Adriana Slavcovici 1, C. Marcu 1, Doina Ţăţulescu 1, Amanda Rădulescu 2, Simona Clichici 3, Adriana Filip 3,
Cristina Cismaru 1, D. Vesbianu 4, Ioana Şoşa 5
1 University Hospital of Infectious Diseases, “Iuliu Haţieganu” University of Medicine and Pharmacy Cluj-Napoca
2 Epidemiology Department, “Iuliu Haţieganu” University of Medicine and Pharmacy Cluj-Napoca
3 Physiology Department, “Iuliu Haţieganu” University of Medicine and Pharmacy Cluj-Napoca
4 Drexel University, Hahnemann Hospital, Philadelphia, PA 19102 USA
5 Pneumology Department, Pediatric Emergency Teaching Hospital Cluj-Napoca
Abstract
Introduction. Tuberculous meningitis represents a severe manifestation of the systemic infection. Clinical and laboratory diagnosis based on classical methods is diffi cult and delayed. Th e current rapid diagnostic tests are costly and sometimes inaccessible. Objectives. We aimed to optimize the diagnostic strategy by establishing the features that are quickly available and have the best predictive value for TBM diagnosis by using two variants of scoring. Methods: We retrospectively studied 58 non-HIV cases of possible TBM hospitalized in the Cluj-NapocaClinic of Infectious Diseases during 2000-2007. Th e inclusion criteria were: meningitis syndrome, nonpurulent cerebrospinal fl uid with lymphocytes predominance, decreased glucose and increased protein levels. We established the signifi cant clinical, laboratory and imagistic features through univariate analysis. Th ese variables were used in a modifi ed Th waite’s and Kumar’s diagnostic scoring (score-1 and-2, respectively). We calculated likelihood ratios for TBM prediction accuracy. Results. Th e diagnosis was microbiologically confi rmed in 33.3% of all patients after 3 weeks in average. Statistical analysis revealed nine variables signifi cantly predictive for TBM diagnosis: prodrome duration≥7 days or ≥14 days, age<23 years, WBC<9000/mm3, cerebrospinal fl uid white-cell count <300/ mm3, CSF neutrophils proportion <50% or <22%, paralysis, extrapyramidal signs / ataxia/ myoclonus / tremor. Using the adapted scoring 1, the posttest probability reached 92%. By using score 2 (based on three variables) we found a signifi cant increase of the TBM prognosis (pLR = 5.2, nLR = 0.2, posttest probability =94%). Conclusions. Th e confi rmatory diagnosis of TBM is delayed and occurs in a small proportion of cases. Early TBM diagnosis may be improved by using Th waite’s and Kumar’s diagnostic scoring adapted to our available variables that were signifi cantly found in the study population.