Abstract Body

Early mortality remains high among HIV-positive people starting ART; TB is the leading cause of death. We hypothesised that an algorithm designed for primary care nurses, using point-of-care tools to identify HIV+ people at high risk of TB, with rapid initiation of TB treatment, then ART, could reduce early mortality. We tested this management strategy in an open cluster-randomised trial, where clusters were South African primary care clinics.

24 primary care clinics were randomised 1:1 to intervention or control arms. We enrolled HIV+ adults with CD4≤150, not taking ART or TB treatment. In the intervention arm, study nurses categorised participants’ TB probability as high (any of Hb <10g/dl, body mass index ≤18.5 or a positive lateral flow assay for urinary lipoarabinomannan), medium (any TB symptom, no high probability criteria) or low (no TB symptoms or high probability criteria). If high probability, participants started TB treatment immediately, then ART two weeks later; if medium probability, further TB investigations were arranged, with re-assessment within a week; if low probability, ART was initiated. The primary outcome was all-cause mortality at 6 months; secondary outcomes included the 6-month risk of hospitalisation. Analyses used methods appropriate for cluster-randomised trials with a small number of clusters. The study is complete: the primary outcome results are final.

3030 participants (55% female, median age 37 years, median CD4 72 cell/mm3) were included in the analysis (intervention: 1508; control: 1522). In the intervention arm, 45.7%, 31.5% and 22.8% were categorised as high, medium and low priority respectively. At 6 months, 98.6% (intervention) vs 97.7% (control) had known vital status: mortality rate was 19.0 (intervention) vs. 21.5 (control) per 100 pyrs; adjusted rate ratio 0.87 (95% CI 0.61, 1.24). By 6 months, 13.9% (intervention) vs 10.8% (control) participants had been hospitalised at least once (adjusted risk ratio [aRR] 1.14, 95% CI 0.74, 1.74), and 16.6% (intervention) vs 15.1% (control) participants had been hospitalised and/or had died (aRR 1.01 (95%CI 0.80, 1.28).

A management strategy prioritising TB treatment for ambulatory HIV+ individuals at high risk of TB did not reduce mortality or the risk of hospitalisation by 6 months. Further work is needed to identify strategies to reduce early mortality among HIV+ people presenting for ART with low CD4 counts.