Analysis of traditional poultry trader networks to improve risk-based surveillance.

Live bird markets and contacts between them through poultry traders are known risk factors in the spread of diseases such as Newcastle disease. A traders' questionnaire survey was used to build networks of chicken movements among 29 shared markets during and outside festive seasons in the Oromia region of Ethiopia. A comparison was made between typologies built using centrality indexes (in-degree, out-degree, in-closeness, out-closeness and random-walk betweenness) and descriptive characteristics of the markets (number of chickens, number and type of sellers and the frequency with which they use different markets). The festive seasons did not appear to have an impact on the network structure, implying that it was not necessary to make structural changes to surveillance targets during these periods. Based on centrality indices, three markets (Meki, Debre Zeit and Adulala) emerged from the typology as being central to the network, which would not have been deduced from their descriptive characteristics alone. These three poultry markets ideally would be chosen in a risk-based type of surveillance system and in targeted control policies.