be differentiated from those with meningitis despite the CSF abnormalities associated with a traumatic LP. However, the clinician should examine all clinical and laboratory information before opting not to treat a child after a traumatic LP.
Key Words: lumbar puncture • traumatic • interpretation
Abbreviations: LP, lumbar puncture • CSF, cerebrospinal fluid • WBC, white blood cell • RBC, red blood cell • O:P ratio, observed to predicted ratio • Hib, Haemophilus influenzae type b • PMN, polymorphonuclear cells • PPV, positive predictive value • ROC, receiver operator characteristic
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Alumbar puncture (LP) is 1 of the most frequently performed procedures in pediatrics. A traumatic LP occurs when the spinal needle passes through the subarachnoid space and penetrates the vascular epidural space, introducing blood into the spinal fluid.1 Traumatic LPs are common, occurring in 15% to 20% of pediatric LPs.2 The contamination of cerebrospinal fluid (CSF) with blood during a traumatic LP makes interpretation of the cell count difficult. Although the presence of white blood cells (WBCs) in CSF usually suggests infection, in a traumatic LP, WBCs are more likely to be from the blood contaminant. Therefore, patients with traumatic LPs are often admitted to the hospital for prophylactic parenteral antibiotics until the CSF culture is negative because their LP results may not be interpretable.
Physicians have searched for decades for a method to differentiate peripheral blood WBCs from true CSF leukocytosis in bloody spinal fluid. Some clinicians use a simple ratio assuming that the proportion of WBCs to red blood cells (RBCs) in the spinal fluid of a normal person is 1:500.3 Other investigators have assumed that the ratio of WBCs to RBCs in the peripheral blood remains constant when the peripheral blood is introduced into the subarachnoid space. Therefore, the expected or predicted CSF WBCs can be calculated on the basis of the peripheral blood count and the CSF RBCs. The predicted value is subtracted from the actual or observed CSF WBC. True CSF leukocytosis exists when the observed CSF WBC count is greater than the predicted WBC count. Some authors have questioned the validity of this correction. In patients with a traumatic LP and later having confirmed negative CSF culture, investigators found that the formula overcorrected the WBC count, as there were fewer WBCs in the bloody spinal fluid than predicted.4–6 Postulated mechanisms for this decrease in observed WBCs included adherence of WBCs to the meninges or lysis of WBCs during transport. There is a possibility that this same phenomenon might be seen with confirmed bacterial meningitis. Thus, these investigators recommended extreme caution in interpreting traumatic LPs.
However, other investigators found the observed to predicted ratio (O:P) to be helpful. Mayefsky and Roghmann7 found that true leukocytosis is rarely masked in blood-contaminated CSF. In their study of 720 CSF specimens in a general medical center, 54, or 7.5%, had culture-positive bacterial or fungal meningitis. Because their study also included patients with meningitis, they were able to compare the O:P ratios in patients with and without meningitis. They focused on the ratio of observed WBCs to the predicted WBCs in the CSF rather than formulas that manipulate WBC counts. An O:P ratio of 10 or greater was both a sensitive and a specific indicator of meningitis. Bonadio et al8 studied exclusively pediatric patients and agreed that CSF abnormalities associated with bacterial meningitis are rarely obscured by blood contamination from traumatic LP. In their study, an O:P ratio of >1 was 100% sensitive and 62% specific for predicting meningitis. Both of these studies focused on predicting meningitis, and Haemophilus influenzae type b (Hib) was the major cause of bacterial meningitis in the pediatric patients, which was consistent with national figures for that time period.9
This study was designed to compare the results of traumatic LPs in children with culture-positive meningitis to traumatic LPs in children with a negative CSF culture in the post-H influenzae vaccine era. The objective was to determine whether an O:P ratio of CSF WBCs after a traumatic LP can be used to predict which patients do not have meningitis and can safely be discharged from the hospital.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
A retrospective medical record review was performed on 2 cohorts of previously healthy children who had received an LP at Children’s Memorial Hospital, an urban, university-affiliated children’s hospital in Chicago, IL. All children were older than 1 month and had an RBC count in the CSF >500/mm3. Patients were identified through computerized microbiology records. Cohort 1 consisted of children who were examined in 1990 through 1999 and had CSF cultures positive for a bacterial pathogen. Cohort 2 consisted of children who were tested during January through December 1999 and had a CSF culture negative for any bacterial pathogen. Exclusion criteria included patients who received antibiotics within 72 hours before evaluation, patients with a previous neurosurgical procedure or CNS bleed, and patients whose complete blood count was not done within 6 hours of LP.
