A deterministic, age-stratified, extended SEIRD model for investigating the effect of non-pharmaceutical interventions on SARS-CoV-2 spread in Belgium

Alleman, Tijs W.; Vergeynst, Jenna; Torfs, Elena; Gonzalez, Daniel Illana; Nopens, Ingmar; Baetens, Jan M.

As a response to the rapidly rising number of SARS-CoV-2 infections, the Belgian governments imposed strict social contact restrictions on March 13th, 2020. After nearly two months, the curve was successfully flattened and social restrictions were gradually relaxed. Unfortunately, pharmaceutical interventions are not yet available so it is expected that preventing COVID-19 outbreaks will depend mostly on the successful implementation of non-pharmaceutical interventions, hence the need for well-informed models. In this study, we built a deterministic, continuous-time, age-stratified-SEIRD model with detailed hospital dynamics. Because the hospitalization data for Belgium are not made publically available by the Belgian Scientific Institute of Public Health (Sciensano), we computed the hospitalization parameters based on data from 370 patients treated in two Ghent (Belgium) hospitals. The basic reproduction number was estimated as R0 = 2.83 in March 2020 and the model fits the hospitalization and ICU admission incidence under lockdown measures well. Despite the relaxation of social restrictions, hospitalizations have been steadily declining. We recomputed the basic reproduction number under lockdown release and found that it had to be as low as R0 = 0.73 to explain the endemic trend. We further found that although the basic reproduction number in the population older than 70 years was smaller than one, this group compromises nearly half of the expected hospitalizations. This indicates that the protection of the elderly may be the most efficient way to reduce strain on the public health care system in case of another SARS-CoV-2 outbreak.