|
||||||||||||||||||||
|
|
Semi-quantitative analysis of cerebrospinal fluid chemistry and cellularity using urinary reagent strip: An aid to rapid diagnosis of meningitis
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.173641
Background: The examination of the cerebrospinal fluid (CSF) for proteins, glucose, and leukocytes still remains the basic gold standard investigation for the initial diagnosis of meningitis. Early diagnosis and initiation of antibiotics within 3 h can reduce the mortality and morbidity associated with meningitis. Hence, we conducted this study to document the usefulness of urinary reagent strip for the semi-quantitative analysis of CSF chemistry and cellularity. Keywords: Cerebrospinal fluid analysis; meningitis; rapid diagnosis; urinary reagent strip
Meningitis is a medical emergency as the infection is in the vicinity of the brain and spinal cord.[1] The disease is more prevalent in rural areas and in low socioeconomic groups. Timely identification and treatment of this condition is essential to prevent permanent neurological deficits and death.[2] Examination of the cerebrospinal fluid (CSF) for leucocytes, glucose, proteins, and culture are the cornerstones in the diagnosis of meningitis. CSF analysis requires trained personnel and an equipped laboratory, which are not available in rural areas. In the centers where the facility is available, the amount of time taken to reach the diagnosis of meningitis using laboratory based CSF analysis is an important factor that causes a delay in starting treatment.[3] Various authors have documented the usefulness of urinary reagent strip for the rapid bedside diagnosis of meningitis, but the technique is not used widely in clinical practice. In this study, we reassessed the diagnostic usefulness of Combur-10 urinary reagent strip for the early diagnosis of meningitis.[1],[3],[4],[5],[6],[7],[8],[9],[10],[11]
This is a prospective study, conducted in the Department of Pathology at a Medical College and Teaching Hospital for a period of 4 months from September to December 2014. All consecutive CSF samples were included in the study. Hemorrhagic CSF samples were excluded from the study. Demographic profiles such as age and sex were recorded. On all the CSF samples collected, both the definitive test and the index test were carried out. Definitive test For leucocytes, the total count was carried out by Neubauer's counting chamber by a technician who was unaware of the strip reading. The differential count was carried out by cytocentrifuged smears stained with Leishman stain. Definitive values for proteins and sugar were obtained from an automated analyzer (COBAS INTEGRA 400 plus, Roche diagnostics). Index test The combur-10 urinary reagent strip, which can detect 10 parameters such as specific gravity, pH, leucocytes, nitrate, proteins, glucose, ketone body, urobilinogen, bilirubin, and hemoglobin, was used to detect the index tests. One of the authors (PC), who was unaware of the definitive findings, carried out the rapid test. With the help of a micropipette, 2–3 drops of undiluted CSF was added to patches of leucocytes, proteins and sugar, and the color changes were recorded. The Combur reagent strip is designed to detect the range of leucocytes from 10 cells/mm 3 to 500 cells/mm 3. It detects the leucocytes by estimation of peroxidase. Depending on the color changes, leucocytes are graded as negative for the cell count of <10 cells/cumm, 1+ for 10–75 cells/cumm, 2+ for 75–500 cells/cumm, and 3+ for cells >500 cells/cumm. The reagent strip detects proteins in the range of 30 mg/dl to 500 mg/dl. The observations for CSF proteins include - no color (negative) if CSF proteins are <30 mg/dl, 1+ if they are between 30 and 100 mg/dl, 2+ if they are between 100 and 500 mg/dl, and 3+ if they are >500 mg/dl. The reagent strip based observation for glucose was: No color for a concentration of glucose <50 mg/dl, 1+ for 50–100 mg/dl, 2+ for a count between 100 and 300 mg/dl, 3+ for a count between 300–1000 mg/dl, and 4+ for a count >1000 mg/dl. Statistical analysis We estimated the diagnostic accuracy of the reagent strip versus the reference standard. The following diagnostic test performance parameters were evaluated: A box--and--whisker plot indicating the distribution of reference standard values for each of the visual categories of strip-color; a set of three analyses of the diagnostic characteristics of the strip – one for each of the three different “cut-off” criteria for strip-color. For each strip-color cut-off value [Table 1], the following were displayed: The 2 × 2 cross-tabulation of strip diagnosis versus reference standard diagnosis; and, sensitivity (Sens), specificity (Spec), positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (PLR or LR +), and negative likelihood ratio (NLR or LR −) along with 95% confidence intervals (CI). A smoothed receiver operating characteristic (ROC) curve showing the relationship between sensitivity and specificity as a function of the strip-color cut-off was plotted; and, the area under the ROC curve (AUC) along with its standard error (SE) were assessed.
