SARS-CoV-2, the coronavirus that causes COVID-19, continues to spread, causing substantial burden on public health worldwide. We are involved in a number of efforts in understanding the spread of SARS-CoV-2 and identifying control opportunities.
We have been working closely with researchers at Institut Pasteur in analysing hospitalisation records in France. We use mathematical models that reconstruct the evolving epidemic, allowing us to estimate the build-up of immunity in the population, the underlying probability of hospitalization, the need for intensive care treatment as well as the underlying probability of death for infected individuals. Further, we estimate the impact of lockdown strategies in containing the epidemic and characterise the changing patient profiles (such as by age and sex) of hospitalised populations.
In addition, we work on understanding who is getting infected in communities but not appearing in hospital surveillance systems. Most infected individuals do not become hospitalised, especially among younger members of the population. This means that surveillance data from hospitals will underestimate the true underlying level of infection in the community. In this context, seroprevalence studies that test for the presence of antibodies specific to SARS-CoV-2 in a subset of the population, are central to understanding the proportion of the population that have been infected. We are using the age profile of COVID-19 patients in combination with the results of seroprevalence studies from different countries around the world. We develop models that explore the consistency of these studies and implied infection fatality ratios. We also estimate location-specific infection fatality ratios that account for the age-structure of the population and observed deaths. This work has highlighted the key role nursing homes have played in the epidemic in European countries. These vulnerable elderly populations have high fatality rates and attack rates in these closed populations can be much higher than the general population.