Enhanced Tuberculosis Diagnosis With Computer-aided Chest X-ray and Urine Lipoarabinomannan in Adults With HIV Admitted to Hospital (CASTLE Study): A Cluster Randomized Trial.

BACKGROUND: People with human immunodeficiency virus (PHIV) admitted to the hospital have high mortality, with tuberculosis (TB) being the major cause of death. Systematic use of new TB diagnostics could improve TB diagnosis and might improve outcomes. METHODS: We conducted a cluster randomized trial among adult PHIV admitted to Zomba Central Hospital, Malawi. Admission days were randomly assigned to: enhanced TB diagnostics using urine lipoarabinomannan (LAM) antigen tests (SILVAMP-LAM, Fujifilm, Japan and Determine-LAM, Alere/Abbot, USA), digital chest X-ray with computer-aided diagnosis (dCXR-CAD, CAD4TBv6, Delft, Netherlands), plus usual care ("enhanced TB diagnostics"); or usual care alone ("usual care"). The primary outcome was TB treatment initiation during admission. Secondary outcomes were 56-day mortality, TB diagnosis within 24 hours, and undiagnosed TB at discharge, ascertained by culture of one admission sputum sample. FINDINGS: Between 2 September 2020 and 15 February 2022, we recruited 419 people. Four were excluded postrecruitment, leaving 415 adults recruited during 207 randomly assigned admission days in modified intention-to-treat analysis. At admission, 90.8% (377/415) were taking antiretroviral therapy with a median CD4 cell count of 240 cells/mm3. In the enhanced diagnostic arm, median CAD4TBv6 score was 60 (interquartile range: 51-71), 4.4% (9/207) had SILVAMP-LAM-positive and 14.4% (29/201) had Determine-LAM-positive urine with 3 samples positive by both urine tests. TB treatment was initiated in 46/207 (22.2%) in the enhanced TB diagnostics arm and 24/208 (11.5%) in the usual care arm (risk ratio, 1.92; 95% confidence interval [CI]: 1.20-3.08). There was no difference in mortality by 56 days (enhanced TB diagnosis: 54/207, 26.1%; usual care: 52/208, 25.0%; hazard ratio. 1.05; 95% CI: .72-1.53); TB treatment initiation within 24 hours (enhanced TB diagnosis: 8/207, 3.9%; usual care: 5/208, 2.4%; risk ratio, 1.61; 95% CI: .53-4.71); or undiagnosed microbiological-confirmed TB at discharge (enhanced TB diagnosis, 0/207 [0.0%], usual care arm 2/208 [1.0%]; P = .50. INTERPRETATION: Urine SILVAMP-LAM/Determine-LAM plus dCXR-CAD diagnostics identified more hospitalized PHIV with TB than usual care. The increase in TB treatment appeared mainly because of greater use of Determine-LAM, rather than SILVAMP-LAM or dCXR-CAD. Poor concordance between Determine-LAM and SILVAMP-LAM urine tests requires further investigation. Inpatient mortality for adults with human immunodeficiency virus remains unacceptability high.