< Modelization

Why are superspreader events crucial to understanding the COVID-19 epidemic?

Text updated on 2020-06-23


The COVID-19 epidemic spreads largely through superspreader events. To avoid a new outbreak, it is important to monitor and limit the most "flammable" places/events and thus prevent the occurrence of super-propagating situations.

The efficiency of COVID-19 transmission from one person to another is extremely variable: the majority of infections are extinguished without offspring, but a small number of infections (10-20%) are responsible for a large number of cases of contamination (up to 80%). This phenomenon could explain why the major epidemic outbreak in France did not occur until March 2020, although the SARS-CoV-2 coronavirus was sporadically present on French territory as early as December 2019.

The occurrence of superspreader events is essential for an outbreak to occur. These situations can be biological, where some people emit more infectious particles than others, and social, where there are events that concentrate large numbers of people in one place. They allow the COVID-19 disease to move from a "background" level to an epidemic stage. Understanding and controlling these superspreader events or locations in advance is essential to prevent a possible second wave from occurring without having to resort to extreme containment measures.

To understand how it COVID-19 spreads, knowing its reproduction number (R0) is not enough. Another characteristic of infections, their dispersal factor k (kappa), must also be taken into account. While the reproduction number reflects the average contagiousness of all infected persons, the dispersal factor measures the variability of this reproduction rate within the population.

When k is high, the epidemic progresses progressively and uniformly, like an oil stain: this was the situation observed during the Spanish flu epidemic in 1918. The closer k is to 0 for the same R0, the more variable the number of people infected by an infected person and the more the epidemic can spread through superspreader events. For example, when k = 0.1 and R0 = 3, 73% of individuals infect less than one person, but 6% infect more than eight. The epidemic then progresses in a discontinuous manner which, in statistical terms, responds to a negative binomial distribution (whereas seasonal influenza is closer to a Poisson-type distribution, which is reached when k tends towards infinity). A mode of diffusion through superspreader events was observed during the SARS epidemic (R0 = 2; k = 0.16) and, to a lesser degree, that of MERS (R0 = 0.6; k = 0.25). Estimates of k are less accurate than those of R0 because k is a measure of dispersion while R0 is an average. This means that many transmission cases are needed to obtain a good estimate of k. At present, studies suggest that the value of k for COVID-19 is around 0.1-0.4.

Under these conditions, it is statistically necessary for a few dozen cases to occur simultaneously to create the necessary conditions for an epidemic to take off. This is where superspreader events play an important role: suddenly, the environment becomes favourable for the spread of COVID-19, despite its low contagiousness in the majority of infected people. This particularity explains why the most important COVID-19 outbreaks have not systematically appeared in large metropolises (as is the case for infections with Poisson-type distribution), but also in less populated places (for example Mulhouse or Codogno), where a super-propagating event had occurred. Furthermore, and still according to the negative binomial distribution, once a focus is established, the growth in the number of cases explodes rapidly, within a few generations of infected patients. Epidemiological models predict that establishing and stabilizing the exponential growth of such an epidemic requires a continuous supply of superspreader events. These results, therefore, suggest that the epidemic can be largely controlled if super-spreading scenarios causing transmission are eliminated.

While a few outbreaks dominate the transmission of the SARS-CoV-2 coronavirus, most of them are self-extinguishing, this means that in order to avoid a new outbreak, it is critical to know and monitor the most "flammable" places/events and, thus, prevent the occurrence of superspreader events.


facebook twitter linkedin

Sources

The SARS-CoV-2 coronavirus was probably sporadically present in France as early as December 2019.

Deslandes, A., Berti, V., Tandjaoui-Lambotte, Y., Alloui, C., Carbonnelle, E., Zahar, J. R., ... & Cohen, Y. (2020). SARS-CoV-2 was already spreading in France in late December 2019. International Journal of Antimicrobial Agents, 106006.

The R0 for seasonal influenza was estimated at 1.3 (range 0.9 to 2.1).

Coburn, B. J., Wagner, B. G., & Blower, S. (2009). Modeling influenza epidemics and pandemics: insights into the future of swine flu (H1N1). BMC medicine, 7(1), 30.

The dispersion factor k for SARS was estimated to be 0.16 (90% confidence interval 0.11-0.64).

Lloyd-Smith, J. O., Schreiber, S. J., Kopp, P. E., Getz, W. M. (2005). Superspreading and the effect of individual variation on disease emergence. Nature, 438(7066), 355-359.

The dispersion factor k for the MERS-CoV was estimated to be 0.26.