Data abstracted from the record included demographic information, CSF cell counts, protein, glucose, culture, Gram stain, and complete blood count. The predicted CSF WBC count was calculated using the formula CSF WBC (predicted) = CSF RBC x (blood WBC/blood RBC). The O:P ratio was obtained by dividing the observed CSF WBC by the predicted CSF WBC. The simple ratio of WBCs to RBCs was also calculated. CSF pleocytosis was defined as a CSF WBC >5/mm3. Polymorphonuclear cell (PMN) predominance was defined as >50% CSF PMNs, hypoglycorrhachia as CSF glucose <40 mg/dL, and elevated CSF protein >40 mg/dL.
Traditionally, screening tests are considered in the context of the ability of a positive or abnormal test to predict disease. However, we were interested in identifying patients without meningitis or predicting the absence of disease. Therefore, we reversed the conventional 2 x 2 table. For example, the first column represents patients without meningitis; the first row represents patients with a normal laboratory value.
Sensitivity (the proportion of children who have a negative culture and a normal laboratory value), specificity (the proportion of children who have meningitis and an abnormal test), positive predictive value (PPV; the probability of the absence of meningitis in a patient with a normal test result), and negative predictive value (the probability of meningitis in a patient with an abnormal test result) were calculated. Ninety-five percent confidence intervals were generated using the exact binomial method. Receiver operator characteristic (ROC) curves were generated for the O:P ratio and the WBC:RBC ratio. Continuous variables were analyzed with Mann-Whitney U test. Categorical variables were analyzed with the Fisher exact test. Calculations were performed using SPSS for Windows (version 10.0.0, 1999; SPSS Inc, Chicago, IL). The study was approved by the hospital’s institutional review board.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Among the 57 patients who fit all of the study criteria, 12 (21%) had positive CSF cultures for bacterial pathogens. The patients with meningitis were significantly older (median: 7.8 months; range: 1–106 months) than the patients without meningitis (median: 1.3 months; range: 1–139 months; P = .01). The majority of the patients in both groups were male (meningitis 60%, without meningitis 66.7%; P = .75). The CSF pathogens isolated were Neisseria meningitidis (4), group B Streptococcus (2), Hib (3) H influenzae non-type b (1), and Streptococcus pneumoniae (2).
The O:P ratio was significantly lower in the patients without meningitis (median: 0.064; range: 0.000054–1.09) as compared with patients with meningitis (median: 1.26; range: 0.045–4.72; P < .001). Forty-four of 45 patients whose CSF cultures were negative for bacterial pathogens had O:P ratios <1, with the 45th having an O:P ratio of 1.09. Seven of 12 patients with positive CSF cultures for bacterial pathogens had O:P ratios >1, and 5 of those 7 had O:P ratios >2 (Fig 1). Similarly, the WBC:RBC ratio was significantly lower in the patients without meningitis (median: 0.001; range: 0–4.46) as compared with patients with meningitis (median: 1.98; range: 0.04–24.45; P < .001).
