A clinical indicator-based prognostic model predicting treatment outcomes of pulmonary tuberculosis: a prospective cohort study

Objectives Identifying prognostic factors helps optimize the treatment regimen and promote favorable outcomes. We conducted a prospective cohort study on patients with pulmonary tuberculosis to construct a clinical indicator-based model and estimate its performance. Methods We performed a two-stage study by recruiting 346 pulmonary tuberculosis patients diagnosed between 2016 and 2018 in Dafeng city as the training cohort and 132 patients diagnosed between 2018 and 2019 in Nanjing city as the external validation population. We generated a risk score based on blood and biochemistry examination indicators by the least absolute shrinkage and selection operator (LASSO) Cox regression. Univariate and multivariate Cox regression models were used to assess the risk score, and the strength of association was expressed as the hazard ratio (HR) and 95% confidence interval (CI). We plotted the receiver operating characteristic (ROC) curve and calculated the area under the curve (AUC). Internal validation was conducted by 10-fold cross-validation. Results Ten significant indicators (PLT, PCV, LYMPH, MONO%, NEUT, NEUT%, TBTL, ALT, UA, and Cys-C) were selected to generate the risk score. Clinical indicator-based score (HR: 10.018, 95% CI: 4.904–20.468, P < 0.001), symptom-based score (HR: 1.356, 95% CI: 1.079–1.704, P = 0.009), pulmonary cavity (HR: 0.242, 95% CI: 0.087–0.674, P = 0.007), treatment history (HR: 2.810, 95% CI: 1.137–6.948, P = 0.025), and tobacco smoking (HR: 2.499, 95% CI: 1.097–5.691, P = 0.029) were significantly related to the treatment outcomes. The AUC was 0.766 (95% CI: 0.649–0.863) in the training cohort and 0.796 (95% CI: 0.630–0.928) in the validation dataset. Conclusion In addition to the traditional predictive factors, the clinical indicator-based risk score determined in this study has a good prediction effect on the prognosis of tuberculosis.