Data on vaccination rollout: what we know, and what we need to know
As the UK wearily begins a third lockdown, the NHS is racing against time to get Covid-19 vaccines to as many people as fast as possible.
So what do we know about how the UK’s vaccine programme is progressing? Few datasets can be of more intense public interest right now.
Currently, the UK is reporting the total numbers vaccinated only. Other countries are reporting considerably more detailed information.
The UK’s vaccine programme has made a brilliant start. But we are concerned that if the UK continues to report only high-level data, increasing that big shiny total could be prioritised over getting the vaccine to harder-to-reach groups, like the very elderly.
So in this post, we survey what other countries publish, and what important UK users are asking for. We end with practical recommendations for statistics producers.
What we know #1: Total vaccinations delivered
The official GOV.UK dashboard charts the weekly doses delivered by nation. Just one month after rollout began, more than 1.25 million people have been vaccinated - an amazing achievement.
The devolved nations publish further breakdowns, although none are very detailed:
Northern Ireland reports that 21% of its jabs have gone to care home residents and 56% to health and social care staff, and breaks this down by region.
England reports that around two-thirds of all vaccinations have gone to people over 80. But there’s no other breakdown, so we don’t know how many jabs have gone to different priority groups or local areas.
Wales breaks counts down by region, but nothing else.
Scotland doesn’t have any further details beyond total vaccinations delivered. (UPDATE 14 Jan: Scotland has published breakdowns by area, age decade, and work status - nice work Scotland!)
So we don’t have a lot of information on who’s getting the vaccine, or where they are.
While it made sense to prioritise speed over analysis in the early days of the programme, as we enter the second month of rollout, some MPs and scientists are starting to raise concerns about the lack of information provided.
How UK data compares to other countries
The UK’s data looks sparse and slow compared to similar developed countries. (Our World in Data has a useful list of vaccine data sources by country.)
Israel, currently leading the world for vaccination rollout, also has the strongest data: it reports daily on total by age group, and publishes detailed breakdowns by age group across 279 regions.
Germany reports daily on the previous day’s vaccination count, also broken down by region and priority status (age, health, key worker, care home resident etc).
The United States reports daily on vaccinations delivered and administered by state.
Italy publishes data on GitHub (🥇) including regional breakdowns of where vaccinations have been supplied and delivered, and breaks down recipients by age group and healthcare worker status.
And tons of other countries - from Denmark to Chile to Bulgaria - publish similar detailed regional breakdowns.
Apparently the UK will provide total daily vaccine counts soon, but no other announcements have been made about any other data.
What we know #2: Where vaccinations are being delivered
We do have some insight into where vaccinations are being administered:
England publishes the locations of vaccine centres - 210 hospital hubs and 784 local services, and growing fast
Scotland publishes its own equivalent list.
However, this is less useful than it sounds, firstly because there is no data on how many jabs each centre has actually delivered - it could be tens of thousands, or none.
In England, there’s no documentation on what inclusion means, and some NHS trusts provide only trust names, while others list individual hospitals, or even separate sites within a hospital.
For both reasons, this data isn’t a great guide to actual local vaccine delivery.
What are others asking for?
On 3 January, the Sunday Times reported on concerns of a vaccine ‘postcode lottery’, having analysed the above data on vaccine centres. It’s unclear how far these concerns are valid - the data isn’t robust, as above, and it’s still early days.
However, better local data on delivery of vaccinations would help reassure everyone that they’re arriving effectively and equitably, as well as fast.
The MPs Liam Byrne and Andrew Mitchell have written to Matt Hancock over concerns about local access to data. They call for MPs to be given constituency-level data on (among other things) vaccine refusals. They also ask for constituency-level data on ethnicity, and priority groups still unvaccinated, to be published.
Meanwhile, the scientist Christina Pagel, who is on the Independent SAGE group, has called for better data on vaccination uptake and for additional regional, ethnic, age and deprivation breakdowns.
(UPDATE: The Royal Statistical Society also issued an excellent list of recommendations in December, calling for statistics on who has had the jab, segmented by region, age and risk/exposure group.)
This is a clear list of user needs from Parliamentarians and experts. So what should the UK’s statistical producers share with the public to support the rollout?
Our priority recommendations on vaccine data publication
The UK’s statistics regulator has issued guidance for statistics producers on vaccine rollouts, but it does not discuss what should be published, or who these statistics are for.
We think these stats exist to:
Support public confidence in the programme.
Support the best possible operation of the programme.
That means publishing enough to reassure the public there are no holdups locally, or holdups reaching the highest-priority groups, as identified by the JCVI.
In practice, NHS England and other stats producers should publish adequate data for the public and Parliamentarians to feel confident there are no:
delays in rollout to high-priority groups
delays in rollout to particular regions
access issues for other groups of people
delivery issues at individual vaccine hubs
and that if such issues occur, they will be tackled quickly and effectively.
Practical recommendations for better data
Thus, statistics producers in each nation should publish daily, timely data on:
The total count of vaccinations administered by area, by first or second dose. Ideally, this should be by MSOA (or the smallest possible non-disclosive census geography) of the recipient’s postcode, to allow for easy comparison with census data on deprivation etc.
This local area count broken down further by JCVI priority group and by ethnicity, to identify any issues with access.
Both of the above datasets aggregated by council, constituency and total, for easy reference.
Statistics producers should also make the following information available to health officials and Parliamentarians, and consider making it available to the public;
Total count of vaccinations supplied to, and administered by, each hospital hub or vaccine centre, broken down by dose stage and priority group, to identify issues in the delivery chain.
Data on take-ups and refusals or failure to respond by priority group and ethnicity, and by MSOA as above, to identify where outreach is needed.
Vaccinations are being managed through central registers - e.g. the National Immunisation Management Service in England - so reporting should not be burdensome.
Robust reporting is important to support public confidence in the rollout. We encourage the UK’s statistics producers to consider these recommendations, and help continue the early success of the vaccination programme.
UPDATE: Since publishing this, the Office for Statistics Regulation has written again to NHS statistics producers on 20 January, asking them to improve what they publish.
Have we got these principles right, and is there any other data that you would like to see reported? Let us know on Twitter, or at contact@centreforpublicdata.org.