From the first embers stirring in China, the SARS-CoV-2 virus has been a stealthy wildfire, flaring with deadly efficiency when it happens upon the opportunistic combination of people and place.
In January, after the novel coronavirus had already begun spreading in Wuhan, one patient there infected 14 health care providers. In March, an infected worker at a Korean call center spread the virus to 96 others, including nearly half of those on his floor. In June, a college bar in East Lansing, Michigan, became the transmission space for 146 cases and those people infected 46 more. In early July, Ohio health officials reported that a man who attended a church service a couple of weeks earlier infected 53 worshipers, and they, in turn, infected another 38 people.
Increasingly, researchers are looking at such clusters—called super-spreading events—as a key component driving the pandemic. There is no generally accepted definition of a super-spreading event, but one database created by researchers at the London School of Hygiene and Tropical Medicine lists more than 200 clusters causing from three to 797 cases each, nearly all of them indoors at churches, nursing homes, schools, funerals and retail stores.
"It's incredibly clear that SARS-CoV-2 is very prone to super-spreading events and that they've really been a dominant factor in the epidemiology as it flares up in locations all around the world," says Jamie Lloyd-Smith, a professor of ecology and evolutionary biology at the University of California at Los Angeles who studies emerging pathogens. He was the author of a landmark 2005 study that focused on the SARS outbreak to show the frequency and effects of super-spreading.
One reason the virus that causes COVID-19 is such a powerful super-spreader is because it's sneaky. Those infected typically have the highest level of virus in their systems before they develop symptoms, a cleaving difference from other diseases where people become most infectious after they show symptoms and are likely to be under care or quarantine. Because carriers of the novel coronavirus may not show symptoms for up to 14 days, and some never show symptoms, they go about their lives, unaware they are infecting others. In addition, the virus spreads by respiratory droplets and smaller particles, and studies show the virus can linger in enclosed spaces—though it’s not clear how responsible that lingering is for infection. As scientists work to understand exactly how past super-spreading events unfolded, they are using what they’ve already learned to make recommendations to prevent future outbreaks.
It’s not surprising that the new coronavirus, SARS-CoV-2, spreads in this way, says Kristin Nelson, an assistant professor of epidemiology at Emory University Rollins School of Public Health. Earlier coronavirus outbreaks were fueled by superspreading. Middle Eastern Respiratory Syndrome (MERS), which has killed 886 since 2012, and Severe Acute Respiratory Syndrome (SARS), which has killed 812 since 2003, spread through clusters.
Nelson and her team analyzed more than 9,500 COVID-19 cases in four urban counties and one rural county from March to May. Their paper, a preprint accepted for publication, found that 2 percent of people were responsible for 20 percent of transmission. Other studies also have uncovered a similarly strong correlation between small numbers of people and wildfire viral spread. Researchers examining outbreaks in Hong Kong found that 20 percent of people created 80 percent of transmissions while about 70 percent did not infect anyone. In Israel, investigators looking at 212 cases concluded that they could be linked back to 1 to 10 percent of people. In a peer-reviewed paper, Adam Kucharski, an associate professor at the London School of Hygiene and Tropical Medicine, has estimated that 10 percent of people may be responsible for 80 percent of the cases.
Super-spreading means the virus spreads in fits and starts. So it’s easy for governments to become complacent about the potential for an outbreak. For example, officials in Allegheny County, home to Pittsburgh, relaxed restrictions in early June shortly after crowing that the city had a day without a single case. People flocked to bars. By June 30, the county reported more than 100 cases, which soon climbed to more than 200 daily.
Discussion about stopping the spread of SARS-CoV-2 tends to focus on what’s called R, the average number of new cases caused by an infected person. When that number is lower than one, the disease stops spreading. But most people don't infect others. That's why researchers also look at a value called k, the dispersion factor, which is how much a disease clusters. The lower the number, the larger the super-spreader likelihood.
