Medical Examiner

The Key Stat in the NYTimes’ Piece About Losing Your Sense of Smell Was Wrong

Anosmia and coronavirus might be connected, but we just don’t have good data yet.

A nose with illustrations of the coronavirus in the background.
Photo illustration by Slate. Photo by CDC and PeskyMonkey/iStock/Getty Images Plus.

Monday morning, as I stood at my “desk”—a spot I have carved out to work atop my dresser—I periodically held up and sniffed a couple unlit scented candles. “Rose, yes,” I thought to myself. “Ocean breeze, yes!” Sometimes, I jazzed up this routine by next checking my temperature.

This little sequence was an effort to self-soothe. If I could still smell my candles, I told myself—deluded myself, really—I probably do not yet have the novel coronavirus, which is currently exploding in unknowable numbers in New York City, where I live. My “test” is obviously flimsy and not to be trusted; you can be a carrier of COVID-19 without any symptoms at all. But like many people, I’d read a piece published Sunday in the New York Times titled “Lost Sense of Smell May Be Peculiar Clue to Coronavirus Infection,” so my candles were reassuring. Monday, on social media, I saw people time and again seemingly confirming the reporting from personal experience: They had the novel coronavirus, and like the people mentioned in the piece, they couldn’t smell things either.

And yet, as with so much about this novel coronavirus, the data that could establish loss of smell as a meaningful sign of infection is lacking. Alongside plenty of anecdotal evidence, the main source for the Times piece is a report of sorts—a letter, really—published by the British Rhinological Society and a group called Ear, Nose, and Throat U.K. arguing that a loss of sense of smell is “a marker of COVID-19 infection.” The letter goes on to explain that anosmia is not uncommon after viral infections, and that in fact “viruses that give rise to the common cold are well known to cause post-infectious loss.” The percentage of people who when infected by coronaviruses like the common cold and COVID-19 then lose their sense of smell is unclear—I checked with a few experts who just said it was a common occurrence, but they couldn’t put a number on it. But looking at it from the other side, various surveys indicate that among folks experiencing anosmia, viruses are thought to be responsible for about a quarter of cases.

In other words, it shouldn’t be surprising that the novel coronavirus could cause smell loss. The question is whether it happens enough to consider anosmia a “clue,” at a time when many Americans don’t have an easy way to find out whether they carry the virus. The authors of the British letter wrote that in South Korea, where testing is widespread, “30 percent of patients testing positive have had anosmia as their major presenting symptom in otherwise mild cases.” In the New York Times piece, reporter Roni Caryn Rabin described these figures in a little more detail, noting that in “a study from South Korea,” “30 percent of some 2,000 patients” experience anosmia. This figure was far and away the most convincing and robust piece of evidence in the piece that the COVID-19 connection to anosmia might be relatively strong.

I wanted to read the study it came from, in order to figure out how reliable that 30 percent number was. I wondered about the methods used for the survey, for example, and whether the study had gone through peer review. I figured it probably hadn’t—a lot of science is now being posted on preprint servers, where authors post findings with little or no outside review, rather than in traditional journals. This is helpful, as knowledge needs be shared quickly right now, but it also means data merits a higher level of skepticism.

I reached out to Rabin, who told me she was relying on information from Claire Hopkins, a nose surgeon in the U.K. and an author of that letter. I reached out to Hopkins, who told me she had, in turn, gotten the figure from a Korean news source, Chosun, which had been translated for her by a colleague. I asked again if there was a study underlying that number that she could point me to. In response, Hopkins told me “we have had real problems with the original link” to the source she’d gotten the figure from. She’s not the only one—a Stat news piece, which ran on Monday, initially stated that the 30 percent figure came from a Nature study, but a corrected version now states that it was a mere estimate, and that it came from a clinician in a March 16 Q&A that ran in Chosun.

Hopkins ended up emailing me screenshots of a Facebook messenger chat with the colleague who had done the translating, now apparently translating from a different source (a link included in the chat went to another news outlet): The actual percentage of patients in South Korea with anosmia, based on a survey from a medical society, was 15 percent. That is half of what the Times reported and a more expected rate of anosmia associated with coronaviruses.

Hopkins wrote me back again to say that it appears the 15 percent figure came from an updated version of the South Korean study, which accounted for patients through Tuesday, two full days after the New York Times ran its original story. It’s possible that the numbers in the letter and Times piece accurately reflected earlier findings. But the bottom line is the same: The rate of anosmia in COVID-19 cases does not appear to be as high as the Times article reported Sunday. Hopkins told me that this new data “reflects what we would largely expect—so likely not sensitive enough to act as a proxy instead of testing.” Though Hopkins still maintains that smell could help target testing or self-isolation, it seems like much less of a clear sign of COVID-19 infection than it did mere days ago.

That a potentially strong signal appeared and then dipped so quickly makes perfect sense in this environment. Data on the novel coronavirus is coming in rapidly, as more and more people are tested. It’s good that doctors are sharing as much information as they possibly can, in real time, via channels that are much faster-response than many traditional forms of science publishing. There was also a secondary point to the New York Times story that remains important, despite this number being off—ear, nose, and throat doctors, along with eye doctors, seem to have been particularly vulnerable to this disease.

But reporting on the prevalence of an unusual symptom like losing one’s sense of smell isn’t just going to reach ENTs. It’s going to reach the rest of us who are stuck in our homes wondering if we have the virus without being able to access testing. We are all so desperate for concrete information that any health story relaying a direct piece of advice, particularly one like “doctor groups are recommending testing and isolation for people who lose their ability to smell and taste,” is going to be taken very seriously. As Thomas Hummel, an ear, nose, and throat doctor in Germany, told the Washington Post: “Many people out there have anosmia. They don’t have a good sense of smell. And if you alert all these people that if you have anosmia you need to stay home, there would be many false positives. That’s the other side of the coin.”

It is still possible that anosmia becomes an important signal and a more prevalent symptom for COVID-19 than for other kinds of respiratory infections; it’s also possible that it ends up being more similar to coughing—a common symptom of COVID-19 but not a clear sign of infection on its own. What we have now is a small inkling of anosmia’s connection to the coronavirus, really meant for doctors and researchers, that has been fast-tracked into a bona fide clue for the public. The doctors in the U.K.—amplified by the Times and everyone who then also wrote articles on smell—have argued that a loss of smell should be a reason to self-isolate for seven days. This is still an applicable directive, since everyone in places where the virus is spreading should be socially isolating as best they can, particularly people who have any symptoms that are coldlike. In a pandemic, you would ideally be able to use any little sign of a respiratory infection to get the piece of data that matters: an actual test. In our current situation, you’re probably just going to have to wait.