Adverse treatment outcomes in multidrug resistant tuberculosis go beyond the microbe-drug interaction: Results of a multiple correspondence analysis

Development and Validation of a Nomogram for the Prediction of Unfavorable Treatment Outcome Among Multi-Drug Resistant Tuberculosis Patients in North West Ethiopia: An Application of Prediction Modelling

Background

Multidrug-resistant tuberculosis (MDR-TB) is a global problem and a health security threat, which makes “Ending the global TB epidemic in 2035” unachievable. Globally, the unfavourable treatment outcome remains unacceptably high. Therefore, this study aimed to develop a risk prediction model for unfavorable treatment outcomes in MDR-TB patients, which can be used by clinicians as a simple clinical tool in their decision-making.

Objective

The objective of this study was to develop and validate a risk prediction model for the prediction of unfavorable treatment outcomes among MDR-TB patients in North-West Ethiopia.

Methods

We used MDR-TB data collected from the University of Gondar and Debre Markos referral hospitals. A retrospective follow-up study was conducted and a total of 517 patients were included in the study. STATA version 16 statistical software and R version 4.0.5 were used for the analysis. Descriptive statistics were carried out. A multivariable model was fitted using all potent predictors selected by the lasso regression method. A simplified risk prediction model (nomogram) was developed based on the binomial logit-based model, and its performance was described by assessing its discriminatory power and calibration. Finally, decision curve analysis (DCA) was done to evaluate the clinical and public health impact of the developed model.

Results

The developed nomogram comprised six predictors: baseline anemia, major adverse event, comorbidity, age, marital status, and treatment supporter. The model has a discriminatory power of 0.753 (95% CI: 0.708, 0.798) and calibration test of (P-value = 0.695). It was internally validated by bootstrapping method, and it has a relatively corrected discrimination performance (AUC = 0.744, 95CI: 0.699, 0.788). The optimism coefficient was found to be 0.009. The decision curve analysis showed the net benefit of the model as threshold probabilities varied.

Conclusion

The developed nomogram can be used for individualized prediction of unfavorable treatment outcomes in MDR-TB patients for it has a satisfactory level of accuracy and good calibration. The model is clinically interpretable and was found to have added benefits in clinical practice.