In late February about 175 executives from around the world came to the biotechnology company Biogen’s leadership conference in Boston. Over two days, attendees shook hands, talked among themselves and shared meals. Also in attendance: the new coronavirus. Several people at the event were unknowingly infected with the microbe that causes COVID-19, and it quickly spread among others there, who then brought it home. At least 99 people ended up sick in Massachusetts alone.

Around the same time, the coronavirus was spreading among more than 100 people who went to a funeral in Albany, Ga. sparking an outbreak that soon led to the surrounding rural county posting one of the nation’s highest cumulative incidences of COVID-19. The next month a single individual with the disease infected 52 people during a two-and-a-half-hour choir practice in Washington State. Two people died. In Arkansas, an infected pastor and his wife passed the virus on to more than 30 attendees at church events over the course of a few days, leading to at least three deaths. And these new cases spread to 26 more people, at least one of whom died.

As scientists have learned more about COVID-19, it has become clear that so-called superspreader incidents—in which one person infects a disproportionate number of other individuals—have played an oversized role in the transmission of the virus that causes the disease. The Boston conference and the funeral in Georgia were among several superspreader events that played “a notable role in the early U.S. spread of COVID-19,” according to a report by Anne Schuchat, principal deputy director of the Centers for Disease Control and Prevention. In fact, research on actual cases, as well as models of the pandemic, indicate that between 10 and 20 percent of infected people are responsible for 80 percent of the coronavirus’s spread.

These numbers mean that preventing superspreader events could go a long way toward stopping COVID-19, says Samuel Scarpino, a network scientist who studies infectious disease at Northeastern University. Scientists have identified factors that catalyze such events, including large crowd sizes, close contact between people and confined spaces with poor ventilation. Current evidence suggests that it is mostly circumstances such as these, rather than the biology of specific individuals, that sets the stage for extreme spreading of the novel coronavirus.

When describing how the SARS-CoV-2 virus spreads, epidemiologists not only use the average number of other people that one individual infects but also employ another key value called the dispersion factor, or “k.” This number describes how much a disease clusters. A small k generally means that a relatively small number of cases are responsible for transmissions, while a larger k indicates that transmissions are more evenly spread. In Hong Kong, researchers calculated that in more than 1,000 COVID-19 cases they examined, the value for k was 0.45. That value was higher than that of SARS or MERS—two previous viral outbreaks that featured superspreading—but much lower than that of the 1918 flu pandemic. In other words, SARS-CoV-2’s transmission is not as reliant on superspreading as SARS and MERS were but is far more dependent on it than influenza, Scarpino says.

The novel coronavirus seems to primarily spread via respiratory droplets produced by an infected individual during coughing, sneezing, talking or breathing. The next person becomes infected by inhaling these droplets into his or her lungs or by getting them in the nose or mouth. If people got sick right away after they were infected, they might stay at home in bed, giving them few opportunities to transmit the virus. Instead individuals with COVID-19 are contagious before they have symptoms, says Lauren Ancel Meyers, executive director of the University of Texas at Austin COVID-19 Modeling Consortium. The CDC estimates that about 40 percent of transmissions occur before the infected person has any symptoms and that symptoms take an average of six days to begin. That time gives an infected individual a long window to come into contact with other people—and to perhaps get into a situation ripe for superspreading.

Researchers have identified several factors that make it easier for superspreading to happen. Some of them are environmental. For instance, poorly ventilated indoor areas seem especially conducive to the virus’s spread. A preliminary analysis of 110 COVID-19 cases in Japan found that the odds of transmitting the pathogen in a closed environment was more than 18 times greater than in an open-air space. And the authors concluded that confined spaces could promote superspreader events. (The study has not yet been peer-reviewed.) Another preliminary preprint study, by researchers in London, examined clusters of COVID-19 cases and found that nearly all of them were indoor or indoor-outdoor settings. The largest clusters were found in indoor spaces such as nursing homes, churches, food-processing plants, schools, shopping areas, worker dormitories, prisons and ships.

Unsurprisingly, another thing these superspreader venues have in common is that they are places where large numbers of people congregate. The more individuals you pile into one place, the greater the opportunity for the coronavirus to infect many people at once, Meyers says. “If you max out at five people, it will be very hard to have a superspreading event,” she adds. But as a group’s size increases, so does the risk of transmitting the virus to a wider cluster. A large group size also increases the chance that someone present will be infectious.

Time matters too. The longer a group stays in contact, the greater the likelihood that the virus will spread among them. Exactly how much time someone needs to pick it up remains an unanswered question, says Syra Madad, a special pathogens expert at NYC Health + Hospitals. She adds that the benchmark used for risk assessment in her contact-tracing work is 10 minutes of contact with an infectious person, though the CDC uses 15 minutes as a guideline. Essential workers such as grocery store checkers and nursing home employees interact with large groups by necessity and work in situations primed for superpreading. Meyers says that if we want to contain COVID-19, we will have to find ways to protect them and make their workplaces less favorable to such events.

What people are doing matters, too, because some activities seem to make it easier to spread respiratory gunk. We have all seen droplets go flying when someone coughs or sneezes. But even when you talk, you emit a “tremendous amount” of particles, says University of California, Davis, chemical engineer William Ristenpart. “Nobody thinks about them, but they’re there,” he says. Ristenpart’s team has found that speech emits more particles than normal breathing. And emissions also increase as people speak louder. Singing emits even more particles, which may partially explain the superspreader event at the Washington State choir practice. Breathing hard during exercise might also help the spread of COVID-19. Fitness dance classes held in small rooms with up to 22 students at a time were linked to 65 cases of the disease in South Korea. But yoga classes at one of the same facilities were not linked to any clusters. A study of COVID-19 clusters in Japan found cases connected to exercising in gyms, karaoke parties, cheering at clubs and holding conversations in bars, providing further evidence that these activities may aid transmission.

Ristenpart and his colleagues have not yet confirmed that the particle-emission changes they saw affect transmission of the novel coronavirus. Their study did not measure SARS-CoV-2 itself. But the airborne particles are presumably important carriers of viral particles. The scientists also have found intriguing evidence that a small subset of people may behave as “speech superemitters”—individuals who consistently broadcast an order of magnitude more respiratory particles than their peers. “It is very difficult to identify who is going to be a superemitter ahead of time,” he says. “One of the superemitters was a very petite young woman. And I was a bigger, bulkier guy and was not a superemitter.”

The evidence about superspreading activities has led researchers to believe they are responsible for much of the new coronavirus’s transmission. “All of the data I’m seeing so far suggest that if you tamp down the superspreader events, the growth rate of the infections stops very, very quickly,” Scarpino says. “We saw in Seattle that there were at least a couple of introductions that did not lead to new cases”—implying that the virus can fade out if it is denied circumstances for spreading.

But in the U.S.—where there have been nearly 2.16 million cases and more than 117,000 deaths—those situations may be on the rise. States are reopening businesses and activities, which means more people are coming in contact with one another in larger groups. So minimizing conditions that allow superspreading events to happen will be crucial for keeping COVID-19 in check. In Japan, health officials have advised people to avoid situations with the three C’s: closed spaces with poor ventilation, crowded spaces and close-contact settings. A virus’s ability to infect is not entirely a property of that pathogen, says Cristopher Moore, a computer scientist at the Santa Fe Institute who models virus-spreading events. “It’s a property of how the virus and human society interact,” he notes, and that’s something we have the power to change.

Read more about the coronavirus outbreak from Scientific American here. And read coverage from our international network of magazines here.