It sounds too good to be true. But a compelling new study and computer model provide fresh evidence for a simple solution to help us emerge from this nightmarish lockdown. The formula? Always social distance in public and, most importantly, wear a mask.
If you’re wondering whether to wear or not to wear, consider this. The day before yesterday, 21 people died of COVID-19 in Japan. In the United States, 2,129 died. Comparing overall death rates for the two countries offers an even starker point of comparison with total U.S. deaths now at a staggering 76,032 and Japan’s fatalities at 577. Japan’s population is about 38% of the U.S., but even adjusting for population, the Japanese death rate is a mere 2% of America’s.
This comes despite Japan having no lockdown, still-active subways, and many businesses that have remained open—reportedly including karaoke bars, although Japanese citizens and industries are practicing social distancing where they can. Nor have the Japanese broadly embraced contact tracing, a practice by which health authorities identify someone who has been infected and then attempt to identify everyone that person might have interacted with—and potentially infected. So how does Japan do it?
“One reason is that nearly everyone there is wearing a mask,” said De Kai, an American computer scientist with joint appointments at UC Berkeley’s International Computer Science Institute and at the Hong Kong University of Science and Technology. He is also the chief architect of an in-depth study, set to be released in the coming days, that suggests that every one of us should be wearing a mask—whether surgical or homemade, scarf or bandana—like they do in Japan and other countries, mostly in East Asia. This formula applies to President Donald Trump and Vice President Mike Pence (occasional mask refuseniks) as well as every other official who routinely interacts with people in public settings. Among the findings of their research paper, which the team plans to submit to a major journal: If 80% of a closed population were to don a mask, COVID-19 infection rates would statistically drop to approximately one twelfth the number of infections—compared to a live-virus population in which no one wore masks.
The mask debate, of course, has been raging for weeks in the States and globally. Pro-maskers assert that the widespread use of face coverings can diminish the spread of COVID-19. Some anti-maskers, including various politicians and public health officials, have insisted that there is no proof of the efficacy of face guards. According to some activists, a blanket mask mandate places a limit on individual liberty and even one’s right to free speech. (Pro-mask advocates are fighting back with #masks4all and #wearafuckingmask Twitter campaigns).
Representatives of the World Health Organization have also been sounding rather anti-mask, fretting that many people won’t wear masks properly, thereby risking infection, or that masks will give people a false sense of security and encourage risky behavior, such as partying up close and personal—none of which seems to have played out, as far as we know, in Japan or Hong Kong or other mask-wearing places. Adding to the brouhaha has been the shortage of medical masks for doctors, nurses, bus drivers, and the guy who delivers burritos to your door.
The muddle over masks is what drove Berkeley’s De Kai to drop everything two months ago and help convene an ad hoc team of scientists and academics: a physician from London, a bioinformaticist from Cambridge, an economist from Paris, and a sociologist and population-dynamics expert from Finland.
“I felt like this was pretty urgent,” said De Kai, who was born in St. Louis, and is the son of immigrants from China. “I saw the country where I grew up, where my family lives [now mostly in the Bay Area], about to face this pandemic without knowing much about something as simple as wearing a mask to protect themselves and others.” In part, this comes from a cultural difference between East Asia, where masks have been routinely worn for decades to fend off pollution and germs, and other parts of the world. This includes the U.S., where people are unaccustomed to wearing masks, and, in the past, have sometimes been insensitive, even stigmatizing East Asians, many of whom had chosen to wear them in public prior to the pandemic, and had continued the practice in the aftermath of the SARS and MERS outbreaks. (In part, this habit was meant to show other people that they were concerned about transmitting the disease—something we in the West would do well to emulate.)
De Kai’s solution, along with his team, was to build a computer forecasting model they call the masksim simulator. This allowed them to create scenarios of populations like those in Japan (that generally wear masks) and others (that generally don’t), and to compare what happens to infection rates over time. Masksim takes sophisticated programming used by epidemiologists to track outbreaks and pathogens like COVID-19, Ebola, and SARS, and blended this with other models that are used in artificial intelligence to take into account the role of chance, in this case the randomness and unpredictability, of human behavior—for instance, when a person who is infected decides to go to a beach. De Kai’s team have also added some original programming that takes into account mask-specific criteria, such as how effective certain masks are at blocking the invisible micro-droplets of moisture that spray out of our mouths when we exhale or speak, or our noses when we sneeze, which scientists believe are significant vectors for spreading the coronavirus.
Along with the masksim site, the team is also releasing a study that describes their model in detail as well as their contention that masksim’s forecasts support a growing body of pro-mask evidence. “What’s most important about wearing masks right now,” said Guy-Philippe Goldstein, an economist, cybersecurity expert, and lecturer at the Ecole de Guerre Economique in Paris—and a masksim collaborator, “is that it works, along with social distancing, to flatten the curve of infections as we wait for treatments and vaccines to be developed—while also allowing people to go out and some businesses to reopen.”
While all models have limitations and are only as good as their assumptions, this one is “a very thorough model and well done,” said William Schaffner, an infectious disease specialist at Vanderbilt University, who reviewed the De Kai team’s paper. “It supports a notion that I advocate along with most other infectious disease experts: that masks are very, very important.” Jeremy Howard, founding researcher at fast.ai and a distinguished research scientist at the University of San Francisco, also assessed the paper. “It’s almost overkill how careful they were with this modeling,” said Howard, who also coauthored and spearheaded a study last month (recently submitted to the journal PNAS) that reviewed dozens of papers assessing the effectiveness of masks.
During a screen-share Zoom from his home office in Hong Kong, De Kai, who has not had to shelter in place (“because nearly everyone in here wears masks”), explained to me how the model works. (Check out this video where De Kai demos the site). On De Kai’s Zoom screen, a box pops up filled with dozens of blue dots, each representing a person who is publicly zipping and zapping about, doing their thing, and sometimes interacting with others. These blue dots denote the “uninfected, but susceptible.” As the simulation rolls along, one of the dots becomes orange, representing a person who has been exposed to the coronavirus. This orange dot then touches a nearby blue dot, which also changes to orange while the original orange dot changes to red. This means that dot-person is now infected. As the model runs and simulated “days” pass by, with the dots continuing to bounce around, some of the oranges and reds turn green, meaning they have recovered—or died.
On the screen-share, De Kai first ran a simulation that shows what happens when COVID-19 strikes a population in which no one wears a mask. The orange and red dots proliferate at a frightening speed; “susceptibles” becoming “exposed/infected,” then recovered or dead. “This is what you don’t want,” said De Kai. He changed the setting to simulate what would happen if 100% of the make-believe population wore masks; almost all of the dots would stay blue—with each of them surrounded by a white square, representing someone wearing a mask.
Next De Kai added another tweak, modeling a situation in which 80% of a given population wore a mask. Here, most of the dot-people stay blue, with a few going orange, red, and green. “This is the goal,” De Kai maintained. “For 80 or 90% of the population to be wearing masks.” Anything less, he added, doesn’t work as well. “If you get down to 30 or 40%, you get almost no [beneficial] effect at all.”
“I started to go out just to buy food in mid-March,” recalled economist Guy-Philippe Goldstein. “I was the only one wearing a mask, and people were making fun of me. They aren’t now, although there still aren’t enough people in Paris wearing masks.” This may be one reason why only a few states in the U.S. currently require people to always wear masks when they are out in public, although many states require masks for certain workers, for entering businesses, and on public transportation. Many cities and counties, including Denver and Los Angeles County, require them too. Whether you’re in a blue state or a red one, you don’t want to become one of De Kai’s red dots.