Long COVID data are being erased, again

Listen to Betsy discuss this essay (and other reporting) on our podcast:

When I think back to 2020 in New York City, there’s one moment that sticks out. But it wasn’t in March when the pandemic was declared. It was a few months after, in June.

I had just boarded the subway in downtown Brooklyn: shaken and shivering on an orange plastic seat, heading home from a Black Lives Matter march that the cops had violently targeted. In those early days of the Uprising, there were spontaneous marches every day: I would log off from my data journalism job around 6 PM, grab my bucket drum, and walk the half-hour from my apartment to Barclays Center, where a crowd would inevitably form by sunset. We’d march through the streets until the New York Police Department got too tired or too angry and started to threaten arrests.

That night on the subway — it was a Wednesday — I turned my phone back on, texted my girlfriend that I was heading home. And then I opened Slack. I had to let my fellow volunteers at the COVID Tracking Project know I would be late for that night’s data entry shift.

At that time, the two most important activities in my life were attending marches and contributing to CTP’s datasets: numbers of COVID-19 cases, tests, hospitalizations, and deaths compiled from state agencies by volunteers across the U.S. Specifically, I helped develop the COVID Racial Data Tracker, which gathered data aggregated by race and ethnicity. Our work was cited in reports and op-eds explaining that the pandemic was hitting Black, Indigenous, Latino, and other people of color harder than white people.

While I was too overwhelmed to articulate it at the time, it’s clear to me now that working on this dataset was politically inseparable from my attendance at Black Lives Matter protests. I was both collecting data that showed how COVID-19 widened the cracks in our awful health systems, and marching to support changes — defund the police, and put that money back into communities.

Data is “a major political determinant of health,” the public health scholar Daniel Dawes once told me, in an interview for a story about undercounted COVID-19 deaths among communities of color. In other words: without numbers to show that a crisis is real, it’s harder to argue for solutions, and thus, the crisis continues.

In that vein, the lack of comprehensive Long COVID case counts may be considered one of public health’s greatest failures. But as we enter the sixth year of this ongoing pandemic, U.S. institutions are sweeping this mass disabling event under the rug perhaps more than ever before. We can’t let them get away with it.

What our COVID-19 datasets never counted

It’s perhaps difficult to imagine now, but back in 2020, the U.S. public health system tracked COVID-19 with more resources and rigor than it had tracked any other singular disease before. State and local health agencies stopped other programming and started contact tracing, aiming to follow every outbreak of the novel coronavirus in detail.

“In the first years of the pandemic, we set up surveillance systems that did count every case, which gave us a very detailed picture of where the virus was circulating, who was most at risk, and whether interventions were working,” wrote epidemiologist Caitlin Rivers, a professor at Johns Hopkins University and author of the popular newsletter Force of Infection, in an email to me.

That practice stands in contrast to seasonal viruses like the flu, she said, which “are not directly surveilled”; rather, public health agencies have “systems that estimate burden of disease or give a sense of trends.” Health officials do use contact tracing for other diseases, such as outbreaks of foodborne illness, but COVID-19 led to the practice on a much wider scale.

The detail of COVID-19 reports, combined with mass media institutions following the coronavirus’ every move, contributed to a new public appetite for infectious disease data. I remember how the COVID Tracking Project’s profile quickly grew on social media, with thousands of people attentively waiting for our team’s daily updates.

“Pre-COVID, there wasn’t much of a demand for [public health] data among the general public,” said Samuel Packard, an epidemiologist who has worked on tracking Long COVID. Now, however, “everybody wants access to real-time data on everything all the time.”

Pre-COVID, there wasn’t much of a demand for [public health] data among the general public… [Now,] everybody wants access to real-time data on everything all the time.

Samuel Packard

But many people who wanted that data back in 2020 failed to realize that the numbers were never really accurate. As much as public health agencies tried to count every case, it was a Herculean task: think of all the rural areas with no testing sites for miles, the workers who couldn’t afford to call in sick, the people who refused to pick up a contact-tracer’s call.

And even for those people with COVID-19 who did get counted, most institutions weren’t calling them back in two weeks, in four weeks, in six months to ask if they had “recovered.”

Long COVID cases did not appear on any state COVID-19 dashboards that my fellow CTP volunteers and I used to compile pandemic statistics. In fact, many states even started to spuriously report numbers of people who had “recovered” from COVID-19. “Recovery” had different definitions from state to state, but in most places was simply based on officials assuming that anyone who had not died within a certain timeframe of their infection and/or was released from the hospital was now doing fine.

After a few months of sharing these “recovered” numbers, CTP made the decision to stop publishing them, as my colleagues Amanda French and Quang P. Nguyen explained in a January 2021 article. “COVID-19 can have many long-term health consequences, and none of the definitions for counting people who have ‘recovered’ from COVID-19 account for latent or ongoing health issues that can be caused by COVID-19,” they wrote.

