Wednesday 19 January 2022

COVID-19: The Beginning of the End?

Today, I’m going to look in a bit more depth at the COVID situation in the UK. And I’ll contrast it with the differing situations of some of its neighbours.

At the outset, I’ll say that the UK right now appears to be at least as well positioned against the virus as any other major Western European country. My informed gut feel is that we’re now not so far away from reaching herd immunity, and so beating the damned thing. Oh, and European countries like Germany, Austria and France, that have implemented high lockdown strategies recently, are starting to look a bit silly.

I took the data for this paper on January 16th, 2022 from the following sources:

·       Our World in Data – [[1]].

·       The Blavatnik School of Government – [[2]].

·       The UK government’s coronavirus data pages:

o   “Cases by date reported” at [[3]].

o   “Deaths within 28 days of positive test by date reported” at [[4]].

o   “Patients in hospital” and “Patients in mechanical ventilation beds” at [[5]].

All the data runs up to and including January 15th 2022. Although the most recent data will necessarily be subject to change as late reports and adjustments come in.

Cases

The graph at the head of this paper tells a story. From a base level of between 500 and 1,000 new cases per day (weekly averaged) per million population in the first half of December, case counts shot up to between 2,500 and 3,000 per day per million in England and Scotland, almost 4,000 in Wales and nearly 5,000 in Northern Ireland.

The peak individual days were: In England, 162,572 new cases reported on January 1st. In Northern Ireland, 30,423 cases covering the four days from January 1st to 4th inclusive. In Scotland, 20,217 cases on January 3rd. And in Wales, 22,317 cases covering January 2nd and 3rd. The UK wide peak was 221,222 cases on January 4th.

Since then, though, new cases have plummeted in all four countries. The latest figures in new cases per day (weekly averaged) per million population are as follows:

Because I’m going a bit later to look at the prospects for herd immunity, I’ll also show the total cases per million for each country:

Here is the graph of new cases and deaths for the UK as a whole (the figures will not match exactly, because I am using Our World in Data’s population estimate for the UK as a whole, as opposed to estimates for the constituent countries from [[6]]):

That’s quite a decline – from a peak of 182,891 new cases per day for the week centred on January 2nd, down to a mere 116,689 for the week centred on January 12th. A drop of more than 36% in 10 days! Since then, the drop has continued – Worldometers is showing the weekly averaged new cases for the week ending January 17th as 99,259. And, as you can see from that first graph, new cases have plummeted in all four of the constituent countries.

Here is the UK wide data looked at in terms of weekly case growth:

That’s amazing. The weekly growth (in new case counts which are already centrally weekly averaged – blue line) has fallen from a local peak of plus 51.8% on December 28th to minus 34.0% on the latest date for which I have data to calculate, January 8th. And the calculated reproduction rate of the virus (grey line) has fallen from a local peak of 1.46 on December 29th to 0.73 on January 13th. That’s a halving in the R-rate over the course of 15 days!

Tests

Some have suggested that the recent drop in cases may be due to a shortage of test kits, particularly PCR tests. But here are the graphs of the total tests conducted in the last month, and of PCR test numbers and capacity, from the UK government’s site [[7]]:


There was a clear peak in testing on January 4th, the first working day after the holiday period. Since then, testing has dropped back to pre-Christmas levels, though there are still results, positive or negative, yet to come in. Resources to process PCR tests were in relatively short supply for a few days around January 8th-10th, but this appears to have been made good since. And when I google “PCR test availability crisis,” the hits I get are media scares dated December 29th or 30th. So, it seems to me that there is no reason to ascribe a significant role in the continued drop in new cases to shortage of test kits, or of resources to evaluate them.

Lockdowns

The textual description of the latest UK national lockdown status (as of December 26th) is as follows: Schools: Recommended closed, Workplaces: Some closed, Events: Mandatory cancelled, Gatherings: Up to 11-100 (Regional), Public transport: Recommended closed (Regional), Stay at home: No measures, Travel: No restrictions, International: Quarantine high-risk, Face covering: Required when with others (Regional).

The UK national level lockdown information has not been updated in the Blavatnik School of Government data feed since December 26th (which is why the dark red stringency line in the weekly case growth graph cuts off at that point). However, the data for the individual constituent countries, which goes up to January 10th in all four, shows no changes since this date. This means that it cannot possibly have been fresh lockdowns which caused the declines in the new case counts!

Here are the latest reported lockdown levels in the four constituent countries:





Hospital Occupancy

Here is the COVID hospital occupancy per million for the four constituent countries:


These figures are based on actual counts of patients in hospital who have recent positive COVID tests – not quite the proverbial “bums on seats,” but close. The figures should, therefore, be complete and accurate for all but the very latest couple of days. The “signature” of omicron is visible in the very steep rise in late December in all four countries, just as the “signature” of delta was visible in July in England and Northern Ireland, and August in Wales and Scotland.

In Northern Ireland (red line) there has been a clear downward trend in COVID hospital occupancy each day, from a peak of 500 on January 5th and 6th to a latest count of 402 on January 13th – a 19.6% reduction. This provides evidence that the drop in new cases in Northern Ireland is real, and not merely caused by a shortage of PCR tests. After all, if you can’t get a PCR test when you’re in a hospital, where else can you get one?

In England the peak was 17,120 on January 10th, had gone down to 16,716 by January 13th – only a 2.4% reduction, but a consistent one over the course of three days. By January 15th, it had gone down again to 16,281, a 4.9% reduction from the peak. In Wales, the peak was 890 on January 11th, which by the 13th had gone down to 861 – a 3.3% reduction. In Scotland, there is as yet no clear evidence of a downturn, but I expect one in the next few days.

Here is the graph of hospital occupancy per case, with the new cases offset by 7 days to allow for the gap between confirmed infection and hospitalization:

Those bumps at the end of December in Northern Ireland and Wales look a bit strange; but they may well be due to their case reporting systems not being as timely over holiday periods as the English or the Scots. Otherwise, this metric seems to have been trending consistently downwards since at least the middle of November. That’s very good news – it confirms that omicron really does cause less hospitalizations per case than its predecessors.

Intensive Care Units

Here’s the ICU occupancy per million population:


It was the ICU situation that triggered the January 2021 lockdown. So, it’s good to see that, in all the countries except Scotland, ICU patients have dropped since late November. Current ICU occupancy in England is only 14% of its peak in January/February 2021. In Northern Ireland it is 40%, in Scotland 23% and in Wales 19%. There is plenty of margin in there, particularly considering the dropping new case levels in all four countries.

And here’s the ICU occupancy per case, with the new cases offset by 14 days:

Again, a steady drop since November. That’s more good news.

Deaths

Here are the deaths per million:

Yes, deaths are going up. But this is a consequence of the recent surge of cases, into which some proportion of deaths is necessarily “baked.” Since the mean time between infection confirmation and death if it comes is 21 days, we should see COVID deaths starting to drop off about 3 weeks after the peak of cases – that is, before the end of January.

Here are the deaths per case, with the cases offset by 21 days:


These percentages are tiny, compared to the death tolls earlier in the epidemic. And if you squint a bit, you can see a decline in deaths per case, except in Wales, since early November.

In a European context

I’ll now put the UK’s recent performance into a European context, by showing some of the graphs and lists from my “Europe 14” group of countries. First, cumulative cases per million over the whole epidemic:


The UK, Ireland and Belgium all seem to have taken the “let’s get to herd immunity as quickly as possible” strategy. The Dutch were doing the same, until they decided to lock down just before Christmas; presumably, for fear of filling up the ICUs. Whether the recent high level of French new cases was planned or not, I wouldn’t like to guess.

Next, daily new cases per million:


The UK (pink line) is the only one displaying clear evidence of a “right hook” towards falling new cases. Ireland (green line second from the top) may have almost peaked, but the rest look still to be headed in a distinctly northerly direction. The UK is now third from bottom in daily cases per million; and case numbers are dropping.

How about lockdowns? Here’s the current list:

Germany and Austria, the two countries below the UK in current cases per million, seem to have achieved that status through heavy lockdowns. Italy, as so often, is heavily locked down, but still doing badly. And while the UK as a whole is in eighth place out of 14 in stringency, only Denmark is more lightly locked down than England. As to France, it is in top spot in daily cases per million, and third in lockdown stringency. That isn’t good at all.

Here are the average lockdown stringencies for each country through the course of the epidemic:

Austria, France, Germany, Ireland and Italy are all in the top six in both current and average lockdown stringency. These countries seem to have a yen to lock people down (though less so in Ireland recently); while Denmark, Sweden and Luxembourg tend to go the other way. Beyond the UK, only Portugal and perhaps Spain seem to be treating this as an “end game” – keep lockdowns low, as long as you don’t risk running out of medical resources, in order to get omicron through the population as fast as possible, and so achieve herd immunity. This is, in my view, the only sane strategy at this point.

Here are the current lockdown statuses of the five most heavily locked down countries:

·       Germany (since December 2nd): Schools: Some closed (Regional), Workplaces: Mandatory closed, Events: Mandatory cancelled, Gatherings: Up to <=10, Public transport: Mandatory closed, Stay at home: Recommended, Travel: Mandatory restrictions, International: Ban some arrivals, Face covering: Required in some places.

·       Italy (since December 6th): Schools: Some closed, Workplaces: Mandatory closed, Events: Mandatory cancelled, Gatherings: Up to >1000, Public transport: Mandatory closed, Stay at home: Required with exceptions (Regional), Travel: Mandatory restrictions (Regional), International: Ban some arrivals, Face covering: Required when with others (Regional).

·       France (since January 9th): Schools: Recommended closed, Workplaces: Some closed, Events: Mandatory cancelled, Gatherings: Up to 101-1000, Public transport: Mandatory closed, Stay at home: No measures, Travel: Mandatory restrictions, International: Ban some arrivals, Face covering: Required when with others. From December 4th, the lockdowns were the same as above, plus a restriction on gatherings to up to 11-100.

·       Netherlands (since January 10th): Schools: Some closed, Workplaces: Some closed, Events: Mandatory cancelled, Gatherings: Up to <=10, Public transport: Open, Stay at home: Recommended, Travel: Recommended not to travel, International: Ban some arrivals, Face covering: Required when with others. The stringency has been above 60% continuously since December 19th.

·       Austria: Schools: Recommended closed, Workplaces: Some closed, Events: Mandatory cancelled, Gatherings: Up to 11-100 (Regional), Public transport: Recommended closed (Regional), Stay at home: No measures, Travel: No restrictions, International: Quarantine high-risk, Face covering: Required when with others (Regional). The stringency has been above 60% continuously since November 22nd.

Here are the current weekly case growths:

Oh, my! Despite the high lockdown stringency, Austrian cases have more than doubled in a week! That runaway case growth – as well as growth of nearly 40% weekly in the Netherlands and Germany, and over 20% in France and Italy – suggests that lockdowns don’t control the spread of omicron. Or, at least, that these particular lockdowns don’t – even though they cover almost the whole spectrum of lockdown options. Meanwhile, Ireland and the UK are the only two countries currently showing negative case growth.

Here are the current hospital and ICU occupancies, both expressed as percentages of the available capacity:


You can see why the French locked down relatively hard; they were worried about running out of ICU capacity. This may also be true for the Dutch, whose peak ICU occupancy in the first wave was over 100% of capacity. (They were able to send patients to Germany, which at the time was less hard hit than neighbouring countries, due to early closure of their borders).

