The most vulnerable parts of the UK have largely been spared from the impact of Covid-19 so far, according to a new analysis of excess death data by the New Statesman.
Despite the UK this week registering more than 50,000 excess deaths since the coronavirus crisis began, the country remains widely susceptible to further and more severe outbreaks in vulnerable areas if lockdown is lifted too hastily, according to the analysis, which highlights the risks that Boris Johnson and his government face as they take steps toward restarting the British economy.
The following animation shows the impact of the virus across England and Wales over the first eight of the Covid-19 outbreak (new data, released every Tuesday, is always a fortnight behind).
Each year around 530,000 people die in England and Wales. Over the eight weeks in question, we would have expected around 81,500 people to have died. In reality, 131,468 did so. Thus in those first eight weeks of the UK’s outbreak, there were around 50,000 excess deaths across England and Wales (detailed ONS data is not available for Scotland and Northern Ireland).
Around 37,300 of these excess deaths, or three-quarters of them, were caused by Covid-19. Any potential links between Covid-19 and the remaining 12,700 are as yet unknown. They may be undiagnosed cases of coronavirus, or they may have been caused by lockdown more broadly, as fewer people attend hospitals.
In the animation below you can, using a simple model we have built, compare the geographical impact of the 37,300 Covid-19 deaths and the 12,700 additional deaths (the number of excess deaths is equal to Covid deaths plus additional deaths). Press “play” in the bottom right of the animation to see how the outbreak has unfolded.
This map, created by the New Statesman, is the first to estimate the impact of excess deaths at a local level. Note that local authorities vary greatly in size; the population of Birmingham is nearly 30 times larger than that of Rutland. The best way to see the impact in each area is per 100,000 people.
The worst-hit areas by this measure are, as one would expect, largely concentrated around major cities: especially London and Birmingham, with some clusters too around Liverpool, Manchester and Cardiff. These include London commuter areas such as Hertsmere, Mole Valley and Epping Forest; and Solihull, Walsall and East Staffordshire around Birmingham.
But these are not the areas most at-risk from coronavirus. In the wake of 50,000 excess deaths, it is easy to lose sight of a simple fact: the most vulnerable parts of the UK have not, at least relatively, been badly hit by Covid-19 so far.
Some parts of the UK have experienced fewer than 1 per cent of the deaths they would incur in an unmitigated Covid-19 epidemic, according to assumptions made by a model built by the New Statesman. Yet these lightly impacted places remain at great risk to the spread of Covid-19.
In contrast, as many as 25 per cent of expected deaths (under an unmitigated pandemic) have already happened in the worst-hit parts of the country – or rather had happened by 8 May, the most recent day for which this excess death data is available.
We have a good idea of which areas are most vulnerable to Covid-19 because the disease, as the statistician David Spiegelhalter has detailed, “seems to take any frailty and multiply it”. Such frailty is in large part determined by age.
Nearly 99 per cent of those who have died with or of Covid-19 in England and Wales have been at least 45 years old, and 83 per cent have been at least 70 years old. Every other factor – healthcare resources, deprivation, ethnicity – pales in comparison. As Spiegelhalter puts it, there is “a fairly precise exponential increase [of dying from Covid-19] with age”.
The New Statesman has therefore built a simple model that calculates each area’s vulnerability to Covid-19 by looking at its typical death rate over the past five years. These typical death rates are largely explained by age (population age accounts for nearly 80 per cent of the difference between areas).
Areas with higher typical death rates in the years prior to coronavirus – such as Tendring, home to Clacton and Jaywick, where almost a third of residents are older than 65 – are rated as highly vulnerable to Covid-19. Areas with lower yearly death rates and younger populations, such as Newham in London, are rated as far less vulnerable.
In a typical year, four times as many people (per 100,000) die in Tendring than in Newham. Our model therefore suggests that, if everyone in both places was to contract the virus, four times as many people would die in Tendring than in Newham. The overwhelming reason Tendring has a higher death rate is that it has around four times the proportion of pensioners.
