Characterising the epidemic spread of Influenza A/H3N2 within a city through phylogenetics

Abstract

Infecting large portions of the global population, seasonal influenza is a major burden on societies around the globe. While the global source sink dynamics of the different seasonal influenza viruses have been studied intensively, it's local spread remains less clear. In order to improve our understanding of how influenza is transmitted on a city scale, we collected an extremely densely sampled set of influenza sequences alongside patient metadata. To do so, we sequenced influenza viruses isolated from patients of two different hospitals, as well as private practitioners in Basel, Switzerland during the 2016/2017 influenza season. The genetic sequences reveal that repeated introductions into the city drove the influenza season. We then reconstruct how the effective reproduction number changed over the course of the season. We find trends in transmission dynamics correlated positively with trends in temperature, but not relative humidity nor school holidays. Alongside the genetic sequence data that allows us to see how individual cases are connected, we gathered patient information, such as the age or household status. Zooming into the local transmission outbreaks suggests that the April 26, 2020 1/23 Author summaryAs shown with the current SARS-CoV-2 pandemic, respiratory diseases can quickly spread around the globe. While it can be hugely important to understand how diseases spread around the globe, local spread is most often the main driver of novel infections of respiratory diseases such as SARS-CoV-2 or influenza. We here use genetic sequence data alongside patient information to better understand what the drives the local spread of influenza by looking at the 2016/2017 influenza season in Basel, Switzerland as an example. The genetic sequence data allows us to reconstruct the how the transmission dynamics changed over the course of the season, which we correlate to changes, but not humidity or school holidays. Additionally, the genetic sequence data allows us to see how individual cases are connected. Using patient information, such as age and household status our analyses suggest that the elderly mainly transmit within their own transmission network. Additionally, they suggest that school aged children, but not pre-school aged children are important drivers of the local spread of influenza. 12Phylogenetics allows us to see how individual cases are epidemiologically connected. 13 This is done by reconstructing the evolutionary relationship between temporally spaced 14 samples of genetic sequence data, isolated from different infected individuals. The 15 resulting phylogenetic tree displays how samples are related to each other, and branch 16 lengths in calendar time display the elapsed time. The phylogenetic tree can therefore 17 be interpreted as an approximation of the transmission chain of the sampled cases. Such 18 a view on part of the influenza transmission chain allows to further quantify the 19 epidemiological dynamics which gave rise to the observed phylogenetic tree using 20 phylodynamic methods [...

Viral genomes reveal patterns of the SARS-CoV-2 outbreak in Washington State

“…In contrast to backwards in time coalescent approaches, we can consider different local outbreak clusters as independent observations of the same underlying population process using birth death models. We infer the effective reproduction number using the birth-death skyline model ( 12 ) by assuming the different local outbreak clusters are independent observations of the same process with the same parameters ( 13 ). We allow the effective reproduction number to change every 2 days.…”

“…The inference of the effective population sizes and rates of introductions is performed using an adaptive multivariate Gaussian operator (36) , implemented here https://github.com/BEAST2-Dev/BEASTLabs and the analyses are run using adaptive Metropolis coupled MCMC (37) In contrast to backwards in time coalescent approaches, we can consider different local outbreak clusters as independent observations of the same underlying population process using birth death models. We infer the effective reproduction number using the birth-death skyline model (12) by assuming the different local outbreak clusters are independent observations of the same process with the same parameters (13) . We allow the effective reproduction number to change every 2 days.…”

“…We then use these local transmission clusters to analyse the spread of SARS-CoV-2 in the state using two phylodynamic approaches. First, we estimate the effective reproduction number ( R e ) using a birth-death approach ( 12 ), where we treat each individual local transmission cluster as independent observation of the same underlying population process ( 13 ). Next, we estimate effective population sizes over time and the degree of introductions using a coalescent skyline approach ( 14 ).…”

Viral genomes reveal patterns of the SARS-CoV-2 outbreak in Washington State

“…In contrast to backward-in-time coalescent approaches, we can consider different local outbreak clusters as independent observations of the same underlying population process using birth-death models. We inferred the effective reproduction number using the birth-death skyline model ( 12) by assuming that the different local outbreak clusters are independent observations of the same process with the same parameters (13). We allowed the effective reproduction number to change every 3.5 days.…”

“…We then used these local transmission clusters to analyze the spread of SARS-CoV-2 in the state using two phylodynamic approaches. First, we estimated the effective reproduction number (R e ) using a birth-death approach (12), where we treated each individual local transmission cluster as independent observation of the same underlying population process (13). Next, we estimated effective population sizes over time and the degree of introductions using a coalescent skyline approach (14).…”