Xpert MTB/XDR implementation in South Africa: cost outcomes of centralised vs. decentralised approaches

Introduction In South Africa, Xpert ® MTB/RIF Ultra (Ultra) is the recommended diagnostic assay for TB with line-probe assays for first- (LPAfl) and second-line drugs (LPAsl) providing additional drug susceptibility testing (DST) for samples that were rifampicin-resistant (RR-TB). To guide implementation of the recently launched Xpert ® MTB/XDR (MTB/XDR) assay, a cost-outcomes analysis was conducted comparing total costs for genotypic DST (gDST) for persons diagnosed with RR-TB considering three strategies: replacing LPAfl/LPAsl (centralised level) with MTB/XDR vs. Ultra reflex testing (decentralised level). Further, DST was performed using residual specimen following RR-TB diagnosis. Methods The total cost of gDST was determined for three strategies, considering loss to follow-up (LTFU), unsuccessful test rates, and specimen volume. Results For 2019, 9,415 persons were diagnosed with RR-TB. A 35% LTFU rate between RR-TB diagnosis and LPAfl/LPAsl-DST was estimated. Unsuccessful test rates of 37% and 23.3% were reported for LPAfl and LPAsl, respectively. The estimated total costs were $191,472 for the conventional strategy, $122,352 for the centralised strategy, and $126,838 for the decentralised strategy. However, it was found that sufficient residual volume for reflex MTB/XDR testing is a limiting factor at the decentralised level. Conclusion Centralising the implementation of XDR testing, as compared to LPAfl/LPAsl, leads to significant cost savings.