Design and protocol for a cluster randomised trial of enhanced diagnostics for tuberculosis screening among people living with HIV in hospital in Malawi (CASTLE study).

BACKGROUND: People living with HIV (PLHIV) have a high risk of death if hospitalised in low-income countries. Tuberculosis has long been the leading cause of admission and death, in part due to suboptimal diagnostics. Two promising new diagnostic tools are digital chest Xray with computer-aided diagnosis (DCXR-CAD) and urine testing with Fujifilm SILVAMP LAM (FujiLAM). Neither test has been rigorously evaluated among inpatients. Test characteristics may be complementary, with FujiLAM especially sensitive for disseminated tuberculosis and DCXR-CAD especially sensitive for pulmonary tuberculosis, making combined interventions of interest. DESIGN AND METHODS: An exploratory unblinded, single site, two-arm cluster randomised controlled trial, with day of admission as the unit of randomisation. A third, smaller, integrated cohort arm (4:4:1 random allocation) contributes to understanding case-mix, but not trial outcomes. Participants are adults living with HIV not currently on TB treatment. The intervention (DCXR-CAD plus urine FujiLAM plus usual care) is compared to usual care alone. The primary outcome is proportion of participants started on tuberculosis treatment by day 56, with secondary outcomes of mortality (time to event) measured to to 56 days from enrolment, proportions with undiagnosed tuberculosis at death or hospital discharge and comparing proportions with enrolment-day tuberculosis treatment initiation. DISCUSSION: Both DCXR-CAD and FujiLAM have potential clinical utility and may have complementary diagnostic performance. To our knowledge, this is the first randomised trial to evaluate these tests among hospitalised PLHIV.