Host-Age Structure in Influenza A(H3N2) HA Lineage Transitions in South Korea, 2009-2025

Background: Influenza A(H3N2) has resurged in South Korea despite a shrinking pediatric population and high vaccination coverage. We examined correlates of host-age state transitions on the H3N2 hemagglutinin (HA) genealogy.

Methods: We analyzed 703 South Korean H3N2 HA sequences (2009-2025) in eight host-age states (ages 0-4, 5-9, then 10-year bands to 60 years and older). Discrete-trait analysis and an epoch-structured hierarchical generalized linear model (GLM) were fitted over 1700 HA trees, with epochs split at mid-2020. Within-year age-label permutation provided the sampling control (one for the discrete-trait analysis, 20 for the GLM).

Results: Children aged 5-9 years showed the largest net source balance (+11.7 vs. +10.5 under permuted labels), reflecting a lower recipient-side rate (0.39 vs. 1.08) rather than a higher source-side rate (12.11 vs. 11.59). Of 56 routes, 18 were supported (Bayes factor ≥ 10) but only three exceeded the permutation control, two from this group. Observed labels fitted the genealogy 400 log-likelihood units better, with 24% fewer transitions. Source-group population size (posterior inclusion probability, 0.74) and vaccination coverage (0.66) were supported, both with negative coefficients; contact rate was not (0.016). Only population size exceeded the permutation null (0.74 vs. 0.65; p = 0.048).

Conclusions: H3N2 HA lineage transitions in South Korea were age-structured after accounting for sampling, most clearly for children aged 5-9 years. Their net source balance reflects a low recipient-side rate rather than a high source-side rate, and therefore describes lineage position rather than transmission intensity. These findings support age-resolved genomic surveillance for interpreting such patterns.