Projecting the transmission dynamics of SARS-CoV-2 through the post-pandemic period

Stephen M Kissler; Christine Tedijanto; Edward Goldstein; Yonatan H. Grad; Marc Lipsitch

To understand the relative contribution of depletion of susceptibles compared to seasonal forcing in the observed data, we adapted a linear regression model as follows (22): where Rsij is the weekly effective reproduction number for strain s in week i of season j, R0 is the basic reproduction number, S0 is the fraction of susceptibles at the start of the first season for the reference strain, and ϵsij is a normally distributed error term. A dummy variable for each strain-season combination (ɑsj) captures differences in the reproduction number and the starting fraction of susceptibles between strains and over time, but is unable to distinguish between the two. The next two terms estimate the impact of depletion of susceptibles due to infection by the same strain (dsij) and the other betacoronavirus strain (drij). Depletion of susceptibles for each strain was estimated up to a proportionality constant by the cumulative sum of the incidence proxy over season j through week i. The coefficient on the first term (γs) represents the scaling factor between the cumulative incidence proxy and true depletion of susceptibles, while δs captures the level of cross-immunity in addition to scaling; both coefficients were allowed to vary by strain. Because specific seasonal drivers (e.g. absolute humidity) of seasonal variation in coronavirus transmission have not been identified, we did not include them in our model but . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.