Understanding spatiotemporal clustering of seasonal influenza in the United States
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Seasonal influenza exhibits distinct spatiotemporal patterns across the United States, yet the geographic clustering of influenza activity remains incompletely understood. This study aims to identify jurisdictions with similar patterns of seasonal influenza epidemics by exploring spatiotemporal dynamics across the United States after the 2009 H1N1 pandemic.
Methods
We analyzed data from U.S. influenza surveillance systems, including outpatient illness surveillance and virologic surveillance. The outpatient illness data included weekly proportions of outpatient visits for influenza-like illness from jurisdictions including all 50 states, while virologic data comprised influenza test positivity results from U.S. public health and clinical laboratories covering all 50 states. We calculated Moran’s I statistics to assess spatial autocorrelation in peak timing. We also performed k-means clustering on z-normalized time series data and determined optimal clusters using the silhouette method. We then conducted an analysis of variance (ANOVA) to evaluate differences among clusters based on the Moran’s I statistics and the relative proportions of influenza virus types and subtypes.
Results
Our analysis revealed distinct spatial clusters with significant geographic patterns. We found a consistent grouping of Southeastern states (Georgia, Alabama, Mississippi, Louisiana, and Florida). This clustering pattern was partially explained by earlier seasonal peaks in these jurisdictions and supported by significant spatial autocorrelation in peak timing. While Southeastern states maintained stable cluster associations, Western and Central states showed greater variation in cluster membership across seasons. We also found significant differences between clusters in the Moran’s I statistics and the proportion of all influenza A virus detections that were influenza A/H1 viruses. However, no significant differences were found in the proportion of all influenza A and B virus detections that were influenza A viruses.
Conclusions
These findings quantify the distinct spatiotemporal patterns of seasonal influenza in the Southeastern United States compared to other regions. Understanding these regional clustering patterns can enhance preparations for upcoming changes in influenza activity and inform targeted public health interventions such as timing of vaccination campaigns. Robust surveillance systems, adaptive approaches, and stable long-term data are essential for effectively addressing regional differences and ultimately strengthening nationwide preparedness for seasonal influenza.
