Scientists have been scrambling to understand long Covid ever since the first SARS-CoV-2-infected patients reported persistent symptoms in early 2020.
The condition includes a wide range of symptoms new or worsened after a SARS-CoV-2 infection that last at least three months post-infection and affect a person’s quality of life. A recent estimate suggests about 65 million people worldwide are affected.
Thousands of long Covid studies have hit the preprint servers over the past few years, and literal reams of new knowledge have emerged. But despite the research interest, disentangling long Covid’s symptoms from those of other conditions like chronic lung diseases, autoimmune syndromes, and neuropsychiatric disorders remains a challenge, and scientists still don’t understand who’s at greatest risk for developing long-term health problems after a Covid-19 infection.
In the last few months, the first in a series of studies from the Inspire group (Innovative Support for Patients with SARS-CoV-2 Infections Registry) have been published, and offer painstakingly collected data aimed at answering big questions about long Covid.
Inspire — a CDC-funded collaboration among eight US academic medical centers — is in many ways better designed than many other long Covid studies. However, while some of the study’s findings are instructive, others are confusing. Some even appear to challenge the reality that people with long Covid face — for example, by suggesting that people who didn’t catch Covid had more persistent, severe fatigue than people who were infected.
These new studies help us understand long Covid better, even if they’re not entirely neat. And looking at them closely can help us understand why long Covid — a condition tied to health inequities — is so hard to study, especially in a country with some of the worst health inequities in the world.
It can also help us understand why good studies sometimes yield perplexing results, and what scientists can and should do when that happens.
The Inspire study originated in 2020 as an effort to understand the lived experience and broader impact of Covid-19. When its investigators noticed other studies suggesting different Covid variants had different risks for long Covid, they used the study data to try to answer why and other tough questions about more persistent symptoms.
A problem with many long Covid studies is that some long Covid symptoms — fatigue, headache, and chest pain — are pretty common and can be caused by other conditions like migraine, asthma, depression, and anxiety, or environmental factors like poor air quality. It can be tricky to figure out if a patient has persistent exhaustion because of a recent Covid infection or if something else is going on.
The Inspire study organizers tried to account for this. They decided any adults who had symptoms of Covid-19 would be eligible for their study — both those who tested positive for Covid-19 and those who tested negative for the infection. This design would ideally help isolate Covid-19 as the cause of even the most common symptoms. (They ultimately enrolled 4,113 Covid-positive people and 1,362 Covid-negative people, all tested between December 2020 and June 2022.)
The study design included a few other features intended to avoid some of the problems baked into other studies.
For starters, rather than using data gathered from electronic medical records, the study uses data gathered from surveys administered directly to participants. “Electronic health record data is often relatively restricted,” said Michael Gottlieb, an emergency medicine doctor at Rush University who co-led the study. Information about key symptoms is often missing, depending on what questions clinicians asked each patient and what each patient reported to their provider.
The survey the Inspire group administered also asked about a much broader range of symptoms than most clinicians would ask about, including questions about sleep quality, ability to focus, and hair loss. Additionally, it asked about these issues using questions that have been proven by the National Institutes of Health to be really good at accurately capturing symptoms often reported by people with chronic illnesses.
Another promising feature of the study is that it gathers data going forward from the moment of enrollment, rather than going backward from that moment (a design technically called “prospective”). This makes it less likely that a symptom will be unreported or reported incorrectly because a patient has forgotten the details.
All told, Inspire was well-positioned to deliver answers. Did it?
So far, three papers from the Inspire study have been published since December 2022. Some of the findings they detail jibe with what many clinicians are seeing in long Covid clinics.
Results the group published in January showed that while more people with Covid developed persistent symptoms if they were infected before the delta variant emerged, it wasn’t because of differences between variants. Rather, it was because of social or demographic factors, like preexisting conditions, hospitalization for Covid, and race and ethnicity. Additionally, they found vaccination was protective against developing long Covid after infection. Researchers had suspected that preexisting conditions and vaccination played into long Covid risk. So this all made sense.
Other findings emerging from Inspire are more confusing.
The big one being: People in the control group — that is, the people who never had Covid — had higher rates of symptoms related to long Covid.
In the January study, people who tested negative for Covid actually had higher rates of severe fatigue three months after testing than people who had Covid. They also experienced many other symptoms — like fever, headache, runny nose, and sore throat — at rates comparable to or higher than Covid-positives.
Also confusing were findings detailed in an earlier publication by the study group, which reported poor well-being more often among Covid-negative patients than among people who tested positive — 54 percent compared with 40 percent.
What to make of this?
It’s possible Covid diagnostics are part of the problem. When it comes to Covid tests, “the chances of a false negative are actually quite high, especially if you look at patients that are symptomatic,” said Alba Azola, a physical medicine and rehabilitation specialist at Johns Hopkins University who specializes in long Covid care, and false-negative rates can be as high as 38 percent. That means a large number of people might have been incorrectly assigned to the non-Covid control group when they actually did have the infection.
This is a possibility the authors acknowledge in the publication, although there are other possible reasons for this finding, said Gottlieb.
One being, other infections can produce “long” symptoms similar to long Covid. “Just because they didn’t have Covid doesn’t mean they couldn’t have had another infection that caused a post-infectious syndrome,” he said. After all, they were tested for Covid-19 because they were symptomatic in some way.
Another possibility: More people in the Covid-negative group dropped out of the study before it was completed than in the Covid-positive group. If they did so because they felt better sooner, that could have skewed the results.
Other differences between the Covid-positive and Covid-negative groups could also lead to differences in outcomes: People without Covid were less likely to be female, white, and have private health insurance. It’s possible these characteristics track with other life experiences that raise the risk of severe fatigue and other symptoms. (Gottlieb noted that the latest study was not set up to directly assess this possibility.)
