Long Covid: How Bungle Reporting on a Thorny Topic
A lesson in moral panics doing a disservice to all
Are we in a moral panic about Long Covid that may actually harm the fight against this very real problem? Observing the coverage over the last few months, I fear that this may well be the case.
Let me start by saying that I do not at all dismiss the threat of long COVID—lingering illness months after the initial acute phase ends. Such post-viral sequelae have been observed for many other pathogens, and it’s well-documented that there is a cluster of debilitating conditions we call myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) in a large number of sufferers who deserve―above all—respect, attention and resources.
I also don’t like the “it’s real and caused by a pathogen” versus “it’s in your head” or “is it psychosomatic” type binaries because... Well where is your mind, if not in your body? The sooner medicine gets rid of the mind/body duality, the better. (Whole other topic: personally, I think the remarkable efficacy of the placebo is among the most interesting and understudied topics in the world of medical science). However, the reality of post-viral sequelae—and the importance of studying it properly—makes the current shoddy science and sensationalist reporting around long COVID especially unfortunate.
To understand how such moral panics work, take the coverage around this preprint that followed up some patients with confirmed COVID positive tests in a hospital system to see how they were doing on day 61.
Here’s how the article summarizes the findings in its lede:
Many people who experience long-term symptoms from the coronavirus did not feel sick at all when they were initially infected, according to a new study that adds compelling information to the increasingly important issue of the lasting health impact of Covid-19.
The study, one of the first to focus exclusively on people who never needed to be hospitalized when they were infected, analyzed electronic medical records of 1,407 people in California who tested positive for the coronavirus. More than 60 days after their infection, 27 percent, or 382 people, were struggling with post-Covid symptoms like shortness of breath, chest pain, cough or abdominal pain.
Nearly a third of the patients with such long-term problems had not had any symptoms from their initial coronavirus infection through the 10 days after they tested positive, the researchers found.
Troublesome, right? Even asymptomatic infection seems to potentially create “Long Covid” sufferers in large proportions.
In fact, I saw this article shared widely, with people worrying that we might never get back to normal life even after vaccines. If even asymptomatic, silent infections can cause Long Covid so frequently, how can we do anything, ever?
Here’s the problem though—and these are things you’d never know if you read only the New York Times or other articles about the study, rather than the study itself. The findings were drawn from electronic health records: i.e. people who were interacting with the hospital system. So the study necessarily excludes anyone who didn’t feel the need to interact with the hospital after their positive test. Who is still interacting with the hospital sixty days after their diagnosis? People who don’t feel great. Who is not? People who don’t feel the need to because they feel fine. Hence, clearly, this is not a representative group whatsoever. These are people who sought follow-up, by definition. Their problems are real, for sure, but they don’t form a basis from which to report percentages, really. This is called a “selection effect” — if you only include people who select themselves into a group, you have little idea of what’s actually going on besides some people selecting themselves into that group.
At most, then, one could say some people who were classified as asymptomatic by this study reported some symptoms sixty days after diagnosis.
But wait, there’s more. Were they really asymptomatic? It’s really not clear one can conclude that from the data they have. As many researchers who work with such hospital records pointed out, it’s unwise to assume that health records alone can determine that their illnesses were completely asymptomatic throughout. Absence of symptoms strong enough to seek medical care—and thus missing from the record—is clearly not identical with being a confirmed asymptomatic case. The way “asymptomatic” is being categorized here comes down to: “The hospital didn’t have an interaction where the person came to the hospital and said they had symptoms in the first ten days (study timing) after their test—therefore they were asymptomatic.” (If they did anything else besides rely on electronic health records retrospectively, this is not reported in the study preprint). That’s really not a strong assumption.
Maybe these people got tested, went home and never interacted with the hospital to say they had mild fatigue or other symptoms? Being asymptomatic at the time of the test is no guarantee that the person wasn’t presymptomatic—developing mild symptoms later that were real but didn’t didn’t require medical care.
I suppose we could say that 32% is the ceiling of the group who were asymptomatic, but it’s quite likely that the real number is substantially lower.
So we have a group of people who were indeed PCR positive but are not at all representative. They may or may not have been asymptomatic. But we’re told they were “long haulers” sixty days after their diagnoses.
But wait, it gets even worse.
If you actually read the study, here’s the symptom cluster on day 61:
Symptoms that are prevalent among long-haulers include (descending order): chest pain, dyspnea, anxiety, abdominal pain, cough, low back pain, and fatigue (See Figure 2B)
Those are real issues for sure, but the question remains: compared to what?
A clear indicator of when things start sliding into moral panics is when baselines and comparison groups disappear from the narratives.