|
The sensitivity, specificity, and PPV and negative predictive value for O:P ratios, WBC:RBC ratios, and CSF laboratory values are given in Table 1. ROC curves were plotted for the O:P and the WBC:RBC ratios. The area under the curve for the O:P ratio (0.981) did not differ significantly from the area under the curve for the WBC:RBC ratio (0.970; P = .657).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
We found a significant difference in O:P and WBC:RBC ratios between patients with and those without meningitis. An O:P ratio
0.01, a WBC:RBC ratio 1:100 (0.01), and the absence of pleocytosis were each highly specific with a high PPV for predicting the absence of meningitis. The O:P ratio and WBC:RBC ratio performed equally well, as evaluated by the ROC methods. A negative Gram stain, the lack of PMN predominance, the CSF glucose, and protein did not perform as well in this study. Previous studies of patients with traumatic LPs focused on the identification of patients with meningitis.7,8 They attempted to determine an O:P ratio that clearly differentiated those patients with meningitis from those without meningitis. Therefore, they sought a highly sensitive test because of the potential morbidity associated with meningitis and the consequences of missing this disease. We chose to focus on the identification of patients without meningitis because we hoped to define a subset of patients who clearly did not have meningitis and could be sent home safely without a costly and unnecessary hospitalization. The treatment of a few patients without meningitis was considered acceptable, provided that no cases of meningitis would be inadvertently discharged from the hospital by our criteria. Therefore, we were most interested in a highly specific test.
Traditionally, a specific test is rarely positive in the absence of disease. In our study, because we reversed the conventional 2 x 2 table, a specific test is rarely normal or negative in the presence of disease. An O:P ratio 0.01, a WBC:RBC ratio 1:100 (0.01), and the absence of pleocytosis were each highly specific. In addition, the sensitivity of an O:P ratio 0.01 and a WBC:RBC ratio 1:100 (0.01) were 91% and 84%. Therefore, the majority of patients without meningitis in this study population could have been discharged without treatment by these criteria. The sensitivity of the absence of pleocytosis was only 33%, so, using this criteria, most patients would still require therapy. Our results agreed with Bonadio et al,8 who found the O:P ratio and CSF pleocytosis most helpful in differentiating patients with meningitis from those without. They also found the glucose, protein, Gram stain, and PMN predominance less helpful.
Sensitivity and specificity are properties of the test and do not take into account prevalence of disease. Once the results of the test are available, sensitivity and specificity are less helpful because they give the probability of a test being positive or negative in a patient known or not known to have the disease. However, the clinician must determine the likelihood of disease in a patient on the basis of the results. Therefore, the predictive value of the test is more helpful once the result of the test is available. In this study, we were interested in the PPV, or the probability of the absence of meningitis in a patient with a normal test result. The PPV of an O:P ratio 0.01, a WBC:RBC ratio 1:100 (0.01), and the absence of pleocytosis were 100%. The confidence intervals around these parameters suggest that an O:P ratio 0.01 and a WBC:RBC ratio 1:100 (0.01) perform better than the absence of pleocytosis.
To test the external validity of our results, we applied the parameter of an O:P ratio 0.01 to the results of the studies by Mayefsky and Roghmann7 and by Bonadio et al.8 Because they did not calculate the WBC:RBC ratio, we were unable to evaluate that guideline. Only 7 traumatic LPs in 5 patients with meningitis in Mayefsky and Roghmann’s study had an O:P ratio <1; with the exception of an elderly woman in septic shock who underwent 2 LPs, 4 other pediatric patients with 5 LPs had O:P ratios of 0.4 or greater. In the study by Bonadio et al, all patients with meningitis had O:P ratios >1. Both of these studies confirm that an O:P ratio of 0.01 is highly specific for predicting the absence of meningitis.
The prevalence of meningitis as confirmed by culture in our study (21%) was less than in the Bonadio et al study (32%) during the pre-H influenzae vaccine era. As expected in our study, the prevalence of Hib meningitis was also lower when compared with the Bonadio et al study (25% vs 73%). This epidemiologic trend did not have an effect on the ability to interpret traumatic LPs.
This study has several limitations. As it was a retrospective review, we depended on the accuracy and completeness of the medical record. In addition, compared with 1 previous study, our results were based on fewer cases of meningitis. This most likely reflects the widespread use of the conjugate vaccine against Hib. Because prevalence affects the predictive value of a test, our high PPVs may be influenced by the relatively few cases of meningitis. However, given the introduction of the pneumococcal conjugate vaccine, we would expect even fewer cases of meningitis in the future. Finally, no guideline should completely supplant clinical judgment. The O:P ratio and WBC:RBC ratio should be assessed with other clinical information.
