COVID-19 mortality rates: Looking for perspective

Never before has reading the runes of the national mortality statistics had such direct relevance to our daily lives. With everyone affected by how we respond to the toll the coronavirus is taking and how that may change, it has become a matter of intense debate as to how to quantify that toll in the first place.

It doesn't help that collation of data at speed isn't easy, nor is really understanding what it represents. Hardest of all is the degree of unpredictability of what comes next because we haven't been here before.

Nonetheless, some startling high level conclusions can already be drawn.

Excess weekly deaths up 76%

We are unlikely to get an accurate break down of actual COVID-19 and related deaths for a long time, if ever. But one thing we can be sure of is that the spike in deaths above expected rates can be attributed to the consequences of the pandemic one way or another.

The most reliable gauge of this appears from the total of excess deaths above those expected, published in the weekly Office for National Statistics (ONS) report of all deaths registered in England and Wales.

We began to see the effect of the virus beginning to really bite in the deaths registered by 3 April - week 14 (many of which actually occurred in the previous week). Overall, these were up by 6,082, nearly 60%, compared to the previous five year average for that week. (The week before, it was up 1,011, just 10% higher.)

This week's ONS update has the week ending 10 April showing 7,996 excess deaths, 76% up on the previous five years. We could actually say it was over 80% higher than if we had no pandemic, as we had begun the year with death rates trending five or 6% lower than the average previous five year figures.

Daily total announced deaths understated by about half

Most confusion and controversy has surrounded the total COVID-19 deaths count. Although for some weeks, the UK's daily announcements of deaths have been specified to be those in hospitals, the realisation has been slow to dawn that a much larger total has been clocking up when deaths elsewhere, especially in care homes or at home, are added in, along with those late in being reported. The same issue applies in many other countries, making meaningful tracking of totals and comparison amongst countries more difficult.

On Tuesday, the headlines featured a 41% disparity between the 9,288 daily hospital deaths announced and registered at the time for England and Wales, and the total 13,121 deaths occurring by 10 April registered by 18 April at all locations with COVID-19 mentioned. Put another way, announced COVID-19 deaths represented only 71% of the totals later registered as such.

In fact, the overall regular understatement is even greater than that, giving us only about 50% of the ultimate total of excess deaths. Apart from the obvious omission of non-hospital deaths, the inevitable time lag in all cases being reported is one factor. Also, various factors, including minimal testing, have led to unreliable reporting of COVID-19 on non-hospital death certificates.

Simply looking at overall excess deaths compared to previous weeks or the same weeks in previous years, it's clear we have substantial numbers of surplus deaths beyond those certified as related to COVID-19. We had a total of about 15,000 excess deaths registered by 10 April, 45% more than the 10,350 noting COVID-19 was involved. The actual occurrences for the latter up to 10 April had grown 27% once late registrations received by 18 April were included. Assuming the same proportion would roll in for overall excess deaths, that would give about 19,000 actual excess deaths occurring up to 10 April - just over double the announced 9,288 total COVID deaths from hospitals.

Where is the surplus coming from?

Comparing death rates in early March, it's clear from the ONS figures split by location of death that COVID-19 hospital deaths account for only 56% of recent excess deaths. The rest were mostly split between care homes (26%) and people dying in their own home (17%), whether certified as COVID related or not. The former group can be expected to be almost entirely COVID-19 victims. The latter group must be a mix of COVID-19 cases along with indirect results of the pandemic - people dying of other ailments due to not getting the care they need for things like heart attack and stroke.

The latest UK hospital data shows 18,100 COVID-19 deaths by 21 April have been reported so far. If the same proportions apply as before, this implies a UK total of COVID-19 deaths of over 36,000 may have occurred by 21 April. Wednesday's estimate in the FT is even higher at 41,000.

Men hardest hit

Ahead of the death registration data coming out, a nasty surprise for men emerged from the patient details for those unwell enough to go into critical care (ICU in old money). 

Of all patients seriously ill enough with COVID-19 to be admitted to ICU, 72% were male out of the 5578 patients reported by 17 April. Even worse, of the men with a reported outcome, only 47% survived, whereas the ladies had a better chance, with 55% of them surviving. Put another way, men accounted for 75% of all ICU deaths.

Middle aged and elderly at similarly increased risk?

It's well known and to be expected that the chances of dying from the virus increases with age. Once reaching ICU, the survival rate for over 70s has been only just over 30%, compared to rates ranging from 43% to 78% for younger age groups. (Even so, over 70s only accounted for 39% of ICU deaths, with relatively fewer of them accessing ICU in the first place.)

What's less well known is the idea that for adults of all ages, whatever our pre-COVID-19 risk of dying in the current year was, it's gone up by a similar proportion across all ages. By how much that will be once this is all over, we don't know. But speculation is that a similar proportion of increased deaths will have occurred across all age groups.

Looking at the excess deaths so far, this is far from being proven, but looks like it might be broadly true for ages over 45.

Deaths registered in England and Wales in the four weeks ending 10 April are up by 37% for all ages and both sexes, compared to the previous five year average. For each 10-year age band starting at 45, comparing the three weeks ending 10 April with the previous three, deaths went up by 32%, 39%, 41%, 49% and 41% respectively.

Will insurers notice a rise in death claims?

Obviously, no-one knows how many extra deaths we'll have due to the pandemic this year. But as we're only just reaching the peak, we know it's going to be more than the c.30,000 we've already probably reached in total.

If we have over 50,000 extra deaths this year, that would be about an extra 10%. If evenly spread across the age groups, that could mean an extra 3,500 death claims under term assurance cover paid by group and individual protection insurers. For whole of life cover, 121,822 claims were paid in 2018. If a similar number of policies are in force this year, that's an additional 12,200 claims to be expected.

Not just like flu

The more we learn about the virus behind COVID-19, the more it becomes clear how different it is from flu. It's not only different in how it works and being much more infectious and more deadly when you get it, it also hits different groups of people the hardest. However, comparing it to our experience of flu is the best benchmark we have to set as a baseline for expectations. 

From making this comparison, the weekly critical care reports from the intensive care national audit and research centre (ICNARC) show that as well as the over-representation of men and younger age groups amongst the greatest sufferers, compared to flu it seems to be more dangerous for:

• Patients with high blood pressure or diabetes
• Patients (likely to be included in the above) with high BMI
• People of Black, Asian and mixed ethnicity (BAME) vs White

A follow up article will explore some of the prevailing medical theories and how they might contribute to explaining some of these discrepancies.

Ruth Gilbert heads up insuringchange.co.uk