Rollout of ART in sub-Saharan Africa has been accompanied by high prevalence of virological failure (VF) and resistance on NNRTI-based first line ART. This has led to increasing numbers being switched onto PI-based second line ART (SLART). The limited data on outcomes on SLART in sub-Saharan Africa are predominantly from urban settings. This study determined the incidence and risk factors for VF amongst second line patients receiving routine HIV care and treatment in a government program in a poor and rural setting in Kwazulu-Natal.
Participants were HIV-1 positive adults >15 years initiating SLART between April 2007 and January 2014. They were identified from the Africa Centre’s ART Evaluation and Monitoring System. Their clinical data were anonymised and linked to the Africa Centre’s demographic information system. Exposures were adherence measured as a medication possession percentage (MPP); the duration on failing first line regimen; non-ownership of a refrigerator for storing soft gel lopinavir tablets. We defined VF as viral load (VL) >1000 copies/ml after 6 months of commencing SLART, or death or loss to follow up within 12 months without evidence of suppression (VL<1000). We used competing risk regression for analysis with other death as a competing risk.
Three hundred and fifteen adults started SLART. The median age was 33 (IQR 28-38) years and 72% were female. Unemployment was high (89.8%). Median CD4 at switch was 220 (IQR 113-342) cells/mm3. Nine patients with < 6 months follow up at January 2014 were excluded from analysis. The overall incidence rate of VF on SLART was 21.4 (95%CI 17.9-25.7) per 100 person-years. The cumulative incidence function of failure by 5 years was 45%. A higher MPP in the first 12 months of SLART was strongly associated with a lower risk of VF (sHR=0.67, 95%CI 0.45-0.99, and sHR=0.22, 95% CI 0.09-0.51, comparing 60-94% and ≥95% MPP, respectively, with <60% MPP, p<0.01). This association remained after adjusting for confounders. There was no evidence of an association with refrigerator ownership or duration on failing first line ART.
There is a high incidence of VF on PI-based SLART in this rural setting with high unemployment. The level of adherence to treatment predicts VF. This has implications for the move to near universal, early ART in resource limited settings particularly as there is a lack of availability of third line ART.