HONG KONG – You must have heard about some of these outbreaks; they’re almost emblematic of the COVID-19 pandemic by now: that megachurch in South Korea, meatpacking plants in the United States, a wedding in Jordan, funerals around the world.
You’ve also probably heard of SARS-CoV-2’s R0 (R-naught), or basic reproductive number, the average number of people to whom an infected person passes on a new virus when no measures to contain it have been taken. This coronavirus’s R0 is thought to range between 2 and 3; an epidemic is curbed when that figure drops below 1, the replacement rate.
But that figure has limitations: It doesn’t convey the vast range between how much some infected people transmit the virus and how little others do.
This is why epidemiologists also look at a virus’s dispersion factor, known as “k,” which captures that range and so, too, the potential for superspreading events. To simplify: The fewer the number of cases of infection responsible for all transmissions, the lower k generally is (though other factors, like the R0, also are relevant).
In the case of SARS-CoV-2, evidence is growing that superspreading is a hugely significant factor of total transmission.
Take Hong Kong, which as of June 2 had 1,088 confirmed or probable cases (and four deaths), for a population of about 7.5 million. The city has managed to largely suppress local outbreaks of COVID-19 without a lockdown or mandatory blanket stay-at-home orders, favoring instead a strategy of testing people suspected of being infected, tracing and quarantining their contacts and isolating confirmed cases in the hospital — coupled with outright bans or other restrictions on large social gatherings.
After these measures were progressively relaxed in recent weeks, a new outbreak of seven cases, possibly a superspreading event, has been reported over the past few days: Three are employees of a food-packing company; the other four live in the same housing estate as one of the employees.
We recently published a preprint (a preliminary paper, still to be peer-reviewed) about 1,038 cases of SARS-CoV-2 in Hong Kong between Jan. 23 and April 28 that, using contact-tracing data, identified all local clusters of infection.
We found that superspreading has overwhelmingly contributed to the transmission of SARS-CoV-2 in the city overall.
Of the 349 local cases we identified — the remaining 689 cases were imported from other territories — 196 were linked to just six superspreading events. One person alone appears to have infected 73 individuals after frequenting several bars in late March. Weddings, temples, hot-pot dinners, work parties and karaoke venues featured in the other clusters.
In our study, just 20% of cases, all of them involving social gatherings, accounted for an astonishing 80% of transmissions. (That, along with other things, suggests that the dispersion factor, k, of SARS-CoV-2 is about 0.45).
Another 10% of cases accounted for the remaining 20% of transmissions — with each of these infected people on average spreading the virus to only one other person, maybe two people. This mostly occurred within households.
No less astonishing was this corollary finding: Seventy percent of the people infected did not pass on the virus to anyone.
Now you might be wondering if our study, or the experience of Hong Kong, with its small number of total infections, is more broadly representative. We think so.
An analysis of early cases in the city of Wuhan, China, the site of the original outbreak, published by researchers in Switzerland in late January, was inconclusive about the frequency of superspreading. But more and more studies support the conclusion that in places other than Hong Kong, too, superspreading is a major driver of overall transmission.
A study published in The Lancet in late April, based on data from Shenzhen, southern China, about suspected cases among travelers from around Wuhan, concluded that 80% of transmissions were caused by 8-9% of cases.
Another (also peer-reviewed) paper from late April found that 94 out of 216 employees on the 11th floor of a crowded call center in South Korea likely were infected by a single index case in late February and early March.
A recent preprint (not yet peer-reviewed) about 212 COVID-19 cases in Israel between late February and late April traced 80% of the transmissions back to just 1-10% of cases.
According to mathematical modeling by Akira Endo, of the London School of Hygiene and Tropical Medicine, and others, about 10% of SARS-CoV-2 cases might account for 80% of transmissions worldwide (and the virus might have a dispersion factor, k, of about 0.1).
With other coronaviruses like SARS and MERS as well, a small group of superspreaders was responsible for a large majority of all transmissions.
During the SARS outbreak of 2002-03, hospitals, airplanes and densely populated housing complexes were all implicated in large superspreading events.
A 2005 study of SARS cases in Singapore — considered seminal in the field — found that just 6% of cases accounted for 80% of all transmissions, while 73% of infected people appeared not to have spread the infection. The k factor seemed to be about 0.16.
In Hong Kong, one patient is thought to have infected 138 people in a single hospital during two to three weeks in March 2003; a cluster of 331 infections was traced back to a single resident in the Amoy Gardens housing complex.
For MERS, which first surfaced in Saudi Arabia in 2012, about 14% of cases are thought to have accounted for 80% of transmissions, with k=0.26, and most MERS superspreading events have been linked to hospitals.
This data in turn raise this crucial question: Why are some cases superspreaders and others not?
Superspreading is a complex phenomenon, and it depends on several factors: an infected person’s degree of infectiousness, the length of other people’s exposure to them, the setting of that exposure.
We are not aware of any study having been published that identifies individual characteristics that might account for an infected person’s degree of infectiousness or could otherwise help predict who may be a superspreader.
This much, though, is known: The infectiousness of SARS-CoV-2 appears to peak within the first few days of the onset of COVID-19 symptoms and then decrease with time. That said, one can be contagious before displaying symptoms or without ever displaying any symptoms. (Hence the importance of face masks.)
It stands to reason, too, that a highly contagious person is more likely to spread the infection in a crowd (at a wedding, in a bar, during a sporting event) than in a small group (within their household), and when contact is extensive or repeated.
Transmission is more likely during gatherings indoors than outdoors. Simply ventilating a room can help. We believe that with the South Korean call-center cluster, the essential factor of transmission was the extent of time spent in a crowded office area.
Also consider this counterexample: Japan. The government recently lifted a state of emergency after controlling its epidemic without having put in place any stringent social distancing measures or even doing much testing. Instead, it relied on largely voluntary measures encouraging people to stay at home and advice to avoid overcrowding in public venues.
In essence, Japan adopted an anti-superspreading strategy. The approach was targeted at limiting what some researchers from Tohoku University have called the “three Cs”: closed spaces, crowds and close contacts.
We believe that despite Japan’s success so far, Hong Kong’s suppression strategy, which includes testing and contact-tracing as well, is preferable in the long run, if only because it’s better preparation for any future outbreaks.
But the record in both places, and elsewhere, points to the same conclusion: It’s not just that superspreading events are happening with SARS-CoV-2; they appear to be driving much of the pandemic.
This fact is alarming and reassuring at the same time.
It’s alarming because it suggests a virus swift and efficient, and so seemingly unstoppable.
But the considerable role of superspreading in this pandemic should be reassuring, too, because it also suggests a way to stop SARS-CoV-2 that is both less onerous and more effective than many of the strategies that have been pursued so far.
The epidemic’s growth can be controlled effectively with tactics far less disruptive, socially and economically, than the extended lockdowns or other extreme forms of social distancing that much of the world has experienced over the past few months.
Forget about maintaining — or, if infections resurge, resuming — sweeping measures designed to stem the virus’s spread in all forms. Just focus on stopping the superspreading.
Dillon C. Adam is a visiting research fellow at the University of Hong Kong, where Benjamin J. Cowling is a professor of infectious disease epidemiology. They wrote this article for the New York Times.