Ambient ozone reduces influenza activity in Hong Kong with effect moderated by epidemic phase

Ambient ozone (O₃) has been reported to be associated with influenza activity, but evidence remains inconsistent, particularly in subtropical settings. Here, we investigated this question in subtropical Hong Kong using weekly and daily data from 2005 to 2016 through a multi-method triangulation methodology integrating convergent cross mapping (CCM) from dynamical systems theory, Peter-Clark momentary conditional independence (PCMCI) algorithm from graphical causal modeling, and regression-based generalized linear model (GLM). We found a consistent inhibitory relationship of O₃ with influenza activity at a one-week lag. This effect strengthened slightly with increasing O₃ concentration across the low-to-moderate range but depended more markedly on epidemic activity itself. The effect magnitude varied with epidemic phase: O₃-associated influenza reduction diminished during immunity-saturated epidemic peaks but increased as new waves emerge when baseline immunity was presumably low. This state-dependent pattern points to a putative immune-priming pathway through which ambient O₃ may reduce influenza activity in real-world complex systems. Our population-level evidence calls for further experimental validation to inform seasonal epidemic and pandemic preparedness.