Imagine that the infection fatality rate (IFR) – the percentage of people who die after being infected – of Covid-19 is 1 per cent. This is a best guess of current estimates. It means that 1,000 people in every 100,000 would be at risk of death across England and Wales if everyone was to contract Covid-19. (We are assuming here that all excess deaths are caused by Covid-19; if lockdown also causes deaths, the nationwide number will exceed 1,000 per 100,000.)
But in Tendring our model, using historic death rates, estimates that rate to be as high as 1,675 per 100,000. In Newham, it is as low as 425. Yet these two areas have been almost equally impacted by Covid-19 thus far: Tendring has had 109 excess deaths per 100,000; Newham 101.
In other words, nearly 25 per cent of the deaths we would expect to happen in Newham have already happened. But only 7 per cent of the deaths that our model “expects” in Tendring (if everyone was to be infected) have actually happened.
The graphic below compares the actual number of excess deaths in each local authority across England and Wales (in blue) with the number of deaths we would expect to see in that area if Covid-19 had infected everyone (in red).
Each line represents a local authority. Tendring is the left-most point, with the highest expected impact. Newham is on the far-right (only Tower Hamlets has a lower expected impact).
This data allows us to see the extent to which the UK has been hit by Covid-19 so far. The upshot is simple: the blue lines are a fraction of the red lines. In other words, Covid-19, despite having caused around 50,000 excess deaths by 8 May, has only lightly affected England and Wales relative to its potential impact.
There have been 85 excess deaths per 100,000 across England and Wales. If Covid has an IFR of 1 per cent, the potential impact is 1,000 per 100,000: a dozen times greater than the impact so far.
We can see at a glance that the areas that have been worst-hit by Covid-19 are not the areas that face the greatest potential threat from the disease. If they were, the blue lines would follow the same curve as the red lines.
In reality, the blue lines – which show the actual impact of Covid-19 – tend, if anything, to be higher on the right-hand side of the graph: in the areas that face less of a potential threat from Covid-19. The most vulnerable local authorities have, on the whole, not yet been badly hit.
These numbers are not precise but indicative. They are a way of illustrating the risk that the UK continues to face, and the great gulf in impact across the country thus far. In the areas of Mid Devon and North East Lincolnshire, for instance, fewer than 1 per cent of the deaths one would expect from an unmitigated pandemic have happened. These areas have not experienced any meaningful surge in excess deaths.
Both have had only 7 excess deaths per 100,000 people. Yet if coronavirus infected everyone in these areas, the former is, according to the model, at risk of 1,098 such deaths, and the latter of 1,214.
Ninety nine per cent of possible deaths in these areas have not yet happened. The hope is that they never happen. But a strict lockdown will likely have to stay in place for that to remain the case, at least for the foreseeable future.
The UK’s peak has passed, as the graphs below suggest. The most recent data, released 19 May, shows half as many excess deaths in the first week of May as in the last week of April (although that May data is 15 to 20 per cent lower than in reality, as no data was collected on VE Day, honourably if unhelpfully). And the most severe outbreaks around the country have been brought under control.
But only the district of South Lakeland, the rural home to Lake Windermere, has a particularly old population among the worst-hit areas. Most of the UK’s areas with large elderly populations have so far been spared.
There is no link, in the data to date, between an area’s proportion of elderly residents and its level of excess deaths. If anything, older areas, being less densely populated, have fared better.
Major cities and their satellite towns have borne the brunt of the UK’s Covid-19 outbreak, or at least its first wave. The risk the country faces is that future waves affect far more endangered areas, and lead to greater surges.
The UK government, having failed to protect care homes, is likely to be more vigilant about protecting these areas. That is likely to mean a summer of limited movement for everyone else, at least until the country acquires the ability to test, trace and isolate every case of Covid-19 across the UK.
With additional reporting by Josh Rayman, the creator of the animation above.