Arch Dis Child 1999;81:221–224 221 A diagnostic rule for tuberculous meningitis Arch Dis Child: first published as 10.1136/adc.81.3.221 on 1 September 1999. Downloaded from Rashmi Kumar, S N Singh, Neera Kohli Abstract onstration of mycobacteria in cerebrospinal Diagnostic confusion often exists between fluid (CSF), by direct staining or culture. tuberculous meningitis and other menin- However, these tests are time consuming and goencephalitides. Newer diagnostic tests seldom positive.2 Recognising this problem of are unlikely to be available in many coun- diagnosis, many newer tests have been devel- tries for some time. This study examines oped to diagnose tuberculous meningitis and which clinical features and simple labora- diVerentiate it from pyogenic meningitis—for tory tests can diVerentiate tuberculous example, enzyme linked immunosorbent assay, meningitis from other infections. Two bromide partition test, tuberculostearic acid in hundred and thirty two children (110 CSF, adenosine deaminase in CSF, polymerase tuberculous meningitis, 94 non- chain reaction, etc.34 However, the sensitivity tuberculous meningitis, 28 indeterminate) of these tests is still under study and they are with suspected meningitis and cerebrospi- unlikely to be available where they are really nal fluid (CSF) pleocytosis were enrolled. needed for at least the next decade. In practice, Tuberculous meningitis was defined as in India, treatment is started solely on the basis positive CSF mycobacterial culture or of clinical features and results of simple labora- acid fast bacilli stain, or basal enhance- tory tests of CSF and blood. Therefore, we ment or tuberculoma on computed tomo- attempted to establish a clinical rule for the graphy (CT) scan with clinical response to diagnosis of tuberculous meningitis, whereby a antituberculous treatment. Non- set of clinical and laboratory findings can be tuberculous meningitis was defined as used to “predict” the diagnosis of tuberculous positive CSF bacterial culture or Gram meningitis. Such a rule would be especially stain, or clinical response without antitu- useful for clinicians working in remote areas, berculous treatment. Thirty clinical/ where laboratory facilities are minimal and laboratory features of patients with diagnosis has to be made on clinical grounds. tuberculous meningitis and non- In addition, identifying a group of children who tuberculous meningitis were compared by have certainly (or almost certainly) not got univariate and multiple logistic regression tuberculous meningitis could avoid exposing analysis. Five features were independently them to an unnecessary and long course of predictive of the diagnosis of tuberculous antituberculous treatment. meningitis (p < 0.007): prodromal stage http://adc.bmj.com/ > 7 days, optic atrophy on fundal exam- Materials and methods ination, focal deficit, abnormal move- We conducted our study in the children’s wards ments, and CSF leucocytes < 50% of King George’s Medical College Hospital, polymorphs. When validated on another Lucknow over a period of 30 months from set of 128 patients, if at least one feature August 1994 to February 1997. This is a large was present, sensitivity was 98.4% and, if teaching hospital in the capital city of India’s three or more were present, specificity most populous state, Uttar Pradesh, and caters on September 30, 2021 by guest. Protected copyright. was 98.3%. This simple rule would be use- for the poor and seriously ill children from the ful to physicians working in regions where city and surrounding districts. Children be- tuberculosis is prevalent. tween the ages of 1 month and 12 years admit- (Arch Dis Child 1999;81:221–224) ted with a diagnosis of meningoencephalitis on Keywords: tuberculous meningitis; pyogenic the basis of history, examination, and pleocyto- meningitis; computed tomography scan sis in CSF, on three selected days of each week were enrolled and investigated according to a predesigned protocol. A complete physical Department of Tuberculous meningitis continues to be an examination was carried out on admission. Pediatrics, King important cause of hospital admissions, death, Investigations included blood counts, CSF George’s Medical College, Lucknow, and neurological disability in children in India. examination for cell count, diVerential count, India 226003 The illness accounts for 1–4% of total paediat- protein, sugar, Gram stain, acid fast bacilli R Kumar ric hospital admissions in diVerent parts of the (AFB) stain, bacterial and mycobacterial cul- S N Singh country.1 Tuberculous meningitis is a serious ture, a skin test for tuberculosis, and a chest illness which, if not diagnosed and managed radiograph. A CT scan was also planned and Department of