At that time, no public health agency was doing the work to track those long-term health consequences. A few Long COVID clinics had started up, but they mostly focused on people who had been hospitalized; we now know that the vast majority of Long COVID cases happen following mild or asymptomatic COVID-19.

“We felt so invisible at the time,” said Hannah Davis, co-founder of the Patient-Led Research Collaborative, who — like millions of others — developed Long COVID in early 2020. “We knew so early that this was happening to so many people, and it was really destroying people’s lives,” she said. But the statistics didn’t reflect that destruction at all.

Minimizing ongoing risks under Biden

Recognizing the importance of Long COVID data, much of PLRC’s early work focused on trying to track the disease, Davis said. The group’s first study, shared online in May 2020, demonstrated that many people did not, in fact, recover from COVID-19 within two weeks, using data from 640 members of the Body Politic Slack server and other early Long COVID support groups.

Two years later, PLRC members helped develop questions about Long COVID for the Household Pulse Survey (HPS), a Census program designed to evaluate how COVID-19 impacted Americans’ daily lives. Even that seemingly simple step took a long time, Davis recalled.

The HPS helped show that, indeed, millions of people had Long COVID: between 5% and 7% of U.S. adults reported “currently experiencing” chronic symptoms in data collected between summer 2022 and 2024. Of those with Long COVID, around one in four reported “significant activity limitations” from the disease.

But these estimates arrived as the Biden administration and mainstream liberal project to downplay COVID-19 was well underway. The data faced immediate scrutiny from researchers and pundits who insisted Long COVID’s prevalence could not possibly be so high.

At this point, hundreds if not thousands of studies — and a whole segment of the National Institutes of Health’s RECOVER program — have spent precious resources on simply trying to count people with Long COVID. The counts vary widely depending on who is doing the counting and their criteria: which symptoms are included, the time period from infection, whether a positive test is required, and so on. It’s easy for any government official who might want to downplay the ongoing pandemic to pick and choose the lowest option.

“There was not only a lack of surveillance. There was an active attempt to not survey,” Davis said. “If [government agencies] don’t have numbers on correct Long COVID cases, that’s easier for them to manufacture that the pandemic is over.”

And indeed, Long COVID case counts, even inaccurate ones, never made it onto the official COVID-19 dashboards. The Centers for Disease Control and Prevention (CDC), rather than update pandemic metrics to include appropriate warnings about long-term health damage, has continually shifted its data presentation over the last three years to make COVID-19 appear less threatening than it truly is. Remember when the CDC changed its threshold for “community levels,” turning its map of the U.S.’s COVID-19 risk from bright red to muted yellow and green overnight?

These data choices have helped government officials, pundits, and journalists make endless SARS-CoV-2 infections seem like no big deal. Mainstream media coverage of the disease these days tends to repeat that hospitalizations and deaths are lower than they were at the “height of the pandemic”; the risk of Long COVID is rarely mentioned.

If [government agencies] don’t have numbers on correct Long COVID cases, that’s easier for them to manufacture that the pandemic is over.

Hannah Davis

Intensified data chaos and uncertainty under Trump

Our public health system was never set up to count new cases of chronic illness following viral infections. I wonder how things might be different if the CDC and state health departments had listened back in 2020, when people with myalgic encephalomyelitis (ME) tried to warn the world about Long COVID. Would we be more prepared now to advocate for the millions of people who are sick?

It’s impossible to know, but the progressive failures of the last four years are clear. The Biden administration set us up for a new era, under Robert F. Kennedy Jr. as health secretary, in which no data can be trusted and any collective health measure is labeled “more dangerous” than the disease it’s trying to prevent.

Immediately upon taking office in January, Trump officials called for a freeze on public communications from federal health agencies. Many data pages weren’t updated on schedule, and the CDC halted publication of its flagship scientific journal. A week later, the administration ordered agencies to remove any references to “gender ideology” — meaning, any acknowledgment of transgender people — from public websites.

Since Trump’s second term started, the CDC “has removed at least 143 datasets and other files from its public data platform,” according to an analysis by STAT News. Many of those files were removed specifically because they use the term “gender” or capture data on gender identity, STAT found.

Disruptions included Long COVID data: the CDC’s page reporting results from the Household Pulse Survey’s Long COVID questions was taken down. It’s now back online, but bears a note in all caps stating that the page is “archived for historical purposes and is no longer being updated.”

Even brief interruptions for datasets like this cause challenges and uncertainty for the people who use them, Packard said: “These data are produced as a public good, and they should be made available to many different types of people, from the general public to researchers, journalists, students, and everybody in between. When they are taken down, it disrupts that whole ecosystem.”

Unfortunately, for the HPS, the interruption was not temporary: Packard also found, in late February, that the survey may have ended its collection of Long COVID data. The Census revamped this survey in fall 2024, and according to results files it shared online last month, the Long COVID questions are no longer included. In addition, those Census files no longer include any data on respondents’ gender identity.

