My couple of posts on Hospital Standardised Mortality Ratios (HSMRs), the clearly and seriously flawed statistics that the government and media are using as a hammer to break the NHS, have started something of a storm. Since I posted my articles, I’ve been accused of:
- being disrespectful to the dead
- being henpecked by my wife (a nurse) into writing the articles (she never knows what I’m writing until I tell her afterward)
- of having no right to an opinion for a) not having had a relative die at Stafford or b) not being a statistician
- most disturbingly, of being a sexist and a friend of paedophiles (neither of which could remotely be inferred from my posts, I trust!)
The nature of most of these demonstrates fairly clearly, I think, the rationality or lack of it of some of the people whose opinion is currently being used and quoted by the media and by government politicians.
Most significantly, with only a couple of exceptions, nobody has actually been able to challenge the actual facts and conclusions stated. One journalist mounted a reasoned counter-argument, but it was fairly easily answered – you can read the exchange here. The statistician who devised HSMRs, Professor Jarman, resorted to a defence on Twitter of the ‘avoidable death’ figures (which even he told the Francis inquiry could not be inferred from the statistics) that involved multiplying the single ‘excess death’ possibly identified by the case note review of 50 deaths into the total number of deaths in the hospital. To do this meant ignoring the fact that most/all of the questionable deaths were already concentrated into the case-note review group – an error that no capable statistician could possibly make without realising it.
What did come out of the discussions that have followed the articles was a fairly consistent thread that people did not really grasp why DFI’s practice of ‘rebasing’ HSMRs every year to a new level and still calling that ‘100’ was a problem. That’s not surprising, since it’s complex enough to be difficult to grasp. So I thought I’d provide another illustration to help clarify why it’s a problem.
The kung-fu masters
Imagine a tournament organised by a kung-fu martial arts discipline, which brings together the world’s top practitioners of the martial art. For the sake of argument, let’s say the top 101 masters are invited.
Those martial artists are tested for various attributes – strength, speed, precision and so on – and then ranked, from 1st to 101st. Martial artist no 51 is the ‘average’ or mean Master. He is worse than 50 of the 101 – and better than 50 others. But he’s still a Master – better than the remaining 7 billion or so people in the world at his particular skill.
Now, let’s say he comes back next year having worked hard on his speed. When the 101 masters are re-assessed, he has moved up by 10 places, to 41. But there are still exactly 50 masters ranked more highly than the new number 50, and 50 who are ranked lower. At least 10 people saw their ranking slip because of the improved performance of our former number 51.
Most importantly, at least 1 has fallen into the ‘worst’ 50 because the other guy moved up. And all 101 are still masters.
Am I saying that every NHS hospital is a world-class master? No, of course not. But the NHS is a world-class health service.
So an average is a very flawed means of measuring quality, because all the participants might be excellent – but some will rank a bit more highly. For measuring mortality, it’s a similar story – because every hospital in the world will have avoidable deaths. Some might have more, but for perfectly valid reasons (e.g. the population they serve is very old, or poor).
And, because one or more hospitals might improve between one measurement and the next, a hospital that is just as good as it was might fall down the ‘table’. It might even be doing better than it was – but others have improved slightly more.
Get the picture?
The health minister
Now, still bearing all that in mind, let me show you something I’ve come across. It’s a copy of a letter from Health Minister Norman Lamb to a whistle-blowing doctor who had written to him about HSMRs. In this letter, Mr Lamb tells what the government knows about HSMRs:
‘The Department [of Health] is aware’ that HSMRs can be used (at best) as a prompt to go and check stuff – and that ‘HSMR data alone does not demonstrate poor care’.
Yet, ‘weirdly’, that has in no way prevented Jeremy Hunt, David Cameron and other ministers, not to mention the right-wing press, from proclaiming ‘THOUSANDS OF AVOIDABLE DEATHS’ to talk down the NHS and to attack Labour.
Since HSMRs do not mean ‘avoidable deaths’ – and, as I’ve shown, the system is so full of holes as to be basically meaningless – the real question is:
Just why are they doing that?
Plainly it has absolutely nothing to do with saving patients or improving the NHS. So there must be other motives.
Hmmmm, wonder what those could be?