Back in April — was it really six months ago? — JK Rowling got Covid-19 (or at least she had the symptoms: she wasn’t tested, because almost no one was getting tested in April). And she credited breathing exercises, recommended by some respiratory doctor, for helping her recover.
I remember at the time thinking “Oh god,” because the exercises themselves sounded a little woo-ish: breathe in five times, holding your breath for five seconds each time, then cough vigorously on the sixth. But, to me, the more interesting thing was it also involved lying on your front: “proning”. Apparently it took pressure off the airways of the lungs.
Like what you’re reading? Get the free UnHerd daily email
Already registered? Sign in
There have been a million short-lived excitements around Covid-19. For a while in the early months people were worried about a correlation with taking ibuprofen; there was the “is smoking protective?” business; there was the incident with hydroxychloroquine, over which we shall draw a discreet veil. Most of them flared up for a moment, then were forgotten: as clinicians and researchers try to do science at speed, there are inevitably going to be a few interesting ideas that go nowhere.
Yet if you were to Google “proning” you would find that it hasn’t quite gone away. A pair of doctors writing in the journal Thorax this month have had a look at the evidence, and found that it might be of some benefit; they cite studies finding apparently large benefits for patients who required intubation and ventilation, and more ambiguous benefits for patients who were awake and not on ventilation.
Join us at our upcoming events in the UnHerd Club, or tune in live online.
Those studies are all small, looking at 20-odd patients each, and I imagine not rigorously controlled or randomised. But according to Rupert Beale, director of the Cell Biology of Infection lab at the Crick Institute, “the evidence is very strong for proning. Doctors in Italy were talking about it having an effect quite early on, and there are good physiological reasons to believe it.”
My point isn’t that proning is some cure-all and if we all go around on our bellies like snakes all day we’ll defeat the virus (“Wriggle for Victory!”). It’s that small and unsexy changes in treatment probably have some non-negligible benefits for a subset of patients. If you were to get the disease now, there is a very slightly smaller chance that you will die than in April, because doctors know that in some cases, it might help to lie you on your front.
In recent weeks, there’s been a little bit of interest in what looks like a decline in the infection fatality rate of Covid-19. To refresh your memory: the “case fatality rate”, CFR, of a disease is the number of deaths divided by the number of confirmed cases. The “infection fatality rate”, IFR, is the number of deaths divided by the number of actual infections. So if 10,000 people get a disease, and you detect 1,000 of them with testing, and then 100 of them die, then your CFR is 10% (100/1000) but your IFR is 1% (100/10,000). Your CFR is dependent on how many people you are testing; the IFR is not, or at least not directly.
It’s very hard to tell what’s going on with the IFR, because, obviously, you don’t know how many people actually have the disease: you only know how many tests have come back positive. And, earlier in the disease, we simply weren’t doing enough tests. We are in a better situation now, and the ONS and others have been using random surveys of the population (a bit like doing an opinion poll, except with nasal swabs and blood samples) to establish what percentage have either the active virus or antibodies at any given time. But it’s still very hard.
Nonetheless, the best guess is that it probably has come down. Adam Kucharski, a lecturer in mathematical modelling of disease at the London School of Hygiene and Tropical Medicine, says that during the first wave in Britain, probably about 1% of all the people who were infected went on to die. Now, he thinks, it’s probably more like 0.5%. “The ONS put it at 10,000 new infections per day a couple of weeks ago,” he says. With an average time between infection and death of about three weeks, an IFR of 0.5% would imply about 50 deaths a day, so we’d expect that next week – and we’re currently seeing a seven-day rolling average of 30, with 70 new deaths announced on Monday, so that’s “sort of the territory we’re getting into now”, says Kucharski.
So, on the face of it, your risk of death if you get infected with coronavirus now is probably about half what it was in March or April. The question is: why?
First things first: it’s almost certainly not because the virus has mutated or anything. “There are some things we know are definitely not true,” says Beale. “We’re convinced that the virus itself isn’t substantially different, that there’s no ‘milder form’ of the virus.” The little package of RNA in its protein-and-lipid wrapper is essentially the same now as it was at the beginning of the outbreak.
And there are some things we know definitely are true. At the moment, the age profile of the people getting infected is much lower than it was in the first wave. “It’s circling more among young people, and less in care homes and hospitals,” says Beale. Young people are much less likely to die than older people, so that simple demographic fact will have brought the IFR down. He thinks that “accounts for the majority of the effect”. Kucharski agrees: he points to Singapore, where there have been about 50,000 cases but only 27 deaths, because it has spread almost entirely among young, healthy, working people. “The IFR is very sensitive to age,” he says.
It probably isn’t the whole story, though. According to Sheila Bird of the University of Cambridge’s MRC Biostatistics unit, there probably has been a “decent reduction in the mortality rate — a 20%, 30% reduction” because of “improved treatment and improved knowledge”.
We all know some of the things which that means. Dexamethasone is the big one, a cheap and widely available drug that can reduce mortality rates in some groups of patients by about 30%. That has been shown in major clinical trials.
But although doctors are, these days, very much evidence-based people who pay attention to clinical trials, it’s not all they pay attention to. Just as they always did, they also pay attention to what works in their clinics, what their colleagues tell them, to what seems plausible given what we know about the human body. “This is what we’re paid to do,” says Beale. “It’s not just massive clinical trials and dexamethasone. It’s about experience, and now we have a lot more of it than we did in April.”
