In March 1977, on a Boeing 737 making a run from Anchorage to Kodiak, a bunch of people got the flu.
That’s not supposed to happen on airplanes. Influenza is a respiratory virus, most likely transmitted at least in part via airborne particles, and airplanes have recirculation, air filtration, and fresh-air injection systems burly enough to purge cabins of smoke if there’s a fire. They ought to do the same with a virus. So how’d those people get sick? That’s the interesting part.
The flight made a stop on the way, in Homer, where a new passenger came on board. That person, the “index case,” had the flu. The 737 took off, but an engine problem forced it to return to the airport. There it waited—engines off, on the tarmac. Some passengers stayed on board; some waited in the terminal. Eventually, other planes took the stranded passengers on to their destinations, where nearly three-quarters eventually became sick with the flu as well—“secondary infections.” Almost all of them were the ones who’d spent their unplanned layover on the plane with the sick passenger rather than in the terminal.
That choice—where they waited—was the crux. With the engines off, the 737’s air conditioner wasn’t working. The airplane cabin turned into the classic indoor, crowded, low-ventilation space you really want to avoid if you don’t want to pick up a respiratory virus like the flu—or, more relevantly today, Covid-19.
It might not seem like it, but this story is actually good evidence for why it’s hard to get sick on an airplane, although folk wisdom might suggest otherwise. If everyone’s wearing a mask and the ventilation is blowing, it is indeed difficult to get Covid-19 when you fly.
But it’s not impossible. And that’s where things get complicated.
As most of the United States tips back into uncontrolled spread of the pandemic virus, with just a week before the traditional holiday travel season begins, it’d be nice to know: How risky are airplanes, exactly? The short answer is: No one knows, but risky enough. If you can avoid flying, avoid it. “The position that a lot of airline executives are stating now, publicly—that the risk is essentially zero—is just unrealistic,” says David Freedman, an infectious disease researcher at the University of Alabama at Birmingham who studies travel medicine. “I’m not saying the risk is high. And the risk is manageable. But to start from the position that the risk is zero on an airplane is like the politicians who are saying this is a hoax. It exists.”
This position isn’t perhaps the sexiest. “Low-ish risk but nonzero, avoid unless you can’t” bears some explaining.
Most scientists and health care workers now agree that the virus that causes Covid-19 can travel through the air, borne aloft in teeny globs of snot. Even people without symptoms can emit these particles, just by talking or breathing. Masks can stop some of them, though probably not all—but hey, nobody really knows how many you have to inhale to get sick, either. Another thing that helps prevent infection is being far enough away from people that whatever virus they’re giving off gets diluted before it gets to you; even better is being outside.
Airplanes are definitely inside, and a fully-booked coach cabin doesn’t allow much leeway for social distance. But actually, the filtration and ventilation systems on airplanes are pretty great. Jet engines inhale great gulps of air, mix it with fuel, and set it on fire, which makes thrust. But some of that air (before the fuel and fire parts) gets diverted into the ventilation system. Because it comes from outside, at cruising altitude, that air is very clean and very cold. It feeds through air compressors (which heat the air way up, to hundreds of degrees) and then into air conditioner “packs” in the fuselage which cool it back down and send it to a manifold that mixes it with air from inside the cabin. It goes through ducts in the ceiling to vents that blow it downward through the cabin, and then it gets sucked into intakes next to passengers’ feet. About half of that then gets sent back outside, and the other half gets run through HEPA filters—very good at getting rid of even viral particles—and back to the manifolds, to meet a new gust of outside air. In a plane where everything’s working up to spec, the whole cabin gets an air change every two to three minutes. Tobacco smells, dust, pollen, germs, halitosis, sweat, farts—all gone. Whoosh.
And yet.
It’s clear that at least some people have gotten sick from air travel. It’s hard to track how many; no nation maintains a database of these infections. Contact tracing of air passengers during the Covid-19 pandemic has proven to be difficult—a plane might carry 300 people, and in the US, airlines, government agencies that deal with travel and border security, and the Centers for Disease Control and Prevention have been fighting since February over whose job it should be to keep track of passengers’ personal and contact information after flights, in case of an outbreak.
