In April 2012, I organised the UK’s first Open-data Cities Conference, in my home city of Brighton and Hove.
As a former Fleet Street journalist, one of my motivations was a belief that data-journalism could transform local newspapers through the use of the internet and other technological innovations.
When I became co-owner of Brighton & Hove Independent in July 2013, I tried to show how “open data” — albeit mainly in print — could engage, inform and empower citizens. Specifically, I focused on data such as property prices, the 2011 census, city council elections, political funding, and planning.
Since my business partner and I sold the newspaper to Johnston Press in July 2015, I have been distracted — far too much! — by the politics of the Labour Party.
Thankfully, such distractions were already fading by the onset of the COVID-19 crisis in early February 2020 (when my local GP surgeries — one of them next door to my grandchildren’s school — were at the centre of the first local cluster of infections, one of the first in the country).
It was, however, not until Monday, May 10 that Covid-related data piqued my interest. That was the day that the government began to publish local data on its online COVID-19 dashboard.
By then, of course, we had all been accustomed to the daily Downing Street press conferences and the “next slide, please” presentations, which never drill down to what is happening at anything other than a national or regional level.
In this article, I seek to outline the (very limited) local COVID-19 data that is available and is relevant to anyone who wishes to assess the (declining) risk of infection — especially to people like me, a 63-year-old grandfather with close family members with significant lung or respiratory disorders. I will also highlight some of the many areas where timely local (and national) data is lacking.
To help validate my assertions — and, if necessary, to correct them — I will later publish a downloadable version of my (large but messy) master spreadsheet on Google Drive. Its contents — manipulated manually rather than programmatically — have been derived from several sources, especially the Office of National Statistics.
My starting point is the data relating to 316 lower-tier local authorities (LTLAs) in England, of which Brighton and Hove is one. Broadly, these exclude upper-tier authorities (UTLAs) such as county councils.
For each of these 316 LTLAs, we have downloadable daily data (CSV or JSON) relating to positive lab-confirmed tests (according to date of specimen).
The specimen date of the first such positive test was January 30 in York. The first in Brighton and Hove was on February 5.
Every day, the cumulative total of cases for each LTLA (and UTLA, region, and nation) is given in tabular form, along with the rate per 100,000 of the authority’s population. Inevitably, the rate goes up each day as the cumulative total increases while the baseline population remains the same.
These baseline population figures are not given explicitly, so I have had to calculate them using the cases and rate data given. For Brighton and Hove, the estimate is 290,456.
Without coding expertise, it is difficult — from the master .CSV file — to generate retrospectively such tables for each of the days since January 30.
Therefore I have taken only “snapshot” extracts of the daily tabulation (for example, to compare the increase in cases in each LTLA in the seven days to May 31.
First, though, let us take a look at the 20 LTLAs with the highest cumulative totals on May 31:
For comparison, the cumulative total for Brighton and Hove on May 31 was 441, putting it in 118th place out of the 316 authorities. (At the time of writing, the number up to June 6 was 468.)
By contrast, these are the 20 LTLAs with the lowest cumulative totals on May 31:
Clearly, these figures are significant only when seen in the context of the total number of people within authorities.
These are the 20 LTLAs with the highest rates of cases per 100,000 people on May 31:
For comparison, the rate for Brighton and Hove on May 31 was 151 cases per 100,000 people, putting it in 270th place out of the 316 authorities. (At the time of writing, the rate up to June 3 was 157.7.)
It is important to remember that this does not mean 151 in 100,000 people have the virus; it means 151 in 100,000 people have had a lab-confirmed positive test since February 5. The question of how many people in Brighton and Hove currently have the virus will be addressed shortly.
Meanwhile, though, these are the 20 LTLAs with the lowest rates of cases per 100,000 people on May 31:
Unfortunately, however, we do not know how many tests are being — or have been — conducted (under various testing regimes that have change over recent weeks and months) in each LTLA.
For now, it must be assumed — rightly or wrongly — that the number is of similar proportions to the populations in each authority. (On June 4, as few as 0.8% of tests in the UK were positive for Covid; more than 200,000 were negative.)
When comparing Brighton and Hove with other LTLAs, I have considered the 10 authorities with similar populations (280,000–300,000):
Density of population may be significant — along with transport connectivities, number of visitors and/or commuters, shopping habits and patterns, and so on.
