Paper # 137
Identification of Localized Clusters of High HIV Incidence in a Widely Disseminated Rural South African Epidemic: A Case for Targeted Intervention Strategies
Frank Tanser*1, T Bärnighausen1,2, and M-L Newell1,3
1Africa Ctr for Hlth and Population Studies, Univ of KwaZulu-Natal, Durban, South Africa; 2Harvard Sch of Publ Hlth, Boston, MA, US; and 3Inst of Child Hlth, Univ Coll London, UK
Background: South Africa contains more than 1 in 7
of the world’s HIV+ persons. Knowledge of local variation in the
rate of new HIV infections is important for prioritization of areas for
prevention. Geographical clustering of infections can decrease the efficacy of
existing population-based intervention measures but also implies that targeted
interventions could be highly effective. We evaluated the potential of
community-based, targeted interventions in a population with high levels of
infection. We followed 13,000 HIV– residents in a population with
high HIV prevalence in rural South Africa to test the null hypothesis that
incidence does not differ geographically across the surveillance area.
Methods: The study uses data from one of the most
comprehensive demographic and health surveillance sites in Africa—the Africa
Centre Demographic Information System. Nested within the demographic
information system is the population-based HIV surveillance, which takes place
annually for all consenting residents ≥15 years of age. We used a
2-dimensional Gaussian kernel of 3-km radius to produce robust estimates of HIV
incidence that vary across continuous geographical space. We also applied a
Kulldorff elliptical spatial scan statistic (survival model) to formally
identify clusters of infections at the micro-geographical level (p
<0.05).
Results: Between 2004 and 2009, we observed 943 HIV seroconversions
over 32,110 person-years of observation, at a crude HIV incidence rate of 2.94
(95 %CI, 2.75 to 3.13) per 100 person-years. Two high-risk, overlapping spatial
clusters (RR = 1.44 to 1.9) were identified by the Kulldorff statistic in
peri-urban communities near the National Road (p ≤0.011). Though
the clusters comprise just 5.7% of the study area, they account for nearly 1 of
every 3 seroconversions observed over the study period (figure).
Conclusions: Targeting efforts at settings where HIV
transmission is most intense is crucial. Our study provides clear empirical
evidence for the localized clustering of new HIV infections. The results show
that even in a severely affected rural African community, interventions that specifically
target, geographically defined, high-risk communities could be highly effective
in reducing the overall rate of new infections.
HIV incidence across the study area with high-
incidence clusters superimposed.
|