Lloyd-Smith in his 2005 paper determined that SARS, fueled by super-spreading, had a k of 0.16. The estimated k for MERS is 0.25. For the flu pandemic of 1918, it’s about one. Research so far for the COVID-19 virus places the k number in the neighborhood of SARS, possibly as low as 0.1.
If an infected person travels to a city, they are likely not to spread the virus as part of the 70 percent of people who do not infect anyone. "So potential outbreaks actually tend to fizzle more often than you'd expect based on the average," Lloyd-Smith notes. "But on the flip side, once in a while the virus hits the jackpot. And then you get this explosive epidemic that actually grows way faster than you would expect based on the average."
That makes creating public health policy a difficult balance, especially as scientists learn more about the small period of time people are infectious. Schools may open without an outbreak. But one is coming. "You're ultimately going to roll snake eyes and get the superspreading event,” Lloyd-Smith says.
The window for any one person to ignite that event may be even smaller than researchers have realized. Joshua T. Schiffer, an epidemiologist who has studied herpes transmission, and his team at Seattle’s Fred Hutchinson Cancer Research Center modeled the spread of COVID-19. In a preprint that has not been peer reviewed, they found that people shed enough virus to infect others for a short period, fewer than two days and perhaps as little as half a day. Transmission after the first week of infection was "quite rare." Schiffer cautions that they had limited data on viral loads making modeling a challenge. "It's very possible that there are a subset of people who shed at much higher viral loads for much longer, and those people might be more effective super-spreaders," he adds.
Scientists still are investigating whether certain people are more infectious than others—and to what degree there are so-called super-emitters. Schiffer, Nelson, and others say a more promising prevention focus is looking at the behaviors and places that are fertile ground for super-spreading.
"We don't really yet have a great idea how variable individuals are in how much they spread," says Morgan Kain, a postdoctoral fellow at Stanford and one of the authors of a study on super-spreading. "So right now our understanding of superspreading is really much more from the behavioral side of things. Are you going to areas in which you have the possibility of becoming a super-spreader versus something being physiologically different from individual to individual?"
Kain's team recently conducted a study that determined eliminating high-risk events like large indoor gatherings had a disproportionate effect on reducing transmission. They created a model using death, case, and mobility data from five places—Seattle, Los Angeles, Santa Clara County, Atlanta and Miami—and showed that targeting super-spreading events could control the epidemic. In Seattle and Los Angeles, for instance, they found combining moderate social distancing and removing the top 0.5 percent of spreaders at 75 percent efficiency—a quarter of super-spreading events would slip through—would lower the R number below one, effectively stopping the spread.
Testing and isolating infected people, they noted, is the best option for reducing transmission, but it's expensive and capacity remains limited. Restricting the highest-risk activities like large gatherings and indoor events in poorly ventilated spaces including bars, gyms, churches, restaurants and funerals would create a large reduction in transmission rates, they concluded. It would also potentially prevent a resurgence.
Looking at controlling the pandemic through the lens of super-spreading has Schiffer thinking about "bad buildings.” He says improving ventilation or wearing an N95 mask in those places where super-spreading can occur could knock down the virus.
In Japan, officials are targeting how super-spreading cases start. They have focused on identifying clusters and then using contact tracing not just to isolate new cases, but to look for patterns to super-spreading sources. In a new preprint, Kucharski argues that such “backward contract tracing” could be a valuable part of the public health response, although he acknowledged the difficulty of finding sufficient resources and getting people to cooperate.
For now, to slow super-spreading, Kucharski recommends following simple guidelines pioneered in Japan. Avoid the three C's—closed spaces, crowded places, and close-contact settings.
"I think there’s increasing evidence of the importance of the three C’s," he says. "Given many European countries reopened outdoor activities and dining without seeing a rapid rise in cases, it suggests that a relatively small collection of environments and interactions are responsible for driving outbreaks. The challenge is what happens in winter, as many of these risky situations are easier to avoid in the summer months."