Risk stratification for MDR/RR-TB: a comprehensive integrated patient-level analysis from TB programs and contemporary clinical trials

Abstract Tuberculosis is the deadliest infectious disease, and treatments for multidrug-resistant tuberculosis are lengthy and toxic. By analyzing a large international database (N = 7,750), we identified patient risk phenotypes linked to unfavorable outcomes. Factors like HIV infection, age, low BMI, smoking, acid-fast bacillus smear positivity, extrapulmonary involvement, and cavitary disease were associated ( p < 0.001) with worse clinical outcomes (ROC AUC 0.728; 95% CI 0.726–0.731) and were used to derive a patient risk score. Low-risk patients had a lower probability of unfavorable outcome over time since treatment initiation compared to high-risk patients ( p < 0.001). External validation with a highly effective 6-month BPaL regimen showed 97% treatment success for low-risk patients and 88% for medium-to-high-risk patients. Stratifying patients by risk phenotype enables more effective, personalized tuberculosis interventions, which are key to improving cure rates.