So things like proning are better-known now. Medics are less likely to put people on mechanical ventilation early, because there’s some evidence that can be unhelpful. Earlier use of anticoagulants has become more widespread: “a lot of the pathology is to do with clotting,” says Beale, “and the use of blood thinners is of ongoing interest.”
The triage system is better now — medics will be better at reading the course of the disease, and pushing resources towards those who are most at risk. It’s also worth remembering that although the NHS was not overwhelmed, at one point there was about one ICU nurse for every three patients, rather than the usual one-to-one. “With the best will in the world, that will lead to worse outcomes,” he says. “Mistakes being made, things that would have been spotted 30 minutes earlier are not spotted, and people deteriorate quickly.”
It’s not just about knowing the things to do — it’s about avoiding the things that cause harm, as well. “We were clueless in March,” says Beale. “People are less incompetent now. All sorts of things were happening — vulnerable patients being exposed in daft ways. The mortality among dialysis patients was horrific — something like 30%.” The staff would protect them carefully while they were in the units, “then they’d put them all in the same taxi to go home, all sitting next to each other, with no masks, gloves, anything.”
There’s another possibility, which Beale calls “informed speculation” rather than a well-evidenced fact: the possibility that things like social distancing and mask-wearing have reduced the initial amount of virus that people are exposed to, and perhaps changed the ways they are exposed to it. Your body finds it easier to fight off a small amount of virus than if a huge load is introduced at the same time, and there is, he says, some evidence from animal models that orally ingesting the virus is less deadly than inhaling it. (“That is in hamsters,” though, he warns sternly. Don’t take it at face value.) If that’s true — and, again, it’s just a plausible guess, rather than scientific fact — then, he says, it’s good reason to push for continued use of masks and social distancing: it may be that as well as reducing the disease’s spread, they make the disease a bit less deadly when you get it.
Beale suggests that, in order of importance, the three factors are 1) the changing age profile of patients, 2) improved management of the disease, and 3) this possible reduction in how much virus the average patient is exposed to.
But Bird adds an interesting extra complication: points 1) and 2) are not independent. It’s not random that fewer old people are getting infected. That’s deliberate — or, rather, the fact that so many of them were infected during the first wave was a sad failure. “There will be a shift in the age distribution of those who develop serious disease,” she says, “because we have got better at protecting the most vulnerable.” We have got better at managing the disease, and that includes managing who gets the disease in the first place.
This all sounds fairly hopeful — and it is, to some degree. But there are a few points worth noting. First, a lot of the improvement is a symptom of the UK having done so badly the first time around. Globally, according to Kucharski, the IFR looks like it was about 0.6%. In Britain it seems to have been more like 1%, because the disease was so rampant among vulnerable populations in hospitals and care homes. It’s easy to do better than last time if you really screwed up last time.
Second, death isn’t the only negative outcome. So while death is very rare among younger people, severe disease isn’t; I personally know four people under 40 (two under 30) who got Covid and suffered for months afterward. “If you spend three weeks in ICU, you’re not going to be too happy about it even if you do survive,” says Beale. “It’s possible that you’ll make a full recovery, but a lot of people don’t. And it’s fine telling people that you’re a fit and healthy 30-year-old so you’ve got a 0.01% chance of dying, but your chance of ending up in hospital is much higher, and people don’t fancy spending a week in hospital, frightened and short of breath.”
We shouldn’t downplay the seriousness of this disease even for many people who don’t die of it. The measures and changes we’ve discussed here will reduce the number of severely affected people, as well as deaths, but to what extent is not clear.
And third, says Kucharski, “if you improve your treatment it scales linearly, whereas if you target transmission it scales exponentially.” Let me explain what that means. Imagine you have a disease where, on average, each infected person gives it to three people. (So the reproductive number, R, is three.) And of every 100 people who get the disease, one person dies. (So the IFR is 1%).
Now imagine you have one person with the disease; they give it to three people; they give it to another three each, and so on. By the 10th generation, almost 90,000 people will have been infected*, and about 900 will have died.
Now imagine you develop some clever treatment that can reduce the IFR by 33%, from 1% to 0.67%. You will reduce the number of dead from around 900 to around 600. That’s a very big deal.
But now imagine that you find some clever behavioural intervention — masks, hand-washing, whatever — that reduces your R by 33%, from three to two. Now, instead of giving it to 90,000 people, you’ll give it to about 2,000 people**, and about 20 of them will die.
Small reductions in the IFR lead to small reductions in deaths, and that’s good. But small reductions in R rapidly lead to very big reductions in deaths, and that’s even better. If 0.5% of a million people die instead of 1%, that’s a huge improvement, but it’s still a lot of people. Better, if reasonably possible, that they don’t get infected in the first place.
Healthcare workers, and society at large, really are better at managing the disease now than they were in April. New medical interventions, better protection, and little tricks like proning really are saving lives — possibly even JK Rowling’s, although she’s relatively young and fit so probably would have survived without proning — as, perhaps, is the fact that people are being exposed to smaller doses of the virus in the first place.
So the reduction in the IFR is real, over and above that caused by the change in the demographic profile of people getting infected — and even that is not a coincidence, but the result of better protection of the vulnerable. But reductions in the IFR are not the real story; it’s still the R number we need to care about.
For those who wish to know: