Exactly how dangerous is the coronavirus?

On Monday, a group of researchers from Imperial College London published important new results on the virus du jour in Lancet Infectious Diseases. Imperial College’s epidemiologists have the close attention of the world right now. If you are following virus news, and what other kind is there, you’ll recall how Imperial College scholars issued a memo that diverted the U.K. from following a course of grin-andbear-it antivirus policy involving little or no behaviour modification by the British public.

Almost every developed country is more or less following the strict Imperial College-recommended path now, and the institution has become a source of reassurance, insisting that social distancing and the outlawing of mass assemblies are working. In another report also issued Monday, researchers estimated that behaviour changes have already saved 59,000 lives in 11 Western European countries. This number, like everything else about a young epidemic, is likely to grow with quasi-exponential force.

The Lancet paper represents the best possible attempt, right now, to quantify the severity of COVID-19. This is a notoriously difficult trick in the early stages of a new epidemic disease. Estimates of case-fatality rates, and other parameters of urgent interest, can start out wildly high because early surveillance is identifying only the most severe cases. Those estimates require a lot of infected people, including ones who have experienced the full course of the disease, before they firm up.

Even then, you are likely to have started out with most of your data coming from only one locale: it might be a place, perhaps a province of China, whose reporting you do not entirely trust. And even if you do trust it, it might be a mistake to extrapolate it to other populations. The Imperial College researchers used data from Hubei province and cross-checked it with cases from other places, such as Hong Kong and Macau, and from the Diamond Princess cruise ship, which produced very detailed data from a group of 3,711 unlucky guinea pigs.

The resulting highly educated guesses are still pretty loose. Their point estimate of the overall case fatality ratio (CFR) — the probability of a random person dying once they have a confirmed case of COVID-19 — is 1.38 per cent, with a 95 per cent confidence interval ranging from 1.23 to 1.53 per cent. The estimated infection fatality ratio (IFR) — the chance of dying if the virus gets in you — is 0.657 per cent, with a confidence interval of 0.389-1.33 per cent. You can see that it’s not yet possible to present an estimate like this without adding, “but it might be half that, or double.” (Their IFR point estimate is about half of their CFR, which concurs with Icelandic testing data suggesting that about half of the infected may never experience symptoms.)

Perhaps the most interesting results in the paper are the estimates of the proportions of infection requiring hospitalization, although these are still derived from a very small universe of cases. As with the IFR estimates, these have wide confidence intervals ranging from about half the mean to double. The hospitalization-on-infection rates are given only for particular age groups. For children aged zero to nine, the estimate is literally zero. For those 10-19, it is 0.04 per cent, or one in every 2,500 infectees. It rises to about one per cent for those in their 20s, 3.4 per cent for those in their 30s and so on up the ladder until it reaches 18.4 per cent for those aged 80-plus.

These numbers are important because preventing hospitals from becoming overrun with coronavirus is a paramount goal of any policy we might adopt. They are, on their face, bad news for the hope that we can exempt young people from behaviour restrictions very soon and let them take their own chances with the virus. If Canada infected everybody aged 20-29 with the virus simultaneously, we (or Imperial College boffins) would expect 53,000 hospital cases just from that group. In the real world some would end up having a herd-immunity shield, or just getting lucky, but the hospital burden could still be an enormous fraction of the 53,000.

The good news is that the effort to derive various mortality and morbidity rates will not stop. With every day that we remain in limbo, more countries gather more data and we should soon be able to guess at these parameters with adjustments for factors other than age. Chronic illness, body mass index and lifestyle variables all seem likely to be involved and could allow for stepped reintroduction of less vulnerable groups to ordinary existence. (There are hints that even ABO blood groups could be a factor.) This is the sort of thing the U.S. Centers for Disease Control and Prevention, in particular, ought to be good at if it can ever get its act together.

The search for COVID-19 treatments remains intense, and we can, of course, also use this breathing space to strengthen the health-care system. Italian doctors encountered a COVID-19 disaster despite having outstanding hospitals, which has been a mystery. Their message to the outer world is now: “it’s not ‘despite’; it’s ‘because of.’ ” They want the rest of us to move COVID-19 care out of existing hospitals as much as possible — to treat suspected cases at home for as long as patients can bear it, using telemedicine and mobile medicine, and to create well-equipped separate facilities, essentially 21st-century pesthouses, for the very ill when they appear in numbers. Here in Edmonton, Alberta Health Services has already commandeered the Butterdome at the University of Alberta for more or less this purpose, which is sinister, sad — and smart.