Effects of Heterogeneous and Clustered Contact Patterns on Infectious Disease Dynamics

• DOI: 10.1371/journal.pcbi.1002042
• Publication Date: 2011-06-03T11:41:02Z
• Authors: Erik M. Volz; Joel C. Miller; Alison Galvani; Lauren Ancel Meyers
• Keywords: Degree distribution; Clustering coefficient; Epidemic model; Preferential attachment; Percolation (cognitive psychology)

The spread of infectious diseases fundamentally depends on the pattern contacts between individuals. Although studies contact networks have shown that heterogeneity in number and duration can far-reaching epidemiological consequences, models often assume are chosen at random thereby ignore sociological, temporal and/or spatial clustering contacts. Here we investigate simultaneous effects heterogeneous clustered patterns epidemic dynamics. To model population structure, generalize configuration which has a tunable degree distribution (number per node) level three cliques). dynamics for this class graph, derive tractable, low-dimensional system ordinary differential equations accounts network structure course epidemic. We find interaction is complex. Clustering always slows an epidemic, but simultaneously increasing variance increase final size. also show bond percolation-based approximations be highly biased if one incorrectly assumes periods homogeneous, magnitude bias increases with amount network. apply approach to high within households, using parameters estimated from survey data social interactions, identify conditions under do not account household will biased.