Development and validation of a nomogram to predict atelectasis in adult lymph node fistula tracheobronchial tuberculosis patients.

BACKGROUND

Lymph node fistula tracheobronchial tuberculosis (TBTB) is a severe respiratory condition that can result in complications such as airway stenosis and atelectasis, posing significant clinical challenges, particularly in adults. Currently, no standardized assessment tools are available to predict the risk of atelectasis in these patients, highlighting the need to develop an effective predictive model to guide early clinical intervention and personalized treatment.

METHODS

A retrospective study was conducted involving 547 adult patients diagnosed with lymph node fistula TBTB at our hospital between January 2017 and December 2023. Diagnoses were confirmed by chest computed tomography, bronchoscopy, and combined etiological or pathological examinations. After applying the inclusion and exclusion criteria, 301 cases were included in the final analysis. Patients were randomly assigned to a development group ( = 211, 70%) and a validation group. Following univariate and multivariable logistic regression to identify significant predictors, we developed a nomogram. Model validation included assessment of discriminatory ability [receiver operator characteristic (ROC) analysis], calibration accuracy, and clinical utility (DCA).

RESULTS

Among the 301 patients with lymph node fistula TBTB, the incidence of atelectasis was 60.13% (181/301). Of those, 72.93% (132/181) had right lung involvement, and 50.28% (91/181) specifically had atelectasis in the right middle lobe. Independent predictors identified by multivariable logistic regression included age, occupation as a farmer, mediastinal lymphadenopathy with ring enhancement, and right middle lobe bronchial involvement. A risk nomogram was developed using these predictors. The area under the curve (AUC) of the nomogram was 0.824 (95% CI: 0.685-0.806) in the development group and 0.857 (95% CI: 0.702-0.877) in the validation group. Calibration plots based on 500 bootstrap resamples showed good agreement between predicted and observed probabilities across both groups. DCA revealed that the model provided a net clinical benefit within threshold probability ranges of 0.2-0.9 for the development group and 0.15-0.85 for the validation group.

CONCLUSION

The predictive model and associated nomogram developed in this study can accurately estimate the risk of atelectasis in adult patients with lymph node fistula TBTB. This tool may assist clinicians in developing individualized intervention strategies.