Background3 In resource-constrained settings, people living with HIV (PLWH) treated for tuberculosis (TB) 4 despite negative bacteriologic tests have a higher mortality than those treated with positive tests. 5 Many PLWH are treated without bacteriologic testing; their mortality compared to those with 6 bacteriologic testing is uncertain. 7 8 Methods 9 We conducted an observational cohort study among PLWH ≥ 15 years of age who initiated TB 10 treatment at clinical sites affiliated with four regions of the International epidemiology Databases 11 to Evaluate AIDS (IeDEA) consortium from 2012-2014: Caribbean, Central and South America, 12 and Central, East, and West Africa. The primary exposure of interest was the TB bacteriologic 13 test status at TB treatment initiation: positive, negative, or no test result. The hazard for death in 14 the 12 months following TB treatment initiation was estimated using the Cox proportional 15 hazard model, adjusted for patient-and site-level factors. Missing covariates were multiply 16 imputed. 17 18 Results 19 Among 2,091 PLWH included, the median age at TB treatment initiation was 36 years, 44% 20 were female, 53% had CD4 counts ≤ 200 cells/mm 3 , and 52% were on antiretroviral treatment 21 (ART). Compared to patients with positive bacteriologic tests, the adjusted hazard for death was 22 higher among patients with no test results (HR 1.56, 95% CI 1.08-2.26) but not different than 4 23 those with negative tests (HR 1.28, 95% CI 0.91-1.81). Older age was also associated with a 24 higher hazard for death, while being on ART, having a higher CD4 count, West Africa region, 25 and tertiary facility level were associated with lower hazards for death. 26 27 Conclusion 28 PLWH treated for TB with no bacteriologic test results were more likely to die than those treated 29 with positive tests, underscoring the importance of TB bacteriologic diagnosis in resource-30 constrained settings. Research is needed to understand the causes of death among PLWH treated 31 for TB in the absence of positive bacteriologic tests. 5 32 Introduction 33 Although tuberculosis (TB) accounted for 300,000 deaths among people living with HIV 34 (PLWH) in 2017, diagnosing TB in resource-limited settings remains a challenge [1]. In 2017, 35 only 56% of the 5.5 million pulmonary TB cases reported to the World Health Organization 36 (WHO) globally were bacteriologically confirmed (i.e. positive for smear microscopy, culture, or 37 nucleic acid amplification test [NAAT]) [1]. Among studies reporting the autopsy prevalence of 38 TB in HIV-related deaths, TB was prevalent in 37% of deaths, but in half of those cases, TB was 39 not diagnosed by the time of death [2]. The rollout of nucleic acid amplification tests (NAAT)40 such as the Xpert® MTB/RIF (Cepheid, Sunnyvale, CA, USA), which are more sensitive and 41 specific than smear microscopy, has helped close this diagnostic gap [3]. However, limited 42 impact on mortality has been observed with the use of Xpert MTB/RIF, in part due to high 43 baseline rates of empiri...