Summary

This study addresses a critical gap in the diagnosis of rifampicin resistant tuberculosis by revealing that Xpert MTB/RIF Ultra's low positive predicted value for rifampicin resistance in low-prevalence settings is primarly driven by paucibacillary samples. In our nationwide study in Rwanda, 89% of patients with very low bacillary loads had a false rifampicin-resistant results. This lead to unnecessary MDR-TB treatment for 53% of patients in our study.

These findings have immediate global implications as WHO guidelines now recommend community screening at 0.5% prevalence thresholds, expanding Ultra's use in low prevalence populations where paucibacillary samples are increasingly common.

The research directly impacts patient care by preventing unnecessary MDR-TB treatment, that in Rwanda mandates clinic isolation until culture conversion, and reducing healthcare costs for the TB national program. The proposed diagnostic algorithm incorporating bacillary load thresholds offers a practical solution for tuberculosis programs in a similar low prevalence situation, potentially transforming how rifampicin resistance is interpreted in molecular diagnostics.