early, leads to a high rate of mortality and per- was usually possible within a week of admission Radiodiagnosis, King George’s Medical manent disabilities. to hospital. The patient’s clinical course was College Today, tuberculous meningitis still poses a carefully recorded. At discharge, the final diag- N Kohli diagnostic problem. The reason is that it nosis of tuberculous meningitis or non- presents in a similar manner to other menin- tuberculous meningitis was made on the Correspondence to: goencephalitides, partially treated pyogenic following criteria. Dr Kumar. meningitis in particular. Definitive diagnosis of Tuberculous meningitis was diagnosed if: (1) Accepted 15 February 1999 tuberculous meningitis can be made by dem- mycobacterial culture/AFB stain was positive 222 Kumar, Singh, Kohli Arch Dis Child: first published as 10.1136/adc.81.3.221 on 1 September 1999. Downloaded from Table 1 Clinical and laboratory features in patients with and without tuberculous sample t test, or Kruksal-Wallis test for data meningitis (TBM) that were not normally distributed) using the Epi-info program. Table 1 lists the variables TBM Non-TBM (n = 110) (n = 94) Odds ratio (95% CI) p Value tested. Length of prodromal stage was defined as period between first symptom and first Age (months) 42.4 39.0 — 0.56* Prodromal stage (days) 19.0 4.0 — < 0.005† neurological manifestation. Weight was ex- Weight for age (%) 66.2 73.4 — < 0.005* pressed as percentage of expected weight for Haemoglobin (g/l) 101 100 — 0.72* age. Fever was graded as low (< 37.5°C), mod- Leucocyte count (109/l) 9.4 10.5 < 0.005† DLC (% polymorphs) 66.7 71.9 — 0.01* erate (37.5–38.5°C), or high (more than CSF cell count 137.0 660.0 — < 0.005† 38.5°C). Frequency of convulsions was divided CSF polymorphs (%) 25.0 90.0 — < 0.005† CSF protein (g/l) 1 1 — 0.38† into 1–3/day, 4–6/day, and > 6/day. A history of CSF sugar (g/l) 0.4 0.3 — 0.01* measles included the previous six months only. Religion Hindu 86 73 1.0 (0.5 to 2.1) 0.936 Coma was graded according to Corey et al.5 Fever grade Low 55 9 Presence of meningeal signs was defined as Moderate 38 19 < 0.005 neck stiVness, Kernig’s or Brudzinsky’s sign. High 17 66 Variables with p < 0.25 were entered into a Vomiting 73 60 1.1 (0.6 to 2.1) 0.82 Convulsions 80 61 1.4 (0.8 to 2.7) 0.29 logistic regression analysis using the multlr Focal convulsions 14 16 0.6 (0.2 to 1.4) 0.29 package to obtain â coeYcients, adjusted odds Convulsion frequency ratios, and “predictors” of the diagnosis of Nil 30 33 1–3/day 32 22 — 0.70 tuberculous meningitis. 3–6/day 27 22 For validation, another set of patients was > 6/day 21 17 enrolled in the same way using the same diag- History of measles 9 4 2.0 (0.5–8.1) 0.39 History of contact 46 8 7.7 (3.2–19.3) < 0.005 nostic (gold standard) criteria. The rule was Lymphadenopathy 46 14 4.1 (1.9–8.7) < 0.005 then applied to them and compared with the Coma grade Nil 27 25 gold standard diagnosis in a two by two table to I 58 41 0.58 calculate its sensitivity, specificity, and the like- II 19 22 lihood ratios for various predictors and their III 6 6 Meningeal signs 95 82 0.9 (0.4 to 2.3) 0.98 combinations. â CoeYcients obtained for the Fundal optic atrophy 30 2 19.7 (4.4 to 123.6) < 0.005 five variables in the logistic regression analysis Focal deficits 48 10 6.5 (2.9 to 15.0) < 0.005 were rounded oV and added up to obtain a Extrapyramidal movements 35 4 10.5 (3.3 to 36.9) < 0.005 Decerebrate posturing 26 5 5.5 (1.9 to 17.4) < 0.005 weighted score for each patient in the valida- Increased tone 67 54 1.1 (0.6 to 2.1) 0.72 tion dataset. Sensitivity and specificity at each Ankle clonus 13 2 6.2 (1.3 to 41.3) 0.02 score were calculated and these values were Cranial nerve palsies 25 3 8.9 (2.4 to 38.6) < 0.005 Skin test positive 36/84 4/28 4.5 (1.3 to 17.0) 0.01 used to construct a receiver operator character- Suggestive chest radiograph 48/66 9/32 6.8 (2.4 to 19.8) < 0.005 istic (ROC) curve and determine the best cut oV score. The best cut oV point is taken as that For continuous variables (1–10), analysis was either by two sample t test* and mean values are shown, or by Kruksal-Wallis test† and medians are shown (see methods). For non-continuous closest to the left upper corner of the ROC variables (11–30) data are numbers of patients with the finding. The ÷2 test was used. curve. CI, confidence interval; DLC, diVerential leucocyte count; CSF, cerebrospinal fluid. http://adc.bmj.com/ Table 2 â CoeYcients, p values, and odds ratios (OR) with 95% confidence intervals Results (CI) of five factors found to be independently associated with tuberculous meningitis on A total of 232 out of the 248 children who sat- logistic regression analysis isfied the entry criteria were enrolled.