One potential point of failure might be on the ground. As in Homer in 1977, planes sometimes get delayed, and without power from the plane’s auxiliary power unit (APU) or ground-based air-conditioning, that shuts off the air exchange on the plane. “If they unhook the plane at the gate and there’s a delay in pushback, they’ll say, ‘Don’t worry that it’s getting hot in here, when we start the engines it’ll cool down,’” Freedman says. “That could be five or 10 minutes without proper air-conditioning.” Because the APU is essentially a generator, it has exhaust—and some airports around the world won’t allow planes to run it near the gate.
But that can’t explain every case. A review article by Freedman and his colleague Annelies Wilder-Smith in the Journal of Travel Medicine described all the known cases up to its September publication, including three flights in March—before wearing masks was a thing—that led to what they call “mass transmission.” A charter flight to bring home 11 Israeli passengers from the outbreak on board the Diamond Princess cruise ship, with at least one infected person on board, led to no secondary infections. Mask-wearing was rigorously enforced, and crew and passengers basically weren’t allowed to interact. But a March flight in Australia with at least six infected people on board, all coming back from various cruise trips, and with minimal mask enforcement, led to nearly 60 people getting sick—eight with virus confirmed by genetic sequencing to have come from another passenger.
Even after masking was more common, though, reports of rare midair infections have continued. Freedman and Wilder-Smith note a flight out of Italy in April with mandatory masking and one possible transmission. A more recent case study hinted at “widespread transmission” sparked by infections on a flight to Ireland, but other researchers pointed out that the rapid onset of symptoms suggested infection before the flight, not on it. There just aren’t a lot of studies to go on here. But let’s stipulate that masks are now at least expected on a plane. “What happened in February, March, and April when nobody was wearing a mask is academic,” Freedman says. “The functional point right now is: Are we safe with precautions?”
Masks themselves might not be enough just because of how people use them in real life. On long flights, people eat. They get up to go to the bathroom, where the mask might come off. They might take off the mask to speak, to be better heard over the noise of the engines—and speaking more loudly actually emits more viral particles, if you’re infected. (It’s also probably worth noting that airplane cabins tend to be very dry, because the outside air at altitude is low humidity—and that seems to make fighting infections harder.)
Plus, masks are only part of the equation. Seat maps of which people got sick on a given flight tend to show that the closer someone was to an index case, the more risk. And if that index case is asymptomatic, any seat could be a jackpot. “They do have high air change rates and use high-efficiency filters, but the closer they get to full occupancy on planes, the more close contact is going to be a concern,” says William Bahnfleth, an architectural engineer at Penn State and chair of the Epidemic Task Force of the American Society of Heating, Refrigeration, and Air-conditioning Engineers.
By October, this was all nervous-making enough that the air industry started being public about efforts to make people feel safe. “With only 44 identified potential cases of flight-related transmission among 1.2 billion travelers, that’s one case for every 27 million travelers,” said David Powell, medical advisor to the International Air Transport Association, in a press release. “We think these figures are extremely reassuring.”
Actually, those figures may be misleading. The data came from Freedman’s review, but he doesn't agree with that conclusion—too few people have actually been tested, so it’s impossible to know whether all those 1.2 billion fliers were virus-free. When he and I spoke, Freedman had a bit more faith in other studies that the IATA press release was touting—computer simulations of airflow through plane cabins conducted by Airbus, Boeing, and Embraer, that showed how good their planes were at keeping the air clean. The air flowing from ceiling to floor creates an “invisible barrier,” according to a press release from Airbus. Thanks to that and the HEPA filters, 1 foot of separation on an airplane is like 6 feet on the ground, says the release. To be sure, these companies have the most vested of vested interests in finding that their aircraft are safe.
