14 June 2011

Professor Jarman. First impressions.

This is a preliminary and perhaps impressionistic account of Professor Jarman’s  evidence. There will be much more in the way of important evidence this week, so I wanted to make an attempt to capture first impressions.  I hope that other people with a much more detailed grasp of the statistical systems involved will also choose to analyse the case that he has made.
Professor Jarman has played a very prominent role in the course of the Midstaffs Inquiry. It is the use – or possibly abuse of the statistical systems that he created that led to the highly contested “excess death figures”. These figures, which were set to be included in the health care commission report at the very last moment and then were withdrawn after a major row, have completely dominated the press and media coverage of the Midstaffs story, and are the reason why I am now involved in a campaign for press reform.
With me there are key witnesses that I needed to see in person. Professor Jarman was one of these. I needed to see the person in order to see what his system means to him. This may sound like a strange approach to a matter of Statistics, but been discovering from the Inquiry is that it is the statisticians who are most passionately involved with what has happened here, and that the different systems for monitoring and regulating are always built on a very individual approach. People and personalities matter!
Like everything else in the Midstaffs Inquiry this is not as simple as it might seem!
First – what did we learn about Professor Jarman’s personal history.
It is clear that Professor Jarman’s career is both long and distinguished. One of the things that he is most committed to now is his involvement with the Institute for Health improvement in Cambridge and Boston Massachusetts USA. Since 2001 This is a body of people who are deeply committed to improving the quality of health care. His work with them is a reflection of his own deep commitment.
Professor Jarman’s early background is interesting – He did not begin medical training until the age of 31 after early training as a geologist. His first PHD was in seismic wave propagation.  He then went on to develop an interest in Socio economic indicators and has developed computer systems for CAB and Social services.
He became a GP in 1970. He was a member of the Community health council for Bloomsbury in the 1970s, and thinks they managed things very well.  When visiting a hospital he witnessed a serious untoward incident, which he believed could have led to a death. He does not think this was reported. This may have led to his wish to look for ways of improving quality of health care through better training.
He talked about the unintended Impact of the Griffiths report in 1983, which he believed divorced the health service from control by the clinicians and put it into the control of Health service managers. He clearly feels that this was a mistake.
He was involved in the Bristol Inquiry, where his statement gives us an interesting insight into his relationship with the government – He believes he would not have been the Government’s first choice for this task!  He set up the Dr Foster Unit as a response to what he saw at Bristol.
 He has been working on Mortality rates since 1990 so this forms a very significant part of his life’s work.
It became clear as he gave his evidence that this is something about which he feels very very deeply, and that he sincerely believes that he has developed the tools that have the potential to save many lives.

