Using Big Data to Monitor the Introduction and Spread of Chikungunya, Europe, 2017

Joacim Rocklöv; Yesim Tozan; Aditya Ramadona; Maquines O. Sewe; Bertrand Sudre; Jon Garrido; Chiara Bellegarde de Saint Lary; Wolfgang Lohr; Jan C. Semenza


Emerging Infectious Diseases. 2019;25(6):1041-1049. 

In This Article

Short-range Dispersion: Geocoded Tweets

The spatiotemporal analysis of geocoded Twitter data showed strong human mobility from Lazio (Figure 4) and the Var department in France (Appendix 2 Figure 3) toward several larger cities where Ae. albopictus mosquitoes are present. The top 10 estimates of mobility out of the 2 outbreak zones of Var and Lazio showed the strongest pattern for potential dispersion of chikungunya virus not only into the areas geographically close to the outbreak zones but also to several relatively large cities in Italy, France, and Spain (Table). The monthly mobility patterns during the study period varied between months; for example, the vacation month of August showed a stronger mobility pattern out of Var to areas not in direct connectivity, most notably to Rome (Appendix 2 Figure 4). When we contrasted the mobility proximities between the 2 outbreak zones, we observed the highest proximities within countries (Figure 4; Appendix 2 Figure 3). Although the Var and Lazio outbreak zones experienced high mobility proximity to Barcelona, Lazio was also highly connected to southern Italy (e.g., Catania and Palermo), in close proximity to the chikungunya outbreak in the Calabria area, which was also observed in the International Air Transport Association (IATA) flight passenger data (Figures 3, 4). In Italy, cases were first notified in Anzio at the end of June, followed by notifications in Rome later in July, and in Calabria in early August in order of temporal appearance (Figure 2). In our mobility analysis, we identified the mobility links to all outbreak regions (Figure 4), with the exception of the Emilia-Romagna region, although the region neighboring Emilia-Romagna was positive in our analysis. The mobility patterns correlated more strongly to the outbreak regions in July and August.

A closer look at the Lazio outbreak zone in Italy revealed strong connectivity between Anzio (where the first cases in Italy were confirmed) and Rome (where a higher number of cases were notified) (Figure 5). We compiled the top 10 mobility proximity areas from the outbreak zones of Anzio and Rome in August and September (Table). Although the highest mobility proximity from Anzio was to Rome in August and September, the mobility proximity from Rome to Anzio was also found among the top 10 destinations. Overall, Rome had higher connectivity to many more areas than Anzio.

Figure 5.

Estimated areas of risk for chikungunya spread from the outbreak areas of Anzio and Rome in the Lazio region, Italy, based on combined VC and estimates, August–October 2017. Heavy outlines indicate the outbreak areas. MP, mobility proximity; VC, vectorial capacity.