Establishment Of Predictive Model For Pulmonary Tuberculosis Complicated With Tracheobronchial Tuberculosis

Objective: To explore the risk factors of pulmonary tuberculosis(PTB)complicated with tracheobronchial tuberculosis(TBTB),and establish a clinical risk prediction model for the early diagnosis of TBTB.Methods: Totally 1489 patients with pulmonary tuberculosis from January 1,2019 to January 31,2021 on the Yi Du Cloud platform were analyzed.The patients were divided into PTB group(744 cases)and TBTB group(745 cases).The differences of initial baseline features,clinical mainfestations,laboratory findings and imaging feastures between of the two groups were compared.The values were assigned according to the odds ratios(OR)of each factor to establish the risk prediction model.The model was evaluated using Hosmer-lemeshow test and receiver operating characteristic(ROC).Results:(1)Univariate analysis: Compared with PTB group,TBTB group showed age,gender,occupation,education level,household income,cough,expectoration,WBC,RBC,HB,tuberculosis antibody,T-SPOT,PPD,sputum etiology,m MRC and St.George’s score were statistically significant factors(P < 0.05).(2)Logistic regression analysis reveals that the age less than 40 years old(OR = 1.731,95% CI = 1.279-2.343),female(OR = 2.831,95% CI =2.090-3.835),cough(OR = 2.951,95% CI = 1.925-4.526)and positive in sputum etiology(OR = 3.496,95% CI = 2.565-4.764)were independent risk factors for PTB complicated with TBTB(P < 0.05).According to the OR value of each factor,the risk prediction model is established.The area under the ROC curve(AUC)of the model was 0.745.The sensitivity of the model was 65.5%,and the specificity was 70.3%.Conclusions: The age less than 40 years old,female,cough and positive in sputum etiology are independent risk factors of PTB complicated with TBTB.The risk prediction model might be an effective tool for predicting whether the PTB will be complicated with TBTB.