Background:  High mortality rates in the first year after initiation of ART remain a major challenge for roll-out programs across Sub-Saharan Africa. We compared outcomes of African patients initiating ART in London and in Uganda, to determine whether the higher mortality in Uganda was due to more advanced disease at ART initiation, lower rates of virological suppression, a higher rate of opportunistic infections, or a higher death rate with an opportunistic infection, indicating poorer access to opportunistic infection management.

Methods:  We compared 432 African patients who initiated ART (January 2000 to December 2004) at an HIV clinic in south London, with 559 patients who initiated ART (April 2004 to April 2005) as part of a research cohort at the Infectious Diseases Institute, an urban center of excellence in HIV care in Uganda. The majority received an NNRTI-based regimen. Clinical and CD4 assessment was every 3 months, and virological testing every 6 months. Key outcomes were:  cumulative percentage with viral load <400 copies/mL and CD4 rise at 6 and 12 months; rate of opportunistic infections and mortality in first 2 years after ART. Cox regression models were used to compare relative hazard (RH) of death in Uganda vs London, after stepwise adjustment for key confounders.

Results:  African patients in Uganda initiated ART at a lower CD4 count (98 vs 180 cells, p <0.001) and more advanced disease (WHO 4 34.5% vs 22.6%, p = 0.001), but achieved comparable levels of virological suppression (84% vs 83% at 6 months) and immune reconstitution (median rise 108 vs 81 cells at 6 months, p = 0.051, adjusted for baseline CD4). Rate of opportunistic infections was 1.92 (95%CI 1.37 to 2.68) and mortality 13.23 (95%CI 5.29 to 33.07) times higher (95 and 17 deaths in Uganda and London) in first 2 years of ART in Uganda. After adjustment for lower baseline CD4 count, and higher viral load, RH of death was reduced by 57% to 5.69 (95%CI 1.49 to 21.73). Adjustment for higher rate of opportunistic infections, and other confounders had no effect on RH of death. Deaths from serious opportunistic infections, especially cryptococcal and mycobacterial disease accounted for this remaining excess mortality in Uganda.

Conclusions:  Approximately half of the 13-fold higher mortality among African patients in Uganda compared to London is explained by their more advanced disease at ART initiation. The remaining excess mortality is due to a higher death rate among patients with serious opportunistic infections in Uganda, reflecting poorer access to opportunistic infection therapy. Key strategies to address this high mortality include both earlier HIV diagnosis and ART initiation, and improving access to and management of life-threatening opportunistic infections, especially cryptococcal meningitis.