Estimating the effect of self-protection on transmission dynamics of SARS-CoV-2 in Germany in 2021: a modelling study

During the COVID-19 pandemic, German states implemented non-pharmaceutical interventions, while individuals also adopted self-initiated protective behavior. Most epidemiological studies tend to focus on one of these aspects, but in reality, both factors influence transmission dynamics simultaneously. In this study, we investigate the effect of self-protection and NPIs on the transmission dynamics of SARS-CoV-2 in Germany during 2021 and identify the corresponding model parameters based on publicly available data. We present a unified mathematical model that integrates both self-initiated protective behavior and mandated policies. By using infection and intensive care unit data from four German states, we identify all behavioral and some viral parameters, while some are set according to literature values. Since the data alone do not reveal the cause of reduction, we use the different functional structures of self-protection and non-pharmaceutical interventions to determine their respective influence. Based on these parameters, we conduct counterfactual simulations, modeling the absence of one of the mechanisms, respectively, while assuming that the other mechanism is left unchanged. Our findings indicate that both mechanisms substantially reduced transmission. Self-protection reduced the transmission less than mandated policies most of the time, but provided between 67.6 (± 6.9) % and 81.9 (± 2.1) % further reduction of the transmission rate at highest reported values of intensive care unit occupancy. Through counterfactual simulations, we demonstrate that the absence of policies or self-protection would have resulted in higher case numbers or the need for stronger adaptations. The results emphasize the crucial role of self-protection in addition to mandated policies in controlling the spread of the virus. Our research highlights the importance of incorporating self-protective behavior and mandated policies jointly in epidemiological models, which are used for policy evaluation.