Marston M, Todd J, Glynn JR, Nelson K, Rangsin R, Lutalo T, Urassa M, Biraro S, Van der Paal L, Sonnenberg P and Zaba B.
AIDS 2007. vol 21 sup 6: S65-S71
Objectives: To estimate mortality directly attributable to HIV infection in HIV-infected adults in low- and middle income countries and to discuss appropriate methodology.
Design: An illustrative analysis of pooled data from 4 different studies across sub-Saharan Africa and 2 studies from Thailand with data on individuals with known dates of seroconversion to HIV.
Methods: Five of the studies also had data from uninfected HIV negative subjects and one study had verbal autopsies. Data for HIV negative cohorts were weighted to have the initial age and sex distribution of the seroconverter cohorts. Using the life table technique of competing risks, and taking the survival of the weighted HIV negative group as representative of background mortality, the net survival from HIV was calculated for the seroconverters. Mortality from all causes and “net” mortality, the mortality due to HIV, were modelled using piecewise exponential regression. Alternative approaches are explored in the dataset with no information on mortality of uninfected persons.
Results: The overall effect of the net mortality adjustment was to increase survivorship by between 2% and 5% after 6 years post infection. The increase was 2% in the 15-24 year age group, increasing to 22% in the 55 year and over group for the East African sites. The mortality rate ratios for net mortality between sites were similar to the corresponding ratios for all cause mortality. Adjustment methods based on cause of death data and use of national life tables gave similar results.
Conclusions: Differences between HIV mortality in different populations and age groups are not explained by differences in background mortality, although background mortality does appear to contribute to the excess in the oldest age groups. In the absence of data from uninfected persons in the same population, model life tables can be used to calculate background rates.