Construction of prediction model for treatment outcomes of tuberculosis patients treated with linezolid-containing regimens and its predictive efficiency

OBJECTIVE To explore the influencing factors for treatment outcomes of the patients with tuberculosis who were treated with linezolid and construct the prediction model for treatment outcomes of the tuberculosis patients who were treated with linezolid-containing regimens. METHODS A total of 382 patients with pulmonary tuberculosis who were treated with linezolid-containing regimens in Changsha Central Hospital from 2019 to 2024 were retrospectively analyzed. The enrolled patients were randomly divided into the training set with 267 cases and the validation set with 115 cases in a 7∶3 ratio. The patients of the training set were divided into the favorable prognosis group with 208 cases and the poor prognosis group with 59 cases according to the treatment outcomes. Influencing factors for the treatment outcomes of the tuberculosis patients were determined by means of multivariate logistic regression analysis, and the prediction model was established. The diagnostic value and clinical application value of the model were evaluated through receiver operating characteristic (ROC) curves, calibration curves and decision curve analysis (DCA). RESULTS The multivariate logistic regression analysis showed that the prognosis of the tuberculosis patients was associated with drug resistance (OR=2.594), coronary heart disease (OR=4.028), C-reactive protein (CRP) (OR=1.010) and hepatoprotective drugs (P＜0.05). Nomogram showed that the risk of adverse prognosis exceeded 90% when the total score was no less than 143 points. The prediction model had favorable discrimination degree in both the training set and the validation set, the areas under the curves(AUCs) were 0.826 and 0.750, respectively. Hosmer-Lemeshow test(the training set P=0.156, the validation set P=0.520) further proved that the model had good calibration. The DCA curve indicated that the model has clinical practicability when the threshold value of probability was within the scope of 0.05 to 0.55. CONCLUSIONS The drug resistance, coronary heart disease, CRP and hepatoprotective drugs have the values in prediction of prognosis of the tuberculosis patients. The prediction model shows favorable diagnostic efficiencies.