Data were analyzed and graphed using the R statistical software package, version 3.0.1. The ROC curves, AUCs, and their SE's were obtained using the pROC add-in package for R (available@ http://www.r-project.org).
Our study included 103 cases dispersed in a wide range of age from 2 days to 75 years, with the maximum number of cases in the age group of 41–50 years. Of the 103 cases, 27 (26.2%) were <12 years and 14 cases were <1 year of age. Males outnumbered the females with the male: female ratio being1.86:1 [Figure 1].
We have analyzed three cardinal parameters to diagnose meningitis namely leucocytes, proteins, and glucose. The findings of the reagent strip were compared with the reference standard. In 19 (18.5%) cases, CSF was normal in all the three parameters; and, in 9 (8.5%) cases, all the three parameters were abnormal. The detailed results of all the CSF samples are given in [Table 2].
Leucocytes leucocytes were detected by the esterase activity. The sensitivity and specificity for leucocytes by the strip method for >10 cells/cumm were 96.6% (95% CI: 82.7–99.9%) and 94.5% (95% CI: 86.5–98.4%), respectively. Proteins The reagent strip had a high sensitivity of 94.9% (95% CI: 87.5–98.6%) and a low specificity of 45.8% (95% CI: 25.5–67.1%) for the protein levels ≥30 mg/dl. If the cut-off level of the proteins was increased to ≥100 mg/dl, both the sensitivity and specificity were acceptable (sensitivity: 96% [95% CI: 79.6–99.8%] and specificity 87.1% [95% CI: 77.6–93.6%]). Glucose The reagent strip test for glucose at cut-off levels of ≤40 mg/dl and ≤50 mg/dl was highly specific with a specificity of 100% but it was less sensitive [Table 3].
The box- and- whisker plot for the reagent strip vs. references standard for the three parameters and the ROC curves with AUC for various cut-off levels are shown in [Figure 2].
Meningitis is defined as inflammation of the leptomeninges namely, the subarachnoid layer and the pia mater. Bacterial meningitis (BM) is a medical emergency requiring immediate diagnosis and treatment. The mortality rate recorded for BM is 27%, and up to 50% of the patients develop psycho-neurological sequelae due to damage to the hippocampal nuclei and other neurological structures.[12] The mortality rate for tuberculous meningitis at stage III is 55% among immunocompetent patients and 75% among immune-compromised patients.[13] Imaging of the brain followed by lumbar puncture is the initial diagnostic modality. It has been documented in the literature that an early diagnosis of the disease and initiation of the antibiotic therapy within 3 h will reduce the morbidity and mortality considerably.[14] In the present study, we have used Combur-10 urinary reagent strip for a rapid analysis of the CSF. The strip had a high sensitivity and specificity for leucocytes ≥10 cells/cumm. With respect to proteins, the results are acceptable for a higher cut-off level of ≥100 mg/dl, where the sensitivity and specificity were 96% and 87.1%, respectively, while at ≥30 mg/dl, the strip was more sensitive and less specific. With respect to glucose, the strip was highly specific (100%) and less sensitive at both the cut-off levels. Observations of our study were comparable to those reported in the literature. Joshi et al.,[3] using the same reagent strip, observed a sensitivity of 85.2% and a specificity of 89.6% for leucocytes >10 cells/cumm. For proteins, at a cut-off of ≥30 mg/dl, the sensitivity was high (98.1%) but the specificity was low (57.1%); however, at a cut of ≥100 mg/dl, both the sensitivity and specificity were high and acceptable (as observed in our study). The strip based analysis for glucose at a cut-off of ≤40 mg/dl was more specific (96.5%) than sensitive (61.1%); and, even at a cut-off level of ≤50 mg/dl, the test was more specific (98%) than sensitive (46.2%). The AUC values observed in