While excess deaths everywhere are far below Covid peaks, many countries are still seeing a sustained excess this summer, suggesting a common factor is causing unexpected deaths in dozens of countries.
But all may not be as it seems...
As the likes of @ActuaryByDay, @john_actuary and @ONS have demonstrated, we can do better than raw, unadjusted excess mortality.
We can use age-standardisation to adjust for the changing age structures of populations, and for which age groups our excess deaths are found in.
Once we standardise mortality rates over time, we see that shifts in population age structures since 2015 explain a large amount of the observed excess.
The corrected data show only a handful of countries now recording an excess, where before the majority were above zero.
This is good news! 🎉
What we thought was months-long unexplained excess mortality in dozens of countries can be explained in most cases by population growth and demographic churn.
But let’s look closer. England is still recording an age-standardised excess, and it’s not alone.
Zooming in on recent months, comparing raw to age-standardised mortality, a handful of countries have a sustained excess from late spring thru summer.
England & Wales are joined by Spain & Portugal. Given negative excess elsewhere, this is worrying, so what could be driving it?
Some of you may be wondering: could it be the heatwave?
Well, yes and no. I’ve deliberately cut off the time series before the record-breaking English heatwave of July 18-19th, so heat can’t explain the English excess.
But what about Portugal and Spain?
It’s highly likely that the Iberian peninsula’s prolonged period of extreme heat has played a major role.
All focus in England was on the record-breaking July 18-19th heatwave, but Portugal & Spain saw temperatures at or above 40ºC for extensive periods earlier in the summer.
So with extreme heat likely to be playing a major role in Iberian excess mortality, that leaves England & Wales as the only part of Europe that has recorded sustained and rising excess mortality over the last few months with no obvious natural explanation.
This suggests we’re dealing with English exceptionalism 🙃, and is arguably sufficient to refute the original idea, hinted at by raw non-standardised excess deaths, that a common factor was causing unexpected deaths in dozens of countries.
So what’s causing the English excess?
To answer that question, we turn to data published by @OHID, which breaks down English mortality by cause of death, allowing us to split out Covid from non-Covid deaths for the first time in this analysis gov.uk/government/sta…
For the first year of the pandemic, Covid deaths and excess deaths were one and the same (very early in the pandemic, some Covid deaths were missed by coroners), but in summer 2021 and 2022 we’ve seen sustained excess deaths over and above those where Covid played a direct role.
Here’s the same data but subtracting Covid deaths from total excess deaths, leaving non-Covid excess
During the last two winters we actually saw far fewer non-Covid deaths than usual, with flu largely at bay, but summers have seen more people dying than usual, and not from Covid
So what’s going on?
One theory is that it still *is* Covid, just not acute Covid. Elevated risk of heart problems in the months following a Covid infection have been posited as one explanation, for example nature.com/articles/d4158…
Here are non-Covid excess deaths by cause of death mentioned on the death certificate.
At first glance, the indirect-Covid theory is borne out, with a much larger rise in the number of additional deaths from heart and circulatory diseases than from other conditions.
But...
That fails to take account of what people usually die from, and once we switch from nominal to relative excess, i.e expressing increased deaths as a % of typical levels, the cardio-vascular dominance disappears, instead leaving a broad-based increase across many conditions.
Multiple key patterns here:
1) We know there was significant disruption to cancer treatment and screenings during pandemic, but given the relatively slow progression of cancer, that is not yet feeding through into increased cancer mortality. Unfortunately it may well do later.
2) The increases in deaths across cardio-vascular, liver and urinary diseases, diabetes and beyond suggests that rather than a single specific condition being exacerbated by e.g prior Covid infection, we’re looking at a more broad-based mechanism here...
...the sort of mechanism that might be explained by the sudden collapse of emergency care, leaving people suffering from these conditions having to wait hours for urgent treatment, if they are treated at all.
And that brings me onto the direct evidence for the role of England’s current NHS emergency care crisis in the country’s sustained non-Covid excess mortality.
Last year, a team of healthcare researchers including @sib313 published a vitally important study showing the increased mortality risk resulting from exceptionally long waits in A&E before receiving treatment emj.bmj.com/content/emerme…
Patients who waited 8-12 hours had a 16% higher chance of dying in the subsequent 30 days than average.
This was after adjusting for a huge range of possible confounders, i.e this was not due to those patients’ characteristics, conditions etc, but due to the length of the wait.
Crucially, we can overlay that elevated risk gradient onto A&E waiting times data, and use it to estimate how many additional deaths may have been caused by exceptional A&E waits in recent months.
