COVID-19 pandemic interventions reshaped the global dispersal of seasonal influenza viruses

Abstract

The global dynamics of seasonal influenza viruses inform the design of surveillance, intervention, and vaccination strategies. The COVID-19 pandemic provided a singular opportunity to evaluate how influenza circulation worldwide was perturbed by human behavioral changes. We combine molecular, epidemiological, and international travel data and find that the pandemic's onset led to a shift in the intensity and structure of international influenza lineage movement. During the pandemic, South Asia played an important role as a phylogenetic trunk location of influenza A viruses, whereas West Asia maintained the circulation of influenza B/Victoria. We explore drivers of influenza lineage dynamics across the pandemic period and reasons for the possible extinction of the B/Yamagata lineage. After a period of 3 years, the intensity of among-region influenza lineage movements returned to pre-pandemic levels, with the exception of B/Yamagata, after the recovery of global air traffic, highlighting the robustness of global lineage dispersal patterns to substantial perturbation.

Fig. 1.. Virological and genomic surveillance intensity, and positivity rates of seasonal influenza viruses, from January 2017 to March 2024.

(A) Intensity of virological surveillance of influenza indicated by rolling numbers of specimens processed globally for influenza virus testing. (B) Intensity of genomic surveillance of influenza, as indicated by rolling percentages of reported influenza cases sequenced at high quality. (C-I) The weekly count of high-quality hemagglutinin (HA) genetic sequences (stratified by continent) and global positivity rates among tested specimens, for H1N1pdm09 (C), H3N2 (E), B/Victoria (G) and B/Yamagata (I). The positivity rates of H1N1pdm09 (D), H3N2 (F), B/Victoria (H) are also presented separately for regions that experienced influenza waves during the acute phase of the pandemic period (Africa, Southeast Asia, and South Asia; see text). The color scheme is illustrated in panel (J). Positivity rates are presented as central-aligned rolling averages (5-week window) and the 95% intervals indicate uncertainty in inferring the specific subtypes or lineages using a Bayesian framework. The light orange and light blue shaded areas represent the acute and transition phases of the COVID-19 pandemic period, defined as April 2020 to March 2021, and April 2021 to April 2023, respectively.

Fig. 2.. Predictors of global movements of seasonal influenza virus using a 4-epoch phylogeographic GLM model.

(A-C) Average monthly air passenger traffic network between 12 geographic regions across the three periods. Here, only those routes with >100,000 average monthly air traffic passengers are presented, for clarity. (D) Relative air traffic from and to each region over time, by dividing the numbers by the maximum value of each region. Air traffic between southern China and northern China was not included because we only considered between country mobility. Colors correspond to those used in the maps in panels A-C. (E) MDS visualization of the similarity of among-region origin/destination (O/D) absolute air passenger volumes for different time windows. Here, each time window refers to the range from 1 April of each year to 31 March of the following year, except for 2022/2023 (April 2022 to April 2023) and 2023/2024 (May 2023 to March 2024) where the time window is aligned to the WHO’s declaration of the end of the pandemic. The arc is used to show the sequence of the air passenger network. (F) MDS visualization of the similarity of among-region origin/destination (O/D) air passenger travel frequencies for different time windows. Frequency refers to the fraction of air volume of a specific O/D journey during each time window. (G) Posterior summaries of the product (reported as log effect size) of the log constant-through-time predictor coefficient and the predictor inclusion probability (pooled across the time periods), for H1N1pdm09, H3N2, and B/Victoria lineages. B/Yamagata analyses were performed under a time-homogeneous (1-epoch) GLM due to the lack of post-March 2020 gene sequences. Points and ranges represent the posterior mean and 95% highest posterior density (HPD) intervals, respectively. Location-specific predictors were included as both origin (O) and destination (D) predictors of the pairwise transition rates.

Fig. 3.. Global migration dynamics of seasonal influenza virus lineages through time.

(A) Rolling weekly Markov jump counts (location transition events) over time for the four influenza lineages. (B-E) Estimates of the number of location transition events per year between each pair of geographic regions during the pre-pandemic, acute pandemic, transition pandemic and post-pandemic periods. Analyses are based on the posterior summaries of the Markov jumps under a time-inhomogeneous (4-epoch) GLM with only air traffic data as the predictor of overall and relative transition rates, except for B/Yamagata, which was analyzed with a time-homogeneous GLM-diffusion phylogeographic model.

Fig. 4.. Dynamics of measures of genetic diversity of seasonal influenza virus lineages.

(A) The maximum clade credibility (MCC) tree for the B/Yamagata lineage. Tip colors represent the sampling location of each sequence. The inset shows the MCC tree with tips annotated to show the main B/Yamagata clades. (B) Relative genetic diversity of influenza viruses, as inferred by Bayesian skygrid population reconstruction. (C) Mean pairwise diversity of influenza virus, measured as average branch length distance (patristic distance) in units of years between pairs of tips in phylogeny at monthly intervals.