Network structure and disease risk for an endemic infectious disease

TL;DR

This study analyzes cattle transportation networks in Turkey to understand how network structure influences the risk and spread of endemic foot-and-mouth disease, highlighting key network properties and potential surveillance targets.

Contribution

It provides a detailed characterization of the cattle shipment network and links specific network centrality measures to infection risk for endemic disease.

Findings

High centrality farms are more likely to be infected or at risk.

Network exhibits small-world and scale-free properties with spatial constraints.

Surveillance should focus on farms with high centrality measures.

Abstract

The structure of contact networks affects the likelihood of disease spread at the population scale and the risk of infection at any given node. Though this has been well characterized for both theoretical and empirical networks for the spread of epidemics on completely susceptible networks, the long-term impact of network structure on risk of infection with an endemic pathogen has been less well characterized. Here, we analyze detailed records of the transportation of cattle between farms in Turkey to characterize the global and local attributes of the shipments network between 2007-2012, building an aggregated static directed - weighted network. We study the correlation between network properties and the likelihood of infection with, or exposure to, foot-and-mouth (FMD) disease over the same time period using recorded outbreaks. The shipments network shows properties of small-worldness and scale-freeness (intermediate degrees), with an exponential cut-off for high degrees. The shipments network illustrates strong modularity and lack of assortativity. The shipments network shows signs of spatial constraints, with few long-distance connections, and a strong similarity to other spatially constrained networks. We find that farms that were either infected or at high risk of infection with FMD (within one link from an infected farm) had higher values of centrality; farms that were never less than 2 links from an infected farm had disproportionately low centrality. However, the correlation of the rankings of farms shows that central farms (high eigenvector centrality) are not necessarily those with more connections to/from it (in/out degree). Several central farms serve as bridges of densely connected farms (high betweenness centrality). These results suggest that to detect FMD spread, surveillance efforts could be focused preferentially on farms with centralities greater than the mean.