The infectious disease research community isn’t as confident. As with everything about this pandemic, the situation is complex. “Most of us who are following this think that transmission can happen on airplanes, and it's almost certainly higher the closer you're sitting to the infected person, but there aren’t sufficient data to quantify that risk,” says A. Marm Kilpatrick, an infectious disease researcher at UC Santa Cruz. “It obviously also depends on the level of infection in the community of people where the flight is originating. A flight from New Zealand or Australia right now has almost almost zero risk, whereas a flight from North Dakota would be a very scary thing.”
More striking, perhaps, than the computational fluid dynamics simulations was a real-world study conducted by the US Defense Department, with aid from Boeing and United Airlines. Run by the US Transportation Command, or Transcom, which flies military personnel around the world, the new study used mannequin heads rigged to emit fluorescent particles that mimicked the behavior and approximate count of viral particles in the cabin air, estimating a release of 4,000 infectious viral particles per hour. They tried it with the head masked and unmasked, in various seats and in the jetway that connects the gate to the plane. And they did it in the air and on the ground, using sensors in other seats to see how much pseudo-virus got to other passengers’ potential personal space. The researchers found that the amount of particles that made it even to the closest seats was barely detectable, only enough to account for one infection per 54 hours of time in the air. But they acknowledged that even their real-world methods weren’t really real. The viral particles given off and infectious dose were still just estimates, and they couldn’t say what happened when people were eating or talking, or feeling free to move about the cabin.
“It was an excellent study,” Freedman says. “But the Transcom study just doesn’t account for the vagaries of human behavior. What they were measuring was people in their seats.”
That leaves out not just maskless eating, but also interactions with flight attendants—who are exposed to many more people over a day or week of work on airplanes than a passenger—and the chaos of boarding and deplaning, when the ventilation system might not be running at peak efficiency and people are even closer together.
Kilpatrick agrees that the results aren’t clear. For one thing, the Transcom report wasn’t published in a journal or peer reviewed. For another, having Boeing and United on board presents a potential conflict of interest. But even on the raw science side, since nobody actually knows how many particles an infected person emits, or how many it takes to infect someone, it’s hard to build a study that depends on those numbers. “Without data on infectious dose and how much virus is actually expelled by an infected person, we can’t quantify the risk using an experiment like this,” Kilpatrick says. “If it was 10 times lower, and you still had a 4 percent chance of getting infected if you were seated next to an infected person on a flight, would you consider that low risk? I wouldn’t.”
It seems like even the Transcom researchers agreed. Sometime after publication, they added a statement to their report. “The viral aerosol production rates, infectious dose and general assumptions used to estimate a flight time of 54 hours to produce an infection are hypothetical and were not designed to provide actionable information about viral risk during flight, safe flight times or seating capacity,” they wrote. Your mileage, they seemed to be saying, may vary. The DOD public affairs person working on the study didn’t return my emailed requests to speak with the researchers for follow-up.
It gets worse. Sitting on the airplane isn’t the only potentially risky part of air travel. No one has yet studied what happens in crowded airports, in waiting areas at gates, at baggage claim carousels, or riding rental car shuttles—in short, all the other crowded, uncomfortable parts of flying. “You have to look at travel overall—the process, and all of the different risks associated with different parts of it,” Bahnfleth says. “You may take public transit to get to the airport. You’re in the airport and may have a hard time avoiding close contact with people. Crowds just go along with airports.”
A major report from Harvard University’s Aviation Public Health Initiative concluded much the same thing at the end of October. The researchers there found a small but not insignificant risk from flying, somewhat mitigated by rigorous mask use. The next phase of their research is supposed to look at the context and environment of air travel, the airport itself. A spokesperson says they’re “in the middle of the research” on that now.
In the meantime, air travel is another one of those things that—for people who could afford it, anyway—used to range from innocuous to unpleasant, and now requires each of us to do some complicated risk assessment based on our own tolerance, need, and incomplete information. The risk is there, and the best way to mitigate it is not to fly. Stay masked if you do. Any better sense of the risk is still circling the runway.