Analysis of clinical characteristics of smear-negative pulmonary tuberculosis and construction of a prediction model
Objective To analyze the clinical characteristics of smear-negative pulmonary tuberculosis and establish a prediction model.Methods A total of 118 patients with suspected pulmonary tuberculosis and sputum acid-fast staining smear negative for three times hospitalized in the Shenzhen Bao'an District Songgang People's Hospital from December 2018 to January 2021 were selected and divided into tuberculosis group(n=68)and non-tuberculosis group(n=50)according to final diagnosis.Clinical data including general information,symtoms,tuberculosis-specific enzyme-linked immunospot assay(T-SPOT.TB)results of bronchoalveolar lavage fluid(BALF)acid-fast staining smear,real-time PCR and gene Xpert Mycobacterium tuberculosis/Rifampin(Gene Xpert MTB/RIF)were collected.Logistic regression analysis was performed to analyze the clinical characteristics of smear-negative pulmonary tuberculosis and a prediction model was constructed based on the regression results.Results Based on the univariate analysis,significant differences were found in age,weight,erythrocyte sedimentation rate,BALF real-time PCR,BALF Gene Xpert MTB/RIF,T-SPOT.TB,fever and cough symptoms between the two groups.Multivariate logistic regression analysis showed that smear-negative pulmonary tuberculosis presented with low weight,positive BALF real-time PCR(OR=25.887,95％CI:1.826-367.005)and Gene Xpert MTB/RIF(OR=30.553,95％CI:2.078-449.199),positive T-SPOT.TB(OR=53.739,95％CI:4.547-635.121)and no cough symptom(OR=0.049,95％CI:0.004-0.599).The AUC and C-index of the model were 0.955 after internal validation of the model by bootstrap confidence analysis.The AUC of the nomogram prediction model was 0.973;the calibration curve basically coincided with the diagonal line,and there was good consistency between the predicted value of the model and the actual observed value.The DCA curve showed that the clinical benefit of the predictive model was always higher than that of"full intervention"and"no intervention",showing that the model had good clinical applicability.Conclusion The diagnostic model constructed for predicting smear-negative pulmonary tuberculosis based on weight,BALF real-time PCR,BALF Gene Xpert MTB/RIF,T-SPOT.TB and cough symptoms showed good discrimination and accuracy and could be used for identification of smear negative tuberculosis.
NETL
NSTL
万方数据
