BMC Pulmonary Medicine (Feb 2023)

Clinical characteristics and predictive model of pulmonary tuberculosis patients with pulmonary fungal coinfection

  • Hongxuan Yan,
  • Li Guo,
  • Yu Pang,
  • Fangchao Liu,
  • Tianhui Liu,
  • Mengqiu Gao

DOI
https://doi.org/10.1186/s12890-023-02344-4
Journal volume & issue
Vol. 23, no. 1
pp. 1 – 11

Abstract

Read online

Abstract Background In clinical settings, pulmonary tuberculosis (PTB) patients were often found to have pulmonary fungal coinfection. This study aimed to assess the clinical characteristics of patients suffering from coinfection with TB and pulmonary fungal and construct a predictive model for evaluating the probability of pulmonary fungal coinfection in patients with pulmonary tuberculosis. Methods The present case–control study retrospectively collected information from 286 patients affected by PTB who received treatment from December 6,2016- December 6,2021 at Beijing Chest Hospital, Capital Medical University. As control subjects, patients with sex and address corresponding to those of the case subjects were included in the study in a ratio of 1:1. These 286 patients were randomly divided into the training and internal validation sets in a ratio of 3:1. Chi-square test and logistic regression analysis were performed for the training set, and a predictive model was developed using the selected predictors. Bootstrapping was performed for internal validation. Results Seven variables [illness course, pulmonary cavitation, broad-spectrum antibiotics use for at least 1 week, chemotherapy or immunosuppressants, surgery, bacterial pneumonia, and hypoproteinemia] were validated and used to develop a predictive model which showed good discrimination capability for both training set [area under the curve (AUC) = 0.860, 95% confidence interval (CI) = 0.811–0.909] and internal validation set (AUC = 0.884, 95% CI = 0.799–0.970). The calibration curves also showed that the probabilities predicted using the predictive model had satisfactory consistency with the actual probability for both training and internal validation sets. Conclusions We developed a predictive model that can predict the probability of pulmonary fungal coinfection in pulmonary tuberculosis patients. It showed potential clinical utility.

Keywords