124
Measuring the Force of the HIV Epidemic in a Rural Area of South Africa: The Africa Centre
Till Barnighausen*1,2, M L Newell1,3, F Tanser1, C Mbizana1, C Wallrauch1, G Cooke1,4, Z Gqwede1, and K Herbst1
1Africa Ctr for Hlth and Population Studies, Univ of KwaZulu Natal, Somkhele, South Africa; 2Harvard Sch of Publ Hlth, Boston, MA, US; 3Inst of Child Hlth, Univ Coll London, UK; and 4Imperial Coll, London, UK
Background: Very few demographic surveillance sites
collect longitudinal information on HIV status in residents and migrants linked
to demographic surveillance data and ART cohorts. Such information is essential
to analyze complex relationships between HIV epidemic dynamics, social and
economic changes.
Methods: The AC Demographic Information System
(ACDIS) provides longitudinal demographic, socioeconomic, behavioral, and
spatial (GIS) data on 85,000 individuals and their households and communities; linked
to data from annual population HIV surveillance. Incidence estimates are based
on data from resident and non-resident women (aged 15 to 49 years) and men
(aged 15 to 54 years) who tested HIV– in the first round and tested
either positive or negative subsequently. We adjusted for selection effects due
to non-participation with multiple imputations and investigated risk factors
for acquiring HIV in Weibull multiple regression.
Results: HIV prevalence in this community is high
(22.5%); in 2003 and 2004 prevalence peaked at 51% among 25- to 29-year-old women
and 44% among 30- to 34-year-old men; and was significantly higher in
non-residents than residents. Between the first and second round (2003 to 2005),
we observed 170 seroconversions in 5253 person-years at risk; giving crude HIV
incidence rates per 100 person-years of 3.8 (95%CI 3.2 to 4.6) in women and
2.3 (95%CI 1.8 to 3.1) in men. Multiple imputations significantly increased the
HIV incidence in women (7.9, 95%CI 7.4 to 8.4) and men (5.1, 95%CI 4.1 to 6.2).
Holding other factors constant, Weibull regression showed a 53% higher hazard
for HIV seroconversion in women than men (p <0.001) and twice the
hazard of seroconversion among individuals with a partner who were not married
compared to married couples (p <0.001). The hazard of HIV acquisition
increased with increasing distance (in kilometers) between the next government
clinic and the individual’s household (adjusted hazard ratio (aHR) 1.17, p
= 0.051) and decreasing distance between the closest primary road and the
individual’s household (aHR 0.856, p = 0.002).
Conclusions: We find very high HIV incidence in a
high HIV prevalence community. The Africa Centre informs understanding of the
dynamics of HIV infection epidemic at both individual and population level. Effective
HIV prevention interventions are urgently needed in young women and men, and in
geographically targeted communities such as those situated in close proximity
to major roads.
|