A little of the detail.
Because I am rushing to get to today’s inquiry I am now going to focus on the points that stood out from the evidence – all of which I will try to come back to later. I would of course advise people to read the transcripts for themselves. Do not take it from me! I am not objective, and neither is anyone else who is reporting this matter!
There was discussion about the Doctor Foster Unit and Dr Foster Intelligence. The Inquiry tried its best to clarify this difficult distinction. Dr Jarman is keen to emphasise that there are points of real difference between them, and that he is often at pains to point out where he thinks they are going wrong.
There was a long discussion about HSMR and the basics of coding. What struck me from this is Professor Jarmans emphasis on the importance of Primary coding. The system depends on people getting this right. It was made clear that this was a matter of clinical judgement – often by junior Doctors working under stress. Professor Jarman sees this as a simple matter- the question which is still unanswered from my point of view, and I think from the point of view of Robert Francis is how simple is this in practice.
The discussion got bogged down in “primary coding” actually meant. It was clear from board minutes in Midstaffs that the hospital did not have the same understanding of this as Professor Jarman. He seemed genuinely surprised by their degree of incomprehension.
Robert Francis asked a hypothetical question about how you could code a case where a person was admitted with a broken Hip and then developed CDiff. Professor Jarman’s answer was that it was a matter of clinical judgement how this would in fact be coded.
What was I felt notable by its absence was any prolonged discussion on Secondary coding. We know clearly from the Health care commission report that this was something that was being done entirely wrong in Stafford. The coding manager had been off sick for a long time, the stand-ins found difficulty getting information about co-morbidities. They therefore had to rely on Primary coding (which may or may not have been accurate).
To my mind, and I am of course no expert on this, any system which sets out to tell us if a death is “expected” or “not expected” has to reflect the total medical condition of a patient. Let’s take a case study from my own family. My mother in law was admitted to hospital (in Scotland) because she had had a massive stroke, brought on by the medication for the rheumatoid Arthritis which she had suffered from for 15 years. The Stroke left her immobile, and she was in acute pain whenever staff had to move her because of the arthritis and eventually needed morphine to enable her to cope with being moved. Her breathing patterns became very irregular, and she needed heavy medication to cope with this. She had kidney infections and needed antibiotics. Her condition stabilised.  She at one point was told that she would be fitted with a peg feeding tube and discharged to the only nursing home that could cope with her complex conditions. She did not want this and went down hill rapidly at this point. She then got C Diff. She eventually died from heart failure.  I do not think she died of “a primary cause” I think she died of all these reasons combined.  I have no idea how her case was coded.
To form any judgement of how well the HSMR system works, I think we need to look at a large number of case studies and understand the variation of coding that occurs in practice.
There was a lot of discussion about a factor that came as complete news to me after three years of looking at the Stafford Case.  Professor Jarman makes it clear that he is always open to suggestions about how to improve his system, and as a result of concerns from some of the users he changed the statistical model in 2007.
The “customers” for the system had pointed out the need for a palliative care code, which would make it easier to show cases where a patient was essentially just being kept comfortable, and was unexpected to survive. The introduction of this code came in 2007 and this along with the failure of Stafford to “rebase” their coding explains why there was a huge jump in their HSMR. The Hospital had been expecting to see HSMR at 114, High but not alarmingly high, It in effect came in as a result of their failure to use the new palliative care code at 127. The figure that sparked all the fuss.  
It is clear that West Mids hospitals as a whole had missed the introduction of the Palliative care code, and that therefore those hospitals which had a high case load of elderly patients – with complex health problems, who were not expected to survive, were being wrongly coded. This may – and I repeat may – be the explanation for the higher than usual HSMRs on the 6 hospitals in the West mids region.
A slide which drew a lot of attention from the Inquiry, and very much pleased the press, related to the work that the SHA did with its hospitals at this point to sort out their coding problems. The three hospitals that Professor Jarman pointed out began using the palliative care code extensively and their HSMR consequently fell rapidly. 
What Professor Jarman clearly felt, and the press picked up on was that this was “gaming” the system.  I think it is perfectly possible that there was some overcorrection going on, but I think it is also very important to view this in the light of the genuine undercoding that had been going on previously.  This was not discussed and I think it should have been.
There was quite a lot of discussion of the way in which the SHA had commissioned the Birmingham university group to carry out research into HSMR. Dr Jarman clearly interprets this as a personal slight. His anger about this was tangible.
He also feels immensely strongly that the DoH have not valued his work in the way that they should and that if they had done so thousands of lives would have been saved.  He puts out the HSMR data to all hospitals now, not because he is paid to do so, but because he feels it is his moral duty.
One thing which did impress me is that he described work which he had done with a hospital to help them study their processes and help them bring down mortality.  I think this may be in contrast with the work of DFI where the emphasis is on offering training and consultancy to hospitals to help them use the system better.
Dr Jarman is a physician. He wants people to do things better, and what his system can do at its best is to help people identify areas of concern. It is only at that point that the work begins. People have to analyse what is happening, and work out how to do things better.
Dr Jarman’s system can help people begin this intellectual journey. So can all the statistical systems that have been shown to us by the Trust, The SHA and the CQC.  It is also clear from the work highlighted by Ben Goldacre, that systems that this government has wished to scrap – involving the Targets, can also help people begin this journey very effectively.

Getting beyond the politics.
To my mind the big problem that all of us are struggling with is how statistical systems, which need to be carefully evaluated, and carefully used, have found themselves at the heart of what is essentially now a political battle.
The Conservatives took a political position on this. Targets were out, Outcomes were in, and Professor Jarman’s system, alongside his belief in the importance of a GP led Health care system would do the job.  This annexation of professor Jarmans work may have been damaging to him. He clearly expected that his work would finally now be recognised, but yesterday he expressed his disappointment with the contents of the Health bill.