Seasonality of Human Influenza and Co-Seasonality with Avian Influenza in Bangladesh, 2010-2019

Abstract

Purpose: Seasonality and periodicity of influenza in human and animal populations is not well characterized in tropical and subtropical regions globally. In Bangladesh, where over 90% of poultry products are marketed through live poultry markets, avian influenza is prevalent in poultry and epidemics of influenza occur in humans. We sought to empirically examine the timing and intensity of influenza seasonality in humans and avian influenza in poultry in Bangladesh using time series methods. Methods & Materials: We evaluated 8,790 cases of human influenza occurring between January 2010 and December 2019 obtained from 32 hospital-based sentinel surveillance sites across Bangladesh. Avian influenza positive samples (n=2,274) were obtained from environmental surveillance across 106 live poultry markets in Dhaka between January 2016 and December 2019. We applied wavelet analysis to determine influenza periodicity, and conducted Poisson regression with harmonic terms adjusting for trend and an offset for samples tested to estimate the average timing and intensity of the primary influenza peak in each population. We examine co-seasonality by estimating the time lag between the annual influenza peak in humans and the annual avian influenza peak in poultry. Results: Over the 10-year period there was consistent annual periodicity of influenza in humans with peak incidence occurring in early July (peak calendar-week: 27.3, 95%CI: 27.0-27.5) and intense seasonality (amplitude: 14.7, 95%CI: 13.4-16.2). Large metropolitan centers displayed earlier epidemic timing, with epidemic spread following a spatial diffusion pattern based on geographic proximity. Comparatively, avian influenza displayed weak seasonality, with moderate year-round transmission and only a small peak in late March to early April (peak calendar-week: 13.5, 95%CI: 11.5-16.8; amplitude: 1.5, 95%CI: 1.3-1.7). Peak avian influenza preceded peak human influenza by 13.8 calendar-weeks (95%CI: 10.7-15.5). Conclusion: Our findings suggest that while influenza epidemics in humans are highly seasonal and take off faster in metropolitan areas, avian influenza in poultry does not display a strong seasonal trend in Bangladesh. Although influenza peaks do not coincide, endemic avian influenza continues to pose a risk for viral reassortment throughout the year. These co-seasonality results provide evidence for more quantitative risk assessments of viral reassortment and inform timing of sequencing-based surveillance.