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
A WBC:RBC ratio of
1:100 (0.01) and an O:P ratio of 0.01 identified a large group of patients without meningitis. Because the WBC:RBC is a less cumbersome calculation, we recommend that traumatic LPs be screened by this calculation first. With the use of these methods in children older than 1 month, the majority of patients without meningitis can be differentiated from those with meningitis despite the CSF abnormalities associated with a traumatic LP. However, the clinician should examine all clinical and laboratory information before opting not to treat a child after a traumatic LP.
|
ACKNOWLEDGMENTS |
|---|
We thank Helen Binns, MD, MPH, for thoughtful review of the manuscript.
|
FOOTNOTES |
|---|
Received for publication Jul 26, 2002; Accepted Sep 12, 2002.
Reprint requests to (S.S.M.) 2300 Children’s Plaza, Box 62, Chicago, IL 60614. E-mail: s-mazor2{at}northwestern.edu
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
- Kooiker JC. Spinal puncture and cerebrospinal fluid examination. In: Roberts JR, Hedges JR, eds. Clinical Procedures in Emergency Medicine. Philadelphia, PA: WB Saunders; 1998:1054–1075
- Yogev R. Meningitis. In: Jenson HB, Baltimore RS, eds. Pediatric Infectious Diseases. Philadelphia, PA: WB Saunders; 2002:630–650
- Fuchs S. Neurologic disorders. In: Barkin RM, ed. Pediatric Emergency Medicine. St Louis, MO: Mosby; 1997:972–1024
- Osborne JP, Pizer B. Effect on the white blood cell count of contaminating cerebrospinal fluid with blood.
Arch Dis Child.1981; 56
:400
–401
[Abstract/Free Full Text] - Novak RW. Lack of validity of standard corrections for white blood cell counts of blood-contaminated cerebrospinal fluid in infants. Am J Clin Pathol.1984; 82 :95 –97[Web of Science][Medline]
- Rubenstein JS, Yogev R. What represents pleocytosis in blood-contaminated ("traumatic tap") cerebrospinal fluid in children? J Pediatr.1985; 107 :249 –251[CrossRef][Web of Science][Medline]
- Mayefsky JH, Roghmann KJ. Determination of leukocytosis in traumatic spinal tap specimens. Am J Med.1987; 82 :1175 –1181[CrossRef][Web of Science][Medline]
- Bonadio WA, Smith DS, Goddard S, Burroughs J, Khaja G. Distinguishing cerebrospinal fluid abnormalities in children with bacterial meningitis and traumatic lumbar puncture. J Infect Dis.1990; 162 :251 –254[Web of Science][Medline]
- Quagliarello VJ, Scheld WM. Treatment of bacterial meningitis.
N Engl J Med.1997; 336
:708
–716
[Free Full Text]
PEDIATRICS (ISSN 1098-4275). ©2003 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
This article has been cited by other articles:
|
|
 |
|
  R. L. King, S. A. Lorch, D. M. Cohen, R. L. Hodinka, K. A. Cohn, and S. S. Shah Routine Cerebrospinal Fluid Enterovirus Polymerase Chain Reaction Testing Reduces Hospitalization and Antibiotic Use for Infants 90 Days of Age or Younger Pediatrics, September 1, 2007; 120(3): 489 - 496. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. R. Haase Acute Bacterial Meningitis in Children Journal of Pharmacy Practice, December 1, 2004; 17(6): 392 - 406. [Abstract] [PDF]
|
|
|
|
|
|
I. R. Holzman, J. A. Berkley, I. Mwangi, B. Lowe, C. R. J. C. Newton, and S. Mazor Traumatic Lumbar Punctures Pediatrics, January 1, 2004; 113(1): 172 - 172. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
