Replication Researchers: Jiangtao Luo
Original Paper Title: HIV Development Assistance and Adult Mortality in Africa
Original Researchers: Eran Bendavid, Charles Holmes, Jay Bhattacharya, Grant Miller
Original publication: JAMA
Replication Plan: Luo’s Replication Plan
Current Status: Completed Replication Study
The Original Study
The study assesses the effect of the US President's Emergency Plan for AIDS Relief (PEPFAR) on adult mortality from 2003 to 2008. The programme aimed to increase prevention and treatment of HIV in 15 countries. To evaluate whether the programme had effects on health outcomes beyond HIV, the researchers conducted a difference-in-difference analysis across countries, with data from 27 countries between 1998-2008 and a subnational analysis in two countries. Data was obtained for 1,538,612 adults from 41 surveys. The analysis across countries found that in 2003 the age-adjusted mortality was 8.3 in the nine focus countries for PEPFAR -Ethiopia, Kenya, Mozambique, Namibia, Nigeria, Rwanda, Tanzania, Uganda, and Zambia- and 8.5 in the 18 non-focus countries. In 2008, the mortality was 4.1 in the focus countries and 6.9 in the non-focus countries. The subnational analysis in Tanzania and Rwanda were consistent with these results. The study concludes that adult mortality rates decreased more in PEPFAR-focus countries.
The first objective of this replication study is to conduct a pure re-analysis of the original study using the methods and data from the original paper. The replication study will verify whether the original study results and conclusions can be generated using its own data and methodology. The second objective is to compare the methods used by the original paper and that used by Duber et al. in their paper, ‘Is there an association between PEPFAR funding and improvement in national health indicators in Africa? A retrospective study’ in the Journal of the International AIDS Society. Duber et al. report no significant difference in 13 of 14 health indicators between PEPFAR-focus and non-focus countries. Subsequently, the replication study will examine the limitation of Duber et al.’s paper. Finally, the replication study will test the generalisability of the methods used in the original paper by using updated data posted on the WHO Statistical Information System (WHOSIS). These data allow us to test the updated dataset using the same analysis plan.