Many countries around the world are or have been under lockdown for varying periods, due to the need to limit the spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus that is causing the current COVID-19 pandemic.
A new study published on the preprint server medRxiv* in May 2020 reflects the need to rethink and selectively impose lockdown on some but not all social activities, in the light of the impact of superspreaders on disease transmission.
Superspreaders and Epidemics
Scientists have discovered that the current pandemic reveals a pattern of superspreader events. For instance, one superspreader at a South Korean nightclub led to 50 new cases even as the country relaxed its lockdown with great care. Similarly, outbreaks in crowded places like hospitals and prisons and following the Mardi Gras carnival in New Orleans, USA, illustrate the power of superspreaders to perpetuate viral transmission.
In fact, one recent study reports its estimate that just 10% of infected individuals cause 80% of new cases. Thus, viral transmission cannot be seen to be a uniform event, but one which is heterogeneous. In other words, using an average figure for R0 can hide the crucial fact that all infected individuals are not equal when it comes to spreading the infection.
Such disparities in disease spread, extending across geographic regions and countries, is a common occurrence in infectious diseases and is thought to be the result of superspreading events. In the current pandemic, some scientists think that some individuals spread the virus more effectively because the viral load in their blood is many orders of magnitude higher than in most others. Exposure then becomes relatively unimportant, compared to the source of exposure.
Understanding the Role of Superspreaders in Mitigation
The current study by researchers at Copenhagen University and Roskilde University describes an agent-based model constructed to analyze the importance of superspreading in mitigation strategies at home, the workplace, and other places.
The model is age-stratified, to allow the need for intensive care units (ICU) to be a measure of epidemic impact. The simulation involves separate time steps of 6 hours, allowing each infected individual (agent) to infect another agent and thus propagate the disease possibly. This period was chosen as the largest that produces the same results.
Over time, each infected agent may undergo a change of state. The infection rates are adjusted to fit an exponential growth model to the earlier reported rate of 23% per day. This is equivalent to an R0 of 2.9 in a randomly selected population.
The researchers measured R0 as the mean number of attempts for an infected individual to spread the disease. The population in each simulation is typically 200,000, with 50 infected people as the nidus at time zero.
The first simulation envisages a population where each person can have three equally weighted interactions, namely, at home, at work, and others. Of course, in real-time social activity is chiefly with those at home, but the previous model is more straightforward.
The “Home” interactions are built around a home with 2.1 people on average, with children having parents who are 20-40 years older, and people in the home who are above 20 years are within one age group of each other.
The “Work” interactions are constructed as clusters of 8 people on average, with each person being further connected to two randomly selected people outside the cluster. People in this setting are between 20 and 70. For people under 20, school classes are used with 24 people on average.
The scale of social activity is the same across Home and Work. However, the interactions are seen to follow a cluster pattern, with the smallest clusters by far being for home.
The “Others” type of interaction involves the entire population, with each contact being chosen at random. Each of these three groups is selected for interaction with equal probability, in an age-dependent manner.
The eventual superspreaders are picked at the start and assigned an individual activity, where there will be an assigned number of infection attempts within the 6-hour time interval, with equal chances of infection each time.
Superspreaders Targeted By Limiting Contact with “Others”
Superspreaders, No Social Structure
The introduction of superspreaders into a model without social structure shows that reducing the number of contacts between the superspreader and susceptible people by 75% halts the epidemic. This entails restricting the social movement of only superspreaders but is realistically non-feasible given the lack of knowledge about the identity of these individuals.
Social Structure, No Superspreaders
If a socially structured model is used without superspreaders, using the three sectors of social interaction, it is seen that social activity can be limited to restrict spread. Once the direction and pace of the epidemic are set to fit a growth rate of 23%, the outbreak is almost identical to that modeled by the classic SEIR (Susceptible, Exposed, Infected, Recovered) model.
When lockdowns are introduced in each of the three social sectors, the effects are found to differ. When social contact is prevented at home, the peak of the epidemic and ICU use come down by a third.
When workplace contacts are disallowed, the peak is about half of the original, but with a smaller reduction in the ICU usage. When the “Others” sector was limited as to social contacts, the peak was reduced to a third, with the ICU usage being even more dramatically reduced.
The implications are serious: the extensive network of social contacts in the “Others” category is responsible for the majority of viral transmission, and the older people in society are, to a considerable extent, connected to society only through such contacts.
According to this model, while 80% to 100% of the population will be exposed to infection by work, school and home contacts, the attack rate drops to only a quarter of this if “Other” contacts are minimized, including between the elderly and their family members, and between people from outside and those at nursing homes for elder care.
Both Social Structure and Superspreaders
Finally, both social structure and superspreaders were modeled together, where only about 10% of the population are superspreaders, causing about 75% of the spread. Here, limiting contacts in the “Home” and “Work/School” sectors does not have as much effect on epidemic peak size as avoiding contacts in the “Others” sector. This latter intervention halts the epidemic.
The researchers say, “This finding suggests that limiting diffuse/random social contacts such as that occurring in transportation, gatherings, weddings, religious meetings, and frequent trips to shops, bars, and restaurants is what drives the COVID-19 epidemic. This dramatic finding only obtains when superspreaders are in the model.”
The Implications for Successful Mitigation
It is essential to understand that the success of lockdowns and the impact of exit strategies depend to a great extent on the percentage and potency of superspreaders. When mitigation is not performed, this is not relevant, of course.
Fitting the model to the real-time scenario in Denmark, it is found that the presence of superspreaders actually offers a better opportunity to control the outbreak provided by social contacts in the “Others” category are limited. The study suggests, “The way to optimize mitigation is probably to limit large gatherings where superspreaders can do their deed, but also multiple visits to smaller gatherings.”
In short, in this model scenario, say the researchers, “Superspreaders matter greatly in a mitigation scenario. Epidemics driven by superspreaders are less sensitive to reductions in close contacts at home or work/school, but highly affected by changes in their random contacts.”
Importantly, they also point out that this is an important feature for more accurate modeling:” It is in this perspective that we propose that COVID-19 models used to model the epidemic trajectory and forecast the effect of mitigation strategies and re-openings are inaccurate unless they include superspreaders. Without this element, such models will easily overestimate the epidemic size as well as the mitigation intensity needed in order to obtain control.”
medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.