Effectiveness of interventions for unstable patients on antiretroviral therapy in South Africa: results of a cluster-randomised evaluation.

Abstract

As loss from HIV care is an ongoing challenge globally, interventions are needed for patients who don't achieve or maintain ART stability. The 2015 South African National Adherence Guidelines (AGL) for Chronic Diseases include two interventions targeted at unstable patients: early tracing of patients who miss visits (TRIC) and enhanced adherence counselling (EAC). As part of a cluster-randomised evaluation at 12 intervention and 12 control clinics in four provinces, intervention sites implemented the AGL interventions, while control sites retained standard care. We report on outcomes of EAC for patients with an elevated viral load (>400 copies/ml) and for TRIC patients who missed a visit by >5 days. We estimated risk differences (RD) of 3 and 12-month viral resuppression (<400 copies/ml) and 12-month retention with cluster adjustment using generalised estimating equations and controlled for imbalances using difference-in-differences compared to all eligible in 2015, prior to intervention roll-out. For EAC, we had 358 intervention and 505 control site patients (61% female, median ART initiation CD4 count 154 cells/μl). We found no difference between arms in 3-month resuppression (RD: -1.7%; 95%CI: -4.3% to 0.9%), but <20% of patients had a repeat viral load within 3 months (19.8% intervention, 13.5% control). Including the entire clinic population eligible for EAC with a repeat viral load at all evaluation sites (n = 934), intervention sites showed a small increase in 3-month resuppression (28% vs. 25%, RD 3.0%; 95%CI: -2.7% to 8.8%). Adjusting for baseline differences increased the RD to 8.1% (95% CI: -0.1% to 17.2%). However, we found no differences in 12-month suppression (RD: 1.5%; 95% CI: -14.1% to 17.1% but suppression was low overall at 40%) or retention (RD: 2.8%; 95% CI: -7.5% to 13.2%). For TRIC, we enrolled 155 at intervention sites and 248 at control sites (44% >40 years, 67% female, median CD4 count 212 cells/μl). We found no difference between groups in return to care by 12 months (RD: -6.8%; 95% CI: -17.7% to 4.8%). During the study period, control sites continued to use tracing within standard care, however, potentially masking intervention effects. Enhanced adherence counselling showed no benefit over 12 months. Implementation of the tracing intervention under the new guidelines was similar to the standard of care. Interventions that aim to return unstable patients to care should incorporate active monitoring to determine if the interventions are effective.