Finding the Correlation between COVID-19 and Influenza Obscured by Seasonal Patterns: A Mathematical Modelling Study

Abstract

Background: Although significant seasonality had been reported among common human coronaviruses (NL63, 229E, HKU1, and OC43) and influenza viruses (IVA and IVB), the potential seasonal pattern of coronavirus disease 2019 (COVID-19) transmission and its interaction with winter influenza remains to be seen. Understanding the seasonality of and correlation between severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and influenza viruses may contribute to the prevention and control of the COVID-19 pandemic in the coming flu season.Methods: We used time series decomposition to separate the seasonal factors of four common human coronaviruses and two influenza viruses in the United States from 2014 to 2020. Test results and confirmed time series in the United States from the 14th to 45th weeks of 2020 were used to analyse the seasonality of SARS-CoV-2. We tried to eliminate the seasonality of SARS-CoV-2 by applying the average seasonal factors of the other beta-coronaviruses. After seasonal adjustment, we identified a correlation between COVID-19 and influenza. Additionally, we collected and analysed the weekly percent positive of influenza virus cases in nine census divisions in the United States between the 1st and 45th weeks of 2020, and the same period in 2016-2019, to verify this correlation.Findings: We identified a moderate correlation between betacoronaviruses and influenza viruses from 2014-2020. The seasonally adjusted Spearman correlation coefficient between coronavirus OC43 and influenza A virus (IVA) was 0.39, and 0.41 between coronavirus HKU1 and influenza B virus (IVB), while no similar relationship was identified between alphacoronaviruses and influenza viruses. The seasonally adjusted Spearman correlation coefficient between SARS-CoV-2 and influenza viruses among nine census divisions in the United States from the 14th to 45th weeks of 2020 showed a moderate negative correlation in most census divisions, with mean correlation coefficients of -0.352 between SARS-CoV-2 and IVA, -0.327 between SARS-CoV-2 and IVB. The weekly percent positive of influenza virus cases in the United States showed a significant reduction between the 14th and 45th weeks of 2020 and the same weeks in 2016-2019, which partly verified the negative correlation between COVID-19 and influenza.Interpretation: As synchronous seasonality commonly exists among human coronaviruses and influenza viruses, seasonal adjustment is necessary before the pathogen-pathogen correlation analysis can be performed. The present study revealed a negative correlation between SARS-CoV-2 and influenza viruses, which was obscured by strong synchronous seasonality. According to our study, COVID-19 may unfortunately increase in winter due to its seasonal pattern, but a simultaneous outbreak of influenza and COVID-19 may not occur this winter due to the negative interaction between these two pathogens.Funding Statement: This study was funded by the Science and Technology Innovation Committee of Shenzhen Municipality (JCYJ20190808152613121) and the Shenzhen-Hong Kong Institute of Brain Science – Shenzhen Fundamental Research Institutions (2019SHIBS0003).Declaration of Interests: We declare no competing interests.