TB fast track: effect of a point-of-care TB test-and-treat algorithm on early mortality in people with HIV accessing ART

Treatment for HIV (antiretroviral therapy, or ART), has greatly reduced death rates among people with HIV worldwide. However in low-resource settings, death rates continue to be high even for people taking ART, particularly in the first few months on treatment. Tuberculosis (TB) is the most important cause of these early deaths, and people who are due to start ART are recommended to be screened for TB first. However, TB is hard to diagnose because the tests which are most widely available do not work well for people with HIV. As a result, people with HIV may have to go through lots of tests for TB, which delays the start of TB treatment; it also delays the start of ART. Both of these delays increase the risk that the patient dies.
We propose a study to try to reduce this high death rate. In 20 primary care clinics in South Africa, we will enrol people with advanced HIV who have come to start ART to our study. In 10 of the clinics (intervention), selected at random, we will use simple, inexpensive tests (a new urine test for TB, weight and height, and a blood test for anaemia) which can be done on site with immediate results, to identify those patients who are at highest risk of having TB, and of dying early. People who are high risk will start TB treatment straight away, then ART as soon as possible afterwards; people who are low TB risk will start ART straight away. People with medium TB risk will be given antibiotics while a chest X-ray and sputum (spit) test for TB are done, and will come back after a week to decide if they should start TB treatment followed by ART, or just start ART. At the other 10 (control) clinics, patients will be looked after in the normal way, following national guidelines. We will follow all patients up for 6 months. We will compare the death rate at 6 months in the 10 intervention clinics, using our new approach, to death rates in the 10 control clinics, using the routine approach. If our new approach reduces the death rate, clinics throughout low resource settings could use this low-cost strategy to save lives among people with HIV.

Background: Early mortality remains high among adults with HIV accessing antiretroviral therapy (ART) in resource-constrained settings, with tuberculosis (TB) a leading cause of death. 20% people accessing ART in South Africa have undiagnosed TB. However, due to insensitivity of sputum microscopy, and a complex diagnostic pathway for smear-negative TB, the time to TB diagnosis is protracted, delaying both TB treatment (among those diagnosed with TB) and ART (among all investigated, whether or not TB is diagnosed). We hypothesise that a care pathway using point-of-care technology to rapidly identify individuals presenting for ART at high risk of TB and ensure they start TB treatment, then ART, will markedly reduce early mortality.
Objective: To compare 6-month mortality among patients with HIV disease and CD4 150 cells/?l referred to start ART between intervention clinics implementing a point-of-care algorithm to identify individuals with a high probability of active TB, and facilitate rapid initiation of TB treatment, followed by ART; vs. control clinics delivering standard of care management according to South African TB/HIV care guidelines.
Methods: This is a cluster-randomised trial, randomising 20 primary health clinics in Gauteng and Limpopo provinces, South Africa. ART-eligible patients (CD4 150, ambulant, no recent TB treatment) at intervention clinics will be assessed using simple point-of-care assays. Patients with any of: positive urine lipoarabinomannan (LAM), haemoglobin (Hb) 10, body mass index (BMI) 18.5 will be considered high probability TB, and start TB treatment immediately, then ART. Asymptomatic patients with negative LAM, Hb 10, BMI 18.5 will be low probability and will start ART. Others ( medium probability ) will be investigated for smear-negative TB as per South African guidelines with review after one week to re-categorise into high or low probability. In control clinics we will recruit similar patients and obtain permission for follow-up; management will be as per national guidelines. The primary outcome is mortality at 6 months after recruitment; secondary outcomes include severe morbidity, measured by duration of hospitalisation. Diagnostic cost per TB case detected, and cost per DALY, will be estimated. We estimate mortality conservatively at 25/100 person years in the control arm; with 10 clusters per arm, 175 patients per cluster, and 95% ascertainment of vital status at 6 months, using between-cluster coefficients of variation of 0.2 and 0.25, there will be 91% and 85% power to assess 40% reduction in mortality, respectively.
Significance: This strategy may result in a major reduction in mortality for people with HIV.