Unsurprisingly, the 20 LTLAs with the highest densities of population are all in London:
By comparison, Brighton and Hove has a relatively high population density of 3,505 people per square kilometre, putting it in the top fifth of LTLAs. Even this measure, however, may not help signify anything important— since half of Brighton and Hove is in the South Downs National Park, thus increasing the population density in the city centre and its surrounding areas.
To reinforce the point, South Lakeland has one of lowest population densities in England — 69 people per square kilometre — but has one of the highest rates of COVID-19 cases in the country: 527 cases, or 504.2 per 100,000 population, on May 31.
There is little evidence to suggest that any particular age-group — except young children, marginally — is less likely or more likely to catch or spread the virus, either symptomatically or asymptomatically.
When it comes to age, however, we do know that COVID-19 has the most severe impact on those aged 70 or older.
It is perhaps heartening to know Brighton and Hove has a relatively young population.The 2011 census gives these numbers and proportions for each age range:
Nationally, LTLAs can be compared by using the “Old Age Dependency Ratio” (OADR), a simple ratio of the number of people of pensionable age and over per 1,000 people aged from 16 years to the state pension age.
North Norfolk has the highest OADR (62.6), followed by Rother (60.9), East Devon (56.6), and Tendring (55.5). Interestingly, South Lakeland is ninth-highest with 50.2 and nearly three in 10 (28.5%) of its population aged over 65.
Brighton and Hove comes 291st, with an OADR of just 18.7 and fewer than one in eight (13.4%) over 65.
The five lowest OADRs are all in “young” authorities in London: Tower Hamlets (8.8), Hackney (10.8), Newham (11), Lambeth (11.4), and Southwark (11.7).
In addition to population, density, and age, there are many other important differences in the socio-economic composition of LTLAs, including employment, income, ethnicity, and housing (the numbers and proportion living in space-restricted tower blocks — compared with those living in more open, leafy suburbs).
Comparisons between LTLAs (subject to the same government policies) is nearly as difficult as it is between nations (where governments have adopted different approaches). Even at a regional in England, the rate of positive lab-confirmed tests range (on June 5) from 139.5 in the southwest to 388.7 in the northeast; at national level with in the UK, the range is 253.8 in Northern Ireland to 456.1 in Wales.
In Brighton and Hove, it is worth considering the weekly and monthly figures — by specimen date — for the number of positive lab-confirmed tests for Covid:
Another way of looking at the figures is the rolling seven-day average of cases, ie by calculating, on any one day, the average number of new cases reported on that day and the previous six days (by specimen date).
Six new cases on June 3 pushed up the rolling seven-day average in Brighton and Hove to 4.6 cases a day, the highest since May 5 (when it was 5.6 cases a day).
The highest rolling seven-day average was 13.1 cases a day, on April 7 and 8:
This is how the (more volatile) number of daily lab-confirmed tests for Covid looks:
With the “lockdown” measures introduced on March 23, it seems certain that the peak — in Brighton and Hove, and in most places — was the result of infections that occurred some time during middle/late March.
So what has happened since the peak and where are we now?
A comparison of the changes in rates per 100,000 population gives some clues. Here are the 20 LTLAs with the biggest increase in rates between May 24 and May 31— headed by Ashford and Dover in Kent and followed by Blackpool and Fylde in Lancashire:
By this measure, in this period, Brighton and Hove comes 97th out of the 316 LTLAs, with an increase of 7.3 to 149.5 per 100,000 of population. This, however, is on a relatively low base: a cumulative total of 441 cases.
Here are the 20 LTLAs with the highest total cumulative number of positive lab-confirmed cases on May 31 were:
Before turning to death, it might be best here to deal with the R0 number — not least because of the wild speculation recently that “R=1.7” in Brighton and Hove, suggesting every Covid-infected person in our city allegedly was passing on the infection, on average, to 1.7 others.
This would, of course, be terrifying. Because, as we all know, we have to ensure R = <1 to stop the virus spreading exponentially. Thankfully, the reports based on a website called Deckzero were not true, not least because insufficient data exists to calculate R0 at such a local level. For reasons that As Alistair Hill, Director of Public Health for Brighton and Hove City Council, has explained.
According to Sir Patrick Vallance, the chief scientific adviser, the “natural” R = 3 (for reasons, I confess, I do not fully understand).
