The real coronavirus risk on airlines

Some American airlines have upped the ante in describing how safe it is to fly during the pandemic. United Airlines declared that the coronavirus risk is “nearly nonexistent” even on its full flights, while Southwest Airlines has proclaimed the risk “virtually nonexistent” on its aircraft. The airlines argue that high efficiency particulate air filters and universal masking have reduced transmission risk essentially to zero. To say the least, however, their assertions are highly exaggerated.

The airlines are keen to discuss experiments with dummies on widebody Boeing jets, which suggest the transmission of the coronavirus is minimal. They are less eager to mention a slew of recent papers in medical journals that report the experiences of actual passengers. One such paper discussed a flight from London to Vietnam on which a person with the coronavirus who boarded infected almost a dozen business class passengers within two meters of her.

Another told of an Australian flight from Sydney to Perth on which passengers with prior coronavirus infections caused eight new infections on board. Yet a third paper analyzed a flight from Tel Aviv to Frankfurt on which two passengers contracted the coronavirus. A fourth paper considered two rescue flights from Italy to South Korea, on each of which one passenger contracted the coronavirus. Because all of these planes had modern high efficiency particulate air filters, they demonstrate that the filters can reduce transmission of the coronavirus but cannot eliminate it.

On none of these flights were passengers wearing the cloth and surgical masks that US passengers now wear. The best study about the effectiveness of these masks was a “meta-analysis” that synthesized dozens of studies about viral transmission and estimated that such masks reduced by 66 percent the chance of becoming infected. On the Vietnamese, Australian, and Israeli flights masks were not worn, so masks that are 66 percent effective would have reduced new cases on these flights from eleven to four, eight to three, and two to one, respectively.

However, the passengers on the Korean flight wore N95 masks that offer far greater protection (estimated effectiveness 96 percent) than cloth/surgical masks. Because several infected people boarded these flights, it is likely that the number of new infections would have risen from one to five or more with the masks worn on US planes.

The airlines are also citing a paper that found only 42 confirmed coronavirus infections among the 1.2 billion passengers who have flown worldwide in 2020. But huge numbers of these passengers flew before the pandemic decimated air travel: in the US, there were more air travelers in January and February than in the next six months combined. In the countries closest to recent regular flight schedules (e.g. China), the coronavirus has been all but eradicated. Moreover, the authors of the study conceded that the vast majority of coronavirus infections on airplanes may never have been confirmed. Indeed, one of the authors refused to take part in a presentation of the findings because he believed this key point had been improperly deemphasized.

What we resolve the uncertainties about the coronavirus on American flights would be trustworthy data about what happened on them. But such reliable data are virtually nonexistent. American follow-up processes have in general been so weak that we have scant information about how many infections arose in what settings. How many Americans have become infected while riding on buses? In bowling alleys?

The only cases we know much about are “superspreader” events in which large numbers of people simultaneously contracted the coronavirus. But airplane transmission by its localized nature would not produce superspreader events. We know little about which passengers with primary infections boarded airplanes, still less about which passengers became infected in flight, and practically nothing about subsequent infections that ultimately trace back to aircraft cabins.

Under conservative assumptions, the risk of catching the coronavirus on a American flight that is 75 percent full is not zero but one in several thousand. These infections would result in about one death per million passengers. On American flights, plane crashes cause one death per 34 million passengers. The heavy majority of these victims would be people who were not even on the plane.

American airlines are financially desperate, and they are trying hard to make their planes safe for passengers. But they have no excuse for making promises about the near nonexistent risk of the coronavirus that they cannot possibly keep.

Arnold Barnett is the George Eastman management science professor at Massachusetts Institute of Technology. He conducts research in aviation safety and has a White House citation from the Flight Safety Foundation.