Effect of HIV/AIDS-related mortality on household dependency ratios in rural South Africa, 2000–2005

Abstract

With data from a surveillance system that uses verbal autopsies to identify cause of death in rural South Africa, we investigated whether mortality from HIV/AIDS differs from other causes of death in its effect on household dependency ratio, and to what extent the effect is mediated by the baseline dependency ratio. Findings: (i) the impact of death from HIV/AIDS on the dependency ratio in 2005 is marginally positive compared with other causes of death, but (ii) the impact is overpowered by the effect of death at working age, and (iii) the baseline dependency ratio mediates the effects on the 2005 ratio of cause of death and of the individual's sex and age at death. Migration into and out of the household—anticipating or responding to a death—seems to be a key source of change in the household dependency ratio.

Keywords:

• mortality
• HIV/AIDS
• dependency ratio
• South Africa
• household

Notes

1. Sangeetha Madhavan is at the University of Maryland, 2169 LeFrak Hall, College Park, MD 20742, USA and at the University of the Witwatersrand and the University of Colorado at Boulder. E-mail: [email protected]. Enid Schatz is at the University of Missouri, the University of the Witwatersrand, and the University of Colorado at Boulder. Benjamin Clark is at the University Centre for Population Studies, London School of Hygiene & Tropical Medicine.

2. The authors are indebted to the South African Medical Research Council/University of the Witwatersrand Rural Public Health and Health Transitions Unit (Agincourt) for support in conducting this research. In particular, we thank Mark Collinson, Paul Mee, and Samuel Clark for assistance in constructing the data-set. We are also grateful for the guidance provided by Jane Menken and comments on earlier drafts by Michel Guillot, Ulla Larsen, Jani Little, and Georges Reniers. The Agincourt data were collected with generous support from the Andrew Mellon Foundation and the Wellcome Trust. The secondary analysis of the data was funded by a seed grant from the Population Aging Center at the Population Program, Institute for Behavioral Science, University of Colorado Boulder, and NICHD grant #1R03 HD047760-01A.