Please share this widely. The NHS is under propaganda attack yet again – the lie has a head start and the truth needs to catch up.
Apologies to everyone that I’ve been quiet for the last week or so. The arrival of my new granddaughter , but especially the everyday exigencies of trying to put bread on the table since I started my own company with a colleague with the aim (among others) of gaining more flexibility to write, have taken just about every waking hour for a while.
But I’ve been dying to write, about a number of topics – especially the NHS. I’ve wanted to write this post since last week – and I’ve been grinding my teeth in frustration at the patent and utterly irresponsible rubbish once again being pushed about the NHS based on that most vaporous foundation: HSMR mortality statistics.
It was claimed – predictably enough by the right-wing media, but shamefully also by Channel 4, who usually do a better job on the NHS – that a patient’s chances of dying in the NHS are an astonishing 45% higher in the NHS than in the US healthcare system.
Professor Sir Brian Jarman – the man behind the HSMR system – has been doing a nifty two-step: distancing himself from the headlines and saying his findings are
no more than a trigger to further see whether the large differences in adjusted death rates … indicate possible differences in the quality of hospital care in the two countries
while also fuelling the headlines by declaring himself
quite frankly shocked
by the findings. Neat footwork.
My teeth-grinding was because – as usual – the media headlines are absolute bollocks, and the detail of the articles is no better. Here’s why:
Apples and oranges
The saying that you have to ‘compare apples with apples’ was never more pertinent in this case. As well as highlighting the supposed ‘death deficit’ in the NHS compared to the UK, Prof Jarman’s latest ‘findings’ supposedly show that:
over more than 10 years found NHS mortality rates were among the worst of those in seven developed countries.
But there are simply no grounds or justification for this conclusion. For a start – as Prof Jarman has himself commented to me on Twitter – HSMRs in the UK are ‘rebased’, whereas in other countries they are not. Search this article for ‘rebased’ for details of what it means.
The professor infamously commented that this rebasing was necessary because people in the UK are ‘too simple-minded’ to understand the statistics without it. But whether that’s correct or he’s just incredibly condescending, the fact is that different methodologies in different countries make comparisons meaningless.
Not only this, but as the SKWAWKBOX revealed months ago, the head of DFI, the company behind HSMRs and hospital league tables, admitted to the Francis Inquiry that there is no auditing of HSMR results in the UK, nor any training given in how to ‘code’ patients’ diseases consistently.
‘Coding’ – the allocation of codes for patients’ ‘co-morbidities’, the co-existing conditions from which a patient suffers alongside the main one for which s/he is being treated on that particular occasion – is not a standardised methodology, so there are so many potential variations in the data input among countries that no statistician worthy of the name would assume that any two countries are both ‘apples’.
More on coding in a moment. But if the lack of control and training for data input make comparisons among hospitals in the UK meaningless, how much more so to draw comparisons among different countries.
The only way is up..
As my analysis (and that of many others) of Stafford hospital’s HSMRs showed, the completeness – or ‘depth’ in the jargon – of coding of patients has an absolutely massive effect on the apparent mortality statistics, because HSMRs (attempt to) measure actual death rates against ‘expected‘ death rates.
Let’s say a patient is in hospital for a thyroid operation. If s/he is in otherwise good health, the expected mortality rate for that operation might be, say, 1 in 100 (this is just for the sake of clarity).
But if the same patient has, for example, co-morbidities of chronic heart or lung disease, the expected death rate might be 3/100.
If the co-morbidities are not coded, as far as the stats are concerned the complicating conditions didn’t exist. A death rate of 3 in 100 – which should look normal – will then look enormously high. Compared to a false expectation.
By contrast, if that patient is ‘overcoded’ – if more conditions are included as co-morbidities than should be – the same 3/100 mortality rate will look low, because the ‘expected’ figure will be inflated.
Now here’s the crunch: upcoding.
The US situation regarding coding is not merely uncontrolled and of unknown variability – it is rampantly corrupt, and in only one direction.
Google ‘upcoding USA health profits’ and you will very quickly learn that the profit-motive in US healthcare has led to a situation in which huge numbers of ‘patient episodes’ are routinely and deliberately ‘upcoded’ in order to claim more money for treatments provided.
For example, Philips & Cohen, a US law firm specialising in representing whistleblowers, reports a case in which a healthcare firm paid an out-of-court settlement of $2 million to the government after whistleblowers exposed massive upcoding fraud.
Modernhealthcare.com highlights the fact that the Obama administration is having to conduct targeted audits of health providers because of ‘surging’ payments to hospitals caused by suspected upcoding.
The reason for this phenomenon in the US is plainly financial – greed is causing companies to try to milk the system to increase profits.
But a known and documented side-effect of upcoding will be to lower HSMRs, because a patient dying from something with a high death rate won’t raise the HSMR – even when they don’t actually have the condition.
‘Not your average’ idiot
The nonsense of Prof Jarman’s conclusions – and of the media claims exploiting them to damn the NHS – is shown with absolute clarity by an article by ‘NHS Choices’, “the online ‘front door’ to the NHS”.
Ironically, this article makes a worthy attempt to address the imbalance of the latest HSMR claims – but in doing so it repeats, and therefore highlights, a stunning misunderstanding of how HSMRs work. NHSC’s article states:
Comparing the number of hospitals in England and the US that had HSMRs in the different ranges, the majority of US hospitals tended to fall into the less than 100 bracket, meaning their hospital death rates were lower than expected.
The majority of English hospitals tended to fall into the 100 to 150 bracket, meaning their death rates were slightly higher than expected – that is, if they had the average mortality rate for all the hospitals in the countries examined.
The average HSMR for England was 122.4, making it the highest of the seven countries examined. The average HSMR for the US was 77.4.
Why is this idiotic? Because HSMRs are all about averages. The expected death rate is not based on a clinically-calculated mortality probability for a particular set of conditions. It’s based on the average death rate for that condition across all the hospitals in the measured territory. Similarly, a hospital’s overall HSMR is based on its position relative to the average.
It is therefore impossible for the average HSMR in England to be 122.4 – because ‘rebasing’ would turn 122.4 into 100, because 100 is the average.
According to Prof Jarman, HSMRs in the USA are not rebased – so saying the average US HSMR is 77.4 is meaningless unless you know the answer to the question, ‘Compared to what?‘
Comparing a rebased HSMR to an unrebased HSMR is utterly, utterly meaningless. And pointless – unless of course your point is to damn the NHS, and you’re ready to say ‘damn the evidence’ in the process.
Apples and oranges; corruption and fraud; a complete lack of consistency and checks on the quality and methodology of the coding which has a massive effect on HSMRs; a complete and demonstrable ignorance of what HSMRs are and how averages work.
All of these add up to an inescapable conclusion.
It is utterly, utterly – teeth-grindingly – obvious that a comparison of mortality rates in such fundamentally and uncontrolledly different countries and systems is absolutely meaningless. And that any emphatic claims of ‘45% this or that’ are even more so.
All of which is perfectly clear to anyone who bothers to perform even a basic fact-check, or even a bit of commonsense reasoning.
Which means that the latest ‘45%’ claims, if seen properly, do not damn the NHS, but rather those making the claims, and the motives that drive them to make them.