CASTLE is a cluster randomised trial comparing enhanced TB diagnostics (urine SILVAMP-LAM plus urine LF-LAM plus digital chest Xray with computer aided diagnosis plus usual care) to usual care alone, among adults living with HIV admitted to hospital. Clusters are day of admission to hospital. Primary outcome is TB treatment started prior to hospital discharge. Secondary outcomes are all-cause mortality at 56 days from enrolment, undiagnosed TB at time of discharge from hospital, and TB treatment initiation within 24 hours of enrolment. Data was collected using a combination of face-to-face questionnaire, laboratory test results, and information gathered from written medical records by trained research assistants.