Objective: To use observed data to develop a mathematical model that estimates the impact of migration on the spread of HIV in South Africa.
Methods: A deterministic mathematical model was designed to evaluate the dynamic interactions between mobility, sexual behaviour, HIV, and sexually transmitted infections. The model was based on a population study of 488 adults, which included male migrants, male non-migrants and their rural partners in KwaZulu/Natal, South Africa.
Results: The model predicted that the impact of migration depends upon the epidemic's stage and the pattern of migration. Early in the epidemic, frequent migration between populations with different HIV prevalence rates accelerated HIV spread; however, local sexual risk behaviour determined the eventual scale of the epidemic. If migration is coupled with increased sexual risk behaviour by migrant men, as has been reported in the South African communities studied, HIV prevalence would increase 10 times among migrants' female partners (1.8 to 19%). In contrast, if migration were to occur infrequently, with migration-associated risk behaviour assumed to be at current levels, the predicted epidemic would be one fifth that currently observed (2.8 versus 15.1%).
Conclusions: Migration primarily influences HIV spread by increasing high-risk sexual behaviour, rather than by connecting areas of low and high risk. Frequent return of migrants is an important risk factor when coupled with increased sexual risk behaviour. Accordingly, intervention programmes in South Africa need to target the sexual behaviour of short-term migrants specifically, even though these individuals may be more difficult to identify.
From the aMassachusetts General Hospital, Department of Medicine, Harvard Medical School, Boston, Massachusetts, USA
bBrown University Medical School, Department of Community Health, Providence, Rhode Island, USA
cDepartment of Infectious Disease Epidemiology, Imperial College London, St Mary's Campus, London, UK.
Received 1 March, 2006
Revised 11 September, 2006
Accepted 2 October, 2006
Correspondence to Dr M.N. Lurie, Room 221, 121 South Main St, Brown University Medical School, Box GS-121, Providence, RI 02912, USA. E-mail: Mark_Lurie@Brown.edu