Assessing the Population Attributable Fraction of the Acute Stage of HIV in Zimbabwe through Network Modeling
Steven M. Goodreau, University of Washington
Susan L. Cassels, University of Washington
Martina Morris, University of Washington
April A. Greek, Battelle Centers for Public Health Research and Evaluation
Danuta Kasprzyk, Battelle Centers for Public Health Research and Evaluation
Daniel Montano, Battelle Centers for Public Health Research and Evaluation
The proportion of new HIV infections stemming from source partners in the acute stage of infection is disproportionately high, but the importance of relational timing is not yet clear. We use dynamic exponential random graph modeling (implemented in statnet) along with a stochastic simulation model of HIV transmission dynamics to estimate the proportion of infections from three stages of HIV disease (acute, chronic, AIDS). Given behavioral data from Zimbabwe, the estimates at equilibrium are 22%, 49% and 29% respectively. The role of acute infections was more strongly affected by concurrency than by the range of published stage-specific transmission probabilities. Although acute transmission appears to account for a minority of new infections, reducing its impact -- by reducing either acute viral load or the probability someone has additional partners during early infection – could bring the epidemic below the reproductive threshold in populations marked by low rates of partner change but relatively high rates of long-term partner concurrency.