Understandably, there is a lot of focus on the R0 number. But much of it ignores the importance of the level of incidence. For example, if I am the only person in Brighton and Hove to have Covid and I pass it on to three members of my household, then R = 3. If the four of us recover after self-isolating for 14 days, then suddenly R = 0.
R can equal 1, even when only one in a million people has the virus. But that would mean only one in a million continues to have the virus.
In contrast, even if R is virtually zero, if one infected person coughs or sneezes in your face at close quarters, then you are very likely to be infected; the severity and consequences of any symptoms will, of course, vary.
Importantly, the national, regional, or local R0 number does not necessarily map directly onto individual risk of bad outcomes, including death (which depends significantly on age, ethnicity, obesity, and any underlying disorders).
One thing we can be certain of is that England and all its regions — and all its local authorities — have passed the (initial?) peak. Even if — as ONS data has previously suggested — the daily number of positive lab-confirmed tests still represent up to one fifth of all daily Covid infections.
With Brighton and Hove having had 43 positive lab-confirmed cases of Covid in the second half of May, the data could indicate the city has — very roughly! — a maximum(?) of about 215 people who are currently infected, ie 43 x 5, ie 0.07% of the city’s population. This proportion is lower than the the 0.1% in the revised ONS estimate — equivalent to 53,000 individuals infected in England — in the last week of May (when latest statistics suggest the number of new infections nationally were 5,600 a day).
With many of these 215 or so infected people being asymptomatic (and with many others presumably self-isolating), it is possible infections involve people infecting themselves after touching their faces, mouths and/or eyes after coming into contact with infected surfaces.
Differences in hygiene (especially hand-hygiene) and social norms and behaviours (relating to household formation, handshaking, kissing, and respect for personal space) must surely contribute to some of the differences in infection rates in and between nations.
Equally, improvements in hygiene and handwashing will — let us hope — have a beneficial impact in a post-Covid world, especially in schools and workplaces. We may, however, all still want to avoid touching handrails on escalators and door handles in toilets!
Separately, it is still a mystery why up to 99.2% of people — who subsequently turn out not to have Covid — come daily to take a test.
For all those tested —with both positive and negative results — it would be useful to know the breakdown by age, gender, employment, as well as by location or setting. Instead, because of the ill-thought-out application by politicians of the “test, test, test” mantra, we now have vast amounts of testing going on — on June 4, we passed more than five million!— without knowing any details beyond positive/negative and without knowing any lessons to be learned.
My guess is the sizeable majority of positive tests relate — as they probably always have — to elderly people (65+) in hospital and care homes, along with staff in hospital and care homes, as well as public-facing workplaces such as supermarkets.
In his letter to Health Secretary Matt Hancock (dated May 11), Sir David Norgrove — chair of the UK Statistics Authority — said “sole focus on the total national number of tests could mask helpful operational detail”.
Sir David also said: “Further breakdowns would provide more context, for example through showing the levels of testing by geographical area.”
Until this happens, I hope I have covered all the relevant geographical data that is available for tests and infections.
Finally, it is time to turn to the facts of death.
The most significant discussion of deaths caused by — or associated with — COVID-19 usually focuses on “excess” deaths.
It strikes me that we should perhaps refer instead to the number of deaths beyond the average as “premature”. Since the death rate for all of us 100%.
Inevitably, it is too early to calculate “excess” deaths; this cannot be done until the end of the year at the earliest — and, given comparisons are often on five-year averages, maybe not even until 2025.
Certainly, there is no information about local deaths that can estimate the “excess” locally.
Nor do all deaths above the average have to be caused by or associated with COVID-19 to qualify as “excess”. For example, someone with cancer or a heart problem who did not seek diagnosis or treatment but subsequently died — earlier than she would have done otherwise — will have contributed to the excess. As we will see, an above-average number of non-Covid-infected people with dementia or Alzheimer’s Disease and with diabetes appear to have died in recent months.
As we approach the end of the second quarter of 2020, we do, however, have some local data about deaths that may — or may not — be similarly premature.
In other words, the current “excess” in the first half of 2020 may include people who would have died in the second half of 2020 even if there had not been a pandemic.
This may be particularly relevant in care homes, where the median length of “stay” is only 19.6 months — with only 11.9 months for those in nursing beds. Overall, the average is 26.8 months.
