Seasonal patterns COVID-19 and Flu Like Illnesses comparable

Introduction During the first wave of COVID-19 it was hypothesized that COVID-19 is subject to multi-wave seasonality, such as other respiratory viral infections since time immemorial, including earlier respiratory pandemics. It has already been observed that the COVID-19 community outbreaks appear to have a similar pattern as other influenza like illnesses (ILI). One year into the pandemic, we aimed to test the seasonality hypothesis for COVID-19. Methods We gather and calculated the average ILI annual time series based on incidence data from 2016 till 2019 in the Netherlands. And, compared this with two independent COVID-19 time series during 2020/2021 for the Netherlands, plotted on a logarithmic infection scale. We tested our hypothesis by calculating correlation coefficients and, as a sensitivity analysis, by performing univariate regression analysis. Results The COVID-19 time series strongly and highly significantly correlates with the ILI time series r(45) = 0.75 (p < 0.00001) and (r(45) = 0.798, p < 0.00001). Also the univariate regression analyses that we performed as a sensitivity analysis are all highly significant: respectively F(1, 43) = 61.45, p < 0.0001, and F(1, 43) = 81.18, p < 0.0001 and the correlations (r2) are moderate to strong. Conclusions Given the strong, and highly significant, correlations between the ILI and COVID-19 time series, we conclude that COVID-19 behaves as seasonal as ILI in a country in the temperate climate zone, such as the Netherlands. Moreover, the COVID-19 peaks are all during flu season, and lows are all in the opposing period as expected. Furthermore, the COVID-19 time series satisfies the two characteristics of earlier pandemics, namely a short first wave at the tail-end of a flu season, and a longer and more intense second wave during the subsequent flu season.


Introduction 36
During the first wave of COVID-19 it was hypothesized that COVID-19 is subject to multi-wave seasonality 37 (Kissler et al., 2020;Grech et al., 2020), such as other respiratory viral infections since time immemorial the cycles themselves are sufficiently representative for time series analysis, even though the  incidence during the first cycle is most likely underestimated compared to incidence during the second cycle, due infections scale (Steffen, 2018). We added descriptive labels to each scale as an aid for qualitative interpretation and comparison.
Correlations are calculated manually in Excel, and for linear regression Graphpad 2021 is used. The means and standard deviations of the dataset are summarized in Table 1. Figure 1 shows a short first COVID-19 wave at the tail end of the 2019/2020 flu season, and a more severe 114 second wave during the 2020/2021 flu season in terms of total incidence. The peaks are all within the seasonal 115 boundaries between week 33 (± 2 weeks) and week 11 (± 5 weeks), and the nadirs in the opposing period. The 116 hospitalizations based estimates for COVID-19 incidence provide likely a more realistic picture of especially, the 117 first wave, given test bias. However, on a logarithmic scale, the first COVID-19 wave appears more visually 118 comparable in both time series.

119
On our logarithmic infection scale (Fig. 2), the estimated COVID-19 incidence tops around 6 (severe epidemic series, and the average ILI time series. The outcomes were again highly significant: respectively F(1, 43) =

Discussion 138
Given the strong, and highly significant, correlations between the ILI and COVID-19 time series, we conclude

146
Interestingly, all over Europe the COVID-19 cycles were all more or less in sync with the Dutch COVID-19 147 cycle (Reuters, 2021), and thus ILI seasonality, independent of the start of the first cycle, the severity of 148 lockdown measures taken, and given that herd immunity is not yet reached. The seasonality pattern of COVID-149 19 appears to be influenced though not caused by social distancing and lockdown measures as these measures 150 were mainly anti-cyclical and following the trend. They were increasingly applied to flatten the curve after 151 COVID-19 incidence increases, gradually lifted after the sharper than expected COVID-19 downcycles in Spring 152 and Summer, and only re-applied after the second wave seriously kicked in, during Autumn and Winter. It is 153 beyond our research to quantify the considerable impact of lockdown and social distancing measures, although it 154 might explain that COVID-19 incidence on the logarithmic scale (see Fig. 2) starts to rise slightly earlier than 155 what's usual for ILI (week 33 ± 2 weeks) as social distancing and lockdown measures were increasingly relaxed 156 and ignored in this period. seasonality of seasonal allergens and ILI including COVID-19 is independently confirmed by a recent Chicago study that covered not only for pollens but also mold spores (Shah et al, 2021 (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.   (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.