Differential meteorological drivers of influenza subtypes in Shanxi Province, China

Influenza exhibits pronounced seasonal and geographical heterogeneity, suggesting that meteorological factors play a key regulatory role in influenza transmission. However, whether different influenza subtypes possess distinct meteorological drivers remains unclear, and systematic evidence from temperate continental monsoon climate regions is relatively scarce. Based on laboratory-confirmed influenza surveillance data and concurrent meteorological data from Shanxi Province during 2009–2024 (excluding the period of COVID-19 non-pharmaceutical interventions), a multi-step analytical approach was employed to systematically evaluate the meteorological drivers of influenza A(H1N1)pdm09, A(H3N2), and B viruses. First, a two-stage random forest was used to identify key meteorological variables. Second, univariate analysis was performed to determine the optimal lag structures for each factor, and multivariate logistic regression models were constructed to identify independent effects. Finally, joint effect analysis was conducted to explore synergistic and antagonistic effects among meteorological factors. A total of 29,246 laboratory-confirmed influenza cases were identified during the study period, with A(H1N1)pdm09, A(H3N2), and B accounting for 28.7%, 39.6%, and 31.7%, respectively. Absolute humidity emerged as the most robust protective factor across all subtypes. The three subtypes exhibited distinct meteorological driver patterns: A(H1N1)pdm09 was sensitive to low temperature, low humidity, and high wind speed, with the risk effect of high wind speed being reversed by high temperature; A(H3N2) was primarily regulated by higher solar radiation and higher temperature, which exhibited a synergistic protective effect; B was closely associated with the humidity‒pressure system, with high humidity offsetting the risk effect of high pressure. Influenza subtypes exhibit differential responses to meteorological factors, with complex synergistic and antagonistic effects among these factors. These findings suggest that the development of influenza meteorological early warning models should consider subtype specificity and multi-factor interactions, rather than relying on single universal indicators. Not applicable. (As this study is not a clinical trial.)