population-level AIDS mortality following the introduction of antiretroviral therapy in Addis Ababa, Ethiopia.

population-level AIDS mortality following the introduction of antiretroviral therapy in Addis Ababa, Ethiopia.

Assessments of population-level effects of antiretroviral therapy programmes in Africa are rare. Reniers and colleagues use data from burial sites to estimate trends in adult AIDS mortality and the mitigating effects of antiretroviral therapy in Addis Ababa. Antiretroviral therapy has been available since 2003, and for free since 2005. To substitute for deficient vital registration, the authors use surveillance of burials at all cemeteries. They present trends in all-cause mortality, and estimate AIDS mortality (ages 20-64 years) from lay reports of causes of death. These lay reports are first used as a diagnostic test for the true cause of death. As reference standard, the authors use the cause of death established via verbal autopsy interviews conducted in 2004. The positive predictive value and sensitivity are subsequently used as anchors to estimate the number of AIDS deaths for the period 2001-2007. Estimates are compared with Spectrum projections. Between 2001 and 2005, the number of AIDS deaths declined by 21.9% and 9.3% for men and women, respectively. Between 2005 and 2007, the number of AIDS deaths declined by 38.2 for men and 42.9% for women. Compared with the expected number in the absence of antiretroviral therapy, the reduction in AIDS deaths in 2007 is estimated to be between 56.8% and 63.3%, depending on the coverage of the burial surveillance. Five years into the antiretroviral therapy programme, adult AIDS mortality has been reduced by more than half. Following the free provision of antiretroviral therapy in 2005, the decline accelerated and became more sex balanced. Substantial AIDS mortality, however, persists.

vital registration systems do not function well enough to provide accurate cause-specific adult mortality estimates, burial surveillance can provide data for realistic estimates of the impact of antiretroviral treatment programmes. During 2003-2005, the Ethiopian national programme required a co-payment of 28-80USD per month and AIDS mortality fell 15.8% in Addis Ababa, more in men than in women. The decline was far sharper between 2005 and 2007 (40.6%) when antiretroviral treatment was free, with greater declines observed for women. This study demonstrates the utility of using other information sources to monitor programme effects when vital registration is deficient and suggests a restraining effect of co-payments on antiretroviral treatment uptake, particularly for women who may have less access to resources for financing treatment.