Multinational patterns of seasonal asymmetry in human movement influence infectious disease dynamics

Amy Wesolowski,Kenth Engø-Monsen,Christopher Lourenço,C J E Metcalf,Vivek Charu,Andrew J Tatem,Caroline O Buckee,Nathan Eagle,Elisabeth Zu Erbach-Schoenberg,Cecile Viboud,Taimur Qureshi

doi:10.1038/s41467-017-02064-4

Abstract

Seasonal variation in human mobility is globally ubiquitous and affects the spatial spread of infectious diseases, but the ability to measure seasonality in human movement has been limited by data availability. Here, we use mobile phone data to quantify seasonal travel and directional asymmetries in Kenya, Namibia, and Pakistan, across a spectrum from rural nomadic populations to highly urbanized communities. We then model how the geographic spread of several acute pathogens with varying life histories could depend on country-wide connectivity fluctuations through the year. In all three countries, major national holidays are associated with shifts in the scope of travel. Within this broader pattern, the relative importance of particular routes also fluctuates over the course of the year, with increased travel from rural to urban communities after national holidays, for example. These changes in travel impact how fast communities are likely to be reached by an introduced pathogen.

Highlights

Seasonal variation in human mobility is globally ubiquitous and affects the spatial spread of infectious diseases, but the ability to measure seasonality in human movement has been limited by data availability
Using previously described methods to extract mobility[1,5] from mobile phone call data records (CDRs, see Methods, Supplementary Fig. 1), we first characterized country-wide seasonal changes in the magnitude of travel using data from Kenya, Namibia, and Pakistan (Table 1), defined as the proportion of subscribers that traveled on subsequent days
We observe differences in the magnitude of travel between countries: the mean percentage of the population traveling between subsequent days for the entire country was lowest in Kenya (mean: 5%, 95% quantile interval (2–8%)), intermediate in Namibia (13% (11–18%)), and substantially higher in Pakistan (33% (31–36%)), see Supplementary Table 1, Supplementary Figs. 2–3

Summary

Introduction

Seasonal variation in human mobility is globally ubiquitous and affects the spatial spread of infectious diseases, but the ability to measure seasonality in human movement has been limited by data availability. Despite the connection between travel and infectious disease spread, one major feature of human mobility has been neglected in most applications to date: human mobility is generally seasonal[9,10,11,12] This phenomenon has been attributed to the intersection of climatic, economic, and social drivers and occurs in countries across the globe. Using the location and timing of mobile communications, subscriber travel patterns have been inferred and used to understand general mobility patterns[38,39] as well as the spread of infectious diseases[2,5,24,37,40,41] These data sets provide a powerful window onto spatio-temporal variation in travel[1].

Methods

Results

Conclusion