Read on Twitter
The NHS in England has rapidly delivered mass vaccination. Here we share an OpenSAFELY analysis that gives insights on exactly which patients received the COVID-19 vaccine in the earliest parts of the vaccination campaign. https://www.medrxiv.org/content/10.1101/2021.01.25.21250356v1.full.pdf+html
The data comes from OpenSAFELY-TPP, a secure analytics platform created by our team in huge collaboration now running across the full, linked, pseudonymised electronic health records of over 23 million people in England for urgent COVID-19 research (e.g. https://www.nature.com/articles/s41586-020-2521-4 )
In this very detailed data, we developed algorithms to identify key demographic and clinical sub-groups. and generated descriptive statistics on proportion of eligible patients receiving the vaccine among key Joint Committee on Vaccination and Immunisation (JCVI) target groups.
The data is current as of Jan 13th: we will share weekly updates. Between December 8th and January 13th 961,580 people out of 23.4m in our dataset received a COVID-19 vaccine.
Note: we see data 4-11 days after clinical event (we can build the datasets faster if there is clear operational need). This data is *not* the place to pore over daily vaccine rollout with “yesterday’s numbers”; it *is* the place to rapidly examine coverage in detailed subgroups.
Of 1,160,062 patients aged 80 or over and not identifiably living in a care home, 476,375 had been vaccinated in total by Jan 13th (41.1%).
We observed a substantial divergence in vaccination by ethnicity within the over 80 group (White 42.5% vaccinated, Black 20.5%, South Asian 29.5%, mixed 27%, other 27%, Unknown 39.7%). Finer ethnic groupings in paper. https://www.medrxiv.org/content/10.1101/2021.01.25.21250356v1
This reflects previous vaccination campaigns for other vaccines, and is consistent with survey data on intention to accept the COVID-19 vaccine. https://www.medrxiv.org/content/10.1101/2020.12.27.20248899v1
We also observed different vaccination coverage by postcode deprivation within the over 80 group (least deprived 44.7%, most deprived 37.9%).
https://www.medrxiv.org/content/10.1101/2021.01.25.21250356v1
https://www.medrxiv.org/content/10.1101/2021.01.25.21250356v1
(Note: these figures are for single dose vaccination; numbers for double dose vaccination are given separately in the paper https://www.medrxiv.org/content/10.1101/2021.01.25.21250356v1 )
Next we looked at clinical subgroups, within the over 80s, using the detailed data in OpenSAFELY.
For most pre-existing medical conditions, people were equally likely, or more likely, to have received a vaccine. This is good news: it means the patients at risk are being reached.
https://www.medrxiv.org/content/10.1101/2021.01.25.21250356v1
https://www.medrxiv.org/content/10.1101/2021.01.25.21250356v1
However there were some important exceptions. We saw lower vaccination coverage in over 80s with severe mental illness (30.3%), dementia (30.9%) and learning disability (28.1%).
https://www.medrxiv.org/content/10.1101/2021.01.25.21250356v1
https://www.medrxiv.org/content/10.1101/2021.01.25.21250356v1
Some *important* caveats for all the data. You can read about the challenges in creating the variables (which mostly reflect trade-offs applicable to *all* analysts working on these questions) in the paper. https://www.medrxiv.org/content/10.1101/2021.01.25.21250356v1
Secondly, these are rapid, early insights from the earliest phases of the vaccination programme: coverage for many of the different subgroups is likely to change over time.
Thirdly, variation in vaccination coverage in different ethnic, IMD, and clinical subgroups will have many complex drivers: in our view this data should be interpreted cautiously. It may reflect peoples preferences (“vaccine hesitancy”), features of the service, or both.
Fourthly: variation in vaccination coverage, by and within regions, will have many complex drivers. The vaccination teams across the country are heroic, and moving mountains. This data is here to help, with early warning of possible coverage issues.
On which: we are happy to help NHS staff in STPs, RVOCs and so on; we can share regional level data for your patch, with detailed breakdown of vaccine coverage for different demographic and clinical subgroups. Please do get in touch: [email protected]
So, thanks: http://OpenSAFELY.org is a huge collaboration: the developers and researchers of http://thedatalab.org (my group at Oxford); EHR researchers at LSHTM; developers and researchers at EHR vendors TPP and EMIS; and NHS England, who are Data Controllers for the project
Props to @helencebm @brianmackenna and @jessRmorley at http://thedatalab.org and NHSE who made the running; @sebbacon @_evansd @bloodearnest @themadwort @ghickman @dr_c_morton and @inglesp at t’datalab for building the OpenSAFELY software (and caroline/peter for code review)...
… AND all at @TPP_SystmOne inc @drchrisbates @SamPDHarper @DrJohnParry @jonnycockburn and all for the epic database backend and more; TPP and @ehr_lshtm for codelists and insights; NHSE and NHSX for their support and unswerving excellence; and so many more.
As always, all code for the OpenSAFELY platform is shared and open for review and re-use at http://GitHub.com/OpenSAFELY , along with all codelists and all code for all data management, all variable creation, and all analyses, including this one, here: https://github.com/opensafely/nhs-covid-vaccination-coverage
If you would like to learn more about the OpenSAFELY platform you can read the full user manual here http://docs.opensafely.org/ and, if you’re brave, download the full software suite to develop your own analysis code against dummy data here https://opensafely.org/code/
Lastly, if you you would like to see a full walk through of how you develop code for data management, variable creation, dummy data, and data analysis, you can read this notebook by our @wjchulme here https://nbviewer.jupyter.org/github/opensafely/os-demo-research/blob/master/rmarkdown/Rdemo.html
These are difficult times. People across the country are moving mountains, in the NHS, in business, in schools, in govt, at home, everywhere. Please follow the rules and be good to each other. /ends.
