Background:  Adherence to antiretroviral therapy (ART) improves survival after HIV infection. Our hypothesis is that pharmacy claims predict survival rates in HIV-infected South African adults.

Methods:  We studied 7812 HIV-infected adult South Africans enrolled in a private-sector HIV/AIDS disease management program and who started triple-combination therapy between January 1999 and March 2003. ART adherence was calculated by dividing the number of months that patients has submitted claims by number of months since commencement of ART. χ² analysis and Cox proportional hazard models were used to identify variables associated with adherence and survival, respectively. Survival curves were compared between groups with the log-rank test.

Results:  The mean age at ART initiation was 37 ±8  years; 4605 patients (56%) were female and 7554 (96.7%) were black Africans. The mean CD4⁺ count and log₁₀ HIV viral load at enrollment were 145 ± 102 cells/mm³ and 5.1 ± 0.7 log₁₀ copies/mL, respectively. ART adherence was > 70% for 3908 patients (50.0%), 40 to70% for 1837 (23.5%) and, < 40% for 2067 (26.46%). Patients with > 70% adherence were more likely to be females than males (62% vs 38%; p < 0.01), but age < 40 years and CD4⁺ count < 200 cells/m³ did not predict mean adherence (p > 0.4). As of March 2003, a total of 718 patients died, yielding a crude mortality rate of 9.2%. In the multivariate Cox proportional hazard model, a threshold hazard ratio at ART adherence equal to 80% was identified. The variables significantly associated with decreased survival were:  low ART adherence (< 80%) (relative hazard [RH]), 1.31, confidence interval (CI) 1.19 to1.43); male gender (RH 1.34, 95%, CI 1.15 to 1.56); age ≥ 40 years (RH 1.21, 1.04, to 1.41), baseline CD4⁺ count < 200 cells/mm³ (RH 3.35, 2.62 to 4.26) and high baseline viral load (RH 1.75, 1.54 to1.98 per log₁₀ increase).

Conclusion:  Poor ART adherence as assessed by ART claim data is associated with decreased survival. Pharmacy claims may be a simple and effective tool for monitoring adherence as ART programs in sub-Saharan Africa are scaled up. Reasons for poor adherence in males in our study population need to be explored further.

Keywords: HAART Adherence; Pharmacy Claims; Survival