It’s a pity the Germans aren’t reporting hospital occupancy. But those ICU numbers suggest that they and the Austrians may have lost the plot. The German lockdown stringency has been above 70% since late November, and the Austrian above 60%. Why such heavy and sustained lockdowns, when they don’t seem to be in any danger of running out of medical resources? Could it be no more than “We have ways of making you obey the Vorschriften der Weltgesundheitsorganisation?” (That last word, German for the UN’s World Health Organization, is 27 letters long. Only one less than the longest accredited word in English, “antidisestablishmentarianism”).

The WHO’s 200 per million per day threshold, at which the virus is to be considered “endemic” and unlocking is frowned on, looks to be well on the low side. And to be serious for a moment, why didn’t anyone see that beforehand? For historically recent (since 1900) pandemics like the “Spanish ’flu” have burned themselves out in two, three or at most four years. Four years at 200 new cases per million per day would infect 29.2% of a population – not nearly enough to get to herd immunity against a virus like omicron, or even delta.

Lastly, the lists of cumulative deaths per case, and of current deaths per case with a 21-day offset:


That’s interesting, again. Germany and Austria are both near the top in current deaths per case, and Italy and Germany have the two highest cumulative deaths per case. Lockdowns may control cases (even if only sometimes), but at what cost in eventual deaths?

Oh, and the UK is now not far from the bottom in this particular league; as it has been for some time. Right now, it looks to me as if the UK is at least as well positioned against the virus than any of the other countries; even Denmark, Sweden and Luxembourg.

Comparison with South Africa

Another country worth comparing the UK with is South Africa, the first to report the omicron variant:


You can see the ramp-up of omicron in late November 2021, and what look like earlier ramp-ups of other variants (beta in December 2020? delta in June 2021?) The apparently regular spawning of a new variant every six months is a pattern I haven’t noticed before; I wonder if it has epidemiological significance? If you look with a rather jaundiced eye, you can see something a little bit similar in the UK data too, although the six-month periodicity is not nearly as regular. Oh, and I hear that SAGE, advisors of the UK government on these matters, are predicting another omicron wave in early summer: [[8]].

Vaccinations are not a major feature in South Africa, with only 27% of the population fully vaccinated, and another 5% vaccinated once.

Recent lockdown history:

·       From September 13th 2021: Schools: Recommended closed, Workplaces: Recommended closed, Events: Recommended cancelled, Gatherings: Up to 101-1000, Public transport: Recommended closed, Stay at home: Required with exceptions, Travel: No restrictions, International: Screening, Face covering: Required when with others.

·       On December 6th: Workplaces: Some closed, Events: Mandatory cancelled, Gatherings: Up to >1000, Public transport: Mandatory closed.

·       On December 21st: Public transport: Recommended closed, Stay at home: No measures.

The South Africans seem to have got over their “omicron wave” with only one somewhat panicked-looking lockdown, and that for 15 days only. Otherwise, their recent lockdown stringencies have been comparable with Wales, Scotland or Northern Ireland. And since the South Africans have so few vaccinated, that suggests to me that vaccinations are not a big factor in helping control omicron.

How close is the UK to herd immunity?

Herd immunity is achieved against a particular variant of the virus when the actual reproduction rate of the virus (R) drops below 1 and stays there, either until the virus is beaten, or a more transmissible variant appears. The actual R-rate, in its turn, is a function of the basic reproduction rate (R0) of that variant and the proportion of people in the population who are still susceptible to the virus (S). (Where different groups of people have different risks, for example due to different levels of vaccination or to some having already had COVID, each can be weighted by an appropriate risk factor when calculating the effective S).

The relation I would expect between the two is R = R0S. That is, the actual reproduction rate should be the basic reproduction rate multiplied by the proportion of susceptibles. The criterion for herd immunity is then R0S < 1, or S < 1/R0.

