Purpose: The aim of this research was to develop a methodology to efficiently identify the source of an outbreak while contamination-caused illnesses are still occurring, thereby resolving investigations earlier and averting potential illnesses.
Methods: A network-theoretical approach was developed for rapid identification of the source of foodborne contamination events. Given the spread of outbreak-related cases and limited data on the distribution network, the inference approach uses backward induction and network analysis to determine the probability that any location is the outbreak source; the greater the number and dispersion of cases, the fewer locations are suspect due to the topological properties of the network. A probabilistic simulation approach involving distribution network models of selected food products was developed to model the accuracy and robustness of the approach across multiple outbreak scenarios.
Results: In extensive simulation testing across a variety of realistic distribution network structures, the outbreak source is robustly ranked within the top 5% (1%) of feasible locations after 5% (25%) of the cases have been reported, reducing by up to 45% (25%) the eventual total number of illnesses in the simulated outbreaks. The method greatly outperforms all heuristics, which can be viewed as representative of investigation strategies applied in practice.
Significance: Our results suggest this methodology that can form the basis of a “tool” to supplement existing traceback processes, enabling regulators to identify high probability sources of an ongoing outbreak, and to make strategic recommendations regarding allocation of investigative resources and search effort.