December 14, 2021   7 mins

In my more hopeful moments, I allow myself to be convinced that Omicron will not be as bad as all that. There are optimistic signs: Dr Angelique Coetzee, a South African doctor, recently told reporters that Omicron “causes mild disease, whether you’re a child, 80 years old, vaccinated or unvaccinated, or have comorbidities”. And the most recent hospital surveillance report found only 230 people in ICU in Gauteng province, which has been hardest hit by Omicron. At a comparable point during the Delta wave, there were well over a thousand. Omicron seems to be putting far fewer people in intensive care.

More ambiguously, last week there were about 6,000 hospital admissions with Covid in South Africa, all of which were presumably Omicron since the variant is now dominant there. That’s well below the July peak of about 15,000 a week, although there’s a lag between infection and hospitalisation.

This sort of news has led to people thinking that it’s much less deadly than Delta. As the Omicron wave looms over Britain, that sort of hope keeps us sane.

But how much can we trust that hope? The Prime Minister has promised to ramp up the booster programme, saying that everybody over 18 would be vaccinated by 31 December – about a million people a day. Can that really be possible? And if it were, would that be enough, or are we going to need something more than the “Plan B” restrictions that are already in place.

First, we need to know if it really is less deadly than the other variants. It’s not as clear-cut as all that. South Africa has only fully vaccinated about a quarter of its population, but it has had three large waves of infection: very probably, by now, most of the population has had the disease. So it’s not possible to directly compare the Omicron wave with the Delta wave: Omicron is working its way through a population with much higher immunity.

The trouble is, that won’t be true in the UK. When Delta came here, we were already highly vaccinated: almost every adult had had at least one jab. At least some of the reasons why Omicron appears to be less of a threat to South Africans don’t apply here. We’re also a much older population.

Closer to home, Danish data looking at 2,417 cases of Omicron found only 19 people had been admitted to hospital. But again, it’s hard to know what to make of it. The patients involved were mainly young – the median age is about 30. And given an average lag of about a week between infection and hospitalisation, and since about 2,000 of the cases were detected in the last week, it might not be as promising as it sounds on the surface.

The most recent UK Health Security Agency technical briefing similarly finds that “None of the cases is known to have been hospitalised or died”. But their data only goes to 6 December. By 6 December, there had only been 260 cases of Omicron in the country, most of which had been detected in the previous few days. Since then, there have been at least 10 hospitalisations and at least one death linked to the new variant. Things may become clearer in the coming weeks.

It’s perfectly possible that Omicron is less dangerous than Delta to a person with prior immunity. But we can’t be sure yet. And we should also remember that while deadliness — virulence — is important, it’s not as important as transmissibility.

Imagine the following scenario. You have a disease with an R of four — that is, on average, each person with the disease gives it to four more people. And it also has an infection fatality rate (IFR) of 4%, so it kills one person in every 25 that it infects. (These are made-up numbers for illustration.)

You follow that disease through 10 generations. The first person would infect four more; those four would infect a further 16, and so on. By the time you reached the 10th generation, you would have about a million new cases – and adding all the generations together, you’d get a total of 1.4 million cases. Of those 1.4 million, about 56,000 would die.

Imagine that a new variant arrived which was more deadly: which kills 5% of victims. It would be worse. You’d expect about 70,000 deaths, 25% more than before.

But now imagine that a new variant arrived which was more transmissible, by the same margin. Each person infects five people, on average. Suddenly the first one infects five, those five infect 25, those 25 infect 125. By the time you reach the 10th generation, it’s almost 10 million new cases, and about 12 million in total. Even if it still only had an IFR of 4%, it would kill about 480,000 people – nine times as many as the original.

So even if Omicron is less deadly, if it’s very good at spreading through the population, it could quickly overwhelm the health service just through sheer weight of numbers.

The question, then, is: is it more transmissible? There was much excitement over the weekend, when the case numbers appeared to be slowing there — after zooming up to 19,000 a day, they dropped back to 17,000, and people started to wonder whether it had peaked. But then the South African health service released a new tranche of data on Sunday, with another 37,000 cases — the apparent slowdown appears to have been a reporting backlog.

