By Liam Mannix
Examine, a weekly newsletter covering the latest developments in science, is sent every Tuesday. Below is an excerpt – sign up to get the whole newsletter in your inbox.
Long COVID is an awful thing. It has put scientists in wheelchairs, struck down fit people with chronic fatigue, caused nerve dysfunction, heart and lung problems, headaches and brain fog.
Some estimates are that up to 54 per cent of those infected with COVID will develop long COVID, but that number is highly contentious – and estimates are all over the place.
18 year old Payton Jacobs with physiotherapist Anne Tanner at NSW’s first long COVID clinic.Credit: Louise Kennerley
“A lot of it is pretty dodgy,” one scientist told me. “The high estimates really don’t stand up to even basic anecdata.”
Hitting the bullseye
The best local data we have comes from a 2021 Lancet study of 2904 Australian cases which found that three months after infection, 5 per cent of patients surveyed said they had not recovered.
UK national survey data reports that 2.8 per cent of the entire population (!) had self-reported long COVID in May. Among the general US population, it’s 36 per cent – or maybe 23.2 per cent. A systematic review – supposed to be the cream of the research crop – published in the American Medical Association’s open access journal JAMA, put the number at 54 per cent.
It’s... a bit of a mess. Why?
First, because science is accumulative.
Imagine the “true number” for long COVID risk is a red bullseye, says University of Sydney epidemiologist Professor Alexandra Martiniuk. Each attempt at measuring that number is like throwing a dart.
Imagine the ‘true number’ for Long COVID risk is a red bullseye, says University of Sydney epidemiologist Professor Alexandra Martiniuk. Each attempt at measuring that number is like throwing a dart.Credit: James Davies JGD
“When studies start, they are all over the place, the arrows land in the bushes,” she says. Eventually, study problems will be fixed and larger numbers of people will be included. “And over time, you get closer to the truth.”
Problem two: scientists are yet to fully agree on a definition of long COVID, in part because it presents so differently in different people (one study included 115 different symptoms).
Problem three: there’s no consensus across studies of how “long” long COVID is. Is it symptoms that last two weeks? A month? A year?
Problem four: many studies ask people to self-report their symptoms rather than objectively measuring them. Some people are less likely to notice and report symptoms than others – think of your grumpy old uncle who refuses to ever go to the doctor.
And the final, big problem: the lack of a control group. Without one, we can’t tell whether the persistent symptoms are caused by COVID or something else. “If you survey anyone, some are going to say they feel fatigued,” says Associate Professor Bette Liu, an epidemiologist at the University of NSW.
Martiniuk’s perfect study of long COVID would look like this: randomly select a few thousand people who have never had COVID and follow them over several years. Then compare the people who get long COVID to those who get the virus but don’t get long-term symptoms and, voila, you can work out the true risk.
Three years and a bunch of new variants into the pandemic, such a study is now impossible. That means our data is necessarily imperfect – making it hard to know if we’ve ever truly hit the bullseye.
Let’s return to the Lancet’s 5 per cent estimate of local prevalence, which has made its way into a lot of media reporting. That estimate is based on self-reported symptoms and there’s no control group. As such, “it would be reasonable to conclude that ‘real’ prevalence in the population may well be less than that,” says Liu, who was first author on the study.
A study hot off the presses in Nature tries to solve many of these problems. It concluded that three months after infection with COVID, 4.9 per cent of people reported at least three symptoms, compared to 4 per cent of people who had never been infected.
”I think,” says Martiniuk, “that we are a long way off the bullseye.“
One disease or several?
There’s another potential problem: what if we’re not looking at one condition?
The wide variety of symptoms reported hint that there might be multiple different disease processes going on, says Martiniuk. This may explain why different studies come to different conclusions: they’re actually looking at different diseases.
“You don’t try to understand breast cancer prevalence by looking at pancreatic cancer prevalence,” she says.
It also seems at least plausible that a proportion of long COVID is really just recovering from a nasty illness.
COVID can leave us badly damaged. So can its cure. An intensive care unit in general is not a place you ever want to be. Patients in ICU lose a huge amount of muscle; their lungs need to recover from having a machine breathe for them. Many are left with psychological damage. Some 25 per cent of patients in ICU – this is without COVID - will get “Post-intensive Care Syndrome”, according to one estimate.
Now consider this large study published in peer-reviewed medical journal PLOS Medicine. It found 73.2 per cent of people who were in intensive care for COVID ended up with long COVID.
”Some amount of long COVID is just what we’d expect from people who’ve had an ICU admission,” says University of Wollongong epidemiologist Gideon Meyerowitz-Katz.
Where does all this land us? To be honest, I’m not sure.
None of the estimates we have seem likely to be very accurate – although every expert I spoke to was leaning towards the true rate being low, not high.
At this stage, the true risk of long COVID seems to still be a pandemic mystery.
The Examine newsletter explains and analyses science with a rigorous focus on the evidence. Sign up to get it each week.