Applying a Combined Model to Evaluate the Risk of Poor Treatment Outcomes in Rifampicin Resistant Tuberculosis...

Treating and managing rifampicin resistant tuberculosis (RR-TB) patients in Yunnan, China, are major challenges. This study aims to evaluate the risk of poor treatment outcomes in RR-TB patients, allowing clinical doctors to proactively target patients who would benefit from enhanced patient management.

Four RR-TB care facilities in different regions of Yunnan province as the data collection points were selected. A total of 524 RR-TB patients were included in this study and randomly assigned into a training set (n=366) and a validation set (n=158). In the training set, four significant factors were screened by using a random forest model and a Lasso regression model, and then included in a logistic regression model to construct a nomogram for internal validation.

The successful treatment rate of RR-TB patients in training set was 42.6% (156/366), and the main poor treatment outcomes were loss to follow-up (66.7%) and death (18.1%). Low hemoglobin (HGB) (OR=0.977, 95% CI: 0.964-0.989), long-regime (OR=2.784, 95% CI: 1.634-4.842), poor culture results at the end of the 6th month (CR6TM) (OR=11.193, 95% CI: 6.507-20.028), pre-extensively drug-resistant tuberculosis (pre-XDR) (OR=3.736, 95% CI: 1.294-12.034) were risk factors for poor treatment outcomes in RR-TB patients. The Area Under Curve (AUC) of this model was 0.829 (95% CI: 0.787-0.870), and there was good consistency between the predicted probability and the actual probability. The DCA curve showed that when the threshold probability was 20-98%, the use of nomogram to predict the net benefit of poor treatment outcomes risk in RR-TB patients was higher.

We combined multiple models to develop a nomogram for predicting poor treatment outcomes in RR-TB patients. This would help clinical doctors identify high-risk populations and enable them to proactively target RR-TB patients who will benefit from strengthened patient management.