You may have heard almost 30,000 people waited 12 hours in English A&Es in July, but that figure actually only refers to the wait *after* initial assessment
Thanks to @Rebeccasmt’s reporting, we know that if you include all time spent waiting, 100,000 people waited 12+ hours ‼️
So we can use that data to calculate the number of additional long A&E waits that we’re currently seeing over and above the pre-pandemic level, and multiply those by the increased mortality risk associated with such waits, which gives us...
Our estimate of as many as 500 non-Covid excess deaths every week in England, due to the dire situation in emergency departments.
These will be people with conditions including but not limited to heart problems, complications from diabetes or liver disease... sound familiar?
The prior demonstration of a direct and precise association between A&E waiting times and mortality risk makes these data the best for estimating the mortality impact of the current NHS crisis, but other data show very similar patterns.
Take ambulance delays, for example: for emergency situations including suspected strokes and heart attacks, the *average* wait for an ambulance to arrive on the scene is now one hour, and 40,000 people with these sorts of emergencies waited 2 hours last month.
@tomhcalver has shown that these delays cause real harm to patients, in some cases directly contributing to loss of life thetimes.co.uk/article/6311fc…, and as with A&E waits we can use the data to get an idea of how much "excess ambulance-delay-related harm" is currently being caused
As before, the peaks in ambulance-related harms broadly coincide with peaks in England’s non-Covid excess mortality.
It’s a grim picture, and an increasingly conclusive one.
So if we can be increasingly confident that the crisis in emergency departments is the proximate cause of a substantial portion of England’s excess mortality, then what is the underlying cause?
Why are A&S so full? Why are ambulances so stretched?
One common theory is that more people must be getting sick, and those who are, must be getting sicker. English A&Es must simply be overrun, dealing with the fallout from the pandemic.
But the data don’t bear this out.
First, we can look at how many people are turning up to A&E. Numbers have just about recovered to the pre-pandemic trend, but they’re certainly no higher.
Second, we can look at the share of A&E attendees who end up being admitted to hospital, as a proxy for the severity of the conditions people are presenting with.
This is now below the pre-pandemic trend, suggesting no increase in acuity.
If the number of people needing urgent care is not abnormally high, and nor is the severity of their conditions, this underscores that the problem is what happens when people get to A&E / call an ambulance.
So why are delays and therefore outcomes so bad?
The answer is not found in A&E, it’s found in the rest of the hospital.
Specifically, in the acute shortage of available beds for emergency arrivals.
Just 7% of English hospital beds are currently unoccupied, the lowest figure since records began, and far short of safe levels.
You can’t run hospitals like a "just in time" operation, keeping spare capacity to a minimum.
You might be able to plan and schedule elective appointments, but you can’t plan and schedule emergency arrivals. They require a healthy amount of spare capacity
So, the problem driving the A&E crisis is a lack of available beds elsewhere. Well, what’s causing that?
Delayed discharges of patients.
You’ll recall just 7% of beds are unoccupied. Well almost double that — 13% — are occupied by patients who are waiting to be discharged.
This is another record high, and amounts to more than 13,000 beds that are occupied by people who no longer need to be in hospital.
Why have they not been discharged yet? This one is a little more nuanced.
The most commonly cited reason is a shortage of social care resources, be it spaces in care homes, home care packages, support/funding etc, with these insufficiencies worsened by a decade of stagnant social care funding and the rapid growth in the population needing its services
But while social care is absolutely a major factor in delayed discharges, it’s far from the only one, and many others are arguably lower-hanging fruit.
Many patients are awaiting decisions from hospital staff, and the wider issue of patient flow is a long-running concern.
Patient flow is hampered by factors including woefully outdated IT systems, whose impacts are both direct — on the speed of moving patients through the system — and indirect — keeping doctors from treating patients thetimes.co.uk/article/nhs-pa…
Outdated IT systems are just the tip of the iceberg when it comes to chronic under-investment in the long-term of the NHS
While UK spends a similar share of GDP on healthcare vs peer countries, this overwhelmingly goes on short-term, neglecting both preventive and social care...
...and also under-investing in infrastructure, leaving British hospitals far more poorly-resourced than those of peer countries
And don’t just take my word for it on all of this:
A wealth of reports on past A&E pressures have concluded that the causes stem more from hospital capacity shortages and issues with social care and patient flow assets.publishing.service.gov.uk/government/upl…
Finally, one other area that will undoubtedly be having an impact, is the ongoing unseasonably high levels of NHS staff sickness absence, due in part to Covid.
Bed availability, patient flow and A&E staffing levels themselves all suffer when the staff are too ill to work.
So, to conclude a gargantuan thread:
(1/4)
• Excess deaths are currently over-stated in many countries due to a failure to age-standardise
• But in England there remains a significant non-Covid excess which cannot be explained by extreme heat, unlike other countries
(2/4)
• Delays in urgent & emergency care are almost certainly a large factor
• These stem from a shortage of capacity elsewhere in the hospital system
• Which in turn stems from under-investment in social care, IT and other infrastructure, plus other issues with patient flow
(3/4)
• This is not an exhaustive list, either on the global excess death picture or on the pressures facing the NHS
• Please let me know if I’ve missed anything, or if there are other areas I should be digging into
• I’m now going to lie down for approximately a week
(4/4)
• To state the obvious, in the weeks of work that went into this, I’ve not encountered any evidence for either vaccines or lockdowns playing any role in this excess mortality
• This is not to dismiss wider impact of lockdowns (cc @snj_1970), just saying no clear role here
Oh, and to pre-empt what I suspect may be an FAQ:
After searching high and low, I don’t believe US publishes sufficiently granular all-cause mort data to allow for full age-standardisation. I ideally need weekly deaths by 5-year age-group, 2015 to now. Any pointers appreciated!
Shout-outs
• @sib313 for being an immensely patient sounding-board
• @ActuaryByDay for making sure I didn’t screw up the age-standardisation
• @rebeccasmt for being so helpful with A&E waits data
• @AnitaCTHF for vast amounts of help on several stories featured in this thread
• @NHSEngland & @NHSDigital, for putting out so much granular data on so many facets of the NHS. Just like with @UKHSA@ONS during pandemic, I’m astonished at how much NHS data is available to UK public. We may be in a crisis, but at least we can see what’s happening.
<ends>
Oh, and one aside:
Due to the nature of the job, I’m writing more pieces than ever, but I have a lot less time for tweeting them out, so please check here for the latest, including health, climate, immigration, obesity and most recently energy bills
👇 ft.com/jbm
What else should I write about over the coming weeks?
Health, climate, economics, you name it, what are the biggest under-reported stories today? Especially ones that extend beyond/outside UK?
Reply, DM or john.burn-murdoch@ft.com
Always immensely grateful for suggestions 🙏
• • •
Missing some Tweet in this thread? You can try to
force a refresh
My wish for the next election is that poll trackers look like the one on the right 👉 not the left
This was yet another election where the polling showed it could easily go either way, but most of the charts just showed two nice clean lines, one leading and one trailing. Bad!
Pollsters and poll aggregators have gone to great lengths to emphasise the amount of uncertainty in the polls in recent weeks...
But have generally still put out charts and polling toplines that encourage people to ignore the uncertainty and focus on who’s one point ahead. Bad!
The thing about human psychology is, once you give people a nice clean number, it doesn’t matter how many times you say "but there’s an error margin of +/- x points, anything is possible".
People are going to anchor on that central number. We shouldn’t enable this behaviour!
We’re going to hear lots of stories about which people, policies and rhetoric are to blame for the Democrats’ defeat.
Some of those stories may even be true!
But an underrated factor is that 2024 was an absolutely horrendous year for incumbents around the world 👇
Harris lost votes, Sunak lost votes, Macron lost votes, Modi (!) lost votes, as did the Japanese, Belgian, Croatian, Bulgarian and Lithuanian governments in elections this year.
Any explanation that fails to take account for this is incomplete.
Many of the NHS’s difficulties can be traced back to the deep cuts in manager numbers.
Fixing this doesn’t just unblock waiting lists, it also gives doctors more time to be doctors, and alleviates the stress and poor morale that come from having to do things that aren’t your job
Here’s another fun NHS low hanging fruit example:
A trial last year found that by running two operating theatres side by side, they cut the time between operations from 40 minutes to 2, and were able to do a week’s worth of surgeries in one day thetimes.com/uk/article/lon…
In what might be one of the most significant trends I have ever charted, the US obesity rate fell last year.
My column this week is about this landmark data point, and what might be behind it ft.com/content/21bd0b…
We already know from clinical trials that Ozempic and other GLP-1 drugs produce sustained reductions in body weight, but with mass public usage taking off — one in eight US adults have used the drugs — the results may now be showing up at population level.
It’s really striking how the Corbynite left has migrated to the Greens.
The result is a curious coalition between the older and more Nimby environmentalist base, and the new hard left/progressive influx.
These are quite different people with quite different politics!
In 2019, one in ten Green voters was from the most progressive/left segment of voters; now that’s one in four.
Big difference in policy preferences, priorities and pressure on the leadership, as we’ve seen in e.g reaction to Denyer’s Biden statement.
The most glaring tension between these two types of Green is on decarbonisation, where the older Nimby base doesn’t want pylons *or even onshore wind farms* but many of the new progressive Green vote do.
Greens are actually less keen on wind farms than Labour and Lib Dem voters!