Image: Pexels, maps/visualisation by The Working Group
COVID-19 disproportionately affects people who live in lower income areas. Those findings are clear from the latest Office for National Statistics (ONS) release.
But we wanted to dig further into what various datasets tell us. And see whether any other contributing factors stood out.
ONS figures released on 12 June
Firstly, let’s examine those updated statistics. Particularly which datasets have been used to compile them.
They’re based on provisional COVID-19 related death figures, between 1 March and 31 May - in England and Wales. And deaths are recorded in each MSOA (Middle Super Output Layer) via postcodes on death certificates.
The parallel between income and impact has been highlighted in an ONS led analysis. But some interesting anomalies appear, on further investigation.
Let’s start by looking at the range of datasets available. And the various stages our own analysis of those took.
Using the ONS data release, we first made an OpenStreet interactive map of COVID-19 related fatalities https://weirdos-and-misfits.gitlab.io/covid-map/):
Animation - weekly number of deaths
We then created an animation of weekly death rates, using NHS and private trusts’ statistics (published on the ONS website),
It’s worth noting that locations of death, in this case, are recorded in terms of the hospital or care-home they occurred in. And we’ve used this as the main dataset for our animation below. (The hex map is taken from Oli Hawkin's blog post)
Worst affected regions
The worst affected area is shown to be Crabtree and Fir Vale in Northern Sheffield. And this might initially point straight to the link with deprivation we’ve discussed. Or, perhaps it’s related to the Northern General Hospital (NGH) being located here (deaths are attributed by home address postcode).
However, the large spike in the area has been widely reported to be linked with local care homes. (An investigation is ongoing into those issues themselves).
So you can see straight away that initial assumptions can be problematic. And often several factors are actually at play. What is probably most likely in Crabtree and Fir Vale is that all of these issues, in combination, are responsible.
|ONS geography MSOA name||House of Commons Library MSOA Names||Covid Total||Non-Covid Total||All Total|
|County Durham 008||Stanley South||31||35||66|
|Gateshead 002||Crawcrook & Greenside||31||21||52|
|Bromsgrove 011||Bromsgrove Central & Sanders Park||31||32||63|
|Watford 007||Nascot Wood||32||40||72|
|Leeds 073||Colton, Austhorpe & Whitkirk||32||16||48|
|Cheltenham 007||Alstone & St Mark's||32||37||69|
|Warrington 017||Fairfield & Howley||32||35||67|
|Mid Sussex 009||Haywards Heath West||33||44||77|
|Brent 025||Church End||36||21||57|
|Sheffield 020||Crabtree & Fir Vale||66||38||104|
For all of these MSOAs, the number of deaths can be compared with ONS data on Multiple Deprivation and Total Population.. An example case for Sheffield is plotted below:
And although it's clear that deprived areas are, on average, the most affected (in terms of deaths per 100k people). Further analysis throws up a range of areas for further study. For example, those with a younger population have clearly felt the most impact.
The focal point of our next investigation was to group data into parliamentary constituencies. And, interestingly, a clear visual pattern emerged across the country between April and May.
This is apparent here - on the ODI Leeds' "Hex Map" of parliamentary constituencies.
By far the biggest overall impact (in terms of total number of deaths and total % number of deaths) has been highlighted across areas of North and North-West London.
But, when data is split into Labour or Conservative voting areas (based on 2019 general election data), some other interesting patterns start to emerge. Clear correlations and trends are highlighted - directly linked to constituency demographics.
Labour voting, urban areas, often with a higher % of non-white communities, are typically more affected in terms of the % number of deaths from COVID-19. They also have a higher IMD (deprivation) score.
These areas have, however, seen a greater decrease in percentage of total deaths between April and May 2019, And were also those affected first, and most severely, by the crisis.
We then looked at the levels of funding each Local Authority received, to cope with the impact of COVID-19. Figures released by the government on 28 April show just how uneven this allocation has been.
The £3.2 billion total was shared in two tranches of £1.6 billion (on 19 March and 18 April). Approximately £670 million (£540 million in the second tranche) was given to county and fire authorities.
The remaining amount was distributed across local authorities. This was around £1billion in total (£1,031,336,432 in the first tranche, £1,156,133,598 in the second).
These authorities were first given a total of £9,742,549 on the 19th March, followed by £214,031,274 on the 18th of April. The areas, which had a higher number of COVID-19 cases were first given £1,021,593,883 followed by £942,102,324.
A summary of these figures is given in the following table:
|First Tranche of Covid-19 Funding||Second Tranche of Covid-19 Funding||Total Covid-19 Additional Funding||COVID-19 Deaths (March)||COVID-19 Deaths (April)||COVID-19 Deaths (May)||Total COVID-19 Deaths|
|No funding boost||£ 1,021,593,883||£ 942,102,324||£ 1,963,696,207||3,275||18,864||6,524||28,663|
|Large funding boost||£ 9,742,549||£ 214,031,274||£ 223,773,823||1,219||9,864||4,243||15,326|
When you look at the geography and demographics of the regions that received more in the second tranche, it’s clear they’re typically more rural - and conservative voting:
There’s lots more to be examined - including the ways each tranche of funding was calculated. But that’s for another post...
What is clear, though, is that the government distributed monies quite differently - between these two local authority ‘classes’.
The first - generally more urban, with a high density population (and more labour voting) received least funding. And was hardest hit in terms of total numbers of deaths associated to COVID-19 during March and April.
The second - more rural - received a large bump in funding, and appears to have a higher proportional number of deaths from COVID-19 in March.
The following plot shows that the second group suffered a higher number of deaths per 100k people (red solid line v.s. red dashed line) despite having a lower total number of COVID-19 deaths (blue solid line versus blue dashed line) in April:
It was also quite clear that the % funding change between the two tranches was greater or smaller for certain authorities:
So what does this all tell us? Well, the situation is complex - with many underlying factors and trends underpinning it. It’s true, of course, that deprivation is a major causal factor in high Covid 19 death rates. But there also appears to be a clear correlation with local funding levels.