Development and validation of a clinical risk score nomogram for distinguishing nontuberculous mycobacteria from Mycobacterium tuberculosis pulmonary disease

Background: The prevalence of nontuberculous mycobacterial pulmonary disease (NTM-PD) has increased substantially in recent years, positioning it as a major global public health concern. This rise is compounded by its clinical and radiographic similarities to pulmonary tuberculosis (PTB), encompassing shared manifestations such as chronic cough, hemoptysis, and cavitary lesions on imaging, which pose significant challenges in differentiating infections caused by nontuberculous mycobacteria from those attributable to Mycobacterium tuberculosis. As a result, NTM-PD is frequently misdiagnosed as PTB, prompting the administration of antituberculous regimens that are ineffective against Mycobacterium tuberculosis (NTM) species and may foster antimicrobial resistance. These diagnostic errors not only delay the initiation of appropriate targeted therapies but also exacerbate disease progression, adversely affect patient outcomes, and impose unnecessary burdens on healthcare systems through prolonged hospitalizations, superfluous diagnostic evaluations, and inefficient interventions. Methods: The data from patients diagnosed with NTM-PD or PTB at Peking University People's Hospital and Peking University International Hospital (China) between January 2021 and January 2024 was retrospectively analyzed. The patients from Beijing University People's Hospital were assigned to the training group, while the patients from Beijing University International Hospital were assigned to the validation group. The univariable and multivariable logistic analysis gradually identified the independent risk factors for NTM-PD. A nomogram was constructed incorporating variables with statistical significance (p < 0.05) in the multivariate logistic analysis and those deemed clinically relevant. Calibration curves and clinical decision curve analysis (DCA) were used to assess the model's fitting performance and clinical value. Harrell's C-index and the area under the receiver operating characteristic curves (AUC) were used to evaluate the predictive effectiveness of models. Results: A total of 233 patients were ultimately included. There were 169 patients in the training group and 64 patients in the validation group. The nomogram incorporated the following predictors: Sex, Tumor, Pulmonary nodule size, Mediastinal lymph nodes enlarged (MLOE), Bronchiectasis, Honeycomb lung, Pleural effusion. The C-index of nomogram was 0.895 (95% CI:0.831-0.960) in the training group and 0.855 (95%CI:0.760-0.991) in the validation group, indicating robust predictive performance. Conclusions: We have established and validated a nomogram to predicting the risk of NTM-PD and PTB in patients. The clinical features and radiographic-including Female, Tumor, Bronchiectasis, Honeycomb lung, absence of MLOE and no Pleural effusion-are hallmark indicators of NTM-PD. It helps clinicians to distinguish PTB from NTM-PD, especially junior physicians.