Origin
This is an attempt to describe the history of the evolution of this dashboard for people who arrive at it long after various aspects of it were created.
This dashboard originated in early 2020 when Open Innovations (ODI Leeds as was) were asked for a local authority view of covid by various people who work in local authorities. At that time England, Wales, Scotland, and Northern Ireland all reported in different places and no agency apparently yet tasked with making a UK-wide consistent view of the data. The gradually changed as PHE (as they were then called) were tasked with making a UK dashboard. We used PHE data (for England) but, early on, downloads were removed and data had to be manually downloaded from a different URL every day. There wasn't always local-authority-level data or not for all the figures we needed. We took part in several #OpenDataSavesLives sessions during the first months of the pandemic where we highlighted what we thought should be improved in terms of which data were provided and how. Part of this was a technical ask but it seemed to be largely a political ask to allow the PHE dashboard to publish certain things.
During those early months the only way we could get reliable, machine-readable, updated, lower-tier-local-authority-level data on cases was via the huge efforts of Tom White who curated it by hand from the four nations. He closed that effort down in July/August 2020 for various reasons. PHE (as they were called back then) started providing local-authority-level data for cases across the UK so we switched to using their API for the cases data despite some issues with the API that would cause it to fail sometimes.
As time went on we started showing ONS local-level death data. At that point in time PHE (as it was called) did not appear to be providing death data at lower-tier-local-authority-level via the API.
As with all things, PHE's (as they were called) API has evolved over time; some things may be available now which weren't in the past at the lower-tier-local-authority-level we need for this dashboard.
Levels of risk
The column colours and "risk" labels were chosen in early September 2020. They were inspired by experiments by Jeni Tennison who used California's four levels of risk. We had previously been using the House of Commons data on "Covid risk levels" but they had stopped publishing that around the time the UK Government implemented various regional lockdowns/"second lockdowns". The various UK-level dashboards/maps/data sources kept (understandably) changing their definitions of risk so settling on one (which has stayed fixed) provided a level of stability in expectation. At the start of September 2020 the levels seemed to make sense but quickly everything got stuck on "purple" as cases numbers shot up (and largely stayed up apart from in May 2021). Several times we've asked users of this dashboard if we should change these levels/colours and each time people have said to stick with them, so we have.
Daily cases
Daily cases shows a smoothed 7-day rolling average (+/3 days of the day in question) of confirmed cases by specimen date. The smoothing hides some of the "noise" in the daily data and smooths out the weekend effect. The large number we show is this smoothed value as of 5 days ago (in the data). We ignore the most recent days because the data is still coming in and the current value under-estimates the "final" figures for those dates. During late 2020 we showed a dashed-line segment for the recent days however, as people were not appreciating that these values are nearly always an under-estimate, we stopped showing those recent days by default. At the start of April 2022 the Government stopped providing free lateral flow tests so we expect the number of tests conducted to reduce and therefore for the number of cases to become under-reported.
Deaths
Total deaths gives the total of all deaths in a Local Authority. That means deaths by every cause including COVID-19. Total COVID-19 deaths gives all deaths that the Office of National Statistics has noted as being related to COVID-19 specifically. Total COVID-19 deaths as a percent of total deaths takes the Total COVID-19 deaths and divides it by Total deaths to find the percentage of deaths which were related to COVID-19. In pre-COVID times it is known that some areas have more deaths per capita than others because they may have older populations. Therefore the percentage may help compare across authorities with very different figures. Ideally we'd have been calculating the "excess deaths" but that wasn't (things may have changed since) available by local-authority on a regular basis and neither were the data required to calculate that. It may be possible to do a better job of "excess deaths" now but that is still affected by statistical noise, week shifts, and bank holidays affecting figures so we haven't switched to that.
Vaccinations
When NHS England started publishing vaccine data in late 2020 this wasn't available from the PHE API at lower-tier-local-authority-level by age group. Local authority data by age and date has since been added - vaccinationsAgeDemographics - so we are now able to use the API instead (we switched on 7th December 2021). We are showing values by vaccination date (which may mean values change over time) rather than by reporting date.
Population
Initially we used ONS mid-year 2019 population estimates. Then we switched to National Immunisation Management Service (NIMS) estimates. For the vaccine sections we now use the mid-year 2020 population estimates published by the Office for National Statistics (provided via the UKHSA API) for each age group.
Credits
© 2020-21 Open Innovations. Cases data comes from the UK Health
Security
Agency (formerly Public Health England), death data comes from ONS, vaccination data comes from NHS England (along with NIMS population data), and restrictions come from the House of Commons Library (until late December 2020). Code released under an MIT license. Source on Github.
