Rapid Urine-Based Screening for Tuberculosis to Reduce AIDS-Related Mortality in Hospitalized Patients in Africa (STAMP) Trial

Sub-Saharan Africa bears the brunt of the global HIV/AIDS epidemic, with 23.5 million people living with HIV and 1.2 million deaths in 2012. Tuberculosis (TB) is the leading cause of AIDS-related illness and deaths worldwide and 75% of this disease burden is in sub-Saharan Africa. Studies from across the continent have shown that 30%-67% of HIV-infected adult hospital in-patients who die have evidence of TB at post-mortem. Much of this disease is neither clinically suspected nor diagnosed before death. This indicates abject failure of current approaches to diagnosis, which is the key problem addressed by this trial. The very large number of HIV-TB deaths means that Africa is not on track to meet the WHO TB control targets linked to the 2015 MDGs. This has therefore become a high-level strategic priority, with calls for action from WHO, UNAIDS, STOP-TB, PEPFAR and other agencies. We believe that recent advances in TB diagnosis can be harnessed to address this challenge in a fundamentally new way. Background studies conducted by the PI in South Africa found that the burden of confirmed TB among unselected HIV-infected patients needing hospital admission was extremely high (32.6%). Symptoms were so poorly predictive for screening, however, that in day-to-day clinical practice, much of this disease remains 'under the radar' and undetected. We propose that, regardless of symptoms, all such patients should be investigated for TB on admission. In these studies the diagnostic yield using the traditional approach of sputum-based testing was, however, very limited as fewer than half of the patients could produce sputum samples and much disease involved organs other than the lungs. In contrast, urine samples could be obtained from almost all patients and the yield of diagnoses from testing these with rapid diagnostics was far greater than that obtained from sputum. Urine was first tested using a simple 30-minute strip-test which is commercially available (Determine TB-LAM). Urine was also concentrated by simple centrifugation and tested using the new WHO-approved rapid Xpert MTB/RIF assay (Xpert). Compared to Xpert testing of sputum alone, addition of this urine-based testing strategy to the initial screen increased the early diagnostic yield of TB 3.0-fold - an extraordinary improvement. Some recent studies have reported that implementation of new TB diagnostics with superior sensitivity did not improve patient outcomes. Thus, since use of new, costly tests might divert limited resources from other important healthcare needs, it is imperative that the impact, cost and cost-effectiveness of new diagnostic strategies such as this one are assessed in trials. This randomised controlled trial will enrol adult HIV-infected medical inpatients admitted to two regional referral hospitals in South Africa and Malawi. On admission, patients will be randomised to one of two TB screening strategies (1,300 patients in each arm), comparing TB diagnosis by Xpert testing of sputum (standard arm) with an intervention arm in which urine will in addition be screened with a combination of the two urine-based diagnostic tests. The care of patients provided by the routine medical team will not be otherwise altered. The main study outcome will show whether additional urine-based screening results in greater survival due to improved yield and speed of TB diagnosis. To more fully assess the overall impact on patients and the efficiency of the healthcare received, a range of additional outcomes will be assessed. These include the total yield of TB diagnoses; times to diagnosis; the proportions starting TB treatment, other antibiotics and antiretrovirals; the length of hospital stay and the need for readmission. Implementation of this intervention will be further justified by comprehensive cost-effectiveness and budgetary impact analyses. If impact is demonstrated, this intervention could be readily implemented and scaled-up.