Structured epidemic models and the spread of influenza in the central Canadian subarctic.

Patterns of transmission of infectious diseases within and among populations are strongly affected by population structure, which can either facilitate or limit interactions among people from different groups. Results from several theoretical studies show that nonrandom mixing among subgroups can affect the time when an infectious disease is introduced to the population, the speed of propagation of the disease, and the severity of an epidemic. Because many of these models focus on the effects of population structure, they are functionally similar to models used to describe the genetic structure of a population. One major difference between genetic models and epidemic models is that genetic models, with a time scale of the order of generations, incorporate migrations (or permanent movement) among subgroups, whereas epidemic models, with a time scale of the order of days or weeks, must incorporate short-term mobility among subgroups. Such mobility can be included in models for epidemic spread by explicitly incorporating the process by which residents from different locations interact with one another. We present a derivation of a mobility model for epidemic processes and apply it to the spread of the 1918-1919 influenza epidemic among the Cree and Métis people associated with three Hudson's Bay Company posts in the central Canadian Subarctic. The model distinguishes mobility from population effects. Results indicate that social organization (population effects) and social responses to the epidemic were more important than movement patterns (mobility) in explaining the differential impact of this virgin soil epidemic on the three study communities.