Thread by @ewanbirney

COVID thoughts from beautiful, misty Northumberland. TL;DR As expected, Omicron is exploding in numbers in cities and beyond worldwide. Reassuringly Omicron infections are less likely to end up in hospital, but whether this reduction in severity is enough to be safe is unclear

Context: I am an expert in human genetics and computational biology. I know experts in infectious epidemiology, viral genomics, immunology and clinical trials. I have some conflicts of interest: I am a consultant and sharehold of Oxford Nanopore and I was on the Ox/Az trial.

Reminder: Omicron is the first "antigen drift" variant with fast transmission from SARS-CoV-2. This drifting antigen presentation on spike is one of the ways Coronaviruses shift their appearance to our immune system, so it was expected, though always not fully appreciated.

It's shifted immunological profile means Omicron can reinfect people who were previously infected (just as 'the common cold' - often other coronavirues - cycles over the year) and, frustratingly, will infect and transmit for double dosed individuals.

With fast transmission and plenty of susceptibles it has spread like wild-fire worldwide, often starting in connected cities (London, New York, Copenhagen, Oslo, Paris) in the younger generation. The rate is eye wateringly fast, with doubling times of 2 to 3 days.

Work from primarily South Africa, Scotland and England have all pointed to substantial reinfection levels; far lower double dose vaccine effectiveness on symptomatic infection and (thankfully) that booster (3rd doses) suppress symptomatic infection, hence the urgency in boosters

Over the last couple of days the same groups aimed to answer the more and more important question of the rate of hospitalisation given infection. All three groups found it to be less than Delta, but at different levels. Here it gets complex.

One level of complexity is the classic problem of observational data; the people who get infected with Omicron aren't at random; they are younger; they probably socialise more; most of them live in big cities. All these things confound the analysis.

Deeper still than this is exactly what do we mean by hospitalisation given infection; for example, the larger reinfection levels means there are people who are cases but previously infected, and often not even they know they were previously infected.

The ability for Omicron to re-infect (and infect on top of vaccination which will protect) thus changes the denominators of cases to hospitalisation, regardless of the biology.

At some level this is fine - one wants the aggregate rate of people going into hospital in the UK (or France, or Germany) - accurately modelling the rate is the key thing, not necessarily unpicking the reasons. ...BUT...

The headache is that early on to estimate this one has these biased (younger, big city) samples which one wants to use to model the consequence of infection (or not) across UK, or across Saxony with different age profiles and different vaccination statuses.

Here is a nice thread about this from @_nickdavies one of the authors of the LSHTM models about these headaches and just how hard it is to know if the "severity paramter X" in the model is the same as "severity as estimated in this analysis".

https://twitter.com/_nickdavies/status/1473941671196905478?s=20

(I know, I also wish it was easier to get intuitive understanding of these things, but this not how infectious biology is wired. It is related to some of the classic statistical Simpsom paradoxes which always stop people in their tracks until you break it down).

The "headline" numbers for severity range from 80% lower through to 30% lower; the range here is partly setting (South Africa vs England) but at least as much in this definition "what does 30% lower severity mean in terms of this model".

At the same time, London's Omicron surge seems like it is growing less fast, though testing capacity issues are being reached in the UK. Paris is rapidly catching up with London, and rest of UK is following a growth trajectory.

In Germany response to the late Delta surge has been good (in particular in Bavaria) but there remains a big East/West dichotomy, and old East German still looks like they have the biggest risks in older unvaccinated

(Note: the UK data can't easily estimate much on older unvaccinated beyond "people are still coming to hospital" because the numbers of older unvaccinated people are thin for the denominator).

Germany has announced more restrictions post Christmas; UK has decided not to change restrictions, at least until 27th Dec, but advocating more boosting and lateral flow testing (both clearly good things). I don't believe France has changed anything yet (correct?)

But all European countries will need to plot a way out of Omicron post Xmas. The efficacy of booster vaccines means in some sense the same logic as Delta is present protect everyone one can via vax+boosters, ease off restrictions (maybe light) consistent with healthcare levels

(this strategy is waaaay easier to tweet than to do though. In particular the "protect everyone one can" means taking a view on how many unvaxed one can have to have the ease down ramp work).

Will we be doing this every winter? We don't know, but this is too exhausting and expensive (in multiple measurement ways) to be stable. My own sense is that we will have to improve on multiple axes:

1. Improve the treatmen of COVID the disease. New antivirals are the next weapon here. I still think there is scope for yet more dampening of inappropriate immune response.

2. Improve structural factors of respiratory infection (note: will also help suppress colds, influenza etc) - good ventilation, perhaps routine mask wearing as in Asia, in particular in the winter (this is not so offensive to me).

3. Improve detection and response of new variants and new viruses / bacteria. Omicron was better than Delta, which was better/similar to Alpha. Lots of moving parts to this, a very worldwide problem.

4. Improve healthcare capacity (particularly for UK, which runs very lean)

5. With these improvements, normalise / manage as a year to year process the health burden of this virus (and others) alongside all the other things we do for health.

Back to post Christmas; health and economic strategists will have their work cut out for them in Berlin, Paris, London, Dublin, Copenhagen etc... working out a route through Jan and Feb which doesn't lead to healthcare melting down but minimises other harms.

This is just not easy. For example, schools being off almost certainly reduces transmission now, but almost certainly every society will want schools back in Jan as soon as possible.

Another headache are staff rostas giving fast infection - what is the peak number of people isolating for society to function - regardless of healthcare?

And then the very real impacts on mental health, on livelihoods (hospitality and travel industries in particular) and life-thats-for-living.

It's not easy; I don't envy the people trying to put together the credible options nor the decision makers here.