The influence of host movement on epidemic dynamics: Commuting patterns in cities and their consequences for the spread of influenza

Background/Question/Methods

Populations exposed to the same infectious pathogen can have different epidemic dynamics.  This observation is difficult to explain using standard ecological theory, which posits that a pathogen’s basic reproductive ratio completely determines its rate of spread.   A possible missing ingredient in this theory is host movement.  While most epidemiological models assume that probability of infectious contact between two individuals does not depend on their spatial locations within the population, in reality individual trajectories are often highly organized, causing differential contact rates among individuals in different places.  This heterogeneity may be responsible for differences in epidemic dynamics among otherwise similar populations.

To test this idea we analyzed census data on commuting patterns in 48 Canadian cities, encompassing the home and work locations of 6.7 million individuals.   These data were well predicted by a modified “gravity model” of movement, where individuals tended to commute relatively short distances to workplaces clustered in a few urban centers within each city.  We then contrasted two competing models of the 2009 H1N1 influenza pandemic with incidence data from several of the same cities.  One model was based on the movement patterns in the commuting data and the other was a null-model with the same contact rates but no spatial structure. 

Results/Conclusions

The magnitude and timing of outbreaks in the commuting-based epidemic model were more variable than in the null model, and showed dependence on the location of the initial infection and the configuration of the urban centers within the cities.  The commuting-based epidemic model also better predicted the observed correlations in disease incidence between different areas of the cities.  These results suggest that host movement plays an important role in epidemic dynamics, causing divergent responses to the same pathogen over space and time.  With the majority of the world’s population now living in urban areas, most future epidemic casualties will be located in cities.   Understanding the epidemic consequences of their spatial structure may be important for public health planning.