What is the R0 for omicron? According to a Lancet article from December 17th [[9]], the delta variant has an R0 of just under 7. That means that, to achieve herd immunity against delta, you need(ed) to reach S < about 15% of the population. One expert says the R0 for omicron “could be as high as 10.” That would mean S < 10% as the herd immunity criterion.

What is the value of S in the UK today? To calculate S, all (sic) you need to know is the number of people who have had COVID, the chance someone who has already had it will contract it again, and the effectiveness of different levels of vaccination against the virus.

We know the number of reported UK cases so far – about 22% of the population. But that is far less than the actual number of people who have had the virus, since many early or asymptomatic cases will have been missed. I saw a paper that suggested you should add 28% to the reported cases to allow for the unreported ones, which would lift the 22% to 28%; but I didn’t like the methodology much, because they seemed to be averaging figures from different countries and different stages of the epidemic all into one big pot. Moreover, the UK’s Office for National Statistics (ONS) are now saying that infection rates for omicron may be three times published case counts: see [8]. That means that the correction factor for undetected cases will probably need to be well bigger than 28%!

Meanwhile, we don’t know just how effective the vaccines are in preventing transmission, and it’s said they are less effective against omicron than against earlier variants. If the South African experience is anything to go by, that may well be true. But irrelevant, if herd immunity is close anyway.

One statement in the Lancet article I referenced above is interesting: “the proportion of the population who harbour antibodies against SARS-CoV-2 exceeds 90%.” These antibodies are produced when a person recovers from the virus and acquires T-cell immunity, and can also be produced as an effect of vaccination. Could a very significant fraction of the population indeed be already, at least to some extent, immune?

Then there is this quote from the USA from October 2021 [[10]]: “Fewer than 0.005% of fully vaccinated Americans have experienced a breakthrough case resulting in hospitalization or death — and people who have already had COVID-19 may be even less likely to be reinfected, according to the Cleveland Clinic.” That referred, of course, to delta, not omicron.

Yet Imperial College modellers, publishing on the same day as the Lancet article [[11]], seem to think that “the risk of reinfection with the Omicron variant is 5.4 times greater than that of the Delta variant.” But then, their study also “finds no evidence of Omicron having lower severity than Delta.” I think there is now good evidence, from hospitals and ICUs all over the UK, that omicron is less likely to cause hospitalization than delta. So, I’m inclined to take their estimates with several pinches of salt.

False dawns – and the beginning of the end?

I see from the UK wide new cases graph that we have had three “false dawns” already. On three occasions, new cases have fallen abruptly without any increase in lockdown stringency, but after a while have started rising again. The first was in the second half of July 2021, right after the “great unlock,” and cases fell to about 55% of the peak before coming back up. The second was near the beginning of September, around back-to-school time, with trough about 75% of peak. And the third was towards the end of October, around school half-term, with trough about 70% of peak.

If I look at the latest peak and descent, weekly averaged cases have dropped by 36% in the first 10 days or so of the new year. The big difference, though, is that this time round, the four constituent countries of the UK have moved in lockstep. That may imply a convergence of conditions between the countries, because omicron spreads so fast. It may also mean that the Christmas and New Year period has been a perfect opportunity for the virus to spread, and that opportunity is lessened considerably once life returns to something approaching normal.

Can I be sure this peak won’t go the same way the last three did – cases dropping to 50% or so of the peak, then turning upwards again? No, I can’t. At least, not without estimating the count of susceptibles S with a great deal more accuracy than the data I have is capable of supporting. But I actually think – feel in my bones, if you like – that omicron may well be the “beginning of the end” for COVID in the UK. Get fully over this hump, and herd immunity may prove to be a lot closer than it looks at first sight.

As to SAGE’s idea of another wave in early summer, bring it on! That should seal herd immunity for sure. And not before time.

So, I’m going to do what I usually do… Keep watching the data. If new cases keep on going down for a month or so, we’re on a winner. Until the next variant arrives, of course.