Those Long COVID prevalence data were a vital resource for researchers and advocates seeking to establish the disease’s vast impact. The HPS not only showed that Long COVID impacted millions of Americans; it also revealed that the disease hit already marginalized groups, like transgender people and those with prior disabilities, the hardest.

The survey’s changes were determined before Trump took office. But they add to a broader climate of chaos for anyone seeking to collect and use health data under this administration. Packard expressed concern for how these changes further erode public trust in institutions: “They modify their data sets or take them offline, all of a sudden, you have career epidemiologists who usually say, ‘Go to the CDC’ … Now even that group of people is going to say, ‘You can’t trust the CDC’s data.’”

In response to questions about HPS, a Census spokesperson pointed The Sick Times to existing webpages. A CDC spokesperson noted that the agency continues to collect Long COVID data through other surveys; however, these are conducted only once a year.

At the same time, the Trump administration has ended Long COVID initiatives, illegally fired thousands of federal workers, pulled funding for research and international aid, and called basic health measures into question. RFK Jr. has spread misinformation about measles vaccines during a record outbreak that led to America’s first pediatric measles death in a decade.

Epidemiologists like Rivers are concerned that Trump’s actions could dismantle key disease surveillance systems and leave the country unprepared for new outbreaks. “Those systems are our visibility into health threats,” she said, including seasonal diseases, contaminated food products, and novel pathogens.

Pushing back against erasure

For now, the ship has likely sailed on accurate Long COVID numbers. Everyone has had COVID-19 multiple times, yet most people aren’t testing for it or tracking their symptoms afterward.

The federal government has clearly abandoned us — us being people with Long COVID and related diseases and disabilities, and their allies. Death Panel podcast cohost Beatrice Adler-Bolton told me in November: “The state will not save us. If that’s not something you got from the last four years, you should certainly be getting it from the incoming administration.” 

So, I think the best question to ask is: how do we save ourselves? There are hundreds of answers to that question around the U.S. and globally, in the form of mask blocs, air-purifier lending libraries, Long COVID support groups, COVID-conscious meet-up groups, and more. All these organizers are filling in the gaps left by widespread government failures.

But collecting public health data is harder to do at a small-scale, community level than other efforts like distributing N95 masks. “The infrastructure needed to do something, at scale and correctly, is really, really hard,” Davis said.

Restoring Long COVID questions in the HPS, or another federal survey with similar frequency, may be a worthwhile advocacy goal going forward. Advocates could also consider how Long COVID data collection might be tied to other existing programs; one model is the recent requirement from the Centers for Medicare and Medicaid Services that hospitals must report COVID-19, flu, and RSV hospitalizations to receive Medicare and Medicaid payments.

Groups could also advocate for state and local health agencies to track Long COVID in their jurisdictions. New York City’s health department has done this: one 2022 survey by the agency found that a staggering 80% of adult New Yorkers who had COVID-19 “experienced at least one symptom lasting one month or longer.” Minnesota’s state health department similarly supports a Long COVID research and education program.

There are other ways community groups can get involved with tracking Long COVID and sharing their findings. For example, PLRC recently published a detailed fact sheet about the disease, citing information from 80 scientific papers. The fact sheet aims to “aggregate the data we have in narrative form,” Davis said.

Plus, look at the datasets and reports of CrunchME, a new patient-led organization dedicated to compiling and sharing statistics about ME, Long COVID, and other infection-associated chronic conditions. “Accurate statistics are vital for speaking the language of policymakers, and pressing our case as compellingly as possible,” CrunchME’s founder, Rory Preston, wrote in a recent essay for The Sick Times.

And, I would argue, we don’t need vast government resources to check in on our comrades: to do “long-term follow-up,” as Davis put it. Just as organizers might ask people to wear high-quality masks or test before a protest, what about asking people who report they’ve gotten sick how they’re doing two weeks, four weeks, six months later? What about offering to bring them groceries, to help them get doctor’s appointments, or apply for disability benefits?

Back in 2020, the millions of people who didn’t “recover” from COVID-19 were missing from our detailed datasets. But rather than fix this issue — or address any of the other myriad societal chasms widened by the pandemic — our governments gave more money to the police. Some of us who joined the marches in 2020 are still outside and still making spreadsheets, demanding change.

Some of us who joined the marches in 2020 are still outside and still making spreadsheets, demanding change.

---

Editor’s note: The Patient-Led Research Collaborative, like The Sick Times, has received support from the Balvi and Kanro funds. Our newsroom operates independently of financial supporters.

The Sick Times will continue following Long COVID-related news from the Trump administration. Send us tips at editors@thesicktimes.org, or reach out to Betsy on Signal @betsyladyzhets.25.

All articles by The Sick Times are available for other outlets to republish free of charge. We request that you credit us and link back to our website.