|
Purpose by collecting clinical manifestations, laboratory tests and imaging data, to establish economic, simple and practical diagnostic scoring model used for early diagnosis of central nervous system tuberculosis infection, improve the prognosis. Methods A retrospective collection of January 2004 - December 2010 in Xiangya Hospital, Second Xiangya Hospital, Hunan Provincial People's Hospital, Changsha Central Hospital, Department of Neurology, intracranial infection in patients with four common demographic characteristics, clinical symptoms and signs , laboratory parameters, five categories of imaging data. Study included 910 cases of patients were 349 cases of tuberculous meningitis, 481 cases of viral meningitis, 49 cases of bacterial meningitis, 31 cases of cryptococcal meningoencephalitis. Data were processed using software Epidata entry, SPSS software for analysis. In univariate analysis, Kruskal-Wallis test was used to compare continuous variables (measurement data), X2 (Fisher's) test was used to compare count data. Select the univariate analysis, P value less than .05 test level variables into the multivariate analysis, the use of Forward Stepwise Logistic Regression (P into the lt; 0.05, P out gt; 0.10) gradually filter out predictive diagnosis of tuberculous meningoencephalitis independent variable . Using the ROC (Receive operating characteristic curve) curve to find the diagnostic cut points for continuous variables, and transform it into two categorical variables, all the variables in the model based on logistic equation β value is assigned to diagnostic score factor scoring system to establish the diagnosis. Reuse by ROC curve optimized to achieve the best sensitivity and specificity of the diagnosis when the diagnosis point scoring model and to evaluate the diagnostic performance. The results were obtained four differential diagnosis of tuberculous meningitis and viral meningitis, bacterial meningitis, cryptococcal meningitis, non-tuberculous meningitis diagnosis scoring model. High diagnostic efficiency of each model, the sensitivity and specificity were 92% and 96%, 86% and 90%, 85% and 90%, 93% and 90%. Among them, the longer duration, TB symptoms (anorexia, weight loss), hyponatremia, cerebrospinal fluid sugar, low chloride, high protein, chest radiographic findings of tuberculosis signs and brain imaging tips skull basal cistern enhancement, hydrocephalus , granulomatous lesions of tuberculous meningoencephalitis specific performance. Conclusion This study established the differential diagnosis of central nervous system tuberculosis infection scoring system has a high diagnostic performance, and simple, economical and practical, can be used for early diagnosis of tuberculous meningoencephalitis.
|