Modelling H5N1 in Bangladesh across spatial scales: Model complexity and zoonotic transmission risk

• DOI: 10.1016/j.epidem.2017.02.007
• Publication Date: 2017-02-22T00:28:48Z
• Authors: Edward M. Hill; Thomas House; Madhur S. Dhingra; Wantanee Kalpravidh; Subhash Morzaria; Muzaffar G. Osmani; Mat Yamage; Xiangming Xiao; Marius Gilbert; Michael J. Tildesley
• Keywords: Biosecurity; Pandemic; Spatial epidemiology

Highly pathogenic avian influenza H5N1 remains a persistent public health threat, capable of causing infection in humans with high mortality rate while simultaneously negatively impacting the livestock industry. A central question is to determine regions that are likely sources newly emerging strains pandemic potential. suitable candidate Bangladesh, being one most densely populated countries world and having an intensifying farming system. It therefore vital establish key factors, specific enable both continued transmission within poultry spillover across human-animal interface. We apply modelling framework epidemics Dhaka region occurring from 2007 onwards, resulted large outbreaks sector limited number confirmed human cases. This model consisted separate zoonotic components. Utilising farm spatial population information set competing nested models varying complexity were fitted observed case data, parameter inference carried out using Bayesian methodology goodness-of-fit verified by stochastic simulations. For component, successfully identifying minimal complexity, which enabled accurate prediction size distribution cases outbreaks, was found be dependent on administration level analysed. consistent outcome non-optimal reporting infected premises materialised each epidemic interest, though analysed there substantial differences estimated parameters. The component main contributor Bangladesh differ another. conclude discussing possible explanations for these discrepancies behaviour between epidemics, such as changes surveillance sensitivity biosecurity practices.