the present study for leucocytes ≥10 cells/cumm, proteins ≥30 mg/dl, proteins ≥100 mg/dl, glucose ≤40 mg/dl, and glucose ≤50 mg/dl were 99.05%, 84.86%, 95.69%, 86.51%, and 76.99%, respectively. For the same cut-off levels, Joshi et al.,[3] have observed the AUC values of 87%, 77%, 90%, 78%, and 72%, respectively. Kumar et al.,[1] Parmar et al.,[5] and Romanelli et al.,[11] also tested the Combur-10 urinary reagent strip for a rapid CSF analysis. Kumar et al.,[1] in their study, observed a positive correlation between the strip and laboratory values for the diagnosis of meningitis with κ =0.94, 0.819, 0.819 for cells, protein and glucose respectively, with P< 0.000, which was statistically significant. Parmar et al.,[5] in their study on 63 CSF samples observed a high sensitivity and specificity for the diagnosis of bacterial and tubercular meningitis, both being 100% and 96.55%, respectively. For the diagnosis of aseptic meningitis, the strip was more specific than sensitive. Romanelli et al.,[11] observed the sensitivity of 90.7%, specificity of 98.1%, PPV of 95.1%, NPV value of 96.4%, and accuracy of 96.1% for the results of reagent strip analysis in the diagnosis of bacterial meningitis. The P value observed by them was <0.0001, and κ was 0.9. These values were statistically significant. The lower version of the Combur-10 reagent strip, Combur-9 has also been tested for the rapid CSF analysis and has been found to be useful.[9],[10] Maclennan et al.,[7] in their study tested the usefulness of the nitrate patch for the diagnosis of bacterial meningitis. They concluded that nitrate patch, which is also a component of Combur-10 urinary reagent strip, will become positive in bacterial meningitis in situ ations where the granulocytes are increased and have not released the esterase enzyme. Thus, additional information may be obtained without any extra cost. Even though the technique is simple, economical, easily available, and its usefulness has been documented in the literature, it has not gained popularity among the physicians, pediatricians and neurologists.[3] The disadvantage of the Combur-10 reagent strip assessment is its cost. It is useful only if the CSF is clear and is not applicable in the presence of bloody CSF and in cryptococcal meningitis.[2],[3],[5] Another disadvantage is that it is designed for urine and not for CSF. The lower limit of protein and glucose levels vary between the urine and CSF. The lower level of cut-off of proteins in the urine is 30 mg/dl but for the CSF, it is 45 mg/dl, and for the latter, protein values more than 45 mg/dl signify the presence of meningitis. The lower limit of glucose in the urine is 50 mg/dl, and for CSF it is 40 mg/dl. With respect to leucocytes, 10 cells/cumm may be acceptable in the CSF findings, as there will be unavoidable trauma during lumbar puncture causing blood granulocytes to enter the CSF.[3] Newer reagent strips with three patches for the assessment of proteins, leukocytes, and glucose with cut-off values corresponding to the CSF values are highly recommended. Adding the fourth patch of nitrate for gaining additional information is also advantageous.[3],[5]
It is evident from the observations of our study and also from the review of literature that the technique of rapid diagnosis of CSF by urine reagent strip is useful. Furthermore, a new strip containing three or four patches corresponding to CSF cut-off values is recommended. If this method is used in routine clinical practice, the rapidity of diagnosis will considerably decrease the morbidity and mortality that occurs due to meningitis. Acknowledgments We are grateful to Dr. John C. Pezzullo, Associate Professor, Department of Medicine, Georgetown University, Washington DC USA for his valuable help in the statistical analysis. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3]
|
|
|||||