It seems that as the number of vaccines administers goes up, the rate of CSVT is converging on slightly more frequent than 1/100k. They have now also confirmed

It seems that as the number of vaccines administers goes up, the rate of CSVT is converging on slightly more frequent than 1/100k. They have now also confirmed that there is an age effect which they found in the previous data release as well but which the EMA, with a smaller sample size at the time, had not been able to determine to their satisfaction previously. The JCVI now has the challenging task of making a recommendation for people in their 30s without occupational or health reasons for priority, whose turn it will be over the coming weeks to start receiving vaccines. (People from 18-30 have already been recommended a different vaccine "if available", there is probably just barely enough Moderna on order for Q2 to give most of them that). Making that decision is a very difficult one because it depends on a lot of context. You're not just deciding between giving the vaccine and not ever vaccinating, you first have to decide what the counterfactual scenario is where you don't give a 35 year old an AZ dose next week. This on: a) The incidence rate for each demographic of CSVT-type clotting from the vaccine[s]. b) The expected distribution of clinical outcomes from those events, making reasonable assumptions about likelihood of early detection and outcomes with treatment for heparin induced thrombosis. Those two give you the clinical risk of the vaccine. Then you compare those with the clinical benefit from: 1) The relative likelihood of infection for a person, given their demographics, personal exposure profile etc. 2) The absolute likelihood of infection in the country / region in question based on assumptions about the future of the pandemic 3) The expected distribution of clinical outcomes from infection given the person's characteristics (2) is important because you're comparing a one-time risk of side-effects with a time-related risk of infection. Clearly, the same person in Delhi or New Zealand is not really making the same trade-off even if in the abstract, the side-effects and IFR are the same. You also need the counterfactual: If you decide not to give an ad vectored vaccine, you are not deciding never to vaccinate, you are deciding that this person will get a different vaccine, probably later. How much later depends on the location. If you're in the US, you've got Moderna and Pfizer/Biontech coming out of your ears so it might only delay any individual vaccination by a few weeks. If you're a healthy 25 year old who can work from home, the risk trade-off of having to wait an extra 6 days is very different than that same person if they have to wait another 6 months. As a result, different countries, looking at the same data will rationally make different decisions because their counterfactuals are different. The US has lots of mRNA vaccines so will choose one thing, the EU has been badly hit by the failure of Sanofi/GSK to deliver a working vaccine, by AstraZeneca scaling problems, and by J&J coming along later than anticipated but has quite a lot Pfizer production capacity, the UK has used up its Pfizer stocks and will now only use it for second shots. CureVac and Novavax may be able to supply some doses at some point but aren't in the picture yet. Several European countries are negotiating to get Sputnik but that is also ad vectored so may have the same problem (and according to reports from Anvisa in Brazil appears to have replication competent adenovirus in it which is definitely not the intention). That changes the risk calculation because it changes the time that people will remain unvaccinated in the counterfactual where ad vectored vaccines are not used for their age group. Of course, you also have to consider the risks from the other vaccines. Risks from the mRNA vaccines appear very low but J&J seems to have the same clotting issue though maybe at a lower frequency. So if you're delaying someone's vaccine in order to get them J&J, you have to calculate the (seems to be lower) CSVT rate in the counterfactual as well. All of this is before the complicated public health element of keeping messages for the public simple and the issues caused by changing that message frequently and the long-term need to ensure that public trust in vaccine programmes remains deserved by the honesty with which they are conducted. It's pretty clear that in the UK where there is just about enough AZ to slowly vaccinate the 30-40s and enough Moderna for the 18-30s (but not any more than that), the public health professionals and immunologists on the JCVI are agonising over the right thing to do. I would be too, even though for myself I have happily received AZ in my mid 30s but it must be an awesome and terrible weight to make a recommendation like that for other people, knowing that they trust you to make the right decision.