Shifting patterns of seasonal influenza epidemics

Abstract

Seasonal waves of influenza display a complex spatiotemporal pattern resulting from the interplay of biological, sociodemographic, and environmental factors. At country level many studies characterized the robust properties of annual epidemics, depicting a typical season. Here we analyzed season-by-season variability, introducing a clustering approach to assess the deviations from typical spreading patterns. The classification is performed on the similarity of temporal configurations of onset and peak times of regional epidemics, based on influenza-like-illness time-series in France from 1984 to 2014. We observed a larger variability in the onset compared to the peak. Two relevant classes of clusters emerge: groups of seasons sharing similar recurrent spreading patterns (clustered seasons) and single seasons displaying unique patterns (monoids). Recurrent patterns exhibit a more pronounced spatial signature than unique patterns. We assessed how seasons shift between these classes from onset to peak depending on epidemiological, environmental, and socio-demographic variables. We found that the spatial dynamics of influenza and its association with commuting, previously observed as a general property of French influenza epidemics, apply only to seasons exhibiting recurrent patterns. The proposed methodology is successful in providing new insights on influenza spread and can be applied to incidence time-series of different countries and different diseases.

Characterizing an epidemiological geography of the United States: influenza as a case study

“…Indeed, the wave-like propagation and relative spatial cohesiveness of the epidemiological regionalization produced by our method suggests that counties proximal to each other often experience similar influenza epidemics, suggesting that geographic proximity is an important driver for influenza dynamics. The consistent importance of the distance from source location and latitude in explaining the regionalization clusters further corroborates previous studies that have characterized the hierarchical dynamics of seasonal influenza [14,15,2] in which transmission decays as geographic distance between two locations increases [2] and transmission is mainly driven by local commuting and work flow patterns rather than long-distance or airline mobility [2,5,15,4]. Surprisingly, our regionalization results did not provide any evidence for our mobility-based hypothesis for more dispersed spatial dynamics of ILI among adults.…”

“…The spatio-temporal dynamics of infectious diseases are complex to interpret, but are key to the success of public health responses. Seasonal influenza, for example, exhibits variability in onset, peak time, duration, and geographic distribution between seasons, and is thought to be affected by a number of environmental, socio-demographic, and biological factors [1, 2, 3, 4, 5]. In the United States, surveillance, vaccination policies, and resource allocation for influenza are managed at the level of ad-hoc health administration mega-regions (each consisting of multiple U.S. states) [6].…”

Differences in epidemic spread patterns of norovirus and influenza seasons of Germany: an application of optical flow analysis in epidemiology

“…Unlike other approaches with modeling spatial transmission 42 , our method does not require graph inputs or information about underlying networks offering a more hands-on approach for governmental agencies and other actors in the public health sector. Other research groups have looked at seasonal influenza patterns 43 using graph based calculations. In comparison the German influenza wave data shows a slightly different structure eluding the definition of a hard onset.…”

Effects of mobility and multi-seeding on the propagation of the COVID-19 in Spain

“…Our working hypothesis is that large mobility flows lead to multiple seeding events resulting in several local outbreaks and, finally, to a higher incidence. Although the role of human mobility in shaping epidemic dynamics has been extensively considered [5][6][7][8][9][10][11][12][13][14], even recently for COVID-19 [15][16][17][18][19][20][21], the effect of multiple seeding due to mobility has received sensibly less attention; with only few theoretical [22,23] and applied exceptions [24][25][26][27]. To test this assumption, we focus on the change in mobility among the province of Madridwhere the first sustained outbreak was recorded-and the other provinces in Spain the week before the onset of the local outbreaks.…”