Welcome to The Plague Pit – issue number 5

In his Easter Sunday briefing, the Health Secretary told us that UK deaths in COVID-positive patients had crossed the 10,000 threshold

A word about data

There are a lot of data about now, so that figure merits a closer look. When doctors submit draft articles to peer-reviewed journals, hoping for publication, editors will usually insist that they give a source for any apparent statements of fact. Here’s Matt Hancock’s source [1]

NHS England receive reports from each NHS-commissioned service (hospitals, GP practices, etc) of all COVID-positive patients dying every day. These data are extracted at 5pm, grouped with similar data from Wales/Ireland/Scotland, and published the next day on the Public Health England COVID-19 dashboard, as above .    

I’ve been wondering about these numbers because they sit rather oddly with data from another source. Every few days an Intensive Care National Audit and Research Centre (ICNARC) report gives a cumulative number for deaths of confirmed COVID-19 patients in UK critical care units (excluding Scotland). [2] Up to 9^th April, that number was 871. The corresponding cumulative death figure from the PHE dashboard for the same day was 7,612.

The ICNARC data come from their Case Mix Programme (CMP), set up in 1994. It’s a pretty blue-chip source. Every UK adult intensive care unit is now a CMP participant, submitting anonymized data from the first 24 hours on all patients they admit, celebrities or otherwise. The ICNARC website now lists nearly 100 academic publications using CMP data – many in the most prestigious journals.

So why do the cumulative ICNARC COVID-19 death numbers look so different from those on the PHE graph? 871 vs 7,612.  There are differences in the way the data are collected as we have seen. PHE data are said to be validated so I presume they are not inflated by duplicate death reports on the same patient (e.g. one from GP, one from hospital). ICU staff are pretty busy, so the lag between a death and the subsequent reporting of that death may be longer with ICNARC than with PHE. That would artificially lower the ICNARC cumulative death number on any given day – but would that really account for a nine-fold difference between the two sources?

The implication of the discrepancy is that the vast majority of patients dying with COVID-19 do so outside an ICU. Perhaps people are not getting to hospital in time. Most COVID-positive people in the UK, however, live near enough to an ICU to get there before expiring. And COVID pneumonia rarely progresses fast enough to kill you in the space of a car journey. Boris Johnson waited ten days after his symptoms started before coming to hospital  

Perhaps large numbers of COVID-positive patients get to hospital but then the severity of their condition is not recognised. They are not transferred to ICU when they should be. This seems unlikely.

Could the difference in ICNARC and PHE numbers, then, be the result of large scale rationing? In the early days of the pandemic, many ICU doctors I know were worried about this. Huge numbers of patients requiring mechanical ventilation were expected to coincide with limits on space, trained staff and equipment. Many doctors anticipated the widespread need for enhanced prioritisation of some cases for ICU care over others – on grounds of prognosis and other factors..

In my institution, anaesthetists scheduled for re-deployment to ICU a few weeks ago were reminded of the humane and efficient processes already in place there for ‘do not escalate treatment’ decisions. Such decisions inevitably go hand-in-hand with rationing discussions and are part of everyday life in critical care. Many ICU doctors work with ethicists in interdisciplinary groups and several write about ethics in the academic journals. Could all the COVID-positive patients dying outside ICU – hundreds every day –  be ‘victims’ of ‘do not escalate’ decisions? Again, I think this unlikely  

Then I came across this.  

It’s a graph published by Professor David Spiegelhalter, Chair of the Winton Centre for the Risk and Evidence Communication at Cambridge University. You can see it in a great article he wrote a few weeks ago (https://medium.com/wintoncentre/how-much-normal-risk-does-covid-represent-4539118e1196).

With a few caveats (including one about health care workers), Professor Spiegelhalter’s conclusion was that there is little evidence for substantial excess UK mortality due to COVID-19 . Most COVID-positive people who die do so with COVID-19 and not of COVID-19. I think that may also explain the gulf between ‘COVID deaths’ on ICU and ‘COVID deaths’ overall. I’m checking with a few friends of mine in general practice and ICU and I’ll report back.

If I am right, there are precedents for this sort of thing elsewhere in medicine. The one we were always taught about at medical school was prostate cancer. This is the commonest cancer diagnosed in UK men and usually affects the middle-aged and elderly. It grows very slowly, to the extent that it may not cause symptoms, remaining undiscovered until the patient expires of something else. I’m sure I remember a pathologist telling us it was ‘an almost ubiquitous post-mortem diagnosis in the over nineties’. As many men die with prostate cancer as of prostate cancer [3] – an interesting fact but not one of any comfort to those in the latter group.

Comedy corner

That’s rather a lot of death – maybe it’s time to lighten up.

PPE continues to make the news daily. It is said that masks for staff involved in high-risk, aerosol-generating procedures like intubation should ‘fit tested’ to the individual concerned. Standard masks worn by staff with beards/stubble and/or atypical facial morphology may apparently not adequately prevent exposure of the wearer to virus-containing droplets. A variety of (allegedly) superior and more eye-catching face-wear is available for these individuals. Here is a picture of a great colleague of mine with non-standard physiognomy in his new kit. The caption was added by a mutual friend.

Finally, in the last issue I said I’d be writing today about next steps in treating COVID pneumonia, once oxygen via a face mask isn’t making it any better. I think I’ll leave that for next time. It will involve considering levels of evidence in academic literature. Sounds heavy weather, but perhaps I can start with this excellent parody of a research paper in the New England Journal of Medicine.   

The Good Soldier

[1]  Public Health England. Total UK COVID-19 Cases Update. Data as of Sunday 12th April 2020. https://www.arcgis.com/apps/opsdashboard/index.html#/f94c3c90da5b4e9f9a0b19484dd4bb14. Accessed 13/04/20

[2] Intensive Care National Audit and Research Committee. ICNARC report on COVID-19 in critical care. 10^th April 2020.

[3]  Riihimaki M et al. What Do Prostate Cancer Patients Die Of? The Oncologist 2011;16:175–18