Catastrophic Health Collapse (part 6)

In this article I utilise A&E admissions for 2017 – 2021 from a sizeable NHS Trust to reveal trends in respiratory illness with comparison of rates for weighted and unweighted data

In part 5 of this series I plotted out the incidence of respiratory diagnosis and the incidence of respiratory procedure in terms of percentage of total intake for the emergency departments of a sizeable NHS Trust for the period 2017 – 2022. I then derived a ratio for these that I termed the Respiratory Severity Index (RSI). We noted a rather lean year for severe respiratory illness in 2019 compared to previous years and unearthed a curious situation in which indicators of respiratory illness didn’t go through the roof during 2020 as we would expect from a novel and deadly virus. I concluded that we’d been conned, and by those working in the NHS of all places.

These analyses were based on persons being admitted and treated, which is a good starting point, but we can go one step further. Age is always a confounding factor when it comes to matters of health and so we need some way of accounting for this. Professors Norman Fenton and Martin Neil of QMUL have been producing some cracking articles and YT videos explaining the lather we can get into if we don’t account for age effects and I recommend subscribers digest these with a decent pot of hot coffee at their elbow by clicking here.

There are a few ways we can go about accounting for age. We can slice the data into age bands, we can use age as a covariate in multivariate modelling, and we can resort to using person years at admission, this being the sum of ages of individuals within a sample. What I am going to offer today is a comparison of analyses made using unweighted (persons) and weighted (person years) data so we may see the implications for trends in respiratory illness over time. This is made possible because age at admission is recorded in the CDS 010 dataset.

Unweighted vs. Weighted

Incidence of Respiratory Diagnosis

The green dashed line in this slide marks 2020/w5 (w/e 31 January 2020), being the lull before the officially declared COVID storm. As we can plainly see there wasn’t much of a COVID storm, and using weighted data hasn’t made much of a difference to the only peak that might offer evidence of a novel virus. However, if we remove the controversial CHEC death spike for the period 2020/w13 – 2020/w22 there’s nothing to see. Neither did delta or any other mutant strain kick up a fuss later in the year. In fact, the post-CHEC death spike period looks decidedly limp! And hospitals were supposed to be brimming with severe respiratory COVID cases were they?

Incidence of Respiratory Procedure

By using weighted data we have elevated the incidence rate for use of respiratory procedures, as may be expected (older people require more care when things go wrong). If you asked a Martian when the great pandemic of the modern era was they’d point to 2017 or even 2019. They would consider any Earthling to be a total fruit cake for suggesting 2020. This is real world data from a real NHS Trust folks (at least I hope it is), not some dodgy number fudge produced by His Majesty’s Expertistas to support an insidious political agenda.

Respiratory Severity Index

And so we arrive at my favourite slide again, showing that there RSI thingy. Using weighted data has certainly put an edge on that 2017 anomaly, which suggests we’re talking older folk who required urgent treatment for a breakout respiratory condition. Perhaps a GP might offer illumination as to what went on back then, though we may be looking at a highly localised situation.

Kettle On!