Abstract
The purpose of this study was to investigate the hypothesis that human immunodeficiency virus (HIV) transmission may be facilitated or obstructed by network structure, incorporating a measure of risk that combines true risk and surrogates. Persons at presumed high risk for HIV were enrolled in long-term follow-up studies of urban and rural networks in Atlanta, Georgia, and Flagstaff, Arizona. We focused on respondents who were also contacts to evaluate information on both sides of the observed dyads and constructed a Risk Indicator, based on a four-digit binary number, that permitted assessment and visualization of the overall risk environment. We constructed graphs that provided visualization of the level of risk, the types of relationships, and the actual network. Although some of the findings conform to the hypotheses relating network structure to transmission, there were several anomalies. In Atlanta, HIV prevalence was most strongly related to men with a male sexual orientation, despite the widespread use of injectable drugs. In Flagstaff, an area of very low prevalence and no transmission, the risk environment appeared more intense, and the frequency of microstructures was as great or greater than representative areas in Atlanta. The network hypothesis is not yet sufficiently developed to account for empirical observations that demonstrate the presence of intense, interactive networks in the absence of transmission of HIV.