And there’s yet another, perhaps more uncomfortable possibility: Our medical system under-recognizes that fatigue and other symptoms exist at high rates, even when they’re not related to Covid.
Is there a way to avoid these pitfalls? An ideal study would be one that enrolled all Covid-positive patients in a defined area at the moment of diagnosis — without a selection process — as well as asymptomatic controls, and called all of them a few months after infection, said Jeffrey Martin, an infectious disease doctor and epidemiologist at the University of California San Francisco (UCSF).
That study hasn’t been done. The closest anything has come, Martin said, was a study the Long Beach, California, health department conducted the first year of the pandemic, before widespread home testing led so many Covid-positive people to go unidentified by local health departments.
The investigators randomly selected adults from a list of all positive Covid tests reported to the county and called them to ask whether, two months after their result, they had any persistent symptoms. One-third did.
Big challenges to long Covid research remain, and will only grow.
“Studying long Covid has gotten harder over time,” said Michael Peluso, an infectious disease doctor and researcher who studies long Covid at UCSF.
When scientists first began researching long Covid, the virus was still relatively new, Peluso explained. Everyone was at risk of catching SARS-CoV-2, reinfections hadn’t happened yet, nobody had been vaccinated, and there were no treatments. All of that’s now changed — and because it’s not clear which factors affect long Covid, scientists need to think about all of them.
What’s a scientist to do when a new study about a poorly understood disease contains confusing findings? Several experts I spoke with said the answer is to look more closely.
When study results don’t make sense, said Peluso, sometimes “those observations come to be because of some issue with how a study is designed or recruited or recorded, and that requires a lot of thought and interpretation.”
The American health care system adds a layer of complexity that impedes research: Because our medical records are not held centrally — the way our tax records are — it’s a much bigger lift to do population-wide research in the US compared with countries like the United Kingdom and Israel. Instead of gathering anonymized information from a database that includes everyone who gets health care, researchers have to enroll people in studies.
That introduces bias into the data, said Peluso, because in the US, “research includes people who choose to participate in research” — but often, it doesn’t include people who can’t get time off to participate in a study or can’t afford to travel to a study site. That leaves lower-income and rural people — who are often also racial and ethnic minorities — less likely to be represented in clinical studies.
Recruiting issues aside, the confusing findings could be an important clue.
“This happens all the time — we make surprising observations that turn out to totally upend a field,” Peluso said.
Unexpected results require clinicians to ask themselves: “‘Does this make sense with what I am seeing as a clinician every day interacting with people?’” said Peluso. “If it doesn’t, it doesn’t necessarily mean that it’s wrong — it just means that we need to give it a little bit more thought and understand why.”
Other ongoing studies will hopefully fill in some gaps. Several large long Covid studies, including the National Institutes of Health’s Recover and N3C studies, continue to examine questions about what long Covid looks like and who’s most at risk. There are also large studies conducted using the massive medical records database of the national Veterans Health Administration medical centers; they’re not perfect, said Azola — “we’re looking at old white men” — but their ability to look at a large group before and after Covid-19 infection is instructive.
Meanwhile, patients with long Covid may not be following Inspire’s results as closely as some researchers. “No one cares,” said Diana Güthe, founder of the long Covid advocacy group SurvivorCorps. The study just isn’t about the questions that matter, she said: How does it “offer any help to a patient who is suffering, out of work, unable to care for their children, their parents, unable to pay their mortgage, chasing doctors who are gaslighting them, in severe physical pain?” she asked.
What people with long Covid really want are clinical studies of treatments for the condition, and studies that show whether repeat infections even after vaccination raise the risk of developing persistent symptoms, said Güthe. Gottlieb said his team is working on additional analyses to answer the latter question and several others.
While some patients may not see the utility of this research, clinicians and scientists say it’s important for establishing better understanding about long Covid. And while the Inspire study isn’t perfect, it’s still much better than most long Covid studies, said Martin — few even attempt to include a control group.
“People want answers,” said Peluso. But it’s important for the world to see the complexity of scientific discovery, he said, because that’s how science works. Most people experience science as something that’s already proven right, “but before something makes it into a medical textbook, it is very messy,” he said.
<div class="c-article-footer c-article-footer-cta" data-cid="site/article_footer-1677958732_9568_50577" data-cdata="{"base_type":"Entry","id":23384187,"timestamp":1677931200,"published_timestamp":1677931200,"show_published_and_updated_timestamps":false,"title":"Why it’s so hard to get answers on long Covid","type":"Article","url":"https://www.vox.com/science/23620146/long-covid-research-study-inspire-bias","entry_layout":{"key":"unison_standard","layout":"unison_main","template":"standard"},"additional_byline":null,"authors":[{"id":8904217,"name":"Keren Landman","url":"https://www.vox.com/authors/keren-landman","twitter_handle":"landmanspeaking","profile_image_url":"https://cdn.vox-cdn.com/thumbor/7IMc-WP8yf8IR8BnPY5Ply-CL7o=/512×512/cdn.vox-cdn.com/author_profile_images/195227/IMG_3168.0.jpg","title":"","email":"","short_author_bio":"is a senior reporter covering public health, emerging infectious diseases, the health workforce, and health justice at Vox. Keren is trained as a physician, researcher, and epidemiologist and has served as a disease detective at the US Centers for Disease Control and Prevention."}],"byline_enabled":true,"byline_credit_text":"By","byline_serial_comma_enabled":true,"comment_count":0,"comments_enabled":false,"legacy_comments_enabled":false,"coral_comments_enabled":false,"coral_comment_counts_enabled":false,"commerce_disclosure":null,"community_name":"Vox","community_url":"https://www.vox.com/","community_logo":"rnrn rn