A nomogram for predicting the risk of multidrug-resistant tuberculosis / 中华传染病杂志

ABSTRACT

Objective:To construct a simple, precise and personalized comprehensive nomogram for prediction the risk of multidrug-resistant tuberculosis (MDR-TB) and to evaluate its prediction value among individuals with previous tuberculosis history (PTBH).Methods:A matched case-control study (1∶2 ratios) was performed in 1 881 patients with PTBH treated in 12 designated tuberculosis hospitals in Hangzhou City between January 1, 2005 and December 31, 2019, and there were 1 719 patients in training set, and 162 in validation set. A multivariable Cox regression analysis was used to evaluate independent predictors for the incident of MDR-TB in individuals with PTBH. A comprehensive nomogram was developed based on the multivariable Cox model. The accuracy of the prediction was assessed using concordance index (C-index), calibration curve and area under the receiver operator characteristic (ROC) curve.Results:The nomogram constructed based on the multivariable Cox regression model incorporated 10 independent predictors of the risk of MDR-TB. A history of direct contact (grade 1, 0-100.0 points) ranked on the top of all risk factors, followed by duration of positive sputum culture (grade 2, 0-84.5 points), unfavorable treatment outcome (grade 3, 0-52.0 points), human immunodeficiency virus infection (grade 4, 0-48.5 points), retreated tuberculosis history (grade 5, 0-40.0 points), non-standardized treatment regimens of retreated tuberculosis (grade 6, 0-32.5 points), duration of pulmonary cavities (grade 7, 0-31.0 points), passive mode of tuberculosis case finding (grade 8, 0-25.0 points), age<60 years (grade 9, 0-17.5 points), and standard frequencies of chest X-ray examination (grade 10, 0-14.0 points). The C-indexes of this nomogram for the training and validation sets were 0.833 (95% confidence interval ( CI) 0.807-0.859) and 0.871 (95% CI 0.773-0.969), respectively, indicating that the nomogram had good fitting effect. The calibration curves for the risk of incident MDR-TB showed an optimal agreement between nomogram prediction and actual observation in the training and validation sets, respectively.The areas under ROC curve of the 1-year, 5-year, and 10-year MDR-TB risk probability of the training set were 0.904, 0.921, and 0.908, respectively, and those of the validation set were 0.954, 0.970, and 0.919, respectively. Conclusion:Through this nomogram model, clinicians could precisely predict the risk of incident MDR-TB among individuals with PTBH in the clinical practice.