Predictive Factors for Progression to Severe Pulmonary Tuberculosis in Patients with Concurrent HIV/AIDS and Type 2 Diabetes Mellitus: A Single-Center Retrospective Study

Purpose: The concurrent of HIV/AIDS, type 2 diabetes mellitus (T2D), and pulmonary tuberculosis (PTB) is a considerable public health challenge. This study aims to develop and validate a predictive model for the progression to severe PTB in patients with these three specific comorbidities. Patients and Methods: A total of 114 patients with all three conditions—HIV/AIDS, T2D, and newly diagnosed PTB—who were admitted to the Public Health Clinical Center of Chengdu between 2018 and 2025, were enrolled as study subjects. We collected general demographic information of the research subjects, as well as laboratory indicators, TB etiological test results, and lung CT scans. First, subjects were divided into severe and non-severe PTB groups according to diagnostic criteria met during hospitalization, the Least Absolute Shrinkage and Selection Operator (LASSO) analysis was used to screen for predictors. Second, a multivariate logistic regression was used to build the predictive model. The model was evaluated using the receiver operating characteristic (ROC) curve, and calibration curve. Results: Binary logistic regression revealed that admission ALB, NLR, and D-Dimer were independently associated with progression to severe PTB. ALB ( OR =0.906, 95% CI : 0.822– 0.998) served as a protective factor, while NLR ( OR =1.234, 95% CI : 1.075– 1.417) and D-Dimer ( OR =1.500, 95% CI : 1.113– 2.021) were risk factors. The area under the curve (AUC) of the model incorporating three variables was 0.892 (95% CI : 0.830– 0.955). Internal validation using Bootstrap resampling (1000 iterations) yielded a concordance index of 0.892 (95% CI : 0.825– 0.946), confirming the robustness of its discriminative power. Conclusion: This study established a predictive model for progression to severe PTB in patients with these three specific comorbidities. Due to the lack of external validation, the clinical utility of this model remains to be further validated by multicenter studies.