Barcelona 2013
Menu
AIDSImpact.com
Conference Details
Authors
International Committee
Plenary Speakers
Presenting Speakers
Programme
Sessions
Scientific Committee
Acknowledgements
| Abstract #344 - E-Posters English | ||||
|
Session: 50.87: E-Posters English (Poster) on Sunday in Chaired by Authors: Presenting Author: Ms Vivian Chitiyo - Biomedical Research and Training Institute, Zimbabwe |
||||
| Additional Authors: Dr. Jordi Casabona, Sra Cristina Sanclemente, Dra. Anna Esteve, Dra. Victoria Gonzalez, Grupo HIVITS TS, | ||||
| Aim: BACKGROUND: Summary birth history estimates of child mortality are used extensively in developing countries because of the lack of complete vital registration systems. The generalised HIV epidemic is thought to have introduced bias in summary birth history estimates of child mortality by violating the theoretical underpinnings of the method, but the extent of bias is unknown. OBJECTIVES: To estimate the extent of bias in the summary birth history infant and under-five mortality rates attributable to HIV/AIDS using the longitudinal survey data of the Manicaland HIV/STD Prevention Project in eastern Zimbabwe and to investigate the factors contributing to this bias. | ||||
| Method / Issue: Summary birth history estimates for 1995-2003 were compared with direct estimates corrected for HIV, using birth history and verbal autopsy data from a general population cohort. Components of the bias in the summary birth history estimates were investigated by comparing summary birth history estimates adjusted for: (1) increased correlation between mortality of mothers and children; (2) use of regression coefficients; and (3) use of non-HIV model life tables and were summed to provide an estimate of aggregate bias. | ||||
| Results / Comments: Overall bias due to HIV in the infant mortality rates was small (less than 5%) while bias in under-five mortality rates was estimated at between 1-8%. Aggregate bias was higher, 6-12% and 4-10% for infant and under-five mortality rates, respectively. Increased correlation between the mortality of mothers and their children, induced by HIV, was the principal contributor to aggregate bias. The contributions of use of regression coefficients derived from populations unaffected by HIV, and use of non-HIV model life tables were small. | ||||
| Discussion: The downward bias expected in summary birth history estimates of early child mortality due to HIV may be offset partially by other biases in the method. Further studies from other demographic surveillance sites are needed to assess the wider applicability of our findings and to develop appropriate adjustments to summary birth history estimates in populations affected by HIV/AIDS. | ||||
| Go Back | ||||