Development and validation of a nomogram predicting multidrug-resistant tuberculosis risk in East China.

Abstract

Objective: Multidrug-resistant tuberculosis (MDR-TB) is a global health threat. Our study aimed to develop and externally validate a nomogram to estimate the probability of MDR-TB in patients with TB.

Methods: A total of 453 patients with TB in Anhui Chest Hospital between January 2019 and December 2020 were included in the training cohort. In addition, 116 patients with TB from Anhui Provincial Hospital Infection District between January 2015 and November 2023 were included in the validation cohort. Multivariable logistic regression analysis was applied to build a predictive model by combining the feature selected in the least absolute shrinkage and selection operator regression model. The C-index, calibration plot, and decision curve analysis were implemented to evaluate the predictive model's discrimination, calibration, and clinical practicality. Then, logistic regression and least absolute shrinkage and selection operator (LASSO) models were constructed using R software, and the accuracy, goodness of fit, and stability of the models were verified using the validation cohort.

Results: Eight variables of patients with TB were selected using the best penalization parameter of the LASSO regression method, and the nomogram was established. The model displayed good discrimination with a C-index of 0.752 and good calibration. A high C-index value of 0.825 could still be reached in the validation cohort. The decision curve analysis demonstrated the clinical value of the model.

Conclusion: In this study, we constructed the LASSO regression model based on eight clinical traits and outcomes of laboratory tests, providing a novel insight for evaluating MDR-TB.