Kucharski, A. J., & Althaus, C. (2015). The role of superspreading in Middle East respiratory syndrome coronavirus (MERS-CoV) transmission. Euro surveillance, 20(25), pii-21167.

A study of clusters in Hong Kong (corresponding to 1,037 people who tested positive) carried out in May 2020 estimates that 20% of cases of SARS-CoV-2 contamination are responsible for 80% of local transmission. Social exposures produce more secondary cases than family or work interactions. The dispersion factor k is estimated at 0.45 (95% CI: 0.30-0.72).

Adam, D., Wu, P., Wong, J., Lau, E., Tsang, T., Nightmare, S., ... & Cowling, B. (2020). Clustering and superspreading potential of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections in Hong Kong.

A British COVID-19 study estimates that for R0 values around 2-3, the dispersion factor k is around 0.1 (median 0.1; 95% confidence interval: 0.05-0.2 for R0 = 2.5): this implies that 10% of individuals are responsible for 80% of the cases of contamination.

Endo, A., Abbott, S., Kucharski, A. J., & Funk, S. (2020). Estimating the overdispersion in COVID-19 transmission using outbreak sizes outside China. Wellcome Open Research, 5(67), 67.

A Swiss modelling study of COVID-19 from January 2020 puts R0 between 1.4 and 3.8 (median 2.2) and k between 0.014 and 6.95 (median 0.54). Data from the epidemic up to January 2020 do not allow a more accurate estimate of k. Estimates of k are less accurate than those of R0 because k is a measure of dispersion while R0 is a mean.

Riou, J., & Althaus, C. L. (2020). Pattern of early human-to-human transmission of Wuhan 2019 novel coronavirus (2019-nCoV), December 2019 to January 2020. Eurosurveillance, 25(4), 2000058.

On the basis of SARS-CoV-2 coronavirus genomes, a team from Oxford estimated that the coronavirus had reached the United Kingdom and started to spread within the borders of the country in at least 1,356 introductions, and that this number was probably underestimated. Introductions mainly occurred via travellers from Spain (34%), France (29%) and Italy (14%).

Pybus O., Andrew Rambaut, Louis du Plessis, Alexander E Zarebski, Moritz U G Kraemer, Jayna Raghwani, Bernardo Gutiérrez, Verity Hill, John McCrone, Rachel Colquhoun, Ben Jackson, Áine O'Toole, Jordan Ashworth, on behalf of the COG-UK consortium. (2020) Preliminary analysis of SARS-CoV-2 importation & establishment of UK transmission lineages. Preprint on virological.org

Explanations on the kappa parameter and the importance of superspreader events.

Korsia-Meffre, S. (2020). COVID-19 "The only thing that matters is where it falls" or how to avoid a possible second wave. Vidal

This study shows that the higher the voice volume (amplitude), the higher the number of particles emitted during speech, ranging from 1 to 50 particles per second (0.06 to 3 particles per cm3) for low to high amplitudes, regardless of the language spoken (English, Spanish, Mandarin or Arabic). In addition, a small fraction of individuals behave as "super emitters", systematically releasing ten times more particles than others.

Asadi, S., Wexler, A. S., Cappa, C. D., Barreda, S., Bouvier, N. M., & Ristenpart, W. D. (2019). Aerosol emission and superemission during human speech increase with voice loudness. Scientific reports, 9(1), 1-10.

Article submitted on May 27th which takes stock of the superspreader events in the context of the spread of the COVID-19 epidemic and identifies the key factors of a super-propagating event. If the epidemic of COVID-19 grows and exceeds a few dozen cases, then the dynamics of transmission begins to show a stable exponential growth, with a growth rate approaching that of a model with a Poisson distribution having the same R0. Then, once the epidemic has taken off, it will appear even more explosive than when the distribution follows a Poisson distribution. Establishing and stabilizing the exponential growth of such an epidemic requires a continuous supply of superspreader situations. These results, therefore, suggest that the epidemic can be largely controlled and the effective number of Reff replicates considerably reduced when the superspreader events that cause transmission are eliminated.

Althouse, B. M., Wenger, E. A., Miller, J. C., Scarpino, S. V., Allard, A., Hébert-Dufresne, L., & Hu, H. (2020). Stochasticity and heterogeneity in the transmission dynamics of SARS-CoV-2. arXiv preprint arXiv:2005.13689.

Further reading

Lethality, mortality, excess mortality, R0, kappa: what are we talking about?

What is a superspreader event for COVID-19 ?

What is the purpose of social distancing?

Why are gatherings not recommended during the outbreak of COVID-19 ?