You see this, for example, in the sometimes breathless coverage of dangers children face from tremendously rare threats like stranger abductions. Such abductions are unthinkably horrifying, but they are extremely rare. And yet they seem to have a disproportionate hold on our imagination, if the media coverage is to go by. Conversely, routine but prevalent dangers to children—such as being driven in a car, basically—receive much less attention.
We saw these problems with the coverage of Multisystem inflammatory syndrome in children (MIS-C) and COVID-19, too. In a nutshell, this is an extremely serious but also extremely rare condition among children. It may have affected a few thousand children per year, perhaps half of which linked to COVID. The syndrome may also be associated with other pathogens, and it was already happening before the pandemic—we’re just paying more attention now. The condition resulted, tragically, in 33 deaths that we know of. Most children recover completely. Early on in the pandemic, it made sense to worry about this because we didn’t know how prevalent it might be, or what kind of danger it posed. A year on, though, we have a clearer picture: it is a serious but extremely rare condition that is not solely linked to COVID. It’s something that pediatricians and emergency rooms should be educated about so they don’t miss cases. It’s especially concerning that the syndrome is more prevalent among children of color.
But given that MIS-C is not even clearly linked only to COVID—that it exists as a rare baseline threat to children—why does it come up at all in the school opening debate? Because we’re scared, understandably. Childhood illnesses are among the worst fears of any parent. It can be hard to contextualize the fears, although clearly, for example, pediatric influenza is a much bigger threat in normal years to children than MISC-C. There are 192 known pediatric influenza deaths in 2019. Each one tragic, but they are still relatively rare as a risk to children compared to traffic accidents which are, sadly, the leading cause of death for kids followed in the United States by firearm-related deaths, horribly.
Comparison groups are also important to understand the scale of a problem. To put it bluntly: it’s not very informative to describe people as COVID “long-haulers” if, sixty days after being diagnosed with an illness that is sweeping the world in a devastating pandemic, they report anxiety, fatigue, insomnia, headache etc., without a comparison group!
Let me repeat that I do not at all dismiss anxiety or any of its related symptoms. But a finding that a 15% of a (non-representative) cohort of patients who tested positive for COVID-19 (who may include some who were asymptomatic but that’s not clear) is not a big finding unless it includes a proper comparison group: What percent of people without any COVID diagnosis were feeling anxiety—which often comes with insomnia, fatigue, a feeling of shortness of breath (i.e. dyspnea—one of the top symptoms listed), etc during the same period? What percent of people diagnosed with any potentially serious illness—let alone one during a pandemic—report anxiety and related symptoms two months after diagnosis?
Besides, why wouldn’t we expect some people to feel anxious two months after they were diagnosed with an illness that’s causing a pandemic? Does this mean it’s the aftereffects of the infection itself—post viral sequelae—that is causing the symptoms, especially the anxiety and its related cluster? Or is it understandable anxiety from the conditions we’re living under? That would be a thorny problem in and of itself. But without a comparison group or a baseline, it’s hard to know what to say, if anything.
Despite all these issues, this preliminary study received a significant amount of coverage, making it even to People Magazine. It was amplified by public-health authorities like Kaiser Health Newsand the American Journal of Medical Care, and picked up by, among others, the Sun, a British tabloid. As the New York Times story shows, the coverage rarely included the caveats, context and the weaknesses we’ve discussed.
I’ve already seen worries expressed based on this coverage that vaccines are not a panacea because people will get infected asymptomatically and then end up with Long Covid anyway. Maybe that’s something to worry about, but not really based on this particular study as it stands.
Where does this leave us? Confused.
Most importantly,, all this does a major disservice to Long Covid sufferers. Without proper studies, we cannot figure out what’s happening, let alone start addressing it.
Unrepresentative groups drawn from incomplete, self-selected sections of the population tell us something may happen, maybe. But that cannot inform us about the scale and the scope of the issue.
Such sensationalism and reporting on preliminary studies without understanding what the study can and cannot represent or assert does a major disservice to the anxiety and anguish that this disease has caused. It furthers the mind/body duality that haunts medicine, implying that physical suffering is real and worthy of consideration only if we can directly tie it to a pathogen.
It’s possible for viral infections to create persistent, and even debilitating conditions, in some portion of people infected with them, believe that we should devote resources to studying this carefully, and it’s also simultaneously possible for stressful, terrible conditions of a pandemic to contribute to anxiety, headache, insomnia etc. and many of the rest of the symptoms identified here. We don’t need to dismiss either dynamic to consider both to be important and worth discussing. But the worst way to do that is to put them into one badly-measured basket and draw conclusions unsupported by what data we do have. Unfortunately, this may well be what’s happening, to the detriment of everyone.