Statistical Analysis of Common Respiratory Viruses Reveals the Binary of Virus-Virus Interaction

Respiratory viruses may interfere with each other and affect the epidemic trend of the virus. However, the understanding of the interactions between respiratory viruses at the population level is still very limited. We here conducted a prospective laboratory-based etiological study by enrolling 14,426 patients suffered from acute respiratory infection (ARI) in Beijing, China during 2005 to 2015. All 18 respiratory viruses were simultaneously tested for each nasal and throat swabs collected from enrolled patients using molecular tests. The virus correlations were quantitatively evaluated, and the respiratory viruses could be divided into two panels according to the positive and negative correlations. One included influenza viruses (IFVs) A, B, and respiratory syncytial virus (RSV), while the other included human parainfluenza viruses (HPIVs) 1/3, 2/4, adenovirus (Adv), human metapneumovirus (hMPV), and enterovirus (including rhinovirus, named picoRNA), α and β human coronaviruses (HCoVs). The viruses were positive-correlated in each panel, while negative-correlated between panels. After adjusting the confounding factors by vector autoregressive model, positive interaction between IFV-A and RSV and negative interaction between IFV-A and picoRNA are still be observed. The asynchronous interference of IFV-A significantly delayed the peak of β human coronaviruses epidemic. The binary property of the respiratory virus interactions provides new insights into the viral epidemic dynamics in human population, facilitating the development of infectious disease control and prevention strategies. Systematic quantitative assessment of the interactions between different respiratory viruses is pivotal for the prevention of infectious diseases and the development of vaccine strategies. Our data showed stable interactions among respiratory viruses at human population level, which are season irrelevant. Respiratory viruses could be divided into two panels according to their positive and negative correlations. One included influenza virus and respiratory syncytial virus, while the other included other common respiratory viruses. It showed negative correlations between the two panels. The asynchronous interference between influenza virus and β human coronaviruses significantly delayed the peak of β human coronaviruses epidemic. The binary property of the viruses indicated transient immunity induced by one kind of virus would play role on subsequent infection, which provides important data for the development of epidemic surveillance strategies.