Risk factors for poor prognosis in patients with tuberculous meningitis and establishment of their nomogram predictive models / 中华神经医学杂志

ABSTRACT

Objective:To explore the risk factors for poor prognosis in patients with tuberculous meningitis (TBM), and establish their nomogram predictive models.Methods:Three hundred and fifty-eight patients with TBM, admitted to Department of Neurology, 940 th Hospital of Chinese People's Liberation Army Joint Logistic Support Force from January 2010 to February 2022, were chosen and divided into model group ( n=287) and validation group ( n=71) according to the simple random sampling at a ratio of 8:2. Their clinical data were retrospectively analyzed. Independent risk factors for poor prognosis of TBM were analyzed by least absolute shrinkage and selection operator (LASSO) and Logistic regression, and the risk factors were visualized by nomogram. Bootstrap method was used for 1 000 repeated samples for internal verification, and data from validation group were used for external verification. The discrimination and calibration of the models were evaluated by area under receiver operating characteristic (ROC) curve, calibration curve and Hosmer-Lemeshow goodness of fit test. Results:Multivariate Logistic regression analysis showed that mental symptoms ( OR=3.593, 95% CI: 1.790-7.211, P<0.001), limb weakness ( OR=3.087, 95% CI: 1.551-6.144, P=0.001), pulmonary infection ( OR=5.162, 95% CI: 2.373-11.227, P<0.001), improved British Medical Research Council (mBMRC) staging II ( OR=4.291, 95% CI: 2.037-9.039, P<0.001), mBMRC staging III ( OR=13.073, 95% CI: 3.352-50.975, P<0.001) were independent risk factors for poor prognosis in TBM patients. Consistency indexes indicated by area under ROC curves by internal verification and external verification were 0.880 and 0.823, respectively; the calibration curve coincided with the ideal curve, enjoying good Hosmer-Lemeshow fit ( P=0.546, P=0.401). Conclusion:Based on these 4 prognostic factors (mental symptoms, limb weakness, lung infection and mBMRC stages), the nomogram is helpful in predicting the risk of poor prognosis in TBM patients.