In addition, the ratio between deaths in hospitals and deaths in care homes may be slightly different: some non-Covid residents who die in care homes may have been reluctant to go to hospital, while some terminally-ill care-home residents in hospital (with or without Covid) may have preferred to die in their care homes.
We have comprehensive — but time-lagged — ONS data about deaths. On June 5, ONS reported that, between March 7 and May 1, a total of 130,009 deaths were registered across England and Wales.
This represented an “excess” of 46,380 death registrations compared with the five-year average; 12,900 (27.8%) of these deaths did not directly involve COVID-19 (ie the disease was not mentioned on the death certificate — and, presumably, no positive test had been recorded).
The ONS reported: “The largest increases in non-COVID-19 deaths compared to the five-year average are seen in deaths due to ‘dementia and Alzheimer disease’ and ‘symptoms, signs and ill-defined conditions’ (the latter mostly indicating old age and frailty); overall, there have been 5,404 excess deaths (an increase of 52.2% on the five-year average) due to dementia and Alzheimer disease and 1,567 excess deaths (an increase of 77.8%) due to ‘symptoms signs and ill-defined conditions’ from Week 11 (ending 13 March) to Week 18 (ending 1 May), which together comprise two thirds of total non-COVID-19 excess deaths in this period.
“Deaths due to causes such as asthma and diabetes increased up to the week ending 24 April 2020 and occurred increasingly outside hospital; this could suggest a delay in care for these conditions is leading to an increase in deaths, although this rise could also be related to undiagnosed COVID-19.”
We must always remember a lot of people die in the UK every year, nearly 620,000 in 2018 — the latest full year for which local death data is available. In Brighton and Hove in 2018, a total of 2,231 people died — an average of 186 a month.
Up to May 22 this year, 961 people died in the city — an estimated average of just over 192 a month; the five-year average for people dying in Brighton and Hove for the same 21 weeks is slightly higher, at 969, ie an average of just over 194 a month.
When you divide the deaths up between Weeks 1–11 and 12–21 (the weeks since the first local death associated with Covid), the figures for the five-year average are 560 (57%) and 409 (43%), respectively.
For 2020, however, the figures are 436 (45%) and 525 (55%), respectively.
In other words, the below-average deaths in Weeks 1–11 in 2020 are compensated for by the above-average deaths during the Covid crisis in Weeks 12–21.
Of the 961 deaths up to May 22, 353 (37%) occurred in hospital; 278 (29%) in care homes; 223 (23%) at home; 75 (8%) in hospices; 22 (2%) in “other communal establishments”; and 10 (1%) “elsewhere”.
For the five-year average of 969 deaths in the same 21 weeks, the proportions in each setting is broadly similar: 388 (40%) occurred in hospital; 242 (25%) at home; 221 (23%) in care homes; 78 (8%) in hospices; 27 (3%) “elsewhere”; and 13 (1%) in “other communal establishments”.
In contrast, it is interesting to compare the settings of the 145 Covid-associated deaths in Brighton and Hove, since the first (in hospital) in the week ending March 20: 78 (54%) occurred in hospital; 54 (37%) in care homes; 7 (5%) in hospices; 5 (3%) in “other communal establishments”; and 1 (0.7%) at home. Here is the full weekly breakdown:
Some 149 local upper- and lower-tier local authorities that have responsibility for care homes reported 11,186 Covid-related deaths to the Care Quality Commission (CQC), out of a total of 28,450 care-home deaths, from April 10 to May 29.
Brighton and Hove is 82nd out of the 149. According to the CQC website, the city has 93 care homes — of which 61 care only. for people aged 65 and older.
The 20 authorities with the highest number of Covid-associated deaths in care homes (from April 10 to May 29) are:
The 20 authorities with the highest proportion of deaths associated with Covid in care homes (from April 10 to May 29) are:
Brighton and Hove is 74th out of the 149, with 41% of deaths in its care homes associated with Covid, from April 10 to May 29.
The range, across the country, in the proportion of deaths associated with Covid in care homes raises questions about how accurately and diligently some or all care homes in some authorities have managed to report Covid-associated deaths to the CQC. But this surely cannot explain the wide disparities.
If you have got this far, you probably agree that we have had enough data. It is time to draw a few conclusions. And raise a few questions. But that is for another article: Wash your hands, stay safe — and stay socialist! Some reflections on data and COVID-19.