Even taking that into account, Omicron isn’t doing quite the mad upwards dash that it was in late November. But it’s still escalating, in a presumably fairly immune population. And it seems to be spreading even faster in the UK: Paul Hunter, a professor of medicine at the University of East Anglia, told the Science Media Centre that “In South Africa the latest R estimate is about 2.2. For the UK the estimate is 3.7 which is doubling every 2 to 3 days.” The UK HSA technical report suggests (p19) that it spreads around twice as easily as Delta in the UK population. If that’s true, it would easily overwhelm any gains from reduced virulence, even if they’re real.

So what does it all mean? Over the weekend, the London School of Hygiene and Tropical Medicine released a model of how Omicron might spread. It included scenarios with different assumptions about how good the new variant is at evading immunity, and about how good the boosters are at building that immunity back up. It also assumed that we’d keep the Plan B advice in place (masks in some public spaces, people working from home if they can, some vaccine-status checking) and that the NHS would administer about 500,000 new booster jabs a day.

The most likely scenario is that the variant is pretty good at evading immunity, but that boosters work pretty well. Under that scenario, the LSHTM model’s projections look fairly bleak: a peak of about 6,000 hospital admissions a day, compared to about 4,000 a day at the January 2021 peak. And they think between about 40,000 and 50,000 people would die of Covid between 1 December and 30 April.

They say in the model that they have assumed Omicron is no less deadly than Delta, given prior immunity. They might be wrong about that, but as I’ve shown, we can’t be sure. It’s certainly very possible that we’ll be looking at another wave of deaths comparable to the first or second one.

Will Johnson’s booster boosterism make much of a difference? I doubt it. It takes at least a week for boosters to have a significant effect on your immunity, and for some people more like two. Even if we started vaccinating a million people a day from tomorrow, the impact of that wouldn’t be felt until around Christmas. We’re not going to start vaccinating a million people tomorrow: we’ve been averaging 400,000 a day recently, fewer than the LSHTM model assumes, and the NHS thinks that we’ll still be boosting well into January and February and we’ll only have offered everyone a jab by the new year. Hopefully it’ll ramp up quickly, but if we’re relying on boosters, then most of the January caseload will be baked in already.

And here’s the really worrying thing. Sajid Javid, the health secretary, told MPs last night that 200,000 people have been infected by Omicron yesterday alone. This was such a huge claim that people assumed he’d got mixed up: perhaps he meant 200,000 people are infected? Or have been? But apparently not. There were about 50,000 cases reported in the UK on 10 December. The ONS infection survey suggests that we only detect 40% of them: that is, there were really 125,000. On the 10th, 19% of cases were Omicron. That is about 25,000 cases. But cases detected on the 10th were probably infected on the 7th, six days ago. And if it’s doubling every two days, then that 25,000 has had time to become 200,000.

For reference, that’s about the number of daily cases that we had in early January 2021, the peak of the second wave, by which time we’d already been in full lockdown for a month. We are not in lockdown now, and several more doublings are probably already locked in. If Javid is correct, then Omicron had better be much less deadly, or there could be real trouble ahead. We’ll know if he’s right in about three days. 

So should we be talking about restrictions?

The restrictions we put in place for the first and second waves were intended to avoid something far worse. The 16 March 2020 paper by Imperial College London predicted 500,000 deaths if no measures were taken. That was what sent us into lockdown. The actual death toll of 40,000 or so in the first wave was a fraction of what it could have been.

But we are nowhere near where we were early last year. We’re now talking about something that is on the level of a very bad flu season. Full-on March 2020 lockdown is probably an overreaction.

But the LSHTM model thinks that we could prevent about two-thirds of those deaths and hospitalisations by going to full lockdown,1 and about one-third by introducing, right now, some less strict measures, equivalent to Stage 2 of the “roadmap out of lockdown” from spring: the biggest part of that was probably that only two households or six individuals could meet indoors.

It is a non-trivial intervention that could slow the course of Omicron; it might buy some time for the sped-up booster programme; and it would mean we wouldn’t have to cross our fingers and hope that Omicron is less deadly than Delta. Obviously politicians will be reluctant to ruin another Christmas. But relying on boosters alone is a big risk, especially since no one but Johnson seems to think we can hit the target.

  1. It actually only models the most optimistic and most pessimistic scenarios under the various restrictions. But in both of those scenarios, about two-thirds of deaths and hospitalisations are prevented under strictest lockdown, and about a third under the less stringent Stage 2 restrictions. I have assumed that the intermediate scenarios see roughly similar numbers.

Tom Chivers is a science writer. His second book, How to Read Numbers, is out now.