Bed Loss & Bedlam (part 1)

A look at bed occupancy within NHS England over the last 13 years

There is a factor far more deadly than SARS-COV-2 and that is government policy. The nation was always going to pay for closure of health services in terms of elevated levels of pain, misery and death, and this was always going to have both immediate and long term effects.

Being a potential confounding factor in studies of vaccine harm/benefit I chewed on a German pencil until an idea came… I could derive a bed loss index for NHS England using bed occupancy data provided in a bundle of spreadsheets sitting on this page.

Before COVID (B.C.)

Before COVID came along analysts had to make do with quarterly estimates for bed occupancy that covered day and night beds as well as beds by speciality (general & acute, learning disabilities, maternity and mental illness). When COVID arrived analysts were provided with daily estimates of COVID and non-COVID bed use as well as quarterly estimates. We can merge the two series and take a look at occupancy since 2010. Have a squizz at this pleasing plot:

There are several features we need to chew over here and I suggest we start with the close match between estimates derived from daily bed counts and estimates derived on a quarterly basis. I am hoping we all agree that these figures are essentially counting the same thing, but with differing resolution. The upshot is that the more recent daily COVID and non-COVID bed count (red curve) must necessarily consist of counts of day only as well as overnight beds, together with combined counts for all four primary specialities. In other words they’ve been counting beds in learning disability units, maternity units and mental illness units and calling them ‘COVID’ beds. I’d prefer to call them PTBs (positive test beds), for if anybody was genuinely sick with COVID-like symptoms they’d be trolleyed down the corridor at a rate of knots (at least I hope so). The bed fudge is declared down in the small print in spreadsheets but for some strange reason this über fact never made it into public consciousness. Funny that.

Before we get stuck into the impact of bed loss during 2020 I want to point out the rather curious decline in bed occupancy over the last 13 years. I’m not convinced this is due to improving health of the nation hence in a future article we’ll take a look at bed availability. I’m going to put good money on less and less beds being available with occupancy rates remaining more or less constant over time. This is a bet I’d love to lose.

The Black Hole

This brings us on to the big black bed hole of 2020. Was it strictly necessary? Real world experience suggests not, and we must remember that the government neatly tucked away common sense to listen to computer programmers and mathematicians above all else. Will the government now lie through their teeth and get the same bunch of carefully groomed mathematicians to spin a story with hand waving and more exotic but utterly useless models? You betcha, and a large slice of the public (and an uncomfortable proportion of healthcare professionals) will believe them because the alternative is unthinkable.

I’m not allowed to punch people so I shall crunch through the data to provide an index I am calling the Bed Loss Fraction (BLF), this being based on the proportion of beds occupied each day with reference to the series daily maximum of 128,079 beds. Thus, a week during which mean daily occupancy reached 64,040 beds would yield a fraction of 0.5 (50%). If that sounds like a ridiculous amount of beds to close in a system already under extreme pressure you better take a look at this graph:

Within the space of a few weeks hospitals across NHS England went from normal levels of occupancy to an estimated bed loss peak of 0.59 (59%) during 2020/w14 (w/e 3 April 2020). There is no way that would have happened without a great deal of pain and stress for both patients and staff, and without killing the most vulnerable and elderly. In the business we call it dangerous discharge, yet at the same time the government were issuing daily messages about saving granny.

Bed Loss & Death

It’s at this point that the valuable tool that is cross-correlation comes into play for this enables us to identify the correlation between bed loss fraction and all cause death over a range of time lags. For this purpose I’ve used the T4253H smoothed all cause weekly series of deaths occurring across England to avoid hiccoughs caused by administrative delays in processing certificates. I had absolutely no idea what I’d find - try this for size:

In this first slide we see a palisade of statistically significant red bars sticking above the upper 95% confidence limit at lags of zero to 3 weeks. There’s a rebound effect at a lag of 6 – 7 weeks likely caused by survivorship bias; that is, the most frail and vulnerable were wiped out in the first 3 weeks to leave a hole in the death stats from week 4 onward. The big question is whether this is directly due to bed loss (i.e. dangerous discharge) or the first wave hitting.

We observe a second palisade of statistically significant red bars sticking above the upper 95% confidence limit at a lags around the 40 week mark. This is not necessarily ‘long COVID’ or a delayed effect of initial bed loss and is simply a correlation with the 2020/21 winter peak.

No doubt some people will say, “ah, yes, but… this was obviously COVID killing them and not dangerous discharge”, to which my response is nothing is ever obvious, and especially so in this tragic game of vested interests. I’d also like such folk to explain away this slide for non-COVID death:

We see a lagged correlation at 3 – 4 weeks even for non-COVID death, so it can’t be the first wave unless non-COVID death doesn’t actually mean non-COVID death, in which case our certification process is well up the creek without a paddle and therefore most of what the authorities are claiming is pure tosh. Note the rebound effect of survivorship bias, and note the second and third wave correlations at 40-odd weeks.

Coffee & Cogitation

I must stress at this point that until we look at medical records we can only surmise. If sudden and dramatic bed loss did result in early death of the most vulnerable patients (as we may reasonably expect) then this debacle is not necessarily the fault of nursing or medical staff who have to follow protocols laid down by lead clinicians and clinical managers whether they like it or not. In turn, lead clinicians and clinical managers have to follow orders from above, thus the thread of shame sneaks all the way back to the Secretary of State and our most senior medical advisors up to the Chief Medical Officer.

Did nobody amongst these powerfully placed and allegedly intelligent people not realise the inhuman nature of their acts and criminally negligent thinking? Where are the mathematical models that modelled agonising outcomes for national health service shutdown? Where’s all the sophisticated cost benefit analysis? I can only imagine these rather smart top bods knew precisely what they were doing - and didn’t care.

Those top bods wanted you to think that not wearing a mask, refusing to social distance and obey a myriad of utterly futile rules killed granny. Dangerous discharge may well have killed granny; if so it also killed granddad along with everyone else who was medically vulnerable at the time.

Kettle On!