Accuracy of different Xpert MTB/Rif implementation strategies in programmatic settings at the regional referral hospitals in Uganda: Evidence for country wide roll out

BACKGROUND: Xpert MTB/RIF assay is a highly sensitive test for TB diagnosis, but still costly to most low-income countries. Several implementation strategies instead of frontline have been suggested; however with scarce data. We assessed accuracy of different Xpert MTB/RIF implementation strategies to inform national roll-out. METHODS: This was a cross-sectional study of 1,924 adult presumptive TB patients in five regional referral hospitals of Uganda. Two sputum samples were collected, one for fluorescent microscopy (FM) and Xpert MTB/RIF examined at the study site laboratories. The second sample was sent to the Uganda Supra National TB reference laboratory for culture using both Lowenstein Jensen (LJ) and liquid culture (MGIT). We compared the sensitivities of FM, Xpert MTB/RIF and the incremental sensitivity of Xpert MTB/RIF among patients negative on FM using LJ and/or MGIT as a reference standard. RESULTS: A total 1924 patients were enrolled of which 1596 (83%) patients had at least one laboratory result and 1083 respondents had a complete set of all the laboratory results. A total of 328 (30%) were TB positive on LJ and /or MGIT culture. The sensitivity of FM was n (%; 95% confidence interval) 246 (63.5%; 57.9–68.7) overall compared to 52 (55.4%; 44.1–66.3) among HIV positive individuals, while the sensitivity of Xpert MTB/RIF was 300 (76.2%; 71.7–80.7) and 69 (71.6%; 60.5–81.1) overall and among HIV positive individuals respectively. Overall incremental sensitivity of Xpert MTB/RIF was 60 (36.5%; 27.7–46.0) and 20 (41.7%; 25.5–59.2) among HIV positive individuals. CONCLUSION: Xpert MTB/RIF has a higher sensitivity than FM both in general population and HIV positive population. Xpert MTB/RIF offers a significant increase in terms of diagnostic sensitivity even when it is deployed selectively i.e. among smear negative presumptive TB patients. Our results support frontline use of Xpert MTB/RIF assay in high HIV/TB prevalent countries. In settings with limited access, mechanisms to refer smear negative sputum samples to Xpert MTB/RIF hubs are recommended.