Seasonal influenza epidemics occur each winter season in temperate zones, involving up to 650,000 deaths each year globally. A published study demonstrated that the circulation of one influenza virus type during early influenza season in the United States interferes with the activity of other influenza virus types. However, this finding has yet to be validated in other settings. In the present work, we investigated the interaction among seasonal influenza viruses (A/H1N1, A/H3N2 and B) in Japan. Sentinel and virus surveillance data were used to estimate the type-specific incidence from 2010 to 2019, and statistical correlations among the type-specific incidence were investigated. We identified significant negative correlations between incidence of the dominant virus and the complementary incidence. When correlation was identified during the course of an epidemic, a linear regression model accurately predicted the epidemic size of a particular virus type before the epidemic peak. The peak of influenza type B took place later in the season than that of influenza A, although the epidemic peaks of influenza A/H1N1 and A/H3N2 nearly coincided. Given the interaction among different influenza viruses, underlying mechanisms including age and spatial dependence should be explored in future.
Keywords: epidemics; epidemiology; influenza; statistical model; viral interference.