Background: Empirical studies have the potential to collect data on patterns of sexual mixing and network structures.
Goal: To explore the contribution of different network measures in sexually transmitted disease epidemiology.
Study design: Individual-based stochastic simulations of a network of sexual partnerships and sexually transmitted disease transmission are analyzed using logistic regression.
Results and conclusions: Measures accumulated over times similar to the duration of infection are more informative than are static cross sections. The patterns of sexual mixing and network structure influence patterns of infection. In particular, the establishment of infection is most sensitive to the proportion of nonmonogamous pairs, the component distribution and cohesion among those with high activity. The subsequent prevalence is most sensitive to the assortativeness of mixing in the high-activity class and a measure of cohesion, both of which reflect the decrease in prevalence brought about by less widespread connections. A